Reforming Processes for Sustainable Syngas Production: · PDF file1.1 Catalysts ... B)...

Promotoren: prof. dr. ir. G. B. Marin, dr. V. GalvitaProefschrift ingediend tot het behalen van de graad van

Doctor in de ingenieurswetenschappen: chemische technologie

Vakgroep Materialen, Textiel en Chemische ProceskundeVoorzitter: prof. dr. P. Kiekens

Faculteit Ingenieurswetenschappen en ArchitectuurAcademiejaar 2017 - 2018

Reforming Processes for Sustainable Syngas Production: The Role of Fe

Stavros-Alexandros Theofanidis

ISBN 978-94-6355-053-6NUR 952Wettelijk depot: D/2017/10.500/88

Promotoren:

Prof. Dr. Ir. Guy B. Marin Ghent University

Dr. Vladimir Galvita Ghent University

Examination committe:

Prof. Dr. Ir. Gert de Cooman Ghent University

Prof. Dr. Ir. Joris W. Thybaut Ghent University

Prof. Dr. Ir. An Verberckmoes Ghent University

Prof. Dr. Christophe Detavernier Ghent University

Prof. Dr. Ir. Christoph Müller ETH Zurich

Prof. Dr. Andrei Khodakov UCCS Lille

Universiteit Gent

Faculteit Ingenieurswetenschappen en Architectuur

Vakgroep Chemische Proceskunde en Technische Chemie

Laboratorium voor Chemische Techniek

Technologiepark 914

B-9052 Gent

België

Tel.: ++32 (0)9 331 17 57

Fax: ++32 (0)9 331 17 59

http://www.lct.ugent.be

This work was supported by the FAST industrialization by Catalyst Research and Development

(FASTCARD) project, which is a Large Scale Collaborative Project supported by the European

Commission in the 7th Framework Programme (GA no 604277).

Gold is for the mistress silver for the maid Copper for the craftsman cunning at his trade." "Good!" said the Baron, sitting in his hall,

"But Iron Cold Iron is master of them all."

Rudyard Kipling

Acknowledgements

i

Acknowledgements

Four years of PhD, it was a long journey, rich in experiences. Many challenges, many difficult

situations, but, on the other hand, many people to counterbalance. So I realized that this

Thesis is a result of direct or indirect effort, support and help of numerous people.

I would like to start with my promoters. Firstly, I would like to thank Prof. G.B Marin for

providing me the opportunity, the necessary tools and motivation to achieve my goals. Your

comments were always helpful and an incentive to move forward. I am also very thankful to

my second promoter, Dr. Vladimir Galvita, who was always available for many productive,

long-lasting discussions leading to an endless supply of new ideas. Vladimir, working with

you was not just a wonderful experience, but inspiration as well. Then, I would like to thank

Hilde for her support and patience during these years. Being in the same group and working

with Vladimir and Hilde was very nice experience, demanding when it was needed, but on

the other hand, full of fun, discussions, …iterations and... XAS campaigns. Finally, I would like

to thank Prof. Joris. It was very nice to collaborate with him, during these years, for the

project. This work was supported by the FAST industrialization by Catalyst Research and

Development (FASTCARD) project, which is a Large Scale Collaborative Project supported by

the European Commission in the 7th Framework Programme (GA no 604277). Participating in

FASTCARD project resulted in meeting some people and I would like to thank them for the

nice collaboration that we had.

I would also like to thank the jury members for investing their time in order to provide me

useful comments and suggestions to improve my thesis.

A special mention to the other members of the group: Lukas, Rakesh, Aditya, Jiawei, Jolien

and the newcomer Mostafa. It was pleasure not only to work with you guys but hanging out,

traveling and having fun. Of course the technical team of LCT deserves to be mentioned.

Many thanks go to Marcel, as he was always helpful and open for discussions and of course

to Bert and Wim. I would also like to acknowledge Geert Rampelberg for his assistance

during XRD measurements.

Acknowledgements

ii

During these four years, I had also the opportunity to coach Johannes and Jonas for their

Master Thesis. Their questions and dedication towards their work was an incentive for me.

On a more personal level, I would like to thank the people that supported me day-to-day.

Starting with Chanakya, Guillermo, Daria, Jeroen and Martin that we shared the same office

for 4 years. It was a multinational office and we had many funny moments that I will miss.

Also, I would like to thank Stamatis, Marko and Nenad for the nice time that we shared

inside and outside of the lab. Apart from all these people, the whole PhD community in LCT

made a wonderful atmosphere and ideal working environment and therefore I am very

happy that I was part of that team.

Last but not least, I am grateful to Sofia for her moral and emotional support during these

years. Especially in the difficult situations when my complains about PhD were increasing. Of

course, a special mention to my friends (Vasilis Magl, Grigoris, Vasilis Had, Thomas K, Takis,

Giannis, Christina, Thomas N, Anna, Dimitris, Manolis, Giorgos, Christos, Babis, Foivos,

Antreas, Stavros) who were always available for hours and hours of talking but also for

spending together some memorable time. Guys you are awesome! For the end, I left the

most important people, my parents. Their unconditional support and love are the main

reasons for what I have achieved. Thank you so much for making this possible.

Table of Contents

iii

Table of Contents Acknowledgements .................................................................................................................................. i

Table of Contents .................................................................................................................................... iii

List of Figures .......................................................................................................................................... vii

List of Tables ........................................................................................................................................... xv

List of Abbreviations and Acronyms ..................................................................................................... xvii

List of Symbols ....................................................................................................................................... xix

Glossary of terms ................................................................................................................................. xxiii

Summary ............................................................................................................................................ xxvii

Samenvatting ...................................................................................................................................... xxxv

Chapter 1 ................................................................................................................................................. 1

Introduction ............................................................................................................................................. 1

1.1 Catalysts .................................................................................................................................. 9

1.1.1 Monometallic ...................................................................................................................... 9

1.1.2 Bimetallic ........................................................................................................................... 10

1.2 Catalyst support .................................................................................................................... 14

1.3 Kinetics .................................................................................................................................. 16

1.4 Catalyst preparation .............................................................................................................. 20

1.5 Catalyst deactivation ............................................................................................................. 23

1.5.1 Carbon deposition ............................................................................................................. 24

1.5.2 Poisoning ........................................................................................................................... 28

1.5.3 Sintering............................................................................................................................. 31

1.5.4 Encapsulation .................................................................................................................... 32

1.6 Catalyst regeneration by carbon gasification ........................................................................ 33

1.7 Scope of the thesis ................................................................................................................ 34

References ......................................................................................................................................... 36

Chapter 2 ............................................................................................................................................... 49

Experimental procedures ...................................................................................................................... 49

2.1 Synthesis of heterogeneous catalysts ....................................................................................... 49

2.1.1 Support preparation .......................................................................................................... 49

2.1.2 Active phase preparation .................................................................................................. 50

2.2 Catalyst characterization methods ............................................................................................ 52

2.2.1 Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) ............................ 52

Table of Contents

iv

2.2.2 Surface area and porosity .................................................................................................. 52

2.2.3 Ex-situ X-ray diffraction (XRD) ........................................................................................... 53

2.2.4 In-situ X-ray diffraction (XRD) ............................................................................................ 54

2.2.5 X-ray absorption spectroscopy (XAS) ................................................................................ 54

2.2.6 Electron microscopy techniques ....................................................................................... 57

2.2.7 X-ray Photoelectron Spectroscopy .................................................................................... 58

2.2.8 RAMAN spectroscopy ........................................................................................................ 59

2.3 Activity and stability tests ......................................................................................................... 59

2.3.1 Absence of pressure drop ................................................................................................. 62

2.3.2 Ideal plug flow ................................................................................................................... 63

2.3.3 Mass and heat transport limitations ................................................................................. 64

2.4 Temporal Analysis of Products .................................................................................................. 66

2.5 References ................................................................................................................................. 69

Chapter 3 ............................................................................................................................................... 71

Enhanced carbon-resistant dry reforming Ni-Fe catalyst: role of Fe .................................................... 71

3.1 Introduction ............................................................................................................................... 72

3.2 Experimental methods .............................................................................................................. 75

3.2.1 Catalyst preparation .......................................................................................................... 75

3.2.2 Catalyst characterization ................................................................................................... 75

3.2.3 In-situ time resolved XRD .................................................................................................. 76

3.2.4 Catalytic activity ................................................................................................................ 77

3.2.5 Alternate Pulse experiment ............................................................................................... 79

3.3 Results ....................................................................................................................................... 79

3.3.1 Catalyst characterization ................................................................................................... 79

3.3.2 In-situ XRD time resolved measurements ......................................................................... 82

3.3.3 Activity tests ...................................................................................................................... 85

3.3.4 Regeneration cycles ........................................................................................................... 90

3.3.5 Effect of alloy on carbon formation .................................................................................. 91

3.3.6 In-situ XRD time resolved during DRM .............................................................................. 92

3.3.7 Alternate Pulse experiment ............................................................................................... 94

3.4 Conclusions ................................................................................................................................ 96

3.5 References ................................................................................................................................. 98

Chapter 4 ............................................................................................................................................. 103

Carbon gasification from Ni-Fe catalysts after methane reforming reactions .................................... 103

Table of Contents

v

4.1 Introduction ............................................................................................................................. 104

4.2 Experimental Methods ............................................................................................................ 106

4.2.1 Catalyst preparation ........................................................................................................ 106

4.2.2 Catalyst characterization ................................................................................................. 107

4.2.3 Ageing of catalyst during DRM ........................................................................................ 108

4.2.4 Carbon species temperature programmed oxidation (TPO) ........................................... 108

4.2.5 Operando XRD ................................................................................................................. 108

4.2.6 Isothermal Temporal Analysis of Products (TAP) experiments for carbon gasification .. 109

4.3 Results ..................................................................................................................................... 110

4.3.1 Catalyst Characterization ................................................................................................ 110

4.3.2 Carbon species temperature programmed oxidation ..................................................... 112

4.3.3 Carbon characterization .................................................................................................. 114

4.3.4 Mechanism of carbon removal by CO2. ........................................................................... 118

4.3.5 In-situ time resolved XRD during O2-TPO ....................................................................... 121

4.3.6 O2-TPO over different catalyst bed configurations ......................................................... 123

4.3.7 Temporal Analysis of Products (TAP) experiments ......................................................... 125

4.3.8 Carbon gasification mechanism by O2 ............................................................................. 130

4.4 Conclusions .............................................................................................................................. 132

4.5 References ............................................................................................................................... 133

Chapter 5 ............................................................................................................................................. 137

Controlling the stability of a Ni-Fe reforming catalyst: structural organization of the active

components ......................................................................................................................................... 137

5.1 Introduction ............................................................................................................................. 138

5.2 Experimental methods ............................................................................................................ 140

5.2.1 Catalyst Preparation ........................................................................................................ 140


5.2.3 In-situ time resolved XRD ................................................................................................ 142

5.2.4 Catalytic activity .............................................................................................................. 143

5.2.5 Catalyst models and DFT calculations ............................................................................. 145

5.3 Results ..................................................................................................................................... 146


5.3.2 In-situ XRD time resolved measurements ....................................................................... 149

5.3.3 Effect of catalyst preparation on alloy formation ........................................................... 151

5.3.4 Activity and stability tests................................................................................................ 153

Table of Contents

vi

5.4 Discussion ................................................................................................................................ 161

5.5 Conclusions .............................................................................................................................. 166

5.6 References ............................................................................................................................... 168

Chapter 6 ............................................................................................................................................. 171

Controlling the stability of a Ni-Fe reforming catalyst: down tuning the Fe concentration ............... 171

6.1 Introduction ............................................................................................................................. 172

6.2 Experimental methods ............................................................................................................ 173

6.2.1 Support and catalyst preparation .................................................................................... 173

6.2.2 Support and catalyst characterization ............................................................................. 173

6.2.3 Operando quick-XAS (QXAS)............................................................................................ 177

6.2.4 In-situ time resolved XRD ................................................................................................ 177

6.2.5 Isothermal Temporal Analysis of Products (TAP) experiments ....................................... 178

6.2.6 Catalytic activity .............................................................................................................. 178

6.2.7 Mechanical mixture ......................................................................................................... 180

6.3 Results ..................................................................................................................................... 181

6.3.1 Support Characterization ................................................................................................ 181

6.3.2 Catalyst Characterization ................................................................................................ 187

6.3.3 Temperature Programmed Reduction (TPR) ................................................................... 189

6.3.4 Activity and stability tests................................................................................................ 194

6.4 Conclusions .............................................................................................................................. 197

6.5 References ............................................................................................................................... 198

Chapter 7 ............................................................................................................................................. 203

Conclusions and perspectives ............................................................................................................. 203

Appendix A .......................................................................................................................................... 207

Appendix B .......................................................................................................................................... 215

Appendix C........................................................................................................................................... 221

List of Figures

vii

List of Figures

Figure 1.1: Number of biogas plants and total installed capacity in Europe for the period 2010-2014. Figure based on [9].

Figure 1.2: Greenhouse gas emissions by: (a) gas, (b) economic sector. Reproduced from [30, 31].

Figure 1.3: Thermodynamics of DRM. Limiting temperatures for reactions of the CH4-CO2 system. Figure based on [39].

Figure 1.4: Thermodynamic H2:CO ratio for the produced syngas as a function of pressure. Red line: CH4:CO2:H2O molar ratio of 2:1:1; green: CH4:CO2:H2O molar ratio of 2:1.5:0.5; blue: CH4:CO2:H2O molar ratio of 1:1:0.

Figure 1.5: Conversion via syngas. Modified from [52].

Figure 1.6: Global syngas market. Modified from [35].

Figure 1.7: Sequence of elementary steps for CH4 reactions with H2O or CO2 on Pt surfaces (→ irreversible step; ↔ quasi-equilibrated step). Reproduced from [150].

Figure 1.8: Preparation of sol-gel catalysts. Reproduced from [164].

Figure 1.9: Core-shell catalyst (left) and yolk-shell catalyst (right). Reproduced from [170].

Figure 1.10: Deactivation mechanisms: A) carbon deposition, B) poisoning, C) sintering or thermal degradation, and D) encapsulation of active metal particles by the support. Reproduced from [177].

Figure 1.11: Original model proposed by Lobo and Trimm for carbon formation on Ni. A: Several modes of carbon formation observed on nickel foils; B: detail of carbon formation on a detached crystallite, showing the reaction steps of the proposed mechanism [184].

Figure 1.12: Formation of volatile Ni(CO)4 at the surface of nickel crystallite in CO atmosphere. Reproduced from [178].

Figure 1.13 Approaches for catalyst optimization in order to eliminate its deactivation.

Figure 2.1: Signals generated from an incident beam on a specimen. Reproduced from [13].

Figure 2.2: The overview of the step response setup with the feed, reactor and analytical sections.

Figure 2.3: Schematical representation of TAP system.

Figure 2.4 Analysis of a TAP experiment: inert (internal standard), reactant and product pulse responses.

Figure 3.1: Full XRD scans of (A) MgAl2O4, as-prepared and reduced Ni-5Fe/MgAl (1NmL/s of 10%H2/He mixture at a total pressure of 101.3 kPa and 1123K). The upper-right inset shows the highlighted rectangular area in higher resolution. (B) Ni-5Fe/MgAl after CO2-oxidation (1NmL/s of CO2, at a total pressure of 101.3 kPa and 1123K).

List of Figures

viii

Figure 3.2: EDX element mapping of Ni-5Fe/MgAl. (A) after H2-reduction (1 NmL/s of 5%H2/Ar

mixture at a total pressure of 101.3 kPa and 1123K). (B) after CO2 oxidation (1 NmL/s of CO2 at a total pressure of 101.3 kPa and 1123K). Red and green colors correspond to Fe and Ni elements respectively.

Figure 3.3: In-situ XRD during H2-TPR. (A) 2D XRD pattern for Ni-5Fe/MgAl. Heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He. (B) Integral intensity variation of (A) for diffraction areas 35.5o-36.5o (NiO), 43.7o-44.2o (Ni-Fe alloy).

Figure 3.4: In-situ XRD during CO2-TPO. (A) 2D XRD pattern for Ni-5Fe/MgAl. Heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, 100% CO2. (B) Integral intensity variation of (A) for diffraction areas 35.4o-36.4o (Fe3O4) and 43.7o-44.2o (Ni-Fe alloy).

Figure 3.5: Schematic diagram of Ni-Fe alloy formation, during H2-reduction, and decomposition, during CO2 oxidation.

Figure 3.6: Ratio of CH4:CO2 consumption rates as a function of temperature for methane dry reforming (6 min at each temperature, Wcat:F

0CH4= 0.40-0.71 kgcat·s·mol-1): ◊: Ni-

0Fe/MgAl, □: Ni-5Fe/MgAl, ∆: Ni-8Fe/MgAl, x: Ni-11Fe/MgAl.

Figure 3.7: Ratio of CH4:CO2 consumption rates as a function of Time-On-Stream (TOS, min) for methane dry reforming at 1023K (Wcat:F

0CH4= 0.40-0.71 kgcat·s·mol-1), CH4:CO2=1:1,

total pressure of 101.3 kPa: ◊: Ni-0Fe/MgAl, □: Ni-5Fe/MgAl, ∆: Ni-8Fe/MgAl, x: Ni-11Fe/MgAl.

Figure 3.8: Space Time Yield for H2 and CO (STY, mmol·s-1·gNi-1) with Time-On-Stream (TOS, min)

for methane dry reforming at 1023K (Wcat:F0

CH4= 0.40-0.71 kgcat·s·mol-1), CH4:CO2=1:1 and total pressure of 101.3 kPa. (A): Ni-0Fe/MgAl (B): Ni-5Fe/MgAl.

Figure 3.9: CO:H2 ratio with Time-On-Stream (min) for all studied catalysts. ◊: Ni-0Fe/MgAl, □: Ni-5Fe/MgAl, ∆: Ni-8Fe/MgAl, x: Ni-11Fe/MgAl (error bars not shown).

Figure 3.10: SEM micrographs and EDX analysis of spent catalysts. (A) Ni-0Fe/MgAl SEM image. (B) Ni-0Fe/MgAl EDX. (C) Ni-8Fe/MgAl SEM image. (D) Ni-8Fe/MgAl EDX. Temperature 1023K, CH4:CO2=1:1, reaction time 4h.

Figure 3.11: CH4 consumption rate (mmol·s-1·gNi-1) during three catalytic cycles 1st cycle DRM

TOS=4h, 2nd and 3rd cycle TOS=1h. CO2 oxidation and H2 reduction: each 20 min. The CH4 consumption rate for each cycle was calculated after TOS=30 minutes.

Figure 3.12: CO Space Time Yield (STY, molCO·s-1·gNi+Fe-1) upon CO2 oxidation of deposited carbon for

different catalysts. The carbon was formed after pulsing methane for 6 min at 1023 K for all pre reduced samples (1123 K, 1 NmL/s 5%H2/He). Total amount of CO produced: (1) 0.04 mol CO/gNi+Fe. (2) 0.06 mol CO/gNi+Fe. (3) 0.07 mol CO/gNi+Fe.

Figure 3.13: Time resolved in-situ XRD patterns of Ni-5Fe/MgAl sample. (1) Ni-Fe alloy formation in 10%H2-TPR to 1023 K, heating rate: 30 K/min. Methane dry reforming: (2) CH4:CO2=1:2, (3) CH4:CO2=1;3, (4) CH4:CO2=1:6, at 1023 K and total pressure of 101.3 kPa.

List of Figures

ix

Figure 3.14: CO production during CO2 and CH4 alternate pulse experiment over (A) Ni-8Fe/MgAl;

(B) 5%Fe. Period I: 1 NmL/s of CO2, 1min; Period II: 1 NmL/s of He, 2min; Period III: 1 NmL/s of CH4, 1min at 1023 K and total pressure of 101.3 kPa. molCO/gmetal produced: (1) 0.15, (2) 0.13 and (3) 0.11 (4) 0.0.

Figure 4.1: Full XRD scans of used Ni-5Fe/MgAl (DRM for 1 h, 1023 K, CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa), after CO2-TPO (maximum temperature 1123 K, flow rate 10 NmL/s) and after O2-TPO (maximum temperature 1123 K, flow rate 10 NmL/s).

Figure 4.2: EDX element mapping of Ni-5Fe/MgAl. A) after DRM (1023 K, CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa, reaction time 1 h). (B) after CO2 oxidation (1 NmL/s of CO2 at a total pressure of 101.3 kPa and 1123K). Red, green and blue colors correspond to carbon, Fe and Ni elements respectively.

Figure 4.3: (A) CO2-TPO profile: CO intensity as a function of temperature for used (DRM for 1 h, 1023 K, CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa) Ni-0Fe/MgAl, Ni-5Fe/MgAl and graphite heating rate of 10 K/min, flow rate of 1 NmL/s of CO2. (B) O2-TPO profile: CO2 intensity as a function of temperature for used Ni-0Fe/MgAl, Ni-5Fe/MgAl and graphite, heating rate of 10 K/min, flow rate of 1 NmL/s of 10%O2/He. Black line: Ni-5Fe/MgAl, grey line: Ni-0Fe/MgAl and dashed line: graphite

Figure 4.4: Raman spectrum of the used Ni-5Fe/MgAl sample (DRM for 1 h, 1023 K, CH4:CO2= 1.1, total pressure of 101.3 kPa). Blue line: graphite, black line: used Ni-5Fe/MgAl catalyst, grey line: used Ni-5Fe/MgAl catalyst after CO2-TPO up to 950 K (removal of shoulder peak on Figure 2A), purple line: Ni-5Fe/MgAl catalyst after CO2-TPO up to 1123 K.

Figure 4.5: (A): HRTEM image of a used Ni-5Fe/MgAl catalyst (after DRM at 1023 K, CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa, reaction time 1 h). EDX element mapping of (B): carbon, (C): Ni and (D): Fe.

Figure 4.6: XPS detail windows of the C1s photoline for Ni-5Fe/MgAl : a) reduced b) after DRM reaction in CH4:CO2:He= 1:1:1.

Figure 4.7: In-situ XRD coupled with MS during CO2-TPO (heating rate 20 K/min, maximum temperature 1123 K, flow rate 10 NmL/s) of used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa):(A) 2D XRD pattern; (B) CO produced during carbon species removal as a function of temperature; (C) Integral intensity variation of (A) for diffraction areas 25.8-26.8o (Graphite), 35.4o-36.4o (Fe3O4) and 43.7o-44.2o (Ni-Fe alloy).

Figure 4.8: 2D in-situ XRD pattern during CO2-TPO (heating rate 20 K/min, maximum temperature 1123 K, flow rate 10 NmL/s). (A) Used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4;CO2= 1.1, total pressure of 101.3 kPa). Selected angular region of 35o-40o from Figure 6(A). (B) Reduced Ni-5Fe/MgAl catalyst (heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He). Selected angular region of 35o-40o.

Figure 4.9: Schematic representation of carbon species removal by CO2 over Ni-5Fe/MgAl catalyst. Cs: deposited carbon. Os: surface oxygen, OL: lattice oxygen. Cm: carbon deposited on metals, Cs: carbon deposited far from metals, Os: surface oxygen, OL:

List of Figures

x

lattice oxygen. The carbon illustration is not corresponding to the real carbon structure.

Figure 4.10: In-situ XRD during O2-TPO of a used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4:CO2= 1.1, total pressure of 101.3 kPa). (A) 2D XRD pattern for Ni-5Fe/MgAl . Heating rate: 20 K/min, maximum temperature 1123 K, flow rate: 10 NmL/s, air. (B) Integral intensity variation of (A) for diffraction areas 25.8o-26.8o (graphite), 35.7o-36.4o (Fe2O3) and 43.7o -44.2o (Ni-Fe alloy).

Figure 4.11: O2-TPO profile: CO2 intensity as a function of temperature for (A) mechanical mixture of graphite and Ni-5Fe/MgAl catalyst (solid line) and for graphite only (dashed line), (B): two and three layers catalyst bed configuration, including one layer of Ni-5Fe/MgAl catalyst and one layer of graphite (dot line) and a Ni-5Fe/MgAl catalyst, MgAl2O4 support and graphite (dash-dot dot line), respectively. Heating rate of 10 K/min, flow rate of 1 NmL/s of 10%O2/He.

Figure 4.12: CO2 response during O2 pulses at TAP reactor at 993 K. (A): 2D view for Ni, (B): CO2 molar flow rate produced during selected O2 pulses over Ni, (C): 2D view for Ni-5Fe/MgAl , (D): CO2 molar flow rate produced during selected O2 pulses over Ni-5Fe/MgAl . Ni and Ni-5Fe/MgAl aged by a sequence of 400 CH4 pulses.

Figure 4.13: 2D projected view of O2 response from TAP during O2 pulses at 993 K over: (A) reduced Ni, (B) aged Ni and (C) aged Ni-Fe catalyst.

Figure 4.14: HRTEM image of aged Ni catalyst after 400 CH4 pulses in TAP.

Figure 4.15: 2D TAP spectrum: CO response during CO2 pulses at 993 K over (A): Ni, (B):Ni-Fe catalyst.

Figure 4.16: Schematic representation of carbon species oxidation by O2 over Ni-5Fe/MgAl catalyst. Two different mechanisms are illustrated: (I) carbon gasification on the active metal (II) particles migration followed by carbon gasification through lattice oxygen. Cm: carbon deposited on metals, Cs: carbon deposited far from metals, Os: surface oxygen, OL: lattice oxygen. The carbon illustration does not represent the actual carbon structure.

Figure 5.1: Full XRD scans of (A) Ni-Fe-3Pd supported on MgAl2O4, as-prepared, reduced and oxidized (1 NmL/s of 10%H2/He or CO2 at a total pressure of 101.3 kPa and 1123 K). (B) Ni-Fe-3Pd reduced state along with possible bimetallic combinations (Fe-3Pd, Ni-3Pd) supported on MgAl (1 NmL/s of 10%H2/He, at a total pressure of 101.3 kPa and 1123 K).

Figure 5.2: EDX element mapping of 1.6-Pd. (A): as-prepared (B) reduced (1 NmL/s of 5%H2/He mixture at a total pressure of 101.3 kPa and 1123 K). Red, green and blue colors correspond to Fe, Ni and Pd elements, respectively.

Figure 5.3: 2D in-situ XRD pattern during H2-TPR for Ni-Fe-3Pd. Heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He.

Figure 5.4: Schematical representation of the alloy formation during H2-reduction up to 1123 K.

Figure 5.5: 2D in-situ XRD pattern during CO2-TPO for Ni-Fe-3Pd. Heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, CO2. The highlighted rectangular

List of Figures

xi

area shows the integral intensity variation of trimetallic Ni-Fe-Pd alloy for diffraction angles 41.7-42.2o as a function of temperature.

Figure 5.6: Full XRD scans of Ni-Fe-3Pd and Ni-Fe+3Pd supported on MgAl2O4, as-prepared and reduced (1 NmL/s of 10%H2/He at a total pressure of 101.3 kPa and 1123K).

Figure 5.7: 2D in-situ XRD pattern during H2-TPR for Ni-Fe+3Pd. Heating rate: 30 K/min, maximum temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He.

Figure 5.8: CH4 consumption rate (mmol·s-1·g-1metals) of samples after 4 h time-on-stream (TOS),

under DRM at 1023 K (Wmetals:F0

CH4 = 0.018-0.025 kgmetals·s·molCH4-1, total pressure of

101.3 kPa and CH4:CO2 = 1:1). XCH4 = 50 ± 5% for all studied samples. The x-Pd catalysts are trimetallic Ni-Fe-Pd samples with constant Fe:Ni ratio, containing different wt% of Pd (see Table 5.1). Error bars were calculated after three independent experiments representing standard deviation (68% confidence interval).

Figure 5.9: Calculated NiO crystallite size (nm) from different batches of 0-Pd (●), 0.1-Pd (■) and 0.2-Pd (▲) as determined from ex-situ XRD full scans.

Figure 5.10: Methane conversion as a function of time-on-stream (TOS) over monometallic Pd catalysts supported on MgAl2O4 under DRM at 1023 K (Wcat:F

0CH4 = 0.26 kgcat·s·molCH4

-

1, total pressure of 101.3 kPa and CH4:CO2 = 1:1). x: mono-0.1Pd, *: mono-0.2Pd, +: mono-0.5Pd.

Figure 5.11: Stability tests during DRM at 1023 K (total pressure of 101.3 kPa and CH4:CO2 = 1:1). (A): CH4 consumption rate (mmolCH4·s

-1·g-1metals); (B): CO2 consumption rate

(mmolCO2·s-1·g-1

metals); (C): CO:H2 ratio. ♦: Ni (Wmetals:F0

CH4 = 0.022 kgmetals·s·mol-1CH4), XCH4 from 67% to 53% ●: 0-Pd (Wmetals:

0FCH4 = 0.025 kgmetals·s·mol-1CH4), XCH4: from 62% to 24% ■: 0.1-Pd (Wmetals:F

0CH4 = 0.018 kgmetals·s·mol-1CH4), XCH4 from 64% to 36%; ▲:

0.2-Pd (Wmetals:F0

CH4 = 0.024 kgmetals·s·mol-1CH4), XCH4 from 58% to 44%. Error bars were calculated after three independent experiments representing standard deviation (68% confidence interval).

Figure 5.12: CH4 consumption rate (mmol·s-1·g-1metals) of used (after TOS=21h shown in Figure

5.11) and regenerated 0-Pd (blue) and 0.2-Pd (green) samples under DRM at 1023 K. Catalyst regeneration took place using 10vol%O2/He and 10vol%H2/He at 1023 K and immediately after the reaction.

Figure 5.13: Isothermal conversion of CH4 and C2H4 as a function of space time during bi-reforming at 1023 K and 101.3 kPa. The solid symbols represent the CH4 conversion while the open symbols represent the C2H4 conversion. ♦: Ni, ●: 0-Pd, ▲: 0.2-Pd. C2H4 equilibrium; CH4 equilibrium. Inset table: composition of feed mixture.

Figure 5.14: Schematical representation of the proposed deactivation for 0-Pd due to Ni-Fe surface alloy reconstruction during DRM at high temperature (1023 K).

Figure 5.15: Stability tests during DRM at 1023 K (total pressure of 101.3 kPa and CH4:CO2=1:1) and high initial CH4 conversion (~80%). ●: 0-Pd (Wmetals:F

0CH4= 0.037 kgmetals·s·mol-1CH4);

▲: 0.2-Pd (Wmetals:F0

CH4= 0.031 kgmetals·s·mol-1CH4).

List of Figures

xii

Figure 5.16: Time-resolved in-situ XRD pattern during isothermal treatment of 0-Pd under different reducing/oxidizing ratios (Rc) at 1023 K. The upper right inset shows the intensity variation as a function of time-on-stream (TOS) for the selected diffraction angles 43.2o-44o (Ni-Fe alloy).

Figure 5.17: Schematical representation of the proposed structure of Pd-modified samples (i.e 0.2-Pd) during DRM at high temperature (1023 K).

Figure 6.1: Structure showing the γ2 two- and η3 three-body configurations with angles at 116° (planar configuration) and 95° (perpendicular configuration) used in the modeled EXAFS signal. Blue color corresponds to oxygen, red to Fe and light blue to Al atoms.

Figure 6.2: Pore size distribution calculated by Jura and Harkins model (pressure range 0.25-0.98) for the studied support materials.

Figure 6.3: Support oxygen mobility. (a): Catalyst bed setup with mechanical mixture of MgFe0.18Al1.82O4 and graphite. inset: graphite oxidation by lattice oxygen. (b): Catalyst bed configuration during pulses in TAP reactor, (c): CO2 molar flow rate during CH4 pulses. (d): CO2 production as a function of pulse number.

Figure 6.4: Support characterization. a, HAADF-STEM image of as-prepared MgFe0.09Al1.81O4 along with (b-d): EDX elemental maps. (e): XANES spectra of support 5.0Fe at the Fe-K edge during H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10 K/min). (f): local environment of randomly dispersed Fe inside the spinel framework. Brown corresponds to Mg, light blue to Al, dark blue to oxygen and red to Fe atoms. (g): experimental and simulated XANES spectra for MgFe0.09Al1.81O4 after H2-TPR with (h): contributions from two different Fe sites, non-distorted (top), and reduced and distorted (bottom). ( ) simulated, ( ) experimental , ( ) distorted octahedron, ( ) non-distorted octahedron.

Figure 6.5: (a): FT of the fresh and reduced sample. The arrow highlights the shift in R occurring in the reduced sample. The shift indicates the formation of Fe2+ in the support. (b): Left figure shows the signals used in the fitting of the reduced sample. The γ and η signals represent contributions from the two-bodies and the three-bodies configuration, respectively. The signals in blue correspond to the distorted Fe local environment; the signals drawn in black are related to the non-distorted environment. The EXAFS data, the calculated fit and the residual are reported below. The right figure shows the EXAFS data of the fresh and reduced sample, with their respective fit.

Figure 6.6 CO yield (mol·kgsupport-1) of MgFe0.09Al1.91O4 support as a function of redox cycles at

750°C. The CO yield remained stable during 60 cycles illustrating the stability of the MgFexAl2-xO4 support under reducing/oxidizing environment. Each cycle lasted for 16 min and it was composed of 4 min of reduction (5%H2/Ar), 4 min of inert gas (He), 4 min of oxidation (CO2) and 4 min of inert gas (He). All the gas flow rates were 1.1 NmL/s.

Figure 6.7: Full XRD scans of 7.5Fe support and fresh, reduced (10 NmL/s of 10%H2/He mixture at a total pressure of 101.3 kPa and 1123K) and re-oxidized (10 NmL/s of CO2 mixture at a total pressure of 101.3 kPa and 1123K) Ni/7.5Fe.

List of Figures

xiii

Figure 6.8: Time-resolved in-situ XRD patterns of Ni/7.5Fe during H2-TPR. Conditions: maximum temperature 1123K , flow rate of 10 NmL/s, 5%H2/He, heating rate 20K/min.

Figure 6.9: a,b: HAADF-STEM image of the mapped area of reduced Ni/1.0Fe together with the EDX elemental map. Schematic representation of Ni-Fe alloy formation upon Fe migration from the support during H2-TPR. Green represents NiO, grey Ni, black Ni-Fe alloy while the arrows indicate the migration of Fe from the support towards Ni.

Figure 6.10: XANES evolution during H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10 K/min) of Ni/5.0Fe at the (B) Fe-K edge and (C) Ni-K edge.

Figure 6.11: (a): Full XRD scans of a mechanical mixture between 7.5Fe support and Ni/0Fe before and after an isothermal in-situ treatment (named as “fresh” and “reduced”, respectively) under 10 NmL/s of 5%H2/He mixture at a total pressure of 101.3 kPa and 1073 K. b, HR-STEM of the mechanical mixture after in-situ XRD along with c, EDX elemental map. Green color corresponds to Ni and red to Fe. Yellow dots at the interface indicate initiated alloying between Ni and Fe.

Figure 6.12: XANES spectra of Ni-4.0Fe supported on MgAl2O4 at (a) Fe-K and (b) Ni-K edge during H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10 K/min).

Figure 6.13: Schematic representation of carbon formation over a, b: benchmark Ni catalyst supported on MgAl2O4 and c, d: novel Ni catalyst supported on MgFexAl2-xO4. OL: lattice oxygen.

Figure 6.14: Activity and stability during DRM at 1023 K and 111.3 kPa. a, STY for CO and H2 after 2 h TOS for Ni/zFe (Extended Data Table 1). b, CH4 consumption rate: ■: Ni/2.5Fe, ♦: Ni/1.0Fe, x: Ni/0.5Fe at WNi:F

0CH4 = 0.024 kgNi∙s∙molCH4, x: Ni/0.5Fe at WNi:F

0CH4 = 0.011

kgNi∙s∙molCH4. Error bars: twice the standard deviation from three independent experiments. c, Deactivation due to Fe segregation from the Ni-Fe alloy for high (Fe:Ni > 1:3), low Fe content (Fe:Ni ≤ 1:9). d, Representation of carbon oxidation during DRM. OL: lattice oxygen. Red: Fe or FeOx, green: NiOx, grey: Ni, light grey: magnesium aluminate.

Figure 6.15: Catalyst activity (mol·s-1·kgNi-1) for a 8wt%Ni/MgFe0.007Al1.993O4 catalyst at 1023 K and

total pressure of 301.3 kPa. ●: During DRM using CH4:CO2 = 1:1 (WNi:F0

CH4 = 0.012 kgNi·s·mol-1; XCH4 ~ 45%), ●: during bi-reforming of methane using a feed composition of 40% CH4, 24% CO2, 16% H2O and 20% He (WNi:F

0CH4 = 0.035 kgNi·s·mol-1; XCH4 ~

68%).

Figure 7.1 The effect of Ni:Pd ratio on the catalyst activity (CH4 consumption rate, mmol·s-1·g-

1metals) and price. Lines are guides to the eye.

Figure A.1 The flow diagram of the step response setup with feed, reactor and analysis sections

located at Laboratory for Chemical Technology (LCT), Ghent University.

Figure A.2 The list of symbols used in the flow diagram in Figure A.1.

Figure A.3 Overview of the feed selection section.

Figure A.4 The overview of the reactor with the catalyst bed.

List of Figures

xiv

Figure A.5 The back pressure regulators.

Figure B.1 Ni3Fe (111) unit cell, without segregation (left) and with a Fe atom from the second

layer segregated to the surface layer in exchange for a Ni atom. Perspective view (top) and top view (bottom) (Ni: green, Fe: red)

Figure B.2 Ni2FePd (111) unit cell, without segregation (left) and with a Fe atom from the second layer segregated to the surface layer in exchange for a Ni atom (middle) or a Pd atom (right). Perspective view (top) and top view (bottom) (Ni: green, Fe: red, Pd: blue).

Figure B.3 Oxygen monolayer coverage, adsorbed at the most stable fcc site for all studied surfaces: Ni3Fe (top) and Ni2FePd (bottom). The same adsorption sites hold for a monolayer coverage of H atoms and CO molecules (Ni: green, Fe: bright red, Pd: blue, O: small dark red atom).

Figure B.4 fcc adsorption sites with numbering, on all considered structures, Ni3Fe (top) and Ni2FePd (bottom), non-segregated structure (left) and segregated structures in which an Fe is exchanged with a surface Ni atom (middle) or a surface Pd atom (right) (Ni: green, Fe: bright red, Pd: blue).

List of Abbreviations and Acronyms

xv

List of Tables

Table 1.1 Stoichiometry, reactants H/C ratio, products H2:CO ratio, reaction enthalpy, operating temperature and operating pressure for methane reforming technologies.

Table 1.2 Methane reforming side reactions: WGS, MD and BR.

Table 1.3 Overview of activities over selected bimetallic Ni-based catalysts during reforming reactions.

Table 1.4 Kinetic rate equations for DRM and SRM over a selection of catalysts.

Table 1.5 Overview of activities and carbon deposits of some selected catalysts during DRM.

Table 1.6 Acid gas removal (AGR) technologies.

Table 1.7 Common poisons classified according to chemical structure [55]

Table 2.1 Overview of synthesized support materials.

Table 2.2 Overview of synthesized catalysts.

Table 2.3

Overview of samples synthesized for comparison purposes.

Table 3.1 Catalyst and support properties

Table 5.1 Catalyst and support properties

Table 5.2 Segregation energies without (ΔEseg, kJ/mol) and with adsorbates (ads

segE ) for the exchange of Fe in the subsurface layer of a (111) surface of Ni3Fe or Ni2FePd, with Ni or Pd from the surface layer, for various coverages, representative for DRM. All coverages refer to the species adsorbed on the fcc sites of a periodically repeated unit cell with 4 surface atoms.

Table 6.1 Support properties. Metal loading was determined by means of inductively coupled plasma atomic emission spectroscopy (ICP-AES).

Table 6.2 The results from the EXAFS model data of the as prepared, reduced and re-oxidized sample 5.0Fe. N is the coordination number, σ² the Debye Waller factor, R the radial distance from the central Fe absorber, θ the angle in the Fe-O-M three-body configuration. In reduced state, two Fe contributions are considered: 1: non-distorted, 2: reduced/distorted octahedron. β represents the corresponding fractions. Bold values were fitted, non-bold values were calculated.

Table 6.3 Catalyst properties. Metal loading was determined by means of inductively coupled plasma atomic emission spectroscopy (ICP-AES).

Table A.1 The overview of the mass flow controllers with their feed specifications.


xvi

Table B.1 Location of the adsorbates for the mixed adsorbate layers on the fcc positions, numbered according to Figure B.4.


xvii


Symbol Description

ALD Atomic layer deposition

AMU Atomic mass unit

ATR Autothermal reforming of CH4

BET Brunauer-Emmett-Teller

BIM Bi-reforming of CH4

BJH Barrett-Joyner-Halenda

BR Boudouard reaction

CSZ Ceria-stabilized zirconia

CTL Coal-to-liquids

DFT Density functional theory

DRM Dry Reforming of Methane

EDX Energy Dispersive X-ray spectroscopy

EELS Electron Energy Loss Spectroscopy

EXAFS Extended X-ray Absorption Fine Structure

FT Fischer-Tropsch

GTL Gas-to-liquids

HRTEM High-Resolution Transmission Electron Microscopy

ICP-AES Inductively coupled plasma atomic emission spectroscopy

LHHW Langmuir-Hinshelwood-Hougen-Watson

MD Methane Decomposition

MFC Mass flow controller

MMSI Medium metal-support interactions

MS Mass spectrometer

OCR Oxy-reforming of CH4

PDF Powder diffraction file

PMC Particle migration and coalescence

POM Partial oxidation of CH4

QXAS Quick X-ray absorption spectroscopy

RAMAN RAMAN spectroscopy

RDS Rate-determining step

SEM Scanning electron microscopy

STEM Scanning transmission electron microscopy

SMSI Strong metal-support interaction

SR Steam reforming

SRM Steam reforming of methane

TAP Temporal analysis of products

TEM Transmission electron microscopy

TOS Time-on-stream

TP Temperature-programmed


xviii

TPO Temperature-programmed oxidation

TPR Temperature-programmed reduction

WGS Water-gas shift reaction

WMSI Weak metal-support interaction

XANES X-ray absorption near-edge structure

XAS X-ray absorption spectroscopy

XPS X-ray photoelectron spectroscopy

XRD X-ray diffraction

List of Symbols

xix

List of Symbols

Symbol Description Units

<u> Local flow velocity m·s-1

µ(E) Absorption coefficient m-1

A Reactor cross-sectional area m²

a Constant for radial dispersion criterion -

Aj Amplitude of coordination shell j -

av Pellet volume specific area m2particle ·m

-3particle

b Volume of inert material as fraction of total solids m³inert ·m-3

inert+cat

b Constant for radial dispersion criterion -

b Instrumental width (Scherrer equation) -

Bo Bodenstein number -

c BET constant -

CCH4,b Bulk CH4 concentration mol ·m-3

CCH4,s Surface CH4 concentration mol ·m-3

CFi Calibration factor for compound i -

DCH4,ax Axial diffusivity m² ·s-1

DCH4,eff Effective diffusivity m² ·s-1

dp Pellet diameter m

dt Internal reactor tube diameter m

dXRD Crystallite size m

E1 Adsorption enthalpy for first layer kJ ·mol-1

Ea Activation energy kJ ·mol-1

EL Heat of condensation kJ ·mol-1

ff Fanning friction factor -

Fi0 Inlet molar flow rate mol· s-1

Fi, Outlet volumetric flow rate mol· s-1

Fi,r Difference between inlet and outlet volumetric

flow rate of reactant i m³ s-1

fm Modified friction factor -

h Planck’s constant m² ·kg ·s-1

I0 Incident beam intensity W ·m-2

It Transmitted beam intensity W ·m-2

jH Chilton Colburn factor -

K Scherrer constant -

kg Mass transfer coefficient between pellet and gas

phase m³reactor ·m

2interface ·s

-1

L Reactor length m

n Diffraction order (integer) -

Nu Nusselt number -

List of Symbols

xx

p Equilibrium pressure Pa

p0 Saturation pressure Pa

Pe’ Modified Péclet number -

Pr Prandtl number -

ptot Total pressure Pa

R Universal gas constant J ·mol-1 ·K-1

r Sample thickness m

Rc The ratio between reducing and oxidizing gases

(CO+H2:CO2+H2O) -

Re Reynolds number -

Robsv,CH4 Observed volumetric reaction rate of CH4 mol ·s-1 ·m-3

Si Signal for compound i -

Si,300 K Reference signal for compound i at 300 K -

Tf Film temperature K

Tw Wall temperature K

u Flow velocity m ·s-1

v Adsorbed gas quantity mol

vm Monolayer adsorbed gas quantity mol

XCH4 CH4 conversion -

xdil Fraction diluted material -

Xi Conversion of reactant i -

Xi Fractional conversion of compound i -

Greek Symbols

Symbol Description Units

αp Heat transfer coefficient W ·m-2 ·K-1

β Line broadening at half maximum intensity -

ΔH°298K Standard reaction enthalpy at 298K kJ ·mol-1

Δp Pressure drop through fixed bed Pa

ΔrH Reaction enthalpy kJ ·mol-1

εb Bed porosity -

η Effectiveness factor -

θ Incidence angle °

θs Sulfur coverage -

λ Incoming X-ray wavelength m

λ Conductive heat transfer coefficient W ·m-1 ·K-1

λer Effective radial conductivity W ·m-1 ·K-1

λer,0 Contribution of conductive heat transfer to the

effective radial heat transfer coefficient W ·m-1 ·K-1

λer,conv Contribution of convective heat transfer to the

effective radial heat transfer coefficient W ·m-1 ·K-1

λp Pellet conductive heat transfer coefficient W ·m-1 ·K-1

List of Symbols

xxi

ν Frequency s-1

ρ Fluid density kg ·m-3

Φ Weisz modulus -

Φj Sine argument for coordination shell j -

χ Summation over coordination shells -

ψ Wave state function m-3/2

Glossary of terms

xxiii

Glossary of terms

Alloy A metal that it is made by mixing two or more metals, or a metal and another substance. The structure of alloy can vary according to the environment i.e. well mixed, core-shell etc.

Alternate pulse experiment The sequential pulses of different feed compositions applied isothermally to the catalyst. Typically the sequence of reducing-oxidizing gases were applied.

As-prepared catalyst The catalyst after the calcination step.

Bi-reforming Combination of steam and dry reforming in one step with a CH4:H2O:CO2 of 3:2:1 produces a gas mixture with 2:1 ratio of H2:CO.

Catalytic carbon gasification The removal of the deposited carbon species from the catalyst surface by O2, CO2, H2O or H2.

Catalyst Substance or material, which through repeated cycles of elementary steps, accelerates the conversion of reagents into products. Catalysts are classified into homogeneous, which are in the same phase with the reagents (e.g. acids and bases, metal complexes, etc.), and heterogeneous, which are separated from the reactants by an interface (e.g. metals, metal oxides, etc.).

CH4 consumption rate Normalized measure for the amount of methane that has been reacted. Calculated from the difference between the inlet and outlet molar flow rates, as measured relative to an internal standard.

Coverage A fraction of the adsorption sites on the surface of a solid occupied by the adsorbate (adsorbed substance).

Deactivation Loss in catalytic activity due to chemical, mechanical or thermal process.

Elementary step A single reaction step with a single transition

http://eng.thesaurus.rusnano.com/wiki/article1779

Glossary of terms

xxiv

state and no intermediates.

Induction period The initial slow stage of a process. After this stage the process accelerates. Induction periods are often observed with radical reactions, but they may also occur in other systems (for example before steady-state concentration of the reactants is reached)

Lattice oxygen The oxygen present in the lattice of an oxide catalyst. It can act as reservoir for oxygen releasing it during reactions creating vacancies. The vacancies can be re-filled by gas phase oxidizing agents such as O2, CO2 or H2O.

Mars-van Krevelen mechanism A reaction between the oxide catalyst and a reducing agent (i.e. hydrocarbon), in which the latter is oxidized and the former is reduced, followed by the reaction of reduced oxide with an oxidizing agent (H2O, CO2, O2) to restore the initial state.

Oxygen storage capacity The total amount of mobile/exchangeable oxygen present in support or catalyst.

Poisoning Refers to the partial or total deactivation of a catalyst caused by exposure to a range of chemical compounds. These compounds can be reactants, products or impurities.

Porosity A measure of the void space in a material, expressed as the ratio of the volume of voids to the total volume of the material.

Promoter Substance added to a catalyst in order to improve its performance, such as activity, selectivity or stability, in a chemical reaction. By itself the promoter has little or no catalytic effect. It can interact with the active component of the catalyst and thereby alter its properties.

Reduction The pre-treatment process performed prior to each activity or stability test in order to reduce the metal oxides to metals.

Spent catalyst The recovered catalyst after an activity or stability test

Glossary of terms

xxv

Spillover Spillover involves the transport of active species sorbed or formed on a first surface onto another surface that does not, under the same conditions, sorb or form the active species

Syngas A variable composition mixture of hydrogen and carbon monoxide.

Tri-reforming Combination of three catalytic reforming processes by CO2 (endothermic), H2O (endothermic) and O2 (exothermic) in one step.

Summary

xxvii

Summary

Global climate changes impose the need to investigate new routes for utilization of

available resources such as natural gas and biogas, to produce fuels and chemicals. Methane

originates from petroleum reserves and landfill gas, and its conversion to higher value

products will become increasingly important for the foreseeable future. It can be used

indirectly as a feedstock for the synthesis of chemicals and fuels by converting it into syngas

(a mixture of CO and H2). Syngas is a versatile building block for chemical industry and it can

be further applied in downstream processes (Figure 1) for the production of chemicals, fuels

e.g. via the Fischer-Tropsch process etc.

Figure 1: Schematic overview of syngas production via CH4 reforming and further utilization in

downstream processes like Fischer-Tropsch and alcohol synthesis.

From the different reforming processes to produce syngas, this work focuses more on

methane dry reforming (DRM) and bi-reforming (BIM), rather than on conventional steam

reforming (SRM). DRM utilizes CH4 and CO2, the two most important greenhouse gases, and

has a 20% lower operating cost than SRM while producing high purity syngas, with low CO2

content and H2:CO ≤1. On the other hand, BIM combines steam and dry reforming in one

step, while the presence of H2O offers flexibility to the desired H2:CO ratio, as it can be

2 CO + 2 H2

CH4

H2O, CO2

Ni catalyst

CnH2n+2

CH3OHCH3CH2OH

Carbon deposition

Natural gas Biogas

Reforming SyngasDownstream

processes

Regenerated catalyst

Control of deactivation

rate

Summary

xxviii

altered by changing the H2O:(H2O+CO2) feed ratio. However, one of the major challenges

associated with all reforming processes, is the catalyst deactivation mainly due to carbon

deposition. Controlling carbon deposition will prolong the catalyst lifetime which is of

importance for the chemical industry.

A first attempt to control the catalyst deactivation due to carbon deposition was made

by adding Fe, one of the most abundant earth elements, as a promoter to Ni-based catalysts.

A series of bimetallic Ni-Fe/MgAl2O4 catalysts with Fe:Ni ratios between 0 and 1.5, were

examined from 923 to 1073 K. Methane dry reforming was selected as reforming reaction.

Alloy formation upon intimate interaction between Ni and Fe was found after the reduction

treatment. This alloy remains stable in a gas stream of CO2 during re-oxidation, up to 900 K,

but decomposes to metallic Ni and Fe3O4 above this temperature (Figure 2).

Figure 2: Mechanism of methane dry reforming over an Ni-Fe catalyst along with the Ni-Fe alloy

phase transformation during reduction/oxidation processes.

The process of dry reforming on Fe-Ni could be described by the Mars-Van Krevelen

mechanism, as illustrated in Figure 2. CO2 oxidizes Fe to FeOx, and CH4 is activated on Ni sites

to form H2 and surface carbon. The latter is re-oxidized by lattice oxygen from FeOx,

producing CO. Lower amounts of accumulated carbon were observed after DRM on Fe-

promoted samples compared to the monometallic Ni catalyst supported on MgAl2O4.

CH4+CO2

CO2

CO

Fe FeOx CH4

H2

O

CO

NiFe

NiFeNi

Fe3O4

reduction oxidation

Ni

FeOx

Fe

Ni-Fe alloy

Summary

xxix

Despite all the different ways to control the catalyst deactivation due to carbon

deposition, carbon accumulation during reforming reactions remains an issue. Eventually,

catalyst regeneration is required in order to remove all carbon species by gasification.

Therefore, it is important to know the catalyst regeneration mechanisms. Two different

oxidizing agents were used, CO2 and diluted O2, resulting in one endothermic and one

exothermic reaction with the carbon deposits (Figure 3). A transient response technique,

Temporal Analysis of Products (TAP) , has been used, for the first time, to investigate the

isothermal carbon species gasification processes.

Figure 3: Schematic representation of carbon species oxidation by CO2 and O2 over Ni-5Fe/MgAl2O4

catalyst. Cs: carbon deposited far from metals, Os: surface oxygen, OL: lattice oxygen. The carbon

illustration does not represent the actual carbon structure.

It was concluded that significant differences on the regeneration mechanism exist

depending on the oxidizing gas. More specifically, the CO2-regeneration only resulted in the

removal of carbon deposits that were located on the active metals of the catalysts. On the

other hand, the mechanism of carbon species oxidation by O2 (or air) can completely remove

them and can be described by two consecutive processes: 1) oxidation of active metals to

oxides and subsequent removal of surface carbon via lattice oxygen, resulting in local

Carbon gasification by CO2

Carbon gasification by O2 (air)

MgAl2O4

NiFe3O4

Cs

Os

Cm Cm

Fe

OL

Fe

Fe2O3NiO

NiCs Cs

Particles migration

Particles migration

CmOL OL

MgAl2O4

Summary

xxx

temperature increase, close to Tammann temperature, due to the exothermicity of these

reactions and 2) metal oxides migration to the carbon species that are deposited far from

the active metals and subsequent oxidation through lattice oxygen of the iron and/or nickel

oxides. The contribution of oxygen spillover in carbon gasification, was found to be

negligible, for the Ni-Fe/MgAl2O4 system.

Although Ni-Fe catalysts present high activity in methane reforming reactions, with high

carbon resistance, they do not deliver the required stability, as they suffer from deactivation

due to sintering and Fe segregation, during long time-on-stream (TOS> 4h) experiments. The

ratio between reducing and oxidizing gases [(CO+H2):(CO2+H2O)], Rc, is important for

determining the resulting phase in the iron/iron oxides system and thus for the stability of

Fe-containing alloys. During methane reforming, iron involved in the CO2 or H2O activation is

segregated from the Ni-Fe alloy, which leads to catalyst deactivation by active site blocking.

Two different routes to enhance the stability of Ni-Fe catalysts were investigated: a) by

adding a low concentration of noble metal (Figure 4) and b) by changing the location of Fe

(Figure 5a).

Figure 4: Effect of noble metal addition on the stability of Ni-Fe catalysts during reforming

reactions. CH4 consumption rate (mmolCH4·s-1·g-1

metals) during DRM at 1023 K (total pressure of 101.3

kPa and CH4:CO2 = 1:1). ●: Bimetallic Ni-Fe (Wmetals:F0

CH4 = 0.025 kgmetals·s·mol-1CH4), XCH4: from 62% to

24% ▲: Trimetallic Ni-Fe-0.2wt%Pd (Wmetals:F0

CH4 = 0.024 kgmetals·s·mol-1CH4), XCH4 from 58% to 44%.

Ni-Fe-0.2wt%Pd

Ni-Fe

▲

●

Summary

xxxi

Enhanced control of the stability and activity of Ni-Fe/MgAl2O4 can be achieved by

means of noble metal addition. Pd was selected as a noble metal, systematically

investigating the effect of its loading, while the Fe:Ni ratio was kept constant. During the

reduction process, a core shell alloy forms, consisting of a Ni-Fe core and a Ni-Fe-Pd shell.

During CO2-TPO, iron segregates from both core and shell, leading to alloy decomposition

above 850 K. The addition of Pd stabilizes the Ni-Fe alloy by means of a thin Ni-Fe-Pd surface

layer, which acts as a barrier for Fe segregation from the core during reforming reactions. A

Ni:Pd molar ratio equal to 75 was found to be the optimal promotion as it increased both the

activity and stability of the catalyst. The origin of the improved catalytic activity derives from

higher Ni dispersion along with the creation of extra active sites, due to Pd addition.

The second investigated route to enhance the stability of a Ni-Fe catalyst was by

changing the location of Fe and eliminating the use of noble metals even in small

concentrations. A novel MgFexAl2-xO4 support was synthesized for Ni catalysts, where Fe was

incorporated into the magnesium aluminate spinel, forming a support with redox

functionality (Figure 5) and high thermal stability.

Figure 5: (a): The local environment of randomly dispersed Fe inside the MgFexAl2-xO4 spinel

framework. Brown corresponds to Mg, light blue to Al, dark blue to oxygen and red to Fe atoms.

(b): CO2 production (10-9·mol) as a function of pulse number during CH4 pulses over an “as-

prepared” MgFe0.18Al1.82O4, using Temporal Analysis of Products, showing support’s oxygen

mobility.

Upon reduction of Ni catalyst supported on MgFexAl2-xO4, a Ni-Fe alloy is formed via

migration of Fe from the support lattice. By carefully down tuning the amount of Fe in

ba

Summary

xxxii

MgFexAl2-xO4, we have obtained a catalyst, where the surface alloy has a Fe:Ni molar ratio

~1:10, which fully mitigates carbon deposition at atmospheric pressure, while keeping Ni

sites accessible (Figure 6). It was found that after 65 h time-on-stream under DRM with

CH4:CO2 of 1:1 there was no deposited carbon. Thus, this Ni-Fe/MgFexAl2-xO4 makes a

combination of a stable alloy (Figure 6c), due to low Fe concentration, and an active support.

In this concept, adding a minimal amount of iron “kills two birds with one stone”: it allows

stability comparable to noble metals while controlling carbon deposition.

Figure 6: Stability during DRM at 1023 K and 111.3 kPa. (a): CH4 consumption rate (mmol·s-1·gNi-1):

■: Ni/2.5Fe, ♦: Ni/1.0Fe, x: Ni/0.5Fe at WNi:F0

CH4=0.024 kgNi∙s∙molCH4-1, x: Ni/0.5Fe at WNi:F

0CH4=0.011

kgNi∙s∙molCH4-1. Error bars: twice the standard deviation from three independent experiments. (b),

Deactivation due to Fe segregation from the Ni-Fe alloy for high (Fe:Ni > 1:3), low Fe content (Fe:Ni

≤ 1:9), (c) CH4 consumption rate (mmol·s-1·gNi-1, closed symbols) and conversion (%, open symbols)

during steam-dry reforming, using a feed composition of 15% CH4, 11% CO2, 4% H2O and 70% Ar, at

1023 K and 911.7 kPa. ♦: Ni/1.0Fe at WNi:F0

CH4 = 0.00342 kgNi·s·mol-1CH4.

Iron and its oxides are making a comeback as they start to be widely employed in the

development of catalysts. Specifically for the reforming reactions, the addition of Fe in the

appropriate concentration can control the carbon deposition, due to its unique redox

a b

c

Summary

xxxiii

properties, and thus control the catalyst deactivation rate. This can have different benefits

for the industry, starting from the longer lifetime of the catalytic material to the total cost of

the catalyst, since Fe is one of the most abundant elements in the earth’s crust.

xxxv

Samenvatting

Klimaatveranderingen op aarde maken het noodzakelijk om nieuwe routes te

onderzoeken die de beschikbare grondstoffen, zoals aardgas en biogas, gebruiken bij de

productie van brandstoffen en chemicaliën. Methaan is afkomstig uit petroleumreserves en

stortgas, en de omzetting ervan tot meerwaardeproducten zal in de nabije toekomst steeds

belangrijker worden. Het kan onrechtstreeks gebruikt worden als grondstof voor de

synthese van brandstoffen en chemicaliën door het eerst om te zetten in synthesegas, een

mengsel van CO en H2. Synthesegas is een veelzijdige bouwsteen voor de chemische

industrie en kan verder toegepast worden in downstream processen (Figuur 1) voor de

productie van chemicaliën, brandstoffen, b.v. via het Fischer-Tropsch proces.

Figuur 7: Schematisch overzicht van synthesegasproductie via CH4 reforming en verder gebruik in

downstream processen zoals Fischer-Tropsch en alcoholsynthese.

Van de verschillende reforming processen om synthesegas te produceren uit methaan,

spitst dit werk zich toe op droge reforming (DRM) en bi-reforming (BIM), eerder dan op

conventionele stoom reforming (SRM). DRM maakt gebruik van CH4 en CO2, de twee

belangrijkste broeikasgassen, en heeft een 20% lagere werkingskost dan SRM, terwijl

synthesegas van hoge zuiverheid gevormd wordt, met een laag CO2 gehalte en een

2 CO + 2 H2

CH4

H2O, CO2

Ni catalyst

CnH2n+2

CH3OHCH3CH2OH

Carbon deposition

Natural gas Biogas

Reforming SyngasDownstream

processes

Regenerated catalyst

Control of deactivation

rate

xxxvi

verhouding H2:CO ≤1. Anderzijds combineert BIM stoom en droge reforming in één stap,

waarbij de aanwezigheid van H2O flexibiliteit naar de gewenste H2:CO verhouding toelaat,

aangezien deze aangepast kan worden door keuze van de H2O/(H2O+CO2) verhouding in de

voeding. Een van de belangrijkste uitdagingen verbonden met alle reforming processen,

echter, is deactivering van de katalysator, hoofdzakelijk door koolstofafzetting. Het

controleren van koolstofafzetting zal de levensduur van de katalysator verhogen, wat van

belang is voor de chemische industrie.

Een eerste poging om katalysatordeactivering door koolstofafzetting onder controle te

houden werd ondernomen door het toevoegen van Fe, een van de meest beschikbare

elementen op aarde, als promotor voor Ni-gebaseerde katalysatoren. Een reeks

bimetallische Ni-Fe/MgAl2O4 katalysatoren met Fe:Ni verhouding tussen 0 en 1.5 werd

getest tussen 923 en 1073 K. Droge reforming van methaan werd geselecteerd als reactie. Er

werd vastgesteld dat er zich een legering vormde door intieme interactie tussen Ni en Fe

tijdens de reductiebehandeling. Deze legering blijft stabiel in een gasstroom van CO2 tijdens

reoxidatie tot 900 K, maar ontbindt in metallisch Ni en Fe3O4 boven deze temperatuur

(Figuur 2).

Figuur 8: mechanisme van droge reforming over een Ni-Fe katalysator en Ni-Fe fasetransformatie

tijdens reductie/oxidatie processen.

xxxvii

Het proces van droge-reforming over Ni-Fe kan beschreven worden door het Mars-Van-

Krevelen mechanisme, zoals geïllustreerd in Figuur 2. CO2 oxideert Fe tot FeOx, en CH4 wordt

geactiveerd op Ni sites met vorming van H2 en oppervlak-gebonden koolstof. Deze laatste

wordt geoxideerd door roosterzuurstof van FeOx, met productie van CO tot gevolg. Na DRM

over Fe-gepromote monsters werden lagere hoeveelheden geaccumuleerd koolstof

gevonden, vergeleken met de monometallische Ni katalaysator gedragen op MgAl2O4.

Ondanks alle verschillende manieren om katalysatordeactivering door koolstofafzetting

binnen de perken te houden, blijft de accumulatie van koolstof tijdens reformingreacties een

uitdaging. Uiteindelijk is katalysatorregeneratie onvermijdelijk om alle koolstofspecies te

verwijderen door vergassing. Daarom is het belangrijk om de katalysator-

regeneratiemechanismes te kennen. Dit werd onderzocht met twee verschillende oxidantia,

CO2 en verdund O2, wat resulteerde in een endotherme en een exotherme reactie met de

koolstofafzettingen (Figuur 3). Een transiënte responstechniek, Temporal Analysis of

Products (TAP), werd voor het eerst gebruikt bij het onderzoek van de isotherme vergassing

van koolstofafzettingen.

Figuur 9: Schematische voorstelling van de oxidatie van koolstofspecies door CO2 en O2 over een Ni-

5Fe/MgAl2O4 katalysator. Cs: koolstof die ver van metalen werd afgezet, Os: oppervlakzuurstof, OL:

roosterzuurstof. De voorstelling van koolstof stelt niet de werkelijke structuur voor.

xxxviii

Er werd vastgesteld dat aanzienlijke verschillen in het regeneratiemechanisme bestaan

afhankelijk van het oxiderend gas. Meer specifiek leidde CO2-regeneratie enkel tot

verwijdering van koolstofafzetting aanwezig op de actieve metalen van de katalysator.

Anderzijds kan het oxidatiemechanisme van koolstofspecies door O2 (of lucht) tot een

volledige verwijdering leiden, beschreven door twee opeenvolgende processen: 1) oxidatie

van de actieve metalen tot oxides gevolgd door verwijdering van oppervlakkoolstof via

roosterzuurstof, resulterend in een lokale temperatuurstijging, tot dicht tegen de Tammann-

temperatuur, omwille van de exothermiciteit van deze reacties, en 2) migratie van

metaaloxides naar de koolstofspecies, ver van de actieve metalen afgezet, gevolgd door hun

oxidatie door de roosterzuurstof van ijzer- en/of nikkeloxides. De bijdrage van

zuurstofspillover tot de koolstofvergassing werd verwaarloosbaar bevonden voor het Ni-

Fe/MgAl2O4 systeem.

Ondanks de hoge activiteit van Ni-Fe katalysatoren bij methaan-reformingreacties en

hun hoge weerstand tegen koolstof, leveren ze niet de benodigde stabiliteit, omdat ze lijden

onder deactivering door sinteren en Fe segregatie tijdens langdurige (time-on-stream > 4u)

experimenten. De verhouding tussen reducerende en oxiderende gassen

[(CO+H2):(CO2+H2O)], Rc, is belangrijk voor de bepaling van de resulterende fase in

ijzer/ijzeroxide systemen en bijgevolg voor de stabiliteit van Fe-houdende legeringen.

Tijdens reforming van methaan segregeert ijzer, betrokken bij de CO2 of H2O activering, uit

de Ni-Fe legering, wat leidt tot katalysatordeactivering door het blokkeren van actieve sites.

Twee verschillende methodes werden onderzocht om de stabiliteit van Ni-Fe katalysatoren

te verbeteren: a) het toevoegen van een nobel metaal in kleine concentratie (Figuur 4) en b)

het veranderen van de locatie van Fe (Figuur 5a).

xxxix

Figuur 10: Het effect van toevoeging van een nobel metaal op de Ni-Fe katalysator tijdens

reformingreacties. CH4 verdwijnsnelheid (mmolCH4 s-1 g-1metaal) tijdens DRM bij 1023 K (totaaldruk

101.3 kPa en CH4:CO2 = 1:1). ▲: bimetallisch Ni-Fe (Wmetaal:F0

CH4 = 0.025 kgmetaal s molCH4-1), XCH4: van

62% tot 24%; ● : trimetallisch Ni-Fe-0.2w%Pd (Wmetaal:F0

CH4 = 0.024 kgmetals·s·mol-1CH4), XCH4 van 58 %

tot 44%.

Verhoogde controle over de stabiliteit en activiteit van Ni-Fe/MgAl2O4 kan verkregen

worden door toevoeging van een edelmetaalpromotor. Pd werd geselecteerd als

edelmetaal, waarbij systematisch het effect van de belading onderzocht werd, terwijl de

Fe:Ni verhouding constant gehouden werd. Tijdens het reductieproces wordt een core-shell

legering gevormd, bestaande uit een Ni-Fe kern en een Ni-Fe-Pd schil. Tijdens CO2-TPO

verliezen zowel de kern als de schil ijzer, hetgeen leidt tot ontbinding van de legeringen

boven 850 K. De toevoeging van Pd stabiliseert de Ni-Fe legering door middel van een dunne

Ni-Fe-Pd oppervlaklaag, die optreedt als barrière voor Fe segregatie uit de kern tijdens de

reformingreacties. Een molaire verhouding Ni:Pd van 75 werd gevonden als optimale

promotie aangezien deze zowel de activiteit als de stabiliteit van de katalysator verhoogde.

De oorsprong van de verbeterde katalytische activiteit ligt in de hogere Ni dispersie alsook

de vorming van bijkomende actieve sites door de Pd toevoeging.

De tweede onderzochte methode voor het verhogen van de stabiliteit van een Ni-Fe

katalysator is door het verplaatsen van Fe en het elimineren van het gebruik van nobele

Ni-Fe-0.2wt%Pd

Ni-Fe

▲

●

xl

metalen, zelfs in kleine concentraties. Een nieuwe MgFexAl2-xO4 drager voor Ni katalysatoren

werd gesynthetiseerd, waarbij Fe in de magnesiumaluminaat spinel vervat was, wat leidde

tot een drager met redox functionaliteit (Figuur 5) en hoge thermische stabiliteit.

Figuur 11: (a) De lokale omgeving van willekeurig gedispergeerd Fe in de MgFexAl2-xO4

spinelstructuur. Bruin komt overeen met Mg, lichtblauw met Al, donkerblauw met zuurstof en

rood met Fe atomen. (b) CO2 productie als functie van pulsnummer tijdens CH4 pulsen over een

ongebruikte MgFe0.18Al1.82O4 drager, Zuurstofmobiliteit van de drager bepaald via Temporal

Analysis of Products.

Bij reductie van de Ni katalysator gedragen op MgFexAl2-xO4 wordt een Ni-Fe legering

gevormd via migratie van Fe uit het rooster van de drager. Door de hoeveelheid Fe in

MgFexAl2-xO4 te beperken wordt een katalysator verkregen, waarbij de oppervlaklegering

een Fe:Ni molaire verhouding ~1:10 heeft, wat koolstofafzetting bij atmosfeerdruk volledig

vermijdt, terwijl de Ni sites toegankelijk blijven (Figuur 6). Na 65 u time-on-stream tijdens

DRM met CH4:CO2 van 1:1 was er geen koolstof afgezet. Bijgevolg combineert dit Ni-

Fe/MgFexAl2-xO4 een stabiele legering (Figuur 6c), door de lage Fe concentratie, en een

actieve drager. In dit concept zorgt de minimale hoeveelheid ijzer voor "twee vliegen in een

klap": het zorgt voor een stabiliteit vergelijkbaar met edelmetalen, terwijl koolstofafzetting

gecontroleerd wordt.

ba

xli

Figuur 12: Stabiliteit tijdens DRM bij 1023 K en 111.3 kPa. (a) CH4 verdwijnsnelheid (mmol s-1 gNi-1)

■: Ni/2.5Fe, ♦: Ni/1.0Fe, x: Ni/0.5Fe bij WNi:F0

CH4=0.024 kgNi∙s∙molCH4-1, x: Ni/0.5Fe bij

WNi:F0

CH4=0.011 kgNi∙s∙molCH4-1. Foutenbalk: tweemaal de standaardafwijking van drie

onafhankelijke experimenten. (b) Deactivering door Fe segregatie vanuit de Ni-Fe legering bij hoog

(Fe:Ni >1:3), laag Fe gehalte (Fe:Ni ≤1:9), (c) CH4 verdwijnsnelheid (mmol s-1 gNi-1, gesloten

symbolen) en conversie (%, open symbolen) tijdens BIM, gebruikmakend van een voeding van 15%

CH4, 11% CO2, 4% H2O en 70% Ar, bij 1023 K en 911.7 kPa. Ni/1.0Fe bij WNi:F0

CH4 = 0.00342 kgNi s

mol-1CH4.

Ijzer en zijn oxides zijn aan een wederopkomst bezig en beginnen wijdverspreid gebruikt

te worden bij het ontwikkelen van katalysatoren. Specifiek voor reformingreacties kan de

toevoeging van Fe in een geschikte concentratie, dankzij zijn unieke redoxeigenschappen,

koolstofafzetting en bijgevolg ook de deactiveringssnelheid van de katalysator beperken. Dit

kan verschillende voordelen hebben voor de industrie, te beginnen met een langere

levensduur van het katalytisch materiaal tot de totale kost van de katalysator, aangezien Fe

een van de meest voorkomende elementen in de aardkorst is.

a b

c

1

Chapter 1

Introduction

Fossil fuels remain our major source of energy as they cover approximately 78% of

today’s energy demand [1] and it is foreseen that they will prevail for the next 30-40 years.

However, the increasing energy consumption creates the need to investigate new routes in

order to utilize the available resources, like natural gas and biogas, to produce fuels and

chemicals [2]. Natural gas, which primarily consists of methane (CH4), has an actual market

price of 1.9 USD/GJ. This makes CH4 one of the most affordable carbon feedstocks around

the world [3, 4]. Current worldwide natural gas reserves are over 200 trillion cubic meters

[5]. In the United States, natural gas production has risen by 17 % from 2010 to 2014 [6],

especially due to the increasing extraction of shale gas [7]. Next to the natural gas reserves,

biogas is also a widely available carbon feedstock. It originates from anaerobic digestion of

organic material and consists of approximately equal fractions of CO2 and CH4. In the USA,

266 million tons of municipal solid waste are landfilled each year, of which 67%

anaerobically decomposes into biogas [8]. The European Biogas Association (EBA) published

the 2015 annual report [9] where it states that there are 17,240 biogas plants with a total

installed capacity of 8,293 MW at the end of 2014 (Figure 1.1). CH4 is thus becoming an

increasingly interesting feedstock for chemicals, having already replaced naphtha for the

synthesis of ammonia and methanol [10].

Synthesis of chemicals can be carried out via conversion of methane, either directly or

indirectly. A direct process consists of one-step conversion of methane to e.g. halocarbons,

hydrocyanic acid, acetylene, carbon black, aromatics, methanol or carbon disulfide.

However, these processes suffer from low net yields of the desired products. This is due to

the high C-H bond dissociation energy (416 kJ/mol) of the first C-H bond, making it virtually

impossible to limit bond breaking to a single C-H bond [11]. Therefore, all commercial

processes use methane indirectly as a feedstock for higher value products, by reforming it

into syngas (CO and H2) as an intermediate resource [7].

2

Figure 1.1: Number of biogas plants and total installed capacity in Europe for the period 2010-2014. Figure based on [9].

The most widely studied technologies for methane reforming to syngas are: steam

reforming (SRM) [12-14], dry reforming (DRM) [15-17], partial oxidation (POM) [18, 19] and

autothermal reforming (ARM) [20]. Table 1.1 summarizes all the reforming reactions where

the oxidant used, the final H2:CO ratio, the kinetics and energetics differ in these processes

[2, 21].

Table 1.1: Stoichiometry, reactants H/C ratio, products H2:CO ratio, reaction enthalpy, operating temperature and operating pressure for methane reforming technologies.

React. Stoichiometry Reactants

H/C ratio

Products

H2:CO

ratio

ΔH0

298K

(kJ·mol-1

)

Operating

temp. (K)

Operating

press.

(MPa)

SMR CH4 + H2O 3H2 + CO 6 3 +205 1023–1173 1.5–4.0

DRM CH4 + CO2 2H2 + 2CO 2 1 +247 1123–1273 2.0–4.0

ATR CH4 + 1

2 H2O +

1

4 O2

5

2 H2 + CO 5 2.5 +85 1473-1773 2.0–15.0

POM CH4 + 1

2 O2 2H2 + CO 4 2 -38 - -

OCR CH4 + CO2 + O2 6H2 + 4CO 3 1.5 +58 - -

BIM CH4 + CO2 + 2H2O 8H2 + 4CO 4 2 +220 - -

SMR uses steam to oxidize CH4 and is nowadays the most conventional process for

syngas production resulting in a H2:CO ratios of 3 or higher, which is typically encountered at

2010 2012 2013 20142011

3

hydroprocessing or ammonia plants, where hydrogen is required as a feedstock [22]. The

reaction is carried out in catalyst-packed tubular reactors which are heated by burners. To

increase the hydrogen content a cold-box is following, i.e. a reactor with a Fe or Cu catalyst

promoting a low temperature water-gas-shift step (WGS, see Table 1.2) in the forward

direction. Due to the high endothermicity (+205 kJ·mol-1) of SMR, the energy requirements

for the process are demanding. Typical SMR operating conditions are shown in Table 1.1.

Usually non-stoichiometric steam-to-carbon ratios (> 3) are used to prevent carbon

formation on the catalyst [23, 24]. To overcome the high energy requirements encountered

during steam reforming, a partial combustion zone can be added before the SMR reactor.

This is called autothermal reforming (ATR). Reaction temperatures are higher compared to

SMR (Table 1.1), but ATR offers flexibility in the H2:CO ratio, which can range from 1.8 to 2.5.

This can be controlled by varying the steam-to-carbon ratio and recycling CO2 [23]. Partial

oxidation can also be used to reform CH4 (POM). In a POM reactor, preheated natural gas

and oxygen are fed at a high pressure, and combusted at high temperatures either with or

without a catalyst. The H2:CO yield is theoretically 2 [23].

However, SMR and POM have some distinct disadvantages. SMR yields a H2:CO ratio of 3,

which is not useful for the production of oxo-alcohols or synthesis of FT alkanes. Due to the

low-temperature forward WGS, SMR results in syngas with high amounts of CO2, which is

not desired for hydroprocessing or ammonia production. Because of the high steam-to-

carbon ratios, SMR requires high amounts of water, which may not always be available in

arid areas. Also, the POM process has its drawbacks. It requires purification of air to obtain

pure oxygen feeds [25], while the reactor that is used presents a high capital cost since it

necessitates a refractory lining resistant against high temperatures [23]. Also, the CH4-O2

mixture may yield flammability as well as explosivity hazards and thus meets safety issues,

especially at high space velocities [26].

Ross and co-workers concluded that among the reforming processes, DRM has a 20%

lower operating cost [10], assuming CO2 can be obtained at a negligible cost [27], while

producing high purity syngas, with low CO2 content and a H2:CO ≤1 [28]. Furthermore DRM

has the advantage of converting CO2, a greenhouse gas, into valuable chemicals.

4

CO2 Origin

Carbon dioxide (CO2) is the most important greenhouse gas (Figure 1.2a) originating mainly from anthropogenic activities such as transportation, electricity and heat production, agriculture and other land use (Figure 1.2b) [29]. Approximately 2040 ± 310 Gt of CO2 were emitted to the atmosphere between 1750 and 2011, while in 2011 the average CO2 emissions from fossil fuels combustion, cement production and flaring reached 34.8 GtCO2/y. The impact of CO2, as a greenhouse gas, can be minimized by using it exactly where it is produced on a large scale, like ammonia synthesis or power plants [30].

Figure 1.2: Greenhouse gas emissions by: (a) gas, (b) economic sector. Reproduced from [29, 30].

Although DRM is endothermic, it will reduce the net greenhouse gas emissions if heating

is provided, partially or completely, by solar or nuclear energy [31]. In fact, its high

endothermicity could be advantageous for the storage of solar energy [32]. For natural gas

wells with inherent amounts of CO2, DRM could provide immediate CH4 valorization without

an additional CO2 separation step. The process is chemically highly related to SRM and

involves similar catalysts [33], so previous research on SRM can be useful. DRM results in a

H2:CO ratio of typically 1, which is especially ideal for Fe-catalyzed FT [34] (the FT market

share takes up 8% of the total syngas market and is responsible for approx. 480 PJth, Figure

16%11%

65%6%

2%

(a)

21%

14%

6%24%

25%

(b)

5

1.5 [34]). For oxo-alcohol synthesis or hydrogen production, the H2:CO ratio should be

increased in a subsequent cold-box step.

During reforming, several side reactions may occur. The most well-known side reaction

is WGS in the reverse direction (Table 1.2), resulting in a decrease in the H2:CO ratio in case

of DRM [35]. It will thus be essential to choose a catalyst which minimally promotes the

reverse WGS. Caprariis et al. [36] found that using Rh as an active metal on BaZr(1-x)MxO3

inhibited the reverse WGS, compared to Pt and Ru. Next to the reverse WGS, CH4

decomposition (MD) and the Boudouard reaction (BR) may occur. These are reactions

responsible for the carbon deposition on the active metal sites and thus leading to catalyst

deactivation. This is one of the main hindrances of the industrialization of DMR process. By

synthesizing specific catalysts this challenge can somehow be overcome. The tendency

towards carbon deposition can be explained due to the lower H/C ratio (=2) in case of DRM,

while SRM (=6) and POM (=4) show higher H/C ratios [37]. Namely, for low H/C ratios the

probability to form C-C is higher rather than forming hydrocarbons.

Table 1.2. Methane reforming side reactions: Water gas shift (WGS), methane decomposition (MD) and Boudouard reaction (BR).

Reaction Stoichiometry ∆𝑯𝟐𝟗𝟖𝑲𝒐 (𝒌𝑱 ∙ 𝒎𝒐𝒍−𝟏)

WGS CO+ H2O CO2 + H2 -41

MD CH4 C(s) + 2H2 +75

BR 2CO CO2 + C (s) -172

Wang and co-workers [38] have reported the Gibbs free energies (ΔG°) for DRM, WGS,

MD and BR, which were reproduced later by Pakhare and Spivey in a very comprehensive

review [2]:

ΔG°DRM = 61770 – 67.32·T

ΔG°WGS = -8545 + 7.84·T

ΔG°MD = 21960 – 26.45·T

ΔG°BR = -39810 + 40.87·T

At equilibrium, ΔG is equal to zero. From this, lower and upper reaction temperatures

for DRM, WGS, MD and BR can be determined (Figure 1.3): the DRM reaction occurs in the

6

forward direction from 918 K, while the WGS in the reverse direction can occur up to 1093 K.

Carbon formation due to MD can take place starting from 830 K, while the BR would only

occur up to 973 K [38].

Figure 1.3: Thermodynamics of DRM. Limiting temperatures for reactions of the CH4-CO2 system. Figure based on [38].

The choice of the CH4:CO2 ratio will also greatly influence the carbon formation.

According to Nikoo and Amin [39], for a stoichiometric 1:1 CH4:CO2 feed ratio, the

temperature is required to be higher than 1373 K to avoid carbon deposition. However, this

temperature would be lowered to 973 K, if the feed ratio were equal to 1:2. Lower CH4:CO2

ratios are thus favorable to suppress carbon formation. During the dry reforming reaction, as

shown in Table 1.1, two moles react forming four moles of products, thus the dry reforming

reaction will be thermodynamically favorable at lower pressures. This has also been

confirmed by Nikoo and Amin [39].

However the DRM process has its Achilles’s heel due to the side reaction of the reverse

WGS, which consumes the valuable H2 in order to produce H2O, resulting in syngas with low

H2:CO ratio (Figure 1.4). This problem is not recognized from many researchers since the

majority of the studies used atmospheric pressure, while the reverse water gas shift reaction

starts to dominate at elevated CO2 partial pressures pressures [40]. Therefore DRM can be

combined with co-feeding H2O [41]. The “so-called” bi-reforming of methane (BIM) is

suggested as an improvement to DRM [7, 42-44]. In this perspective, BIM is a process that

-40

-30

-20

-10

0

10

20

30

40

50

0 200 400 600 800 1000 1200 1400

ΔG

0(1

03·k

J·m

ol-1

)

Temperature (K)

7

offers significant advantages compared to SRM, DRM and POM. To begin with, it uses H2O

and CO2 as oxidizing gases yielding in no flammable or hazard mixtures. Secondly, it utilizes

the two most important greenhouse gases, CH4 and CO2, while the combination of them

with H2O offers flexibility to the obtained H2:CO ratio, as it can be altered by changing the

H2O:(H2O+CO2) feed ratio [45, 46]. Figure 1.4 shows the theoretical H2:CO ratio during BIM

as a function of pressure for different feed molar ratios of CH4:CO2:H2O.

Figure 1.4: Thermodynamic calculated H2:CO ratio for the produced syngas as a function of pressure. Light

blue CH4:CO2:H2O molar ratio of 2:0.5:1.5; red line: CH4:CO2:H2O molar ratio of 2:1:1; green: CH4:CO2:H2O molar ratio of 2:1.5:0.5; purple: CH4:CO2:H2O molar ratio of 2:1.75:0.25; blue: CH4:CO2:H2O molar ratio of

1:1:0.

On the other hand, the combination of DRM with O2 or O2 and H2O at the same time

[47], resulting in the oxy-CO2 reforming (OCR) and tri-reforming [48] respectively, have the

practical limitation, which is related with the danger originating from the co-feeding of CH4

along with O2 that may raise safety issues [7]. Further on, membrane reactors can be used to

overcome the thermodynamic limitations of DRM. Tantayon et al. [49] investigated DRM

with a catalytic PdAgCu alloy membrane reactor packed with a Ni/Ce0.6Zr0.4O2 catalyst.

Compared to conventional reactors, the CH4 and CO2 conversions were increased 3.5 and

1.5-fold, respectively.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 2 4 6 8 10 12 14 16 18 20

H2/

CO

rat

io

Pressure (bar)

CH4:CO2:H2O=1:1:0

CH4:CO2:H2O=2:1:1

CH4:CO2:H2O=2:1.75:0.25

CH4:CO2:H2O=2:1.5:0.5

CH4:CO2:H2O=2:0.5:1.5

8

The product of all the reforming processes that were discussed previously, syngas, is a

versatile building block in chemical industry [50]. It can be produced from almost any carbon

source, ranging from natural gas to biomass (Figure 1.5) and thus it is the key for the

flexibility that a chemical industry requires to produce synthetic fuels [51, 52].

Figure 1.5: Conversion via syngas. Modified from [51]

The total global annual use of fossil fuels that are derived from syngas is approximately

6,000 PJth, corresponding to 2% of the total primary energy consumption. Figure 1.6 shows

the current global syngas market distribution. It can be seen that most of the produced

syngas (~53%) is used for the synthesis of ammonia for fertilizers and for hydrogen

production (~23%) which is exploited in refining processes [34]. Finally another important

market for syngas is the “gas to liquid” processes in order to produce fuels for transportation

i.e. the Fischer-Tropsch processes of Sasol in South Africa and of Shell in Bintulu, Malaysia

[50].

Figure 1.6. Global syngas market. Modified from [34].

A few industrial applications will be discussed. The US steel producer Midrex sells

commercial CO2 reformers which produce a CO-rich syngas, typically used for the reduction

of iron oxides [53]. The CALCOR process from Caloric uses a specially developed staged

catalyst to produce high-purity carbon monoxide required for phosgene. Remarkably, the

Natural gasNaphthaCoalBiomass

O2, H2O, CO2

Syngas Products + heat

53%

23%

11%8%4%

1%

9

syngas effluent of the CALCOR process contains below 500ppm of CH4. Further on, the

Sulfur-PAssivated ReforminG (SPARG) process of Haldor Topsøe is identical to a steam

reformer, but with small amounts of hydrogen sulfide added to the feed stream which

inhibits carbon nucleation on the catalyst [54]. The Japan Oil, Gas and Metals National

Corporation have demonstrated a combined steam and dry reforming plant [55]. Recently,

Carbon Sciences also started licensing dry reforming catalysts from the University of

Saskatchewan in Canada, having reported a catalyst with 92% CH4 conversion, no detectable

sintering, and no carbon deposition for at least 2000h time-on-stream (TOS) [56].

The following paragraphs focus on investigations about catalysts for reforming reactions

based on different active metals and promoters (§1.1), carried by particular supports (§1.2),

bringing about several kinetics (§1.3). These parameters as well as the relevant preparation

methods (§1.4) are adjusted in such a way to obtain maximal catalyst activity for reforming

reactions (high syngas yields at low temperatures) and stability (activity during long TOS).

The catalyst stability is negatively influenced by deactivation (§1.5), which often comes

down to carbon deposition.

1.1 Catalysts

1.1.1 Monometallic

Nickel and noble metals have been the most investigated catalysts for methane

reforming processes. Noble metal catalysts like Pt, Rh, Pd and Ru show high activity and

resistance towards carbon formation, but then again they are expensive [57-61]. Active

debate continues about their relative intrinsic reactivities. For example, Rostrup-Nielsen and

Hansen [62] reported turnover frequencies in the following order: Ru, Rh > Ir > Ni, Pt, Pd

when supported on alumina-stabilized magnesia. Bradford and Vannice found the following

order: Ru > Rh > Ni, Ir > Pt > Pd [63]. However, as the same authors [63] have pointed out,

different metal surface geometries as well as metal-support interactions and/or the

participation of O or OH from the support should be taken into account when evaluating the

intrinsic reactivity of an active metal. It is generally accepted that noble metals show better

coking resistance compared to non-noble metals. Hou et al. found that 5wt% noble metals

supported on α-Al2O3 shows little (i.e. for Pt) to no carbon deposition (i.e. for Ru, Rh, Pd, Ir)

during DRM, while 10wt% Ni or Co on the same support showed carbon depositions of 24.0

10

and 49.4 mgcarbon ·(gcat·h)-1 [64], respectively. Nickel-based catalysts have been investigated

as an alternative to noble metals but they are prone to deactivation due to carbon

deposition, especially at steam/C < 3 or CO2/CH4 close to unity [65-67]. However, the alloy

formation between Ni and other metals does lead to improved catalytic activity [68].

Therefore, many investigators add promoters to low-cost metals (§ 1.1.2), like Ni or Co, in

order to obtain an economical catalyst which is carbon resistant. Promoters can add directly

or indirectly to the catalyst activity. They can interact with the existing active phase and/or

support, thereby altering electronic or crystal structures, which can change the mechanism

of the reforming reaction [69].

1.1.2 Bimetallic

The addition of noble metals (Pd, Rh, Pt, Ru) to Ni-based catalysts has been investigated

and alloy formation has been reported after the reduction treatment. The bimetallic Ni-

based catalyst showed enhanced activity and stability compared to the monometallic Ni [70,

71]. The improved performance originates, according to many researchers [72-75], from

higher dispersion, resistance against sintering and higher intrinsic reactivity compared to the

monometallic catalyst. Damyanova and co-workers [76] investigated DRM over a series of

PdNi catalysts. The higher activity observed for bimetallic Pd-Ni in comparison with

monometallic catalysts was attributed to a synergetic effect between Pd and Ni resulting in

higher metallic surface area and higher reducibility. Ferrandon and co-workers[77] studied

steam and autothermal reforming of n-butane over Ni-Rh catalysts. They concluded that a

low Rh loading in the Ni catalysts performed better than monometallic catalysts due to Ni-Rh

alloy formation. Similarly, Steinhauer and co-workers [78] investigated DRM over bimetallic

Ni-Pd catalysts on different supports. They also concluded that the activity of the bimetallic

catalysts was higher than that of monometallics. Rh as a promoter for Ni/α-Al2O3 would

improve the dispersion and retard the sintering of Ni, as well as lower the activation barrier

for CO2 and CH4 [73]. Pawelec et al. [74] studied Pt addition to Ni/ZSM-5 catalysts, which led

to the formation of small nanosized NiO particles and thus easier to be reduced. This

resulted in a higher metallic dispersion of Ni. A Co/TiO2 catalyst deactivates due to oxidation

of metallic Co by CO2. However, the alloy formation of Co with Pt and Ru increased the

activity and stability of the catalyst, while TPR revealed that the noble metals improved the

reducibility of cobalt oxide [75]. Table 1.3 summarizes the use of Ni-based bimetallic

11

catalysts for reforming reactions. To increase the reliability of the catalytic performance

evaluation during stability tests (long TOS), it is important to work in a region that it is

significantly below thermodynamic equilibrium.

Table 1.3: Overview of activities over selected bimetallic Ni-based catalysts during reforming reactions.

Catalyst CH4 Conv. Reforming reaction

/ TOS Conditions Ref

Ni-Pd/MCM-41 20-40% DRM / 10 h 823 K, 101.3 kPa, CH4:CO2 = 1:1 [76]

Ni-Pd/Y2O3 40-70% OCR / 12 h 973 K, 101.3 kPa, CH4:CO2:O2 = 5:4:1 [79]

Ni-Pt/ ZSM-5 15-37% DRM / 2 h 873 K, 101.3 kPa, CH4:CO2 = 1:1 [80]

Ni-(Fe or Co)/Al2O3 5-25% DRM / 3 h 873 K, 101.3 kPa, CH4:CO2 = 1:1 [81]

Ni-Mo/SBA-15 80-98% DRM / 250 h 1073 K, 101.3 kPa, CH4:CO2 = 1:1 [72]

Ni-Fe/ MgzAlyOz - DRM / 10 h 923 K, 101.3 kPa, CH4:CO2 = 1:1 [17]

Ni-Co-CaO-ZrO2 85-90% DRM / 50 h 1073 K, 101.3 kPa, CH4:CO2 = 1:1 [82]

Ni-Rh/ CeZrO2, CeMgOx,

La-Al2O3 - SR and ATR / 150 h 973 K, 101.3 kPa, O2:C= 0.5, H2O:C= 2 [77]

Ni-Co/Al2O3 - SR / 1 h 973 K, 101.3 kPa, H2O:toluene =3.4 [68]

Ni-Fe/ Fe2O3-Al2O3 - SR / 6h 923 K, 101.3 kPa, H2O:C = 1-4 [83]

Ni- (Co or Fe)/ La2O3 - DRM / 10 h 923 K, 101.3 kPa, CH4:CO2 = 1:1 [84]

Nix- (Sn, Ce, Mn or Co)-

Mg1-xO 10-70% DRM / 100 h 823-1033 K, 101.3 kPa, CH4:CO2 = 1:1 [85]

Ni0.03-(Sn, Ge or Ca)-Mg0.97 50-58% DRM / 4 h 1073-1173 K, 1001.3kPa, CH4:CO2 = 1:1 [86]

Ni-Re 30-90% DRM / 5 h 973-1073 K, 101.3 kPa, CH4:CO2 = 1:1 [87]

Ni-Ru/ γ-Al2O3 30-70% DRM / 70 h 873-1073 K, 101.3 kPa, CH4:CO2 = 1:1 [88]

Apart from noble metals, transition metals, like Co, Mo and Fe, can be also used as

promoters for Ni-based catalysts adding directly to the catalyst activity (Table 1.3) [15]. A

synergistic effect of alloy formation between Ni and Co has been reported in steam

reforming of tars [68]. The addition of Mo to Ni/SBA-15 leads to smaller Ni particle size,

increased Ni dispersion and enhanced CO2 activation. In addition, Mo would especially

prevent the formation of shell-like carbon deposition [89]. Iron-containing Ni catalysts are

noteworthy, as they form Ni-Fe alloys [83, 90, 91] which have an outstanding ability to

restrain surface carbon accumulation [15, 83, 92]. It has been reported that the addition of

Fe as promoter to Ni catalysts [15] has a beneficial impact, up to a certain Fe:Ni ratio, by

suppressing carbon deposition during DRM at 1023K.

12

“Iron Master of them all”

Iron is one of the most abundant elements in the earth’s crust composing 5% of it, while the iron oxides have proven valuable materials to mankind over the years, starting from the pre-historic age where iron oxide containing ochre pigments was used to decorate cave walls. Fe3O4 containing rocks were humans’ first experience with magnetism while compass-like instruments, based on Fe3O4, were already exploited for religious purposes in China around 200 BC. The development of Fe3O4-based compasses for navigation occurred in Europe approximately on 850 AD. Through the 20

th century the iron oxides were at the forefront of

discovery in science. For example, Fe3O4 was one of the first spinel structures solved by Bragg in 1915 [93] and Verwey discovered one of the first metal–insulator transitions in Fe3O4 in 1939.

Iron is also involved in several biological processes. Proteins containing iron can be found in all living organisms [94, 95]. In humans, an iron –protein, hemoglobin is responsible for the blood color. Iron oxides, like Fe3O4, aid the navigation of magnetotactic bacteria [96], it thought that they play a similar role in the beaks of homing pigeons, while it has been also discovered in the human brain and many body tissues in an unknown amount.

Looking to the future, iron and its oxides start to become a hot topic for more scientists and their findings are most rewarding. There has been recently a resurgence of research into iron oxide materials [97-103]. Tartaj and co-workers [104] in their article entitled as “The Iron Oxides Strike Back:…”, describe how the exciting properties of iron oxides, coupled to their low toxicity, stability, and economic viability, make them ideal for applications in a wide range of emerging fields. On the other hand, Dharanipragada and co-workers[105] investigate a combined material including Fe along with magnesium and alumina (Mg-Fe-Al-O) for chemical looping process in order to convert CO2 to CO.

As one of the most significant earth oxides, iron oxide can/has been widely employed to development of active and stable catalytic materials that can be exploited for reforming reactions [15], production of chemicals etc [106]. The wide use of iron/iron oxide as active component for catalytic oxidation in exhaust converters [107], in hydrocarbons reforming [15, 83] and water gas shift reaction [108, 109] has been associated with the unique ability of these oxides to be reduced and then re-oxidized by H2O/CO2 [110].

However, despite several years of research Ni-Fe catalysts don’t deliver the catalyst stability offered by noble metals. The ratio between reducing and oxidizing gases [(CO+H2):(CO2+H2O)], Rc, is important for the iron/iron oxides system and thus for the stability of Fe containing alloys [106, 111]. During methane reforming, iron involved in the CO2 or H2O activation is segregated from the Ni-Fe alloy [92], which leads to catalyst deactivation by active site blocking [41]. This Thesis investigates different ways in order to control the stability of Ni-Fe reforming catalysts and the results are presented in Chapters 5 and 6.

Fe oxide species increase the catalyst oxygen mobility, which helps to reduce carbon

formation. Ashok and Kawi [83] studied toluene steam reforming over an Ni-Fe catalyst

supported on Fe2O3-Al2O3. They concluded on the existence of a strong metal-support

interaction which eliminated the sintering and thus resulted in stable catalyst performance.

Koike et al. [112] studied the steam reforming of phenol over Ni-Fe/MgAlOx hydrotalcite

catalysts and found higher activity and stability compared to the monometallic catalysts,

since Fe enabled the adsorption of phenol. Fe might also improve the reducibility of Ni

according to Djaidja et al. [113] who synthesized Ni-Fe/MgO and (Ni-Fe-Mg)2Al catalysts, for

which the reducibility increased due to the presence of Fe. The activity results showed high

performance and a good resistance against carbon formation. Buelens and co-workers[111]

used F2O3 supported on MgAl2O4 as a solid oxygen carrier material for the super-dry

13

reforming process that they developed, where three molecules of CO2 are consumed per

one CH4,resulting in an enhanced CO production. On the other hand, More and co-workers

[114] used Ni-Fe alloy as oxygen carriers in chemical looping dry reforming due to multiple

oxidation states of Fe that enables the tuning of product selectivities when CH4 is used as a

fuel. However, despite several years of research Ni-Fe catalysts don’t deliver the catalyst

stability offered by noble metals. The ratio between reducing and oxidizing gases

[(CO+H2):(CO2+H2O)], Rc, is important for the iron/iron oxides system and thus for the

stability of Fe containing alloys [106, 111]. During methane reforming, iron involved in the

CO2 or H2O activation is segregated from the NiFe alloy [92], which leads to catalyst

deactivation by active site blocking [41]. This Thesis investigates different ways in order to

control the stability of Ni-Fe reforming catalysts and the results are presented in Chapters 5

and 6.

Alkaline (Na, K) and earth alkaline (Mg, Ca, Ba) elements as well as B, Sn, Mn, La or Re

[35] are typical promoters that can add indirectly to the catalyst activity by changing the

geometric and electronic structure of the support and the active metal to enhance catalyst

activity and stability [115]. Alkaline and earth alkaline elements are added to act as Lewis

bases, provoking improved adsorption of CO2 and formation of semi-carbonate species. They

may bring along several other beneficial effects, as increased activity, higher reducibility and

lower amounts of deposited carbon. For example, Alipour et al. [116] found that for Ni/Al2O3

catalyst MgO is better as a promoter compared to CaO and BaO in order to realize all of the

three aforementioned effects. Another possible promoter is boron. When added to Ni, it will

preferentially bind to step and subsurface Ni sites. These are typically the sites where carbon

formation starts, which are thus made unavailable for nucleation of carbonaceous

structures. The boron thus prevents carbon formation. It also causes a surface

reconstruction of the Ni, lowering CH4 activation barriers [117]. Yu et al. [85] have

distinguished Sn to enhance the carbon-resistance of NixMg1-xO catalysts, although Sn lowers

the activity of Ni surfaces. Unlike Ce or Mn, Sn has an atomic radius close to Mg2+ or Ni2+,

and is therefore better dispersed within the NixMg1-xO catalyst. This leads to improved

carbon-resistance. The lowered activity was assigned to the fact that Sn blocks and

passivates active Ni sites. Similarly, Tomishige et al. [86] found that Sn was effective to

decrease carbon formation on Ni0.03Mg0.97O but that it decreased activity of the catalyst. The

14

authors suggest a Ni-Sn alloy prevents complete CH4 dissociation, thus preventing carbon

formation. Jeong et al. [118] investigated the influence of 5wt% Mg, Mn, K and Ca on a Ni/HY

catalyst for DRM and found that the performance was in decreasing order: Ni-Mg/HY > Ni-

Mn/HY > Ni-Ca/HY > Ni/HY > Ni-K/HY. Ni/Mg/HY showed the highest activity and stability

because of the formation of MgOx and MnOx on the surface. Rosen et al. [87] have used Re

as a promoter for electrodeposited Ni foams at low temperature applications. They found

that the performance of Ni-Re alloyed catalysts was better than the monometallic catalysts

during 50 h TOS. Numerous more cases can be discussed. All the above indicate the

promoting effect of combining Ni with another metal, compared to the performance of the

monometallic catalyst. However, this effect is a function of the promoter that is used, as well

as its loading. One of the prominent challenges in heterogeneous catalysis is the

replacement of noble metals, however, all the viable substitutes are suffering from

deactivation. Therefore, in this work, one of the promoters which will be given more

attention is Fe, because of its good redox properties, high availability and low toxicity,

aiming to obtain activity and stability comparable with noble metals.

1.2 Catalyst support

Metals as such can be used as catalysts. However, typically supported metal catalysts are

preferred over unsupported in order to form small metal particles, resulting in higher

specific amount of catalyst active sites [119]. In this respect, mesoporous materials as ZSM-

5, MCM-41, MCM-22, and SBA-15 have become popular as supports. In addition to their high

surface area, the uniform pore size results in more accessible and exposed active sites.

However, active metals on these supports would therefore sinter quickly. Sintering could

possibly be attenuated by treatments to better confine active metals [120].

Different interactions of metals with the support have been reported. A strong metal-

support interactions (SMSI) may be favorable for high metal dispersion and sintering

resistance. The electronic interaction of the support will also result in different binding

reactivity and stoichiometry of CHx or CO2 adsorbates. However, SMSI may lead to low

reducibility of the active metals as well as its encapsulation by or irreversible reaction with

the support. For weak metal-support interactions (WMSI), the previous effects are less

prominent. In that case, active metal species can be reduced more easily but will result in

15

lower dispersion and lower resistance against sintering [121]. Some authors also make an

additional (although arbitrary) distinction of medium metal-support interactions (MMSI). It is

generally accepted that metals supported on SiO2, Al2O3 and MgO show WMSI; small metal

particles in zeolites show MMSI; and metals on certain reducible oxides (especially TiO2)

shows SMSI [122].

However, there is more to a support than just the metal-support interactions. Inert

supports, e.g. SiO2, although ensuring active metal dispersion, will result in quick

deactivation since CH4 and CO2/H2O are activated on the same active site while insufficient

surface oxygen is present to suppress carbon formation [123]. To improve the latter

mechanism, the synthesis of reforming supported catalysts should be adjusted in such a way

that the support inhibits carbon formation by actively participating in the reaction scheme.

This should include (1) basicity which will enhance the adsorption of CO2 [124, 125], (2)

supports which promote H spillover [126], and (3) supports with redox properties [127-131].

Supports that are based on rare earth elements (La2O3 [124]) and alkaline earth elements

(MgO, CaO [125]) enhance H2O/CO2 activation and thus suppress carbon deposition. On the

other hand, acidic supports, like γ-Al2O3, do not enhance adsorption of H2O/CO2. Metal

oxides as CeO2 [127], CeO2-ZrO2 [132], ZrO2 [129], YSZ [130] and Rh substituted La2Zr2O7

pyrochlores [131] show good redox properties and high oxygen mobility due to the presence

of oxygen vacancies. This effect enhances carbon oxidation and thus improves the carbon

resistance of a catalyst. This stands in contrast with the irreducible oxides Al2O3, SiO2, La2O3,

MgO and Y2O3, who do not benefit from high oxygen mobility. For example, upon reduction,

Ce species, in the CeO2 lattice, will be reduced from Ce4+ to the Ce3+ form, upon which

oxygen vacancies are formed [127]. The combination of CeO2 and ZrO2 results in a higher

thermal stability than the pure materials, and also increases the amount of oxygen vacancies

(ceria-stabilized zirconia or CSZ) [128]. For the same reason, Y2O3 is added to a ZrO2 support

(yttria-stabilized zirconia or YSZ) [130]. Pyrochlores crystals also show very high oxygen

mobility. Their crystal structure (A2B2O7) have an A-site, which is usually a large cation, and a

B-site cation with a smaller radius which allow isomorphic substitution. For instance, the Zr

can be partially replaced by Rh to generate a CH4 reforming active catalyst. The Rh dispersed

in the pyrochlore structure proved to be highly sintering-resistant. Replacing the La by Ca

may improve the oxygen mobility through the creation of lattice oxygen defects [131].

16

Supports are required to achieve high dispersions, inhibit irreversible interaction with

active metals and ensure sintering resistance. Guo et al. [133] studied DRM for a

stoichiometric (1:1) CH4:CO2 ratio over Ni/γ-Al2O3, Ni/MgO-γ-Al2O3 and Ni/MgAl2O4 catalysts.

They found that Ni/MgAl2O4 resulted in higher activity and stability due to low acidity of the

support, high dispersion of Ni particles and resistance against sintering. Penkova et al. [134]

have come to similar conclusions. They found that the incorporation of MgO into γ-Al2O3

resulted in the formation of MgAl2O4 spinel, which decreases the acidity of the support,

suppresses the interaction of Ni with Al2O3 and thus the formation of the inactive NiAl2O4

spinel. The MgAl2O4 is also characterized by a high melting temperature, good mechanical

properties and good chemical resistance [135]. Dharanipragada et al. [105] previously found

that Mg-Al-Fe-O spinels showed high stability for oxygen transfer. Also, MgAlFeOx catalysts

showed high activities for ethylbenzene dehydrogenation [136]. All the aforementioned

examples indicate the importance of the support, not only in terms of interaction with active

metals but also from mechanistic point of view. A support with redox properties, sometimes

called “active or redox support” [137-139], can participate in the reaction scheme by

activating the oxidizing gas, and thus increasing the carbon-resistance of the catalyst, as it

oxidizes the deposited carbon species through lattice oxygen. This study will focus on

MgAl2O4 and Fe-modified MgAl2O4 as support materials (MgFexAl2-xO4).

1.3 Kinetics

The type, sequence and reversibility of elementary steps in the reforming mechanism

strongly depends on the catalyst metal, its dispersion, the support and reaction conditions. It

is therefore difficult to find generally accepted kinetic mechanisms [140-143]. Nevertheless,

here, the mechanism of DRM and SRM over a selection of catalysts will be illustrated.

Understanding of CH4 activation step is essential in order to unravel the mechanism of

DRM and SRM. Two predominant mechanisms are claimed to exist: (a) the indirect and (b)

the direct dissociation. Indirect dissociation, which seems to be the most accepted

mechanism, includes the formation of CHx and formyl species [140, 144, 145], while via

direct dissociation, carbon is formed immediately [146]. This may depend on reaction

conditions. For instance, Seets et al. [146] have reported that for an Ir (110) catalyst, CH4

17

activation gradually shifts from an indirect mechanism to a direct mechanism for increasing

temperatures.

For CO2 activation, according to Spivey and Pakhare [147], there would be general

agreement that this happens via the formation of carbonate intermediate species on basic

catalytic sites, producing CO. This typically happens on the support or on the metal-support

interface rather than on the metal alone (except for inert supports as SiO2). For example,

Bitter et al. [148] found that for Pt/ZrO2 catalysts, the CO2 is activated on the support

reacting at the metal-support interface.

For H2O activation, Xu and Froment [14] reported that H2O reacts with surface Ni

resulting in hydrogen and adsorbed oxygen over Ni/MgAl2O4. Wei and Iglesia [145] reported

the same activation step for H2O on the metal surface during SRM, forming adsorbed

hydrogen and OH- group. Using isotopic labeling, they found that this step is reversible and

quasi-equilibrated. In a different work, Wei and Iglesia [149] demonstrated a rigorous kinetic

and mechanistic equivalence for DRM and SRM. On the other hand, the availability of

surface oxygen at Ni/MgO-MgAl2O4 catalyst could play a rate determining role at higher

temperatures during SRM, according to Aparicio [142].

The next question that need to be addressed, is which of the elementary steps in the

reforming reactions would be the rate-determining, if a rate-determining step (RDS) would

exist at all. Many authors state that CH4 activation is the rate-determining step for the

reforming catalysts [17, 140, 150-152]. For instance, based on isotopic labeling, Wei and

Iglesia have proposed (for Rh/Al2O3 and Rh/ZrO2) that CH4 decomposes to chemisorbed C in

a series of H-abstraction steps, which become increasingly fast as H atoms are withdrawn

from CH4. The first C-H bond dissociation would then be kinetically the most significant step

[140]. They also extended their study and they proposed a common sequence of elementary

steps for CH4 reforming reactions on Ru, Rh, Ni and Pt [140, 145, 149]. Figure 1.7 illustrated

the elementary steps of DRM and SRM over a Pt cluster. Step 1 (k1) influences the reaction

rates, while 2-5 include the required steps for the WGS reaction, which must be quasi-

equilibrated during CH4 reforming reactions. Furthermore, Foppa and co-workers [144] used

density functional theory (DFT) in order to perform a mechanistic study of DRM reaction

network at 973 K. They found that the CH4 dissociation energy is metal independent, while

18

CO2 becomes more energy demanding according to the trend: Ni< Pd< Pt. This means that C-

O bond cleavage is less energy demanding for Ni but not for Pd and Pt surfaces [144].

Moving down to the metal series (from Ni to Pt) the CO2 activation mechanism changes

from C-O bond cleavage in case of Ni to hydrogen-assisted carbon dioxide cleavage routes

(CO2(g) + H*→ COOH* for Pd and Pt) [144]. At the same side, Efstathiou et al. have found that

over Rh/YSZ (Rh is below Ni in the metal series), CO2 activation is the kinetically limiting step

[141]. On the other hand, Aparicio [142] suggests that on Ni/MgO-MgAl2O4 catalysts there is

no rate-determining step for DRM. Bradford and Vannice [143] have proposed (for Ni/TiO2,

Ni/C, Ni/SiO2 and Ni/MgO) that DRM proceeds via reversible dissociation of CH4 to form CHx

and H species, followed by quasi-equilibrated steps, where CO2 adsorbs, dissociates, and

reacts with CHx and hydroxyl groups to form CHxO species. CHx and CHxO decomposition

would be the rate-determining steps. It can be seen then that numerous mechanisms exist in

literature and strongly depend on the type of the examined catalyst.

Figure 1.7: Sequence of elementary steps for CH4 reactions with H2O or CO2 on Pt surfaces (→

irreversible step; ↔ quasi-equilibrated step). Reproduced from [149].

Kinetic data during reforming reactions are often regressed against power law models

[62, 153, 154]. This may be correct to predict data for the fixed temperatures and partial

pressures, and at those conditions the reaction orders will give an idea about the reactivity

of CH4 and CO2/H2O. Although those models will be statistically significant, they will be less

reliable to extrapolate data to other temperatures or partial pressures [155-157]. Also, they

do not lead to deeper understanding of the mechanism over a catalyst. More complex

models which take adsorption and intermediate steps into account are physically more

19

relevant, but then the significant estimation of kinetic parameters will become difficult [155,

158].

Table 1.4 shows examples of kinetic rate equations over various catalysts for DRM and

SRM. Múnera et al. [159] (equation 1 of Table 1.4) proposed a four-step mechanism where

CH4 is activated on the Rh site and carbonate on the La2O3 support during DRM over

Rh/La2O3-SiO2. The CH4 and carbonate dissociation were assumed to be the rate-determining

steps. Their mechanism is based on the previous findings of Verykios for Ni/La2O3 catalysts

[160]. Quiroga et al.[161] also suggested a four-step mechanism, where CH4 is adsorbed and

dissociated while CO2 is adsorbed on both the active sites or on the support. A subsequent

surface reaction of both adsorbed species constitutes the rate-determining step during DRM

over Ni-Rh/Al2O3 catalysts. This led to equation 2 of Table 1.4. Richardson and co-workers

[153] derived a rate equation based on LHHW mechanism, where CH4 and CO2 are both

adsorbed and they take part in the rate determining step. They observed that the kinetic

experimental data obtained over Rh/Al2O3 were in agreement with Eq. 3 from Table 1.4. Xu

and Froment [14] carried out a kinetic analysis over Ni/MgAl2O4 during SRM (equation 4 of

Table 1.4). They found that H2O reacts with surface Ni, yielding adsorbed oxygen and

gaseous hydrogen. Adsorbed CH4 either reacts with adsorbed oxygen or dissociates to form

chemisorbed radicals CHx. The adsorbed oxygen and the carbon-containing radicals react to

form CO and CO2 either directly or via formyl intermediates.

Table 1.4: Kinetic rate equations for DRM and SRM over a selection of catalysts.

N° Reaction Catalyst Assumptions Rate equation Ref.

1 DRM Rh/La2O3-

SiO2

CH4 diss. and

carbonate diss. RDS 𝑟 =

𝐾1𝑘2𝐾3𝑘4𝑝𝐶𝐻4𝑝𝐶𝑂2

𝐾1𝐾3𝑘4𝑝𝐶𝐻4𝑝𝐶𝑂2 + 𝐾1𝑘2𝑝𝐶𝐻4 + 𝐾3𝑘4𝑝𝐶𝑂2 [159]

2 DRM Ni-Rh/Al2O3 Surface reaction is

RDS 𝑟 =

𝑘1𝐾𝐶𝐻4𝐾𝐶𝑂2(𝑝𝐶𝐻4𝑝𝐶𝑂2

𝑝𝐻20.5 −

𝑝𝐻21.5𝑝𝐶𝑂

2

𝐾𝑟𝑒𝑓)

(1 +𝑝𝐶𝐻4

𝑝𝐻20.5 𝐾𝐶𝐻4 + 𝑝𝐶𝑂2𝐾𝐶𝑂2)²

[161]

3 DRM Rh/Al2O3 LHHW mechanism 𝑟 =𝑘𝑅·𝐾𝐶𝑂2·𝐾𝐶𝐻4·𝑝𝐶𝐻4·𝑝𝐶𝑂2

(1+𝐾𝐶𝑂2𝑝𝐶𝑂2+𝐾𝐶𝐻4𝑝𝐶𝐻4)2 [153]

4 SRM Ni/MgAl2O4 CH4 activation RDS 𝑟 =

𝑘1

𝑝𝐻22.5 (𝑝𝐶𝐻4𝑝𝐻2𝑂 −

𝑝𝐻23 𝑝𝐶𝑂

𝐾1)

(1 + 𝐾𝐶𝑂𝑝𝐶𝑂 + 𝐾𝐻2𝑝𝐻2 + 𝐾𝐶𝐻4𝑝𝐶𝐻4 + 𝐾𝐻2𝑂𝑝𝐻2𝑂/𝑝𝐻2)2 [14]

The redox or Mars-van Krevelen (MkV) mechanism has been also discussed in literature.

According to this mechanism CO2 or/and H2O are activated on different sites than CH4. The

CH4 activation lead to the formation of H2 and surface carbon. All surface carbon is removed

20

by interaction with lattice oxygen, producing CO and oxygen vacancy, which is re-filled from

the oxidizing gas (CO2 and/or H2O) [162]. From the above, it is clear that the reaction

mechanism and the kinetic rate equation strongly depend on the type of catalyst which is

used.

1.4 Catalyst preparation

Two main preparation methods are widely used: precipitation (one-pot synthesis) and

impregnation. In both methods, a metal precursor, mostly an inorganic salt, is the starting

point to incorporate or deposit the active metal into the support. With co-precipitation,

metal particle growth is induced by supersaturation of the precursor solution, resulting in

nucleation and growth of metal particles in conjunction with the formation of the support.

During simple wetness impregnation, the support is contacted with a precursor solution. Low

active metal loadings are achieved by adsorption of the precursor molecules onto surface

groups of the support (ion adsorption) or through the exchange of ions in, for instance,

zeolites (ion exchange). When higher loadings are required, no washing step is carried out

and the support is directly dried, so that all precursor ends up on the support (impregnation

and drying). Impregnation can be performed to incipient wetness, whereby only the pores of

the support are filled with precursor solution to prevent active metal deposition on the

external surface of the catalyst grains [163].

Co-precipitation, sequential impregnation or co-impregnation (impregnation of multiple

active metals in the same solution) will have a profound impact on catalyst activity and

stability. For example, Roh et al. [164] investigated the effect of preparation method on

MgO-promoted Ni-Ce0.8Zr0.2O2 for DRM. Co-precipitated Ni–MgO–Ce0.8Zr0.2O2 exhibited very

high activity as well as stability (XCH4 > 95% at 1073 K for 200 h) compared to the co-

impregnated and the sequentially impregnated samples. That was attributed to higher

surface area, higher dispersion of Ni, smaller Ni crystallite size, and easier reducibility of the

Ni in case of the co-precipitated catalyst. Aksoylu et al. [165] found that a 1wt% Ce-

promoted 1wt% Pt/ZrO2 DRM catalyst prepared by co-impregnation is better in terms of CO

and H2 production compared to the sequentially impregnated catalyst. The high activity was

explained by intensive interaction between the Pt and Ce phases for the co-impregnated

sample, enhancing the dispersion of Pt. Theofanidis et al. found that during reduction of tri-

21

metallic Ni-Fe-Pd/MgAl2O4, prepared by co-impregnation, a core shell alloy forms at the

catalyst surface, consisting of Ni-Fe core and Ni-Fe-Pd shell. This alloy is not formed when

the catalyst is prepared by sequential impregnation [41].

Many support materials are also produced via sol-gel chemistry, in which a metal

precursor (usually a silicate dissolved in water), is polymerized in solution, forming

polysilicate species. Further condensation leads to the formation of small particles, links

between particles, and eventually the formation of a gel. The drying procedure and further

thermal treatment are critical to determine the final porosity of the support [163]. Solvent

extraction will typically lead to more porous catalysts than by solvent evaporation. Thus the

synthesized gels are called aerogels and xerogels respectively [166] as it is illustrated in

Figure 1.8 [163]. For instance, Zhu et al. [167] investigated Ni/Al2O3 aerogels for DRM in a

fluidized-bed reactor using catalysts prepared via a sol-gel method and subsequent

supercritical drying. The aerogels showed a higher specific area, higher Ni dispersion and

stronger metal-support interactions compared to the impregnated catalysts.

Figure 1.8. Preparation of sol-gel catalysts. Reproduced from [166].

Colloidal synthesis provides unique opportunities to control the size, shape, and

composition. The nanoparticles are generally formed by stepwise induction of a change in

conditions in an otherwise homogenous solution of the metal precursor. Ligands present in

the solution play an essential role in stabilizing the nanoparticles against aggregation,

coalescence, and growth. Simple wetness impregnation is commonly applied to deposit the

colloids onto a support due to the relatively low nanoparticle concentrations in colloidal

solutions necessary to prevent aggregation and precipitation [163]. Garcia et al. [168] used a

colloidal (microemulsion) route and incipient wetness impregnation to prepare

gelation solvent

extraction

solvent

evaporation

Sol Gel Aerogel

Xerogel

22

monometallic (Ni and Pt) and bimetallic (PtNi) catalysts. The microemulsion method

promoted the formation of NiO instead of NiAl2O4, and thus facilitated its reduction to

metallic Ni during catalyst pretreatment. For the bimetallic catalyst, the NiPt synthesized by

the microemulsion method resulted in lower amounts of deposited carbon.

Atomic vapor deposition involves gaseous metal atoms, clusters, or organics selectively

reacting with support surface groups [163]. Gould et al. [169] used atomic vapor deposition

to deposit Ni and Pt on alumina in order to obtain monometallic and bimetallic catalysts.

Those catalysts were more active for DRM than those prepared by incipient wetness

impregnation. No carbon whiskers were formed during reaction due to high metal dispersion

[169].

Confining catalyst particles in a silica gel can reduce the number of corners and edges and

increase Ni dispersion [170, 171]. This can result in suppression of carbon deposition since it

is primarily catalyzed by Ni atoms located at corners and edges. In literature, most often

core-shell and yolk-shell catalysts for DRM use to achieve this effect [163]. A core-shell

catalyst consists of an active core with an inactive shell, while a yolk-shell catalyst allows a

void space within the inactive shell (Figure 1.9) [172]. Zhang and Li [170] synthesized a core-

shell Ni@SiO2 catalyst for DRM. Calcination of the as-synthesized core-shell nanoparticles

created a micro-/mesoporous structure in an amorphous silica shell. The shell significantly

suppresses sintering and coking.

The core-shell catalyst is stable for 40h TOS, while the commercial Ni catalyst was only

stable for 6.4 h. Yang et al. [171] found that the catalytic performance of a Ni@SiO2-yolk-

shell catalyst is better than for Ni@SiO2-core-shell catalyst. The carbon deposited on the

yolk-shell catalyst was mainly filamentous, while only a small amount was encapsulating. On

the other hand, the core-shell catalyst showed layered encapsulating carbon.

Figure 1.9. Core-shell catalyst (left) and yolk-shell catalyst (right). Reproduced from [172].

However, sol-gel routes, colloidal routes, atomic vapor deposition and yolk-shell / core-

shell synthesis may be difficult for industrial scale-up compared to active metal impregnation

23

and co-precipitation [173]. In this study, impregnation and precipitation methods will be

used for catalyst synthesis.

After the catalyst synthesis, the deposited metal precursor is converted into the final

crystallite structure during the calcination step [163]. The choice of the calcination

temperature is important since it will influence the final particle distribution and

composition. High calcination temperatures may ensure high dispersion and thus lower

carbon deposition, although too high calcination temperatures may also result in the

formation of inert phases, e.g. spinels [163]. Ashok and Kawi [174] found that a calcination

temperature of 773 K results in an Ni/Fe-Al2O3 catalyst, enriched with Fe species at the

surface, compared to higher calcination temperatures. This catalyst also showed relatively

more stable catalytic performance during steam reforming. Sokolov et al. [175] found that

for increasing the calcination temperature for a series of Ni/La2O3-ZrO2 catalysts, the Ni

dispersion was increased. Chen et al. [176] found the formation of NiAl2O4 spinel phases

after calcination of NiO-Al2O3 samples at 1023 K while calcination at lower temperatures,

573-873 K, resulted in NiO phases. The aforementioned investigators all use calcination

under air flow. Also non-conventional calcination techniques as freeze-drying or NO

calcination can be used. For example, Aw et al. [177] used freeze-drying or NO calcination to

improve the DRM activity of NiCo/CeZr catalysts. Freeze-drying would improve the

nanoparticle distribution, whereas a NO treatment would lead to NO scavenging oxygen

radicals. This would be beneficial to prevent sintering and particle redistribution.

1.5 Catalyst deactivation

Catalyst deactivation is the loss of catalyst activity and/or selectivity as a function of time-

on-stream (TOS) and it is a great challenge for many industrial catalytic processes. A possible

process shut down due to catalyst deactivation has an impact of billion euros to industry

[178]. Therefore the need to understand the paths of deactivation mechanisms in

heterogeneous catalysis remains a hot topic in the research community.

The most well-known deactivation mechanisms for reforming reactions (i.e. DRM, SRM,

BIM) could be grouped as: (A) carbon deposition on the catalyst, (B) poisoning, (C) sintering

or thermal degradation, (D) encapsulation of the active metal by the support, (E)

attrition/crushing and (F) volatilization. Attrition and volatilization are not discussed in this

24

Chapter as their impact is considered negligible compared to the other deactivation

mechanisms i.e. carbon deposition, poisoning, sintering and encapsulation. Figure 1.10

illustrates the first four aforementioned deactivation processes [178, 179] that will be

discussed in more detail in the following subparagraphs. For a gas phase reaction, catalyst

leaching (extraction of catalyst components into a liquid phase) is typically not of importance

and therefore it will not be further discussed. For a more comprehensive description of

these deactivation mechanisms, the reader is referred to Chapter 3 in the work of Ulla Lassi

[179], the comprehensive review of Argyle and Bartholomew [178] and Chapter 5 in the

work of Rostrup-Nielsen and Christiansen [52].

Figure 1.10: Deactivation mechanisms: A) carbon deposition, B) poisoning, C) sintering or thermal

degradation, and D) encapsulation of active metal particles by the support. Reproduced from [179].

1.5.1 Carbon deposition

Carbon formation on active metals is a complex process including surface catalysis and

solid state reactions. It can lead to breakdown of the catalyst, while the accumulation of the

carbon species can cause blockage of the reforming reactor [52]. It is the most frequently

encountered deactivation mechanism during methane reforming originating from

Boudouard and methane decomposition reactions (Table 1.2) [26, 92, 180]. Typically, noble

metals are less susceptible to carbon deposition compared to Ni-based catalysts [181].

25

Even if the carbon formation and growth is an undesired phenomenon for the reforming

reactions, the synthesis of carbon nanotubes (CNT), one type of carbon material with

graphite layers and tubular structure, plays very important role in the field of

nanotechnology [182-184]. They were firstly identified by Lijma [185], require a source of

elemental carbon and energy in order to be formed. The CNTs have numerous properties

like high surface area, electronic and thermal conductivity, tensile strength, resistance to

acidic/basic chemicals, making them ideal to be used in a variety of applications such as

catalyst supports, air and water filtration, conductive adhesive, fibers and fabrics etc [186].

The mechanism of carbon formation on metal catalysts remains controversial within the

research community. Lobo and Trimm in 1971 [187-189] suggested a mechanism for steady-

state carbon deposition on Ni catalysts. They suggested that the carbon atoms migrate

through the Ni surface, upon which the hydrocarbon decomposition takes place, towards

the active growth regions (carbides) where the process of carbon growth (nucleation) takes

place (Figure 1.11) [188]. The carbide phase decomposes at some stage to give graphite.

However, Helveg et al. [190] in a more recent publication in 2004 ,made time-resolved high-

resolution in-situ transmission electron microscope observations of the formation of carbon

nanofibers from methane decomposition over supported nickel nanocrystals. Carbon

nanofibers were observed to develop through a reaction-induced reshaping of the nickel

nanocrystals. Specifically, the nucleation and growth of graphene layers were found to be

assisted by a dynamic formation and restructuring of mono-atomic step edges at the nickel

surface. Density-functional theory calculations indicated that the observations are consistent

with a growth mechanism involving surface diffusion of carbon and nickel atoms. The

process thus involves transport of carbon atoms towards Ni atoms away from the graphene-

Ni interface and the flux of Ni atoms is directed towards the free Ni surface.

26

Figure 1.11: Original model proposed by Lobo and Trimm for carbon formation on Ni. A: Several modes of

carbon formation observed on nickel foils; B: details of carbon formation on a detached crystallite, showing

the reaction steps of the proposed mechanism [191].

Hence, Halveg et al. [190] have established that it is possible to transport carbon along

the graphene-Ni interface and their experiments clearly point to the necessity for transport

of Ni, which cannot take place through the bulk of the metal. This stands in contrast with the

carbon bulk diffusion proposition of Lobo and Trimm [191], which consists of carbon

dissolving through the bulk of the nickel metal .

Molecular mass and chemical structure of these carbon species may vary depending on

the reaction type, conditions and catalyst [180, 192, 193]. Similarly, the carbon species

deposited during reforming reactions vary in morphology and may be carbidic, amorphous,

graphene-like, graphitic, filamentous [2, 132, 194-196]. Amorphous carbon can typically be

oxidized at lower temperatures, while polymeric carbon is slightly more difficult to oxidize.

The longer the polymer chain, the more difficult to oxidize. The most difficult type of carbon

to oxidize is the graphene type [147].

Several ways have been examined in order to inhibit or control the deactivation

originating from carbon species deposition. Higher dispersion of the active metals on the

support surface [197], increase of catalyst basicity to achieve a higher activation rate of

mildly acidic CO2 [198, 199] and addition of materials that offer an oxygen reservoir through

their redox behavior [110, 162, 200, 201] have been investigated (see § 1.2). Sadykov et al.

[202] used CeO2 as an oxygen reservoir material for DRM and showed that the oxygen

27

mobility of CeO2 can be increased by incorporation of rare earth metals (La, Gd, Pr) as

dopants. They found that catalytic activity is correlated with oxygen near the surface and/or

bulk mobility for elimination of deposited carbon species. Theofanidis et al. [15] on the

other hand used Fe2O3 as promoter for a Ni/MgAl2O4 catalyst because of its good redox

properties [203]. They found lower amounts of deposited carbon after methane dry

reforming on Fe-modified samples in comparison with pure Ni/MgAl2O4. This was attributed

to FeOx formation during DRM reaction and subsequent oxidation of carbon by lattice

oxygen [83, 91, 204]. Zhang and Verykios [199] showed that Ni/La2O3 activity increases after

2-5 h time-on-stream (TOS). This study revealed a CO2 pool, stored in the form of La2O2CO3.

Table 1.5 reports the carbon deposits of some selected catalyst during DRM. A few trends

can be remarked regarding carbon resistance. For 5 wt% Ni, 18 wt% CeO2-SiO2 is better

support than SiO2, CeO2 and 30wt% CeO2-SiO2 [205]. On TiO2, it is better to impregnate Co

alone instead of Ni or Ni-Co [206], Mn found to be the best candidate to promote Ni/MCM-

41 compared to Zr and Ti. On SiO2, it is very clear that Rh is more carbon resistant than Ni.

However, it is difficult to compare the amount of deposited carbon amongst the different

studies, since they were carried out different space times [205-213].

28

Table 1.5: Overview of activities and carbon deposits of some selected catalysts during DRM.

Catalyst Conversion

CH4 [%]

Carbon

deposits [wt%] TOS Conditions Ref.

5 wt% Ni/SiO2 97 40.8 10h 1073 K, 101.3 kPa, CH4:CO2 1:1 [205]

5 wt% Ni/CeO2 91 62.7 10h 1073 K, 101.3 kPa, CH4:CO2 1:1 [205]

5 wt% Ni/18 wt% CeO2-SiO2 97 49.0 10h 1073 K, 101.3 kPa, CH4:CO2 1:1 [205]

5 wt% Ni/30 wt% CeO2-SiO2 93 62.2 10h 1073 K, 101.3 kPa, CH4:CO2 1:1 [205]

10 wt% Ni/TiO2 52 0.9 24 h 1023 K, 101.3 kPa, CH4:CO2 1:1 [206]

1 wt% Co – 9 wt% Ni/TiO2 48 1.1 24 h 1023 K, 101.3 kPa, CH4:CO2 1:1 [206]

5 wt% Co – 5 wt% Ni/TiO2 40 0.1 24 h 1023 K, 101.3 kPa, CH4:CO2 1:1 [206]

9 wt% Co – 1 wt% Ni/TiO2 35 < 0.01 24 h 1023 K, 101.3 kPa, CH4:CO2 1:1 [206]

10 wt% Co/TiO2 5 < 0.01 24 h 1023 K, 101.3 kPa, CH4:CO2 1:1 [206]

2 wt% Ni/MCM-41 88 45.0 30h 1023 K, 101.3 kPa, CH4:CO2 1:1 [207]

2 wt% Ni – 1 wt% Zr/MCM-41 90 34.0 30h 1023 K, 101.3 kPa, CH4:CO2 1:1 [207]

2 wt% Ni – 1 wt% Ti/MCM-41 15 13.0 30h 1023 K, 101.3 kPa, CH4:CO2 1:1 [207]

2 wt% Ni – 1 wt% Mn/MCM-41 5 8.0 30h 1023 K, 101.3 kPa, CH4:CO2 1:1 [207]

5 wt% Ni/SiO2 72 18.8 24 h 973 K, 101.3 kPa, CH4:CO2 1:1 [208]

3.75 wt% Ni – 1.25 wt% Rh/SiO2 81 36.3 24 h 973 K, 101.3 kPa, CH4:CO2 1:1 [208]

2.5 wt% Ni – 2.5 wt% Rh/SiO2 82 36.3 24 h 973 K, 101.3 kPa, CH4:CO2 1:1 [208]

5 wt% Rh/SiO2 83 3.2 24 h 973 K, 101.3 kPa, CH4:CO2 1:1 [208]

5 wt% Ni/H-S1 62 6.0 10h 973 K, 101.3 kPa, CH4:CO2 1:1 [209]

8 wt% Ni/H-S2 63 4.7 10h 973 K, 101.3 kPa, CH4:CO2 1:1 [209]

5 wt% Ni/ITQ-6 77 2.1 10h 973 K, 101.3 kPa, CH4:CO2 1:1 [210]

5 wt% Ni/MCM-41 75 4.4 10h 973 K, 101.3 kPa, CH4:CO2 1:1 [210]

0.3 wt% Ni/MCM-41 16 71.5 12 h 873 K, 101.3 kPa, CH4:CO2 1:1 [211]

0.3 wt% Pd – 0.3 wt% Ni/MCM-41 36 63.5 14 h 873 K, 101.3 kPa, CH4:CO2 1:1 [212]

The balance between carbon forming and carbon removal reactions will determine the

eventual deposition on the catalyst. Sometimes additional carbon removal after reaction

time will be necessary, in other words, the catalyst will need to be regenerated in order to

restore its initial activity (see §1.6). Regeneration aims to do this by removing carbonaceous

species (or poisons, see §1.5.2), without damaging or altering the structure of the catalyst

itself [214].

1.5.2 Poisoning

Poisoning implies the strong chemisorption of components (see Table 1.7) on active sites,

resulting in the loss of catalytic activity. These components can be reactants, products or

impurities [215, 216]. Whether a compound will act as a poison depends on its adsorption

strength compared to the other compounds competing for catalytic sites. For instance,

oxygen might be reactant in partial oxidation of ethylene over a silver catalyst and a poison

in hydrogenation of ethylene over nickel catalyst [178, 217]. Catalyst poisons can be

29

classified as selective and non-selective, related to the fact that some reactions are

selectively inhibited and some are not [54].

One of the strongest poisons that are present during reforming reactions is H2S, which

chemisorbs on metals according to Eq. 1.1. Sulfur impurity concentrations are typically 5-

500ppm for natural gas [218, 219] and 200-20,000 ppm for biogas [220]. Subsequent

desulfurization processes can reduce sulfur concentrations to a range of less than 10 ppm

[218].

H2S + Me Me − S + H2 (1.1)

It is known that a metal surface, both noble and non-noble, can be easily poisoned by

sulfur [178] due to: (1) strong adsorption of a sulfur atom that can result in blocking three or

four topside sites on a metal surface via the formation of S-metal bonds [178, 180, 221], (2)

electronic modification of the nearest neighbor metal atoms implying that their ability to

adsorb or activate reactant molecules will be modified, (3) restructuring of the catalyst

surface.

Surprisingly, only few investigations on the effect of contaminants during reforming

reactions are to be found in literature. Jablonski et al. [222] studied the effect of H2S, SO2

and COS contaminants over Ni/YSZ and Ni/K2O-CaAl2O4 during DRM and SRM. Relative

deactivation rates for both catalysts for the S-species were COS > SO2 ≥ H2S. Reaction

temperatures of 923 K, 1023 K, and 1073 K were evaluated, but temperature did not

strongly affect the deactivation rates for either catalyst. It was concluded that the YSZ

support requires sulfur removal down to sub-ppm concentrations in order to achieve non-

zero catalyst activity. It has been recently demonstrated that Rh-based catalysts suffer from

sulfur poisoning during catalytic processes such as POM and SRM [223-226]. The studies on

the deactivation of Rh during DRM by poisoning show contradictory results.

Desulfurization

Sulfur exists in natural gas mainly as hydrogen sulfide (H2S). The gas is considered “sour” if the H2S concentration exceeds 2 mg/m

3 natural gas (~4ppm). Therefore, the process of removing H2S sometimes is

called as “sweetening” the gas. Most of the catalysts used for syngas production and for the various downstream processes are susceptible to sulfur poisoning. Thus, a purification step of the feedstock is required before the reforming process. Various methods are commercially available: 1) Absorption of impurities using liquid sorbents, 2) Absorption with solid sorbents, 3) Condensation of impurities at very low temperatures, 4) Catalytic hydrogenation [227].

30

Historically the amine treating process was very common for desulfurization and was used in refineries, petrochemical plants, natural gas processing plants etc. However, the increasing environmental awareness has driven to the development of new technologies Table 1.7 [228].

Table 1.6: Acid gas removal (AGR) technologies.

Solid sorbents are used only for the gas containing low concentration of sulfur impurities. Ferrous hydroxide (Fe2O3·xH2O) and zinc oxide (ZnO) are the most commonly used solid sorbents, with ZnO providing the overall best performance (Eq. 1.2) [227].

𝐻2𝑆 + 𝑍𝑛𝑂 𝑍𝑛𝑆 + 𝐻2𝑂 ΔH2980 = −63 𝑘𝐽/𝑚𝑜𝑙 (1.2)

Pritchard et al. [229] did not observe deactivation for a Rh/γ-Al2O3 catalyst during DRM at

1068K with ca. 500ppm of (CH3)2S. However, it was observed by Erdohelyi et al. [230] that Rh

supported on γ-Al2O3, TiO2, SiO2, ZrO2, CeO2 deactivated with TOS at 773 K in the presence of

22ppm H2S in the feed. The extent of S-poisoning changed with the nature of the support

and was lower in case of γ-Al2O3. Mancino et al. [231] investigated sulfur poisoning during

DRM over a Rh/γ-Al2O3 catalyst. Sulfur levels went up to 30 ppm SO2 or H2S to the feed

(CH4:CO2:N2 = 1:1:2) at reaction temperatures in the range 1073-1173 K. Detecting a rapid

drop in syngas production and a corresponding increase in temperature, it was suggested

that the S bonds to the Rh sites. Concentrations as low as 1 ppm of S in the feed adversely

affected the H2 and CO yield. Also, an increase of carbon formation on the catalyst was

detected with respect to S-free conditions. S-poisoning reached a saturation level for

contents ≥ 10 ppm, independently of the S-bearing compound. However, the S adsorption

was found to be reversible. Of course, poisoning can also occur from impurities in the

support. This has been claimed to cause a low activity of a Rh/MgO catalyst during DRM 923

K [232].

31

Further on, few studies investigate how to enhance poisoning resistance of (reforming)

catalysts. Munnik et al. [163] suggested the synthesis of a catalyst with few active sites at

the edge of the catalyst, where poisons could be captured. This could include for example,

yolk-shell and core-shell catalysts [170, 171]. However, their resistance against poisoning has

not yet been investigated [170, 171]. Reaction temperature also has its influence on

poisoning. Increased temperatures will not be favorable to the adsorption of poison.

However, higher temperatures might provoke degradation reactions of the poisons.

Additionally, higher temperatures are not desirable on the industrial scale as this implies a

higher operating cost [215]. Poisoning effects from remnants of active metal precursors

based on e.g. sulfates or chlorides can be circumvented by using metal formates as

precursors instead [233].

Table 1.7: Common poisons classified according to chemical structure [54].

Chemical type Examples Type of interaction with metals

Groups VA and VIA N, P, As, Sb, O, S, Se, Te Through s- and p-orbitals

Group VIIA F, Cl, Br, I Through s- and p-orbitals

Toxic heavy metals and ions As, Pb, Hg, Bi, Sn, Zn, Cd, Cu, Fe Occupy d-orbitals; may form alloys

Molecules which adsorb with multiple

bonds CO, NO, HCN, benzene

Chemisorption through multiple bonds and

back bonding

1.5.3 Sintering

Sintering, as illustrated in Figure 1.10C, is the loss of the catalyst’s active surface due to

crystal growth. Sintering involves complex phenomena that make the understanding of its

mechanistic aspects difficult. Many experimental and theoretical studies can be found in

literature about sintering and redispersion of supported metals [234-238]. Three

mechanisms are suggested: (1) atomic migration (i.e. movement of one active metal atom at

a time, known as Ostwald ripening) (2) Particle migration and coalescence (PMC), and (3)

vapor transport (at very high temperatures) [236]. Atomic migration includes the

detachment/extraction of metal atoms from the crystallites, migration over the support and

eventually being captured by larger crystallites. On the other hand, the crystallite migration

mechanism involves the transport of the entire crystallite over the support followed by

collision and coalescence. Finally, the vapor transport is considered as an insignificant

mechanism at the reforming temperature, including metal loss through direct vaporization

(Figure 1.12) [180].

32

Figure 1.12: Formation of volatile Ni(CO)4 at the surface of nickel crystallite in CO atmosphere. Obrained

from [180].

It is known that sintering is typically rather physical in nature, meaning it has lower

activation energies than carbon formation or poisoning. The temperature at which the solid

phase becomes mobile depends on several factors as texture, size, and morphology [180].

For instance, highly porous γ-Al2O3 is much more sensitive to sintering than non-porous α-

Al2O3. Sintering processes at high temperatures are also affected by the gas atmosphere.

Catalysts sinter more rapidly in presence of steam [179]. Of course, the main factor which

affects sintering is the metal-support interactions. Zhang et al. studied the sintering of Rh

particles on several supports and found that a substantial reduction of Rh dispersion occurs

in case of Al2O3 during DRM. The degree of sintering was found to be significantly smaller for

TiO2, and negligible for YSZ and MgO supports [239].

1.5.4 Encapsulation

Phase transitions of the support, as presented in Figure 1.10D, can be viewed as an

extreme form of sintering occurring at very high temperatures and leading to crystalline

transformation of the support [179]. For example, γ-Al2O3 will transform at high

temperatures into α-Al2O3 [179]. This will decrease the surface area, and thus gas-phase

molecules will not be able to reach the active metal sites. On supported metal catalysts, this

effect will result in the incorporation of the metal into the support, e.g. the reaction of Rh2O3

with Al2O3 to form inactive Rh aluminate at high temperatures [179]. In 1982, Powell and

Whittington were among the first authors to report on the mechanism of encapsulation.

They found that Pt/SiO2 catalysts showed evidence of encapsulation when annealed at 1200

33

K and 1375K. The 100-nm Pt particles became partially immersed in the SiO2 surface with

concurrent formation of a SiO2 ridge around the base of the Pt particles [240].

1.6 Catalyst regeneration by carbon gasification

Despite all the different ways to reduce carbon deposition, carbon accumulation during

reforming reactions remains an issue. Eventually, catalyst regeneration is required, by

removing all carbon species by gasification [241, 242]. The rate of gasification depends on

the structure of the carbon [243], its location [2] and on the nature of the catalyst present

[194, 244, 245]. One important method for carbon species characterization is temperature

programmed (TP) techniques [246, 247]. However, isothermal studies are also required in

order to understand the kinetics [242]. Oxygen is one of the gases most often used [248] to

infer type and location of carbon species on the catalyst while there are also studies on

carbon gasification by CO2, H2O and H2 [195, 249-251]. There are many reports in literature

proposing different mechanisms for catalytic carbon gasification. Some of the mechanisms

that are mentioned are: 1) carbon bulk diffusion, where carbon is transported through the

metal particle to the region where the gasification reaction takes place [188, 249, 252], 2)

oxygen/hydrogen spillover where metal-activated oxygen/hydrogen may migrate over a

considerable distance over the support towards the deposited carbon species and removing

them [250, 253] and 3) the redox mechanism, where the catalyst provides oxygen towards

carbon (reduction step) and is itself oxidized by the gas phase (oxidation step) [254-256]. A

fourth mechanism is suggested in this Thesis and will be presented in Chapter 4.

Figueiredo and Trimm in 1975 [249] studied the gasification of carbon deposits on Ni foil

and supported Ni catalysts. They found a zero reaction order for carbon gasification by

steam, concluding that the reaction is controlled by the diffusion of carbon through Ni

(carbon bulk diffusion mechanism for carbon gasification). Spillover has been reported as a

possible mechanism for carbon gasification [257]. Gas species adsorbed and dissociated on a

metal catalyst can oxidize the carbon atoms by migrating towards them. Spillover is very

common with H2 as a gas [258-260], but the spillover theory has also been suggested for the

CO2 reaction and O2 reaction although these are not as well-established as the H2-spillover

theory. Finally, Machida et al. [261] suggested the redox mechanism, using CeO2 as an active

catalyst for soot oxidation concluding to two possible reaction pathways: i) soot oxidation by

34

adsorbed superoxide species (O2-) at the three-phase boundary (catalyst, carbon, gas) and ii)

soot oxidation by active lattice oxygen at the CeO2/ soot interface.

1.7 Scope of the thesis

The increasing natural gas production and availability of biogas have made CH4 an

attractive feedstock for chemicals. The direct conversion of CH4 is not economical, so mostly

CH4 is reformed to syngas (CO + H2) as an intermediate step. The most well-known reforming

technology is steam reforming, nonetheless dry reforming could become more important in

the future because of its potential to reduce greenhouse gases, and could possibly be

combined with steam resulting in bi-reforming. However, the bi- and dry reforming

processes suffer from catalyst deactivation mainly due to carbon deposition. Up to today,

researchers are fine-tuning catalyst synthesis parameters in order to overcome this

limitation. While complete elimination of catalyst deactivation is not possible, the rate can

be controlled. Figure 1.13 displays the approaches in order to suppress the catalyst

deactivation by modifying the catalyst synthesis parameters.

Figure 1.13: Approaches for catalyst optimization in order to eliminate its deactivation.

The objective of the present work is to control the activity and stability of Ni-based

catalysts for syngas production by eliminating noble metals use in order to succeed it.

35

Replacing noble metals in heterogeneous catalysts by low-cost substitutes has driven

scientific and industrial research for more than 100 years [111, 262, 263]. An effort to

control the activity and stability of this catalyst will be made by adding Fe, an economically

viable and widely available element: (1) as a promoter for Ni-based catalysts and (2)

incorporated in the MgAl2O4 support resulting in a novel MgFexAl2-xO4 material (approaches

displayed in Figure 1.13). The focus will be on elucidating the role of Fe for the reforming

reactions and to unravel the mechanism that dominates the syngas production over

bimetallic Ni-Fe catalysts. The composition of Fe is systematically investigated, using

advanced characterization techniques, creating a detailed catalyst library, in order to reach

solid correlations with the catalyst structure and activity. Finally the effect of reaction

conditions (temperature, pressure, feed composition) on the activity and stability will be

examined.

36

References

[1] W.E. Outlook, International Energy Agency, (2015). [2] D. Pakhare, J. Spivey, A review of dry (CO2) reforming of methane over noble metal catalysts, Chem. Soc. Rev., 43 (2014) 7813-7837. [3] M.V. Iyer, L.P. Norcio, E.L. Kugler, D.B. Dadyburjor, Kinetic Modeling for Methane Reforming with Carbon Dioxide over a Mixed-Metal Carbide Catalyst, Industrial & Engineering Chemistry Research, 42 (2003) 2712-2721. [4] Z. Boukha, C. Jiménez-González, B. de Rivas, J.R. González-Velasco, J.I. Gutiérrez-Ortiz, R. López-Fonseca, Synthesis, characterisation and performance evaluation of spinel-derived Ni/Al2O3 catalysts for various methane reforming reactions, Appl. Catal., B, 158–159 (2014) 190-201. [5] Y. Xu, J. Yang, M. Demura, T. Hirano, Y. Matsushita, M. Tanaka, Y. Katsuya, Catalytic performance of Ni–Al nanoparticles fabricated by arc plasma evaporation for methanol decomposition, International Journal of Hydrogen Energy, 39 (2014) 13156-13163. [6] G. Pantaleo, V. La Parola, F. Deganello, P. Calatozzo, R. Bal, A.M. Venezia, Synthesis and support composition effects on CH4 partial oxidation over Ni–CeLa oxides, Appl. Catal., B, 164 (2015) 135-143. [7] N. Kumar, M. Shojaee, J.J. Spivey, Catalytic bi-reforming of methane: from greenhouse gases to syngas, Curr. Opin. Chem. Eng., 9 (2015) 8-15. [8] M.J.F. Kohn M.P.; Castaldi, R.J., Auto-thermal and dry reforming of landfill gas over a Rh/Al2O3 monolith catalyst, Appl. Catal., B, 94 (2010) 125-133. [9] E.B.A. (EBA), European Biogas Association releases biogas, biomethane report, 2015. [10] J.R.H. Ross, Natural gas reforming and CO2 mitigation, Catal. Today, 100 (2005) 151-158. [11] A. Caballero, P.J. Perez, Methane as raw material in synthetic chemistry: the final frontier, Chem. Soc. Rev., 42 (2013) 8809-8820. [12] S.D. Angeli, L. Turchetti, G. Monteleone, A.A. Lemonidou, Catalyst development for steam reforming of methane and model biogas at low temperature, Appl. Catal., B, 181 (2016) 34-46. [13] E.T. Kho, J. Scott, R. Amal, Ni/TiO2 for low temperature steam reforming of methane, Chem. Eng. Sci., 140 (2016) 161-170. [14] J. Xu, G.F. Froment, Methane steam reforming, methanation and water-gas shift: I. Intrinsic kinetics, AIChE Journal, 35 (1989) 88-96. [15] S.A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, Enhanced Carbon-Resistant Dry Reforming Fe-Ni Catalyst: Role of Fe, ACS Catal., (2015) 3028-3039. [16] M. Yu, Y.-A. Zhu, Y. Lu, G. Tong, K. Zhu, X. Zhou, The promoting role of Ag in Ni-CeO2 catalyzed CH4-CO2 dry reforming reaction, Appl. Catal., B, 165 (2015) 43-56. [17] S.M. Kim, P.M. Abdala, T. Margossian, D. Hosseini, L. Foppa, A. Armutlulu, W. van Beek, A. Comas-Vives, C. Copéret, C. Müller, Cooperativity and Dynamics Increase the Performance of NiFe Dry Reforming Catalysts, J. Am. Chem. Soc., (2017). [18] M.D. Salazar-Villalpando, D.A. Berry, T.H. Gardner, Partial oxidation of methane over Rh/supported-ceria catalysts: Effect of catalyst reducibility and redox cycles, Int. J. Hydrogen Energy, 33 (2008) 2695-2703. [19] A. Scarabello, D. Dalle Nogare, P. Canu, R. Lanza, Partial oxidation of methane on Rh/ZrO2 and Rh/Ce–ZrO2 on monoliths: Catalyst restructuring at reaction conditions, Appl. Catal., B, 174–175 (2015) 308-322. [20] Y. Yan, Z. Zhang, L. Zhang, X. Wang, K. Liu, Z. Yang, Investigation of autothermal reforming of methane for hydrogen production in a spiral multi-cylinder micro-reactor used for mobile fuel cell, Int. J. Hydrogen Energy, 40 (2015) 1886-1893. [21] A. M. Gaddalla, M. E. Sommer, Carbon dioxide reforming of methane on nickel catalysts, Chem. Eng. Sci., 44 (1989) 2825-2829. [22] J.R. Rostrup-Nielsen, J. Sehested, J.K. Nørskov, Hydrogen and synthesis gas by steam- and CO2 reforming, Adv. Catal., 47 (2002) 65-139.

37

[23] B. Albrecht, Reactor Modeling and Process Analysis for Partial Oxidation of Natural Gas, Twente Research School for Process Technology, Enschede, 2004. [24] J.R. Rostrup-Nielsen, Production of synthesis gas, Catal. Today, 18 (1993) 305-324. [25] T. Ishihara, Y. Takita, Partial Oxidation of Methane into Syngas with Oxygen Permeating Ceramic Membrane Reactors, Catalysis Surveys from Asia, 4 125-133. [26] M.-S. Fan, A.Z. Abdullah, S. Bhatia, Catalytic Technology for Carbon Dioxide Reforming of Methane to Synthesis Gas, ChemCatChem, 1 (2009) 192-208. [27] K.S. Mondal, S.; Badgandi, S.; Chowdhury, D.R.; Nair, V., Dry reforming of methane to syngas: a potential alternative process for value added chemicals - a techno-economic perspective, Environmental Science Pollution Research, (2016). [28] J. Galuszka, R.N. Pandey, S. Ahmed, Methane conversion to syngas in a palladium membrane reactor, Catal. Today, 46 (1998) 83-89. [29] I. Intergovernmental Panel on Climate Change, Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, (2014). [30] I. Intergovernmental Panel on Climate Change, Climate Change 2014: Synthesis Report, (2014). [31] J.H. Edwards, Potential sources of CO2 and the options for its large-scale utilisation now and in the future, Catalysis Today, 23 (1995) 59-66. [32] P. Ferreira-Aparicio, M. Fernandez-Garcia, A. Guerrero-Ruiz, Rodrı, amp, I. guez-Ramos, Evaluation of the Role of the Metal–Support Interfacial Centers in the Dry Reforming of Methane on Alumina-Supported Rhodium Catalysts, Journal of Catalysis, 190 (2000) 296-308. [33] P. Van Beurden, On the catalytic aspects of steam-methane reforming: a literature survey, ECN, 2004. [34] H.R. Boerrigter, R., Review of applications of gases from biomass gasification, ECN-C--04-112, (2004). [35] J.-M. Lavoie, Review on dry reforming of methane, a potentially more environmentally-friendly approach to the increasing natural gas exploitation, Frontiers in Chemistry, 2 (2014). [36] B. de Caprariis, P. de Filippis, V. Palma, A. Petrullo, A. Ricca, C. Ruocco, M. Scarsella, Rh, Ru and Pt ternary perovskites type oxides BaZr(1-x)MexO3 for methane dry reforming, Appl. Catal., A. [37] J.H. Edwards, A.M. Maitra, The chemistry of methane reforming with carbon dioxide and its current and potential applications, Fuel Processing Technology, 42 (1995) 269-289. [38] S. Wang, G.Q. Lu, G.J. Millar, Carbon Dioxide Reforming of Methane To Produce Synthesis Gas over Metal-Supported Catalysts: State of the Art, Energy & Fuels, 10 (1996) 896-904. [39] M.K. Nikoo, N.A.S. Amin, Thermodynamic analysis of carbon dioxide reforming of methane in view of solid carbon formation, Fuel Processing Technology, 92 (2011) 678-691. [40] S.T. Oyama, P. Hacarlioglu, Y. Gu, D. Lee, Dry reforming of methane has no future for hydrogen production: Comparison with steam reforming at high pressure in standard and membrane reactors, Int. J. Hydrogen Energy, 37 (2012) 10444-10450. [41] S.A. Theofanidis, V.V. Galvita, M. Sabbe, H. Poelman, C. Detavernier, G.B. Marin, Controlling the stability of a Fe–Ni reforming catalyst: Structural organization of the active components, Appl. Catal., B, 209 (2017) 405-416. [42] A.C.D. Freitas, R. Guirardello, Thermodynamic analysis of methane reforming with CO2, CO2 + H2O, CO2 + O2 and CO2 + air for hydrogen and synthesis gas production, J. CO2 Util., 7 (2014) 30-38. [43] G.A. Olah, A. Goeppert, M. Czaun, G.K.S. Prakash, Bi-reforming of Methane from Any Source with Steam and Carbon Dioxide Exclusively to Metgas (CO–2H2) for Methanol and Hydrocarbon Synthesis, J. Am. Chem. Soc., 135 (2013) 648-650. [44] G.A. Olah, A. Goeppert, M. Czaun, T. Mathew, R.B. May, G.K.S. Prakash, Single Step Bi-reforming and Oxidative Bi-reforming of Methane (Natural Gas) with Steam and Carbon Dioxide to Metgas (CO-2H2) for Methanol Synthesis: Self-Sufficient Effective and Exclusive Oxygenation of Methane to Methanol with Oxygen, J. Am. Chem. Soc., 137 (2015) 8720-8729. [45] H.-S. Roh, Koo, K.Y., Catal Lett, 125 (2008) 283-288. [46] Q.-H.L. Zhang, Y., Catalysis Today, 98 (2004) 601-605.

38

[47] J. Yoo, Y. Bang, S.J. Han, S. Park, J.H. Song, I.K. Song, Hydrogen production by tri-reforming of methane over nickel–alumina aerogel catalyst, J. Mol. Catal. A: Chem., 410 (2015) 74-80. [48] C. Song, W. Pan, Tri-reforming of methane: a novel concept for catalytic production of industrially useful synthesis gas with desired H2/CO ratios, Catal. Today, 98 (2004) 463-484. [49] S. Sumrunronnasak, S. Tantayanon, S. Kiatgamolchai, T. Sukonket, Improved hydrogen production from dry reforming reaction using a catalytic packed-bed membrane reactor with Ni-based catalyst and dense PdAgCu alloy membrane, Int. J. Hydrogen Energy, 41 (2016) 2621-2630. [50] H. Boerrigter, A.v.d. Drift, Biosyngas: Description of R&D trajectory necessary to reach large-scale implementation of renewable syngas from biomass, ECN-C--04-112, (2004). [51] Rostrup-Nielsen J.R., Christiansen L.J., Concepts in Syngas Manufacture, Energy Technology, 1 (2013) 419-420. [52] J.R. Rostrup -Nielsen, L.J. Christiansen, Concepts in Syngas Manufacture, in: I.C. Press (Ed.), 2011. [53] J. Kopfle, Midrex. Technology Development: A Firm Foundation + Continuous Improvement, Midrex Technologies, Inc., 2007. [54] H.H. Gunardson, Industrial Gases in Petrochemical Processing: Chemical Industries, Marcel Dekker, Inc., New York, 1998. [55] O. Muraza, A. Galadima, A review on coke management during dry reforming of methane, International Journal of Energy Research, 39 (2015) 1196-1216. [56] Carbon Sciences Secures Worldwide Exclusive License to Breakthrough Greenhouse Gas Transformation Technology, License for Dry Reforming of Methane Technology Eliminates Major Development Hurdle for Company, 2010. [57] V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, Transient Kinetic Study of the Oxidation and Hydrogenation of Carbon Species Formed during CH4/He, CO2/He, and CH4/CO2 reactions over Rh/Al2O3 Catalyst, J. Catal., 161 (1996) 31-42. [58] V.A. Sadykov, E.L. Gubanova, N.N. Sazonova, S.A. Pokrovskaya, N.A. Chumakova, N.V. Mezentseva, A.S. Bobin, R.V. Gulyaev, A.V. Ishchenko, T.A. Krieger, C. Mirodatos, Dry reforming of methane over Pt/PrCeZrO catalyst: Kinetic and mechanistic features by transient studies and their modeling, Catal. Today, 171 (2011) 140-149. [59] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, Reforming of Methane with Carbon Dioxide to Synthesis Gas over Supported Rhodium Catalysts: II. A Steady-State Tracing Analysis: Mechanistic Aspects of the Carbon and Oxygen Reaction Pathways to Form CO, J. Catal., 158 (1996) 64-75. [60] M.C.J. Bradford, M.A. Vannice, CO2 Reforming of CH4, Catal. Rev. - Sci. Eng., 41 (1999) 1-42. [61] J. Guo, C. Xie, K. Lee, N. Guo, J.T. Miller, M.J. Janik, C. Song, Improving the Carbon Resistance of Ni-Based Steam Reforming Catalyst by Alloying with Rh: A Computational Study Coupled with Reforming Experiments and EXAFS Characterization, ACS Catal., 1 (2011) 574-582. [62] J.R. Rostrupnielsen, J.H.B. Hansen, CO2-Reforming of Methane over Transition Metals, Journal of Catalysis, 144 (1993) 38-49. [63] M.C.J. Bradford, M.A. Vannice, CO2 Reforming of CH4, Catal. Rev., 41 (1999) 1-42. [64] Z. Hou, P. Chen, H. Fang, X. Zheng, T. Yashima, Production of synthesis gas via methane reforming with CO on noble metals and small amount of noble-(Rh-) promoted Ni catalysts, International Journal of Hydrogen Energy, 31 (2006) 555-561. [65] D.L. Trimm, Coke formation and minimisation during steam reforming reactions, Catal. Today, 37 (1997) 233-238. [66] D.L. Trimm, Catalysts for the control of coking during steam reforming, Catal. Today, 49 (1999) 3-10. [67] J. Ashok, S. Kawi, Steam reforming of toluene as a biomass tar model compound over CeO2 promoted Ni/CaO–Al2O3 catalytic systems, Int. J. Hydrogen Energy, 38 (2013) 13938-13949. [68] L. Wang, D. Li, M. Koike, H. Watanabe, Y. Xu, Y. Nakagawa, K. Tomishige, Catalytic performance and characterization of Ni–Co catalysts for the steam reforming of biomass tar to synthesis gas, Fuel, 112 (2013) 654-661.

39

[69] M.P. Andersson, T. Bligaard, A. Kustov, K.E. Larsen, J. Greeley, T. Johannessen, C.H. Christensen, J.K. Nørskov, Toward computational screening in heterogeneous catalysis: Pareto-optimal methanation catalysts, Journal of Catalysis, 239 (2006) 501-506. [70] D. Li, I. Atake, T. Shishido, Y. Oumi, T. Sano, K. Takehira, Self-regenerative activity of Ni/Mg(Al)O catalysts with trace Ru during daily start-up and shut-down operation of CH4 steam reforming, J. Catal., 250 (2007) 299-312. [71] K. Nagaoka, A. Jentys, J.A. Lercher, Methane autothermal reforming with and without ethane over mono- and bimetal catalysts prepared from hydrotalcite precursors, J. Catal., 229 (2005) 185-196. [72] T. Huang, W. Huang, J. Huang, P. Ji, Methane reforming reaction with carbon dioxide over SBA-15 supported Ni–Mo bimetallic catalysts, Fuel Process. Technol., 92 (2011) 1868-1875. [73] Z. Hou, T. Yashima, Small Amounts of Rh-Promoted Ni Catalysts for Methane Reforming with CO2, Catalysis Letters, 89 193-197. [74] B. Pawelec, S. Damyanova, K. Arishtirova, J.L.G. Fierro, L. Petrov, Structural and surface features of PtNi catalysts for reforming of methane with CO2, Applied Catalysis A: General, 323 (2007) 188-201. [75] K. Nagaoka, K. Takanabe, K.-i. Aika, Modification of Co/TiO2 for dry reforming of methane at 2 MPa by Pt, Ru or Ni, Applied Catalysis A: General, 268 (2004) 151-158. [76] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, C. Sener, T. Dogu, MCM-41 supported PdNi catalysts for dry reforming of methane, Appl. Catal., B, 92 (2009) 250-261. [77] M. Ferrandon, A.J. Kropf, T. Krause, Bimetallic Ni-Rh catalysts with low amounts of Rh for the steam and autothermal reforming of n-butane for fuel cell applications, Appl. Catal., A, 379 (2010) 121-128. [78] B. Steinhauer, M.R. Kasireddy, J. Radnik, A. Martin, Development of Ni-Pd bimetallic catalysts for the utilization of carbon dioxide and methane by dry reforming, Appl. Catal., A, 366 (2009) 333-341. [79] U. Oemar, K. Hidajat, S. Kawi, High catalytic stability of Pd-Ni/Y2O3 formed by interfacial Cl for oxy-CO2 reforming of CH4, Catal. Today, 281, Part 2 (2017) 276-294. [80] B. Pawelec, S. Damyanova, K. Arishtirova, J.L.G. Fierro, L. Petrov, Structural and surface features of PtNi catalysts for reforming of methane with CO2, Appl. Catal., A, 323 (2007) 188-201. [81] K. Ray, S. Sengupta, G. Deo, Reforming and cracking of CH4 over Al2O3 supported Ni, Ni-Fe and Ni-Co catalysts, Fuel Process. Technol., 156 (2017) 195-203. [82] C. Wang, N. Sun, N. Zhao, W. Wei, Y. Zhao, Template-free preparation of bimetallic mesoporous Ni-Co-CaO-ZrO2 catalysts and their synergetic effect in dry reforming of methane, Catal. Today, 281, Part 2 (2017) 268-275. [83] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catal., 4 (2014) 289-301. [84] A. Tsoukalou, Q. Imtiaz, S.M. Kim, P.M. Abdala, S. Yoon, C.R. Müller, Dry-reforming of methane over bimetallic Ni–M/La2O3 (M = Co, Fe): The effect of the rate of La2O2CO3 formation and phase stability on the catalytic activity and stability, J. Catal., 343 (2016) 208-214. [85] M. Yu, K. Zhu, Z. Liu, H. Xiao, W. Deng, X. Zhou, Carbon dioxide reforming of methane over promoted NixMg1−xO (1 1 1) platelet catalyst derived from solvothermal synthesis, Appl. Catal., B, 148–149 (2014) 177-190. [86] K. Tomishige, Y. Himeno, Y. Matsuo, Y. Yoshinaga, K. Fujimoto, Catalytic Performance and Carbon Deposition Behavior of a NiO−MgO Solid Solution in Methane Reforming with Carbon Dioxide under Pressurized Conditions, Industrial & Engineering Chemistry Research, 39 (2000) 1891-1897. [87] B.A. Rosen, E. Gileadi, N. Eliaz, Electrodeposited Re-promoted Ni foams as a catalyst for the dry reforming of methane, Catal. Commun., 76 (2016) 23-28. [88] I. Luisetto, C. Sarno, D. De Felicis, F. Basoli, C. Battocchio, S. Tuti, S. Licoccia, E. Di Bartolomeo, Ni supported on γ-Al2O3 promoted by Ru for the dry reforming of methane in packed and monolithic reactors, Fuel Process. Technol., 158 (2017) 130-140. [89] T. Huang, W. Huang, J. Huang, P. Ji, Methane reforming reaction with carbon dioxide over SBA-15 supported Ni–Mo bimetallic catalysts, Fuel Processing Technology, 92 (2011) 1868-1875.

40

[90] K. Gheisari, S. Javadpour, J.T. Oh, M. Ghaffari, The effect of milling speed on the structural properties of mechanically alloyed Fe–45%Ni powders, J. Alloys Compd., 472 (2009) 416-420. [91] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Catalytic performance and characterization of Ni-Fe catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Appl. Catal. A 392 (2011) 248-255. [92] S.A. Theofanidis, R. Batchu, V.V. Galvita, H. Poelman, G.B. Marin, Carbon gasification from Fe–Ni catalysts after methane dry reforming, Appl. Catal., B, 185 (2016) 42-55. [93] W.H. Bragg, The structure of magnetite and the spinels, Nature, 95 (1915) 561-561. [94] B.K. Coltrain, N. Herron, D.H. Busch, Oxygen Activation by Transition Metal Complexes of Macrobicyclic Cyclidene Ligands, in: D.H.R. Barton, A.E. Martell, D.T. Sawyer (Eds.) The Activation of Dioxygen and Homogeneous Catalytic Oxidation, Springer US, Boston, MA, 1993, pp. 359-380. [95] A.C. Dlouhy, C.E. Outten, The Iron Metallome in Eukaryotic Organisms, Metal ions in life sciences, 12 (2013) 241-278. [96] L. Yan, S. Zhang, P. Chen, H. Liu, H. Yin, H. Li, Magnetotactic bacteria, magnetosomes and their application, Microbiol. Res., 167 (2012) 507-519. [97] P. Xu, G.M. Zeng, D.L. Huang, C.L. Feng, S. Hu, M.H. Zhao, C. Lai, Z. Wei, C. Huang, G.X. Xie, Z.F. Liu, Use of iron oxide nanomaterials in wastewater treatment: A review, Sci. Total Environ., 424 (2012) 1-10. [98] R.V. Jagadeesh, A.-E. Surkus, H. Junge, M.-M. Pohl, J. Radnik, J. Rabeah, H. Huan, V. Schünemann, A. Brückner, M. Beller, Nanoscale Fe2O3-Based Catalysts for Selective Hydrogenation of Nitroarenes to Anilines, Science, 342 (2013) 1073-1076. [99] S. Bagheri, N.M. Julkapli, Modified iron oxide nanomaterials: Functionalization and application, J. Magn. Magn. Mater., 416 (2016) 117-133. [100] E. Bykova, L. Dubrovinsky, N. Dubrovinskaia, M. Bykov, C. McCammon, S.V. Ovsyannikov, H.P. Liermann, I. Kupenko, A.I. Chumakov, R. Ruffer, M. Hanfland, V. Prakapenka, Structural complexity of simple Fe2O3 at high pressures and temperatures, Nat Commun, 7 (2016). [101] G. Chen, Y. Zhao, G. Fu, P.N. Duchesne, L. Gu, Y. Zheng, X. Weng, M. Chen, P. Zhang, C.-W. Pao, J.-F. Lee, N. Zheng, Interfacial Effects in Iron-Nickel Hydroxide–Platinum Nanoparticles Enhance Catalytic Oxidation, Science, 344 (2014) 495-499. [102] S. Handa, Y. Wang, F. Gallou, B.H. Lipshutz, Sustainable Fe–ppm Pd nanoparticle catalysis of Suzuki-Miyaura cross-couplings in water, Science, 349 (2015) 1087-1091. [103] M. Zhao, K. Yuan, Y. Wang, G. Li, J. Guo, L. Gu, W. Hu, H. Zhao, Z. Tang, Metal–organic frameworks as selectivity regulators for hydrogenation reactions, Nature, 539 (2016) 76-80. [104] P. Tartaj, M.P. Morales, T. Gonzalez-Carreño, S. Veintemillas-Verdaguer, C.J. Serna, The Iron Oxides Strike Back: From Biomedical Applications to Energy Storage Devices and Photoelectrochemical Water Splitting, Adv. Mater., 23 (2011) 5243-5249. [105] N.V.R.A. Dharanipragada, L.C. Buelens, H. Poelman, E. De Grave, V.V. Galvita, G.B. Marin, Mg-Fe-Al-O for advanced CO2 to CO conversion: carbon monoxide yield vs. oxygen storage capacity, J. Mater. Chem. A, 3 (2015) 16251-16262. [106] J. Hu, L. Buelens, S.-A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, CO2 conversion to CO by auto-thermal catalyst-assisted chemical looping, J. CO2 Util., 16 (2016) 8-16. [107] A. Grossale, I. Nova, E. Tronconi, Study of a Fe–zeolite-based system as NH3-SCR catalyst for diesel exhaust aftertreatment, Catal. Today, 136 (2008) 18-27. [108] M. Zhu, I.E. Wachs, Iron-Based Catalysts for the High-Temperature Water–Gas Shift (HT-WGS) Reaction: A Review, ACS Catal., 6 (2016) 722-732. [109] F. Meshkani, M. Rezaei, A highly active and stable chromium free iron based catalyst for H2 purification in high temperature water gas shift reaction, International Journal of Hydrogen Energy. [110] V.V. Galvita, H. Poelman, V. Bliznuk, C. Detavernier, G.B. Marin, CeO2-Modified Fe2O3 for CO2 Utilization via Chemical Looping, Ind. Eng. Chem. Res., 52 (2013) 8416-8426. [111] L.C. Buelens, V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Super-dry reforming of methane intensifies CO2 utilization via Le Chatelier’s principle, Science, 354 (2016) 449-452.

41

[112] M. Koike, D. Li, H. Watanabe, Y. Nakagawa, K. Tomishige, Comparative study on steam reforming of model aromatic compounds of biomass tar over Ni and Ni–Fe alloy nanoparticles, Applied Catalysis A: General, 506 (2015) 151-162. [113] A. Djaidja, H. Messaoudi, D. Kaddeche, A. Barama, Study of Ni–M/MgO and Ni–M–Mg/Al (M=Fe or Cu) catalysts in the CH4–CO2 and CH4–H2O reforming, International Journal of Hydrogen Energy, 40 (2015) 4989-4995. [114] A. More, S. Bhavsar, G. Veser, Iron–Nickel Alloys for Carbon Dioxide Activation by Chemical Looping Dry Reforming of Methane, Energy Technol., 4 (2016) 1147-1157. [115] S. Sayas, A. Chica, Furfural steam reforming over Ni-based catalysts. Influence of Ni incorporation method, International Journal of Hydrogen Energy, 39 (2014) 5234-5241. [116] Z. Alipour, M. Rezaei, F. Meshkani, Effect of alkaline earth promoters (MgO, CaO, and BaO) on the activity and coke formation of Ni catalysts supported on nanocrystalline Al2O3 in dry reforming of methane, Journal of Industrial and Engineering Chemistry, 20 (2014) 2858-2863. [117] J. Xu, L. Chen, K.F. Tan, A. Borgna, M. Saeys, Effect of boron on the stability of Ni catalysts during steam methane reforming, Journal of Catalysis, 261 (2009) 158-165. [118] H. Jeong, K.I. Kim, D. Kim, I.K. Song, Effect of promoters in the methane reforming with carbon dioxide to synthesis gas over Ni/HY catalysts, Journal of Molecular Catalysis A: Chemical, 246 (2006) 43-48. [119] G.C. Bond, Metal-Support and Metal-Additive Effects in Catalysis, Platinum Metals Reviews, 27 (1983) 16-18. [120] L. Xu, H. Zhao, H. Song, L. Chou, Ordered mesoporous alumina supported nickel based catalysts for carbon dioxide reforming of methane, International Journal of Hydrogen Energy, 37 (2012) 7497-7511. [121] S.J. Tauster, Strong metal-support interactions, Acc. Chem. Res., 20 (1987) 389-394. [122] J. Zhang, Y. Bai, Q. Zhang, X. Wang, T. Zhang, Y. Tan, Y. Han, Low-temperature methanation of syngas in slurry phase over Zr-doped Ni/γ-Al2O3 catalysts prepared using different methods, Fuel, 132 (2014) 211-218. [123] V.C.H. Kroll, H.M. Swaan, S. Lacombe, C. Mirodatos, Methane Reforming Reaction with Carbon Dioxide over Ni/SiO2Catalyst: II. A Mechanistic Study, Journal of Catalysis, 164 (1996) 387-398. [124] Z. Zhang, X.E. Verykios, Carbon dioxide reforming of methane to synthesis gas over Ni/La2O3 catalysts, Applied Catalysis A: General, 138 (1996) 109-133. [125] E. Ruckenstein, H.Y. Wang, Carbon dioxide reforming of methane to synthesis gas over supported cobalt catalysts, Applied Catalysis A: General, 204 (2000) 257-263. [126] Z.X. Cheng, X.G. Zhao, J.L. Li, Q.M. Zhu, Role of support in CO2 reforming of CH4 over a Ni/γ-Al2O3 catalyst, Appl. Catal., A, 205 (2001) 31-36. [127] N. Laosiripojana, S. Assabumrungrat, Catalytic dry reforming of methane over high surface area ceria, Appl. Catal., B, 60 (2005) 107-116. [128] A. Horváth, G. Stefler, O. Geszti, A. Kienneman, A. Pietraszek, L. Guczi, Methane dry reforming with CO2 on CeZr-oxide supported Ni, NiRh and NiCo catalysts prepared by sol–gel technique: Relationship between activity and coke formation, Catalysis Today, 169 (2011) 102-111. [129] H.Y. Wang, E. Ruckenstein, Carbon dioxide reforming of methane to synthesis gas over supported rhodium catalysts: the effect of support, Applied Catalysis A: General, 204 (2000) 143-152. [130] E.S. Hecht, G.K. Gupta, H. Zhu, A.M. Dean, R.J. Kee, L. Maier, O. Deutschmann, Methane reforming kinetics within a Ni–YSZ SOFC anode support, Applied Catalysis A: General, 295 (2005) 40-51. [131] D. Pakhare, V. Schwartz, V. Abdelsayed, D. Haynes, D. Shekhawat, J. Poston, J. Spivey, Kinetic and mechanistic study of dry (CO2) reforming of methane over Rh-substituted La2Zr2O7 pyrochlores, Journal of Catalysis, 316 (2014) 78-92. [132] A. Horváth, G. Stefler, O. Geszti, A. Kienneman, A. Pietraszek, L. Guczi, Methane dry reforming with CO2 on CeZr-oxide supported Ni, NiRh and NiCo catalysts prepared by sol–gel technique: Relationship between activity and coke formation, Catal. Today, 169 (2011) 102-111.

42

[133] J. Guo, H. Lou, H. Zhao, D. Chai, X. Zheng, Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels, Applied Catalysis A: General, 273 (2004) 75-82. [134] A. Penkova, L. Bobadilla, S. Ivanova, M.I. Domínguez, F. Romero-Sarria, A.C. Roger, M.A. Centeno, J.A. Odriozola, Hydrogen production by methanol steam reforming on NiSn/MgO–Al2O3 catalysts: The role of MgO addition, Applied Catalysis A: General, 392 (2011) 184-191. [135] M.F. Zawrah, H. Hamaad, S. Meky, Synthesis and characterization of nano MgAl2O4 spinel by the co-precipitated method, Ceram. Int., 33 (2007) 969-978. [136] M. Ji, X. Zhang, J. Wang, S.-E. Park, Ethylbenzene dehydrogenation with CO2 over Fe-doped MgAl2O4 spinel catalysts: Synergy effect between Fe2+ and Fe3+, Journal of Molecular Catalysis A: Chemical, 371 (2013) 36-41. [137] M. Lei, Z.B. Wang, J.S. Li, H.L. Tang, W.J. Liu, Y.G. Wang, CeO2 nanocubes-graphene oxide as durable and highly active catalyst support for proton exchange membrane fuel cell, Sci. Rep., 4 (2014) 7415. [138] S. Lin, L. Yang, X. Yang, R. Zhou, Redox behavior of active PdOx species on (Ce,Zr)xO2–Al2O3 mixed oxides and its influence on the three-way catalytic performance, Chem. Eng. J., 247 (2014) 42-49. [139] M. Meledina, S. Turner, V.V. Galvita, H. Poelman, G.B. Marin, G. Van Tendeloo, Local environment of Fe dopants in nanoscale Fe : CeO2-x oxygen storage material, Nanoscale, 7 (2015) 3196-3204. [140] J. Wei, E. Iglesia, Structural requirements and reaction pathways in methane activation and chemical conversion catalyzed by rhodium, J. Catal., 225 (2004) 116-127. [141] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, Reforming of Methane with Carbon Dioxide to Synthesis Gas over Supported Rhodium Catalysts: II. A Steady-State Tracing Analysis: Mechanistic Aspects of the Carbon and Oxygen Reaction Pathways to Form CO, Journal of Catalysis, 158 (1996) 64-75. [142] L.M. Aparicio, Transient Isotopic Studies and Microkinetic Modeling of Methane Reforming over Nickel Catalysts, Journal of Catalysis, 165 (1997) 262-274. [143] M.C.J. Bradford, M.A. Vannice, Catalytic reforming of methane with carbon dioxide over nickel catalysts II. Reaction kinetics, Appl. Catal., A, 142 (1996) 97-122. [144] L. Foppa, M.-C. Silaghi, K. Larmier, A. Comas-Vives, Intrinsic reactivity of Ni, Pd and Pt surfaces in dry reforming and competitive reactions: Insights from first principles calculations and microkinetic modeling simulations, J. Catal., 343 (2016) 196-207. [145] J. Wei, E. Iglesia, Isotopic and kinetic assessment of the mechanism of reactions of CH4 with CO2 or H2O to form synthesis gas and carbon on nickel catalysts, J. Catal., 224 (2004) 370-383. [146] D.C. Seets, M.C. Wheeler, C.B. Mullins, Mechanism of the dissociative chemisorption of methane over Ir(110): trapping-mediated or direct?, Chem. Phys. Lett., 266 (1997) 431-436. [147] D. Pakhare, J. Spivey, A review of dry (CO2) reforming of methane over noble metal catalysts, Chemical Society Reviews, 43 (2014) 7813-7837. [148] J.H. Bitter, K. Seshan, J.A. Lercher, On the contribution of X-ray absorption spectroscopy to explore structure and activity relations of Pt/ZrO2 catalysts for CO2/CH4 reforming, Topics in Catalysis, 10 295-305. [149] J. Wei, E. Iglesia, Mechanism and Site Requirements for Activation and Chemical Conversion of Methane on Supported Pt Clusters and Turnover Rate Comparisons among Noble Metals, J. Phys. Chem. B, 108 (2004) 4094-4103. [150] A.M. O’Connor, Y. Schuurman, J.R.H. Ross, C. Mirodatos, Transient studies of carbon dioxide reforming of methane over Pt/ZrO2 and Pt/Al2O3, Catalysis Today, 115 (2006) 191-198. [151] J. Wei, E. Iglesia, Isotopic and kinetic assessment of the mechanism of reactions of CH4 with CO2 or H2O to form synthesis gas and carbon on nickel catalysts, Journal of Catalysis, 224 (2004) 370-383. [152] G. Sierra Gallego, C. Batiot-Dupeyrat, J. Barrault, F. Mondragón, Dual Active-Site Mechanism for Dry Methane Reforming over Ni/La2O3 Produced from LaNiO3 Perovskite, Ind. Eng. Chem. Res., 47 (2008) 9272-9278.

43

[153] J.T. Richardson, S.A. Paripatyadar, Carbon dioxide reforming of methane with supported rhodium, Appl. Catal., 61 (1990) 293-309. [154] A. Erdöhelyi, J. Cserényi, E. Papp, F. Solymosi, Catalytic reaction of methane with carbon dioxide over supported palladium, Applied Catalysis A: General, 108 (1994) 205-219. [155] G.B. Marin, G.S. Yablonsky, Kinetics of Chemicals Reactions: Decoding complexity, in: Wiley-VCH (Ed.), 2011. [156] N. Kageyama, B.R. Devocht, A. Takagaki, K. Toch, J.W. Thybaut, G.B. Marin, S.T. Oyama, Interplay of Kinetics and Thermodynamics in Catalytic Steam Methane Reforming over Ni/MgO-SiO2, Ind. Eng. Chem. Res., 56 (2017) 1148-1158. [157] P. Sadooghi, R. Rauch, Experimental and modeling study of hydrogen production from catalytic steam reforming of methane mixture with hydrogen sulfide, Int. J. Hydrogen Energy, 40 (2015) 10418-10426. [158] C. Sprung, P.N. Kechagiopoulos, J.W. Thybaut, B. Arstad, U. Olsbye, G.B. Marin, Microkinetic evaluation of normal and inverse kinetic isotope effects during methane steam reforming to synthesis gas over a Ni/NiAl2O4 model catalyst, Appl. Catal., A, 492 (2015) 231-242. [159] J.F. Múnera, L.M. Cornaglia, D.V. Cesar, M. Schmal, E.A. Lombardo, Kinetic Studies of the Dry Reforming of Methane over the Rh/La2O3−SiO2 Catalyst, Industrial & Engineering Chemistry Research, 46 (2007) 7543-7549. [160] X.E. Verykios, Catalytic dry reforming of natural gas for the production of chemicals and hydrogen, International Journal of Hydrogen Energy, 28 (2003) 1045-1063. [161] M.M. Barroso Quiroga, A.E. Castro Luna, Kinetic Analysis of Rate Data for Dry Reforming of Methane, Industrial & Engineering Chemistry Research, 46 (2007) 5265-5270. [162] A.S. Bobin, V.A. Sadykov, V.A. Rogov, N.V. Mezentseva, G.M. Alikina, E.M. Sadovskaya, T.S. Glazneva, N.N. Sazonova, M.Y. Smirnova, S.A. Veniaminov, C. Mirodatos, V. Galvita, G.B. Marin, Mechanism of CH4 Dry Reforming on Nanocrystalline Doped Ceria-Zirconia with Supported Pt, Ru, Ni, and Ni–Ru, Top. Catal., 56 (2013) 958-968. [163] P.d.J. Munnik, P.E.; de Jong, K.P., Recent Developments in the Synthesis of Supported Catalysts, Chem. Rev., 115 (2015) 6687-6718. [164] D.-W. Jeong, W.-J. Jang, J.-O. Shim, H.-S. Roh, I.H. Son, S.J. Lee, The effect of preparation method on the catalytic performance over superior MgO-promoted Ni–Ce0.8Zr0.2O2 catalyst for CO2

reforming of CH4, Int. J. Hydrogen Energy, 38 (2013) 13649-13654. [165] Ş. Özkara-Aydınoğlu, E. Özensoy, A.E. Aksoylu, The effect of impregnation strategy on methane dry reforming activity of Ce promoted Pt/ZrO2, International Journal of Hydrogen Energy, 34 (2009) 9711-9722. [166] S.J. Han, Y. Bang, J. Yoo, J.G. Seo, I.K. Song, Hydrogen production by steam reforming of ethanol over mesoporous Ni–Al2O3–ZrO2 xerogel catalysts: Effect of nickel content, International Journal of Hydrogen Energy, 38 (2013) 8285-8292. [167] Z. Hao, Q. Zhu, Z. Jiang, B. Hou, H. Li, Characterization of aerogel Ni/Al2O3 catalysts and investigation on their stability for CH4-CO2 reforming in a fluidized bed, Fuel Processing Technology, 90 (2009) 113-121. [168] M. García-Diéguez, I.S. Pieta, M.C. Herrera, M.A. Larrubia, L.J. Alemany, Improved Pt-Ni nanocatalysts for dry reforming of methane, Applied Catalysis A: General, 377 (2010) 191-199. [169] T.D. Gould, M.M. Montemore, A.M. Lubers, L.D. Ellis, A.W. Weimer, J.L. Falconer, J.W. Medlin, Enhanced dry reforming of methane on Ni and Ni-Pt catalysts synthesized by atomic layer deposition, Applied Catalysis A: General, 492 (2015) 107-116. [170] J. Zhang, F. Li, Coke-resistant Ni@SiO2 catalyst for dry reforming of methane, Applied Catalysis B: Environmental, 176–177 (2015) 513-521. [171] W. Yang, H. Liu, Y. Li, J. Zhang, H. Wu, D. He, Properties of yolk–shell structured Ni@SiO2 nanocatalyst and its catalytic performance in carbon dioxide reforming of methane to syngas, Catalysis Today, 259, Part 2 (2016) 438-445.

44

[172] X. Fang, S. Liu, J. Zang, C. Xu, M.-S. Zheng, Q.-F. Dong, D. Sun, N. Zheng, Precisely controlled resorcinol-formaldehyde resin coating for fabricating core-shell, hollow, and yolk-shell carbon nanostructures, Nanoscale, 5 (2013) 6908-6916. [173] K.P. de Jong, Synthesis of Solid Catalysts, Wiley VCG2009. [174] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catalysis, 4 (2014) 289-301. [175] S. Sokolov, E.V. Kondratenko, M.-M. Pohl, U. Rodemerck, Effect of calcination conditions on time on-stream performance of Ni/La2O3–ZrO2 in low-temperature dry reforming of methane, International Journal of Hydrogen Energy, 38 (2013) 16121-16132. [176] J. Chen, Q. Ma, T.E. Rufford, Y. Li, Z. Zhu, Influence of calcination temperatures of Feitknecht compound precursor on the structure of Ni–Al2O3 catalyst and the corresponding catalytic activity in methane decomposition to hydrogen and carbon nanofibers, Appl. Catal., A, 362 (2009) 1-7. [177] M.S. Aw, I.G. Osojnik Črnivec, P. Djinović, A. Pintar, Strategies to enhance dry reforming of methane: Synthesis of ceria-zirconia/nickel–cobalt catalysts by freeze-drying and NO calcination, International Journal of Hydrogen Energy, 39 (2014) 12636-12647. [178] M. Argyle, C. Bartholomew, Heterogeneous Catalyst Deactivation and Regeneration: A Review, Catalysts, 5 (2015) 145. [179] U. Lassi, Deactivation correlations of Pd/Rh three-way catalysts designed for EURO IV emission limits, University of Oulu, Oulu, 2003. [180] C.H. Bartholomew, Mechanisms of catalyst deactivation, Appl. Catal., A, 212 (2001) 17-60. [181] J.R. Rostrup-Nielsen, Production of synthesis gas, Catalysis Today, 18 (1993) 305-324. [182] J. Kong, H.T. Soh, A.M. Cassell, C.F. Quate, H. Dai, Synthesis of individual single-walled carbon nanotubes on patterned silicon wafers, Nature, 395 (1998) 878-881. [183] B.O. Boskovic, V. Stolojan, R.U.A. Khan, S. Haq, S.R.P. Silva, Large-area synthesis of carbon nanofibres at room temperature, Nat Mater, 1 (2002) 165-168. [184] D. Wang, G. Yang, Q. Ma, M. Wu, Y. Tan, Y. Yoneyama, N. Tsubaki, Confinement Effect of Carbon Nanotubes: Copper Nanoparticles Filled Carbon Nanotubes for Hydrogenation of Methyl Acetate, ACS Catal., 2 (2012) 1958-1966. [185] S. Iijima, Helical microtubules of graphitic carbon, Nature, 354 (1991) 56-58. [186] X. Pan, X. Bao, Reactions over catalysts confined in carbon nanotubes, Chem. Commun., (2008) 6271-6281. [187] L.S. Lobo, Carbon formation from hydrocarbons on metals Department of Chemical Engineering and Chemical Technology, Imperial College of Science and Technology, 1971. [188] L.S. Lobo, J.L. Figueiredo, C.A. Bernardo, Carbon formation and gasification on metals. Bulk diffusion mechanism: A reassessment, Catal. Today, 178 (2011) 110-116. [189] L.S. Lobo, D.L. Trimm, Complex Temperature Dependencies of the Rate of Carbon Deposition on Nickel, Nature Phys. Sci., 234 (1971) 15-16. [190] S. Helveg, C. López-Cartes, J. Sehested, P.L. Hansen, B.S. Clausen, J.R. Rostrup-Nielsen, F. Abild-Pedersen, J.K. Nørskov, Atomic-scale imaging of carbon nanofibre growth, Nature, 427 (2004) 426-429. [191] L.S. Lobo, J.L. Figueiredo, C.A. Bernardo, Carbon formation and gasification on metals. Bulk diffusion mechanism: A reassessment, Catalysis Today, 178 (2011) 110-116. [192] I. Luisetto, S. Tuti, E. Di Bartolomeo, Co and Ni supported on CeO2 as selective bimetallic catalyst for dry reforming of methane, Int. J. Hydrogen. Energ., 37 (2012) 15992-15999. [193] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, Ni-based catalysts for reforming of methane with CO2, Int. J. Hydrogen Energ., 37 (2012) 15966-15975. [194] T.H. Gardner, J.J. Spivey, E.L. Kugler, D. Pakhare, CH4–CO2 reforming over Ni-substituted barium hexaaluminate catalysts, Appl. Catal. A, 455 (2013) 129-136. [195] J. Guo, H. Lou, X. Zheng, The deposition of coke from methane on a Ni/MgAl2O4 catalyst, Carbon, 45 (2007) 1314-1321. [196] C.E. Daza, S. Moreno, R. Molina, Co-precipitated Ni–Mg–Al catalysts containing Ce for CO2 reforming of methane, Int. J. Hydrogen Energy, 36 (2011) 3886-3894.

45

[197] J.R. Rostrupnielsen, J.H.B. Hansen, CO2-Reforming of Methane over Transition Metals, J Catal., 144 (1993) 38-49. [198] A. Penkova, L. Bobadilla, S. Ivanova, M.I. Domínguez, F. Romero-Sarria, A.C. Roger, M.A. Centeno, J.A. Odriozola, Hydrogen production by methanol steam reforming on NiSn/MgO–Al2O3 catalysts: The role of MgO addition, Appl. Catal. A 392 (2011) 184-191. [199] Z. Zhang, X.E. Verykios, Carbon dioxide reforming of methane to synthesis gas over Ni/La2O3 catalysts, Appl. Catal. A 138 (1996) 109-133. [200] A. Kambolis, H. Matralis, A. Trovarelli, C. Papadopoulou, Ni/CeO2-ZrO2 catalysts for the dry reforming of methane, Appl. Catal. A 377 (2010) 16-26. [201] H. Ay, D. Üner, Dry reforming of methane over CeO2 supported Ni, Co and Ni–Co catalysts, Appl. Catal., B, 179 (2015) 128-138. [202] V. Sadykov, V. Muzykantov, A. Bobin, N. Mezentseva, G. Alikina, N. Sazonova, E. Sadovskaya, L. Gubanova, A. Lukashevich, C. Mirodatos, Oxygen mobility of Pt-promoted doped CeO2–ZrO2 solid solutions: Characterization and effect on catalytic performance in syngas generation by fuels oxidation/reforming, Catal. Today, 157 (2010) 55-60. [203] V.V. Galvita, H. Poelman, G.B. Marin, Combined chemical looping for energy storage and conversion, J. Power Sources, 286 (2015) 362-370. [204] A. Djaidja, H. Messaoudi, D. Kaddeche, A. Barama, Study of Ni–M/MgO and Ni–M–Mg/Al (M=Fe or Cu) catalysts in the CH4–CO2 and CH4–H2O reforming, Int. J. Hydrogen Energy, 40 (2015) 4989-4995. [205] Y.H. Taufiq-Yap, Sudarno, U. Rashid, Z. Zainal, CeO2–SiO2 supported nickel catalysts for dry reforming of methane toward syngas production, Applied Catalysis A: General, 468 (2013) 359-369. [206] K. Takanabe, K. Nagaoka, K. Nariai, K.-i. Aika, Titania-supported cobalt and nickel bimetallic catalysts for carbon dioxide reforming of methane, Journal of Catalysis, 232 (2005) 268-275. [207] D. Liu, X.Y. Quek, W.N.E. Cheo, R. Lau, A. Borgna, Y. Yang, MCM-41 supported nickel-based bimetallic catalysts with superior stability during carbon dioxide reforming of methane: Effect of strong metal–support interaction, Journal of Catalysis, 266 (2009) 380-390. [208] W.K. Jóźwiak, M. Nowosielska, J. Rynkowski, Reforming of methane with carbon dioxide over supported bimetallic catalysts containing Ni and noble metal: I. Characterization and activity of SiO2

supported Ni–Rh catalysts, Appl. Catal., A, 280 (2005) 233-244. [209] P. Frontera, A. Macario, A. Aloise, F. Crea, P.L. Antonucci, J.B. Nagy, F. Frusteri, G. Giordano, Catalytic dry-reforming on Ni–zeolite supported catalyst, Catalysis Today, 179 (2012) 52-60. [210] P. Frontera, A. Macario, A. Aloise, P.L. Antonucci, G. Giordano, J.B. Nagy, Effect of support surface on methane dry-reforming catalyst preparation, Catalysis Today, 218–219 (2013) 18-29. [211] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, C. Sener, T. Dogu, MCM-41 supported PdNi catalysts for dry reforming of methane, Applied Catalysis B: Environmental, 92 (2009) 250-261. [212] M. García-Diéguez, I.S. Pieta, M.C. Herrera, M.A. Larrubia, L.J. Alemany, Nanostructured Pt- and Ni-based catalysts for CO2-reforming of methane, J. Catal., 270 (2010) 136-145. [213] A.H. Fakeeha, W.U. Khan, A.S. Al-Fatesh, A.E. Abasaeed, Stabilities of zeolite-supported Ni catalysts for dry reforming of methane, Chinese Journal of Catalysis, 34 (2013) 764-768. [214] S.D. Abotteen, P., Effective use of catalysts through catalyst regeneration. [215] J.W. Oudar, H., Deactivation and Poisoning of Catalysts, 270 Madison Avenue, New York, New York. [216] J.B. Butt, Catalyst Poisoning and Chemical Process Dynamics, in: J.L. Figueiredo (Ed.) Progress in Catalyst Deactivation: Proceedings of the NATO Advanced Study Institute on Catalyst Deactivation, Algarve, Portugal, May 18–29, 1981, Springer Netherlands, Dordrecht, 1982, pp. 153-208. [217] L.L. Hegedus, R.W. McCabe, Catalyst poisoning, New York, 1984. [218] W.J.C. Niemantsverdriet, I., Concepts of Modern Catalysis and Kinetics, Wiley-VCH, Weinheim, 2003. [219] T.H. Gardner, D.A. Berry, K. David Lyons, S.K. Beer, A.D. Freed, Fuel processor integrated H2S catalytic partial oxidation technology for sulfur removal in fuel cell power plants, Fuel, 81 (2002) 2157-2166.

46

[220] B. Tjaden, M. Gandiglio, A. Lanzini, M. Santarelli, M. Järvinen, Small-Scale Biogas-SOFC Plant: Technical Analysis and Assessment of Different Fuel Reforming Options, Energy & Fuels, 28 (2014) 4216-4232. [221] J.R. Rostrup-Nielsen, Chemisorption of hydrogen sulfide on a supported nickel catalyst, J. Catal., 11 (1968) 220-227. [222] W.S. Jablonski, S.M. Villano, A.M. Dean, A comparison of H2S, SO2, and COS poisoning on Ni/YSZ and Ni/K2O-CaAl2O4 during methane steam and dry reforming, Appl. Catal., A, 502 (2015) 399-409. [223] R. Torbati, S. Cimino, L. Lisi, G. Russo, The Effect of Support on Sulphur Tolerance of Rh Based Catalysts for Methane Partial Oxidation, Catal Lett, 127 (2008) 260-269. [224] A. Bitsch-Larsen, N.J. Degenstein, L.D. Schmidt, Effect of sulfur in catalytic partial oxidation of methane over Rh–Ce coated foam monoliths, Applied Catalysis B: Environmental, 78 (2008) 364-370. [225] S. Cimino, R. Torbati, L. Lisi, G. Russo, Sulphur inhibition on the catalytic partial oxidation of methane over Rh-based monolith catalysts, Applied Catalysis A: General, 360 (2009) 43-49. [226] S. Cimino, L. Lisi, G. Russo, R. Torbati, Effect of partial substitution of Rh catalysts with Pt or Pd during the partial oxidation of methane in the presence of sulphur, Catalysis Today, 154 (2010) 283-292. [227] B.K. Sharma, Industrial Chemistry (Including Chemical Engineering), 1991. [228] N. Korens, D.R. Simbeck, D.J. Wilhelm, J.R. Longanbach, G.J. Stiegel, Process screening analysis of alternative gas treating and sulfur removal for gasification, 2002. [229] M.L. Pritchard, R.L. McCauley, B.N. Gallaher, W.J. Thomson, The effects of sulfur and oxygen on the catalytic activity of molybdenum carbide during dry methane reforming, Applied Catalysis A: General, 275 (2004) 213-220. [230] A. Erdőhelyi, K. Fodor, T. Szailer, Effect of H2S on the reaction of methane with carbon dioxide over supported Rh catalysts, Applied Catalysis B: Environmental, 53 (2004) 153-160. [231] G. Mancino, S. Cimino, L. Lisi, Sulphur poisoning of alumina supported Rh catalyst during dry reforming of methane, Catalysis Today. [232] Z.L. Zhang, V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, Reforming of Methane with Carbon Dioxide to Synthesis Gas over Supported Rhodium Catalysts: I. Effects of Support and Metal Crystallite Size on Reaction Activity and Deactivation Characteristics, Journal of Catalysis, 158 (1996) 51-63. [233] M. Behrens, Ki, F. Girsgdies, I. Kasatkin, F. Hermerschmidt, K. Mette, H. Ruland, M. Muhler, R. Schlogl, Knowledge-based development of a nitrate-free synthesis route for Cu/ZnO methanol synthesis catalystsviaformate precursors, Chem. Commun., 47 (2011) 1701-1703. [234] R.M.J. Fiedorow, B.S. Chahar, S.E. Wanke, The sintering of supported metal catalysts: II. Comparison of sintering rates of supported Pt, Ir, and Rh catalysts in hydrogen and oxygen, J. Catal., 51 (1978) 193-202. [235] S.E. Wanke, P.C. Flynn, The Sintering of Supported Metal Catalysts, Catal. Rev., 12 (1975) 93-135. [236] C.H. Bartholomew, J.R. Farrauto, Fundamentals of Industrial Catalytic Processes, 2nd ed., John Wiley and Sons, Hoboken, NJ,2006. [237] C.H. Bartholomew, Sintering and redispersion of supported metals: Perspectives from the literature of the past decade, in: C.H. Bartholomew, G.A. Fuentes (Eds.) Stud. Surf. Sci. Catal., Elsevier1997, pp. 585-592. [238] C.H. Bartholemew, Model catalyst studies of supported metal sintering and redispersion kinetics, in: J.J. Spivey, S.K. Agarwal (Eds.) Catalysis: Volume 10, The Royal Society of Chemistry1993, pp. 41-82. [239] V.A. Tsipouriari, A.M. Efstathiou, Z.L. Zhang, X.E. Verykios, Reforming of methane with carbon dioxide to synthesis gas over supported Rh catalysts, Catalysis Today, 21 (1994) 579-587. [240] B.R. Powell, S.E. Whittington, Encapsulation: A new mechanism of catalyst deactivation, Journal of Catalysis, 81 (1983) 382-393. [241] L.S. Lobo, Intrinsic kinetics in carbon gasification: Understanding linearity, “nanoworms” and alloy catalysts, Applied Catalysis B: Environmental, 148–149 (2014) 136-143.

47

[242] L.S. Lobo, Catalytic Carbon Gasification: Review of Observed Kinetics and Proposed Mechanisms or Models - Highlighting Carbon Bulk Diffusion, Catalysis Reviews, 55 (2013) 210-254. [243] N. Zong, Y. Liu, Learning about the mechanism of carbon gasification by CO2 from DSC and TG data, Thermochim. Acta, 527 (2012) 22-26. [244] W. Gac, A. Denis, T. Borowiecki, L. Kępiński, Methane decomposition over Ni–MgO–Al2O3 catalysts, Appl. Catal., A, 357 (2009) 236-243. [245] A.S.A. Al–Fatish, A.A. Ibrahim, A.H. Fakeeha, M.A. Soliman, M.R.H. Siddiqui, A.E. Abasaeed, Coke formation during CO2 reforming of CH4 over alumina-supported nickel catalysts, Appl. Catal. A, 364 (2009) 150-155. [246] C.A. Querini, S.C. Fung, Coke characterization by temperature programmed techniques, Catal. Today, 37 (1997) 277-283. [247] K.M. Hardiman, C.G. Cooper, A.A. Adesina, R. Lange, Post-mortem characterization of coke-induced deactivated alumina-supported Co–Ni catalysts, Chem. Eng. Sci., 61 (2006) 2565-2573. [248] J.M. Kanervo, A.O.I. Krause, J.R. Aittamaa, P.H. Hagelberg, K.J.T. Lipiäinen, I.H. Eilos, J.S. Hiltunen, V.M. Niemi, Kinetics of the regeneration of a cracking catalyst derived from TPO measurements, Chem. Eng. Sci., 56 (2001) 1221-1227. [249] J.L. Figueiredo, D.L. Trimm, Gasification of carbon deposits on nickel catalysts, J. Catal., 40 (1975) 154-159. [250] A. Tomita, Y. Tamai, Hydrogenation of carbons catalyzed by transition metals, J. Catal., 27 (1972) 293-300. [251] A. Tomita, Catalysis of Carbon–Gas Reactions, Catal. Surv. Jpn., 5 (2001) 17-24. [252] C.A. Bernardo, D.L. Trimm, The kinetics of gasification of carbon deposited on nickel catalysts, Carbon, 17 (1979) 115-120. [253] E. Baumgarten, A. Schuck, Oxygen spillover and its possible role in coke burning, Appl. Catal., 37 (1988) 247-257. [254] F. Kapteijn, J.A. Moulijn, Kinetics of Catalysed and Uncatalysed Coal Gasification, in: J. Figueiredo, J. Moulijn (Eds.) Carbon and Coal Gasification, Springer Netherlands1986, pp. 291-360. [255] D. Fino, N. Russo, C. Badini, G. Saracco, V. Specchia, Effect of active species mobility on soot-combustion over Cs-V catalysts, AICHE J., 49 (2003) 2173-2180. [256] B.R. Stanmore, V. Tschamber, J.F. Brilhac, Oxidation of carbon by NOx, with particular reference to NO2 and N2O, Fuel, 87 (2008) 131-146. [257] A. Tomita, Y. Tamai, Hydrogenation of carbons catalyzed by transition metals, Journal of Catalysis, 27 (1972) 293-300. [258] E.A. Heintz, W.E. Parker, Catalytic effect of major impurities on graphite oxidation, Carbon, 4 (1966) 473-482. [259] R.T. Rewick, P.R. Wentrcek, H. Wise, Carbon gasification in the presence of metal catalysts, Fuel, 53 (1974) 274-279. [260] W. Karim, C. Spreafico, A. Kleibert, J. Gobrecht, J. VandeVondele, Y. Ekinci, J.A. van Bokhoven, Catalyst support effects on hydrogen spillover, Nature, 541 (2017) 68-71. [261] M. Machida, Y. Murata, K. Kishikawa, D. Zhang, K. Ikeue, On the Reasons for High Activity of CeO2 Catalyst for Soot Oxidation, Chem. Mater., 20 (2008) 4489-4494. [262] J. Guo, H. Lou, H. Zhao, D. Chai, X. Zheng, Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels, Appl. Catal. A 273 (2004) 75-82. [263] X.E. Verykios, Catalytic dry reforming of natural gas for the production of chemicals and hydrogen, Int. J. Hydrogen Energy, 28 (2003) 1045-1063.

Chapter 2: Experimental procedures

49

Chapter 2

Experimental procedures

This chapter introduces the main experimental strategy followed in this Thesis. The

heterogeneous catalysts that were used in this work, were synthesized in lab scale setups

using three different methods: 1) co-precipitation, 2) incipient wetness (co)-impregnation

and 3) sequential impregnation. For more details of the synthesis methods the reader is

referred to § 1.4. All the synthesized materials were characterized in different states i.e. “as-

prepared, “reduced”, “spent” and “re-oxidized” and tested under different reaction

conditions (temperature, pressure and feed composition) in a fixed-bed plug flow reactor.

The use of in-situ characterization techniques allowed to relate the catalytic performance

with the evolution of catalyst structure. That proved essential in order to unravel the

reaction mechanisms.

2.1 Synthesis of heterogeneous catalysts

2.1.1 Support preparation

One-pot synthesis was applied for the preparation of the support materials (Table 2.1.1).

The nitrate precursors, Mg(NO3)2·6H2O (99%, Sigma-Aldrich®), Al(NO3)3·9H2O (98.5%, Sigma-

Aldrich®) and Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®), were mixed in an aqueous solution.

NH4OH (ACS reagent, 28.0-30.0% NH3 basis) was added as precipitating agent to the

precursors’ solution to adjust the pH to 10, at 333 K. The produced precipitate was filtered,

dried in air at 393 K for 12 h and subsequently calcined in air for 1023 K for 4 h. The supports

were ground in order to obtain a particle size ≤ 50 μm.

For the synthesis of MgAl2O4 the Mg/Al molar ratio was ½, while in case of MgFexAl2-xO4,

Fe was incorporated in the magnesium aluminate spinel lattice. The Fe content on MgFexAl2-

xO4 was 0 wt%, 0.5wt%, 1.0wt%, 2.5 wt%, 5.0 wt%, 7.5 wt%, 10 wt% and 15 wt%. The

supports will be labeled as zFe, where z is the wt% of Fe in the magnesium aluminate lattice.


50

These iron contents were chosen based on the study of Dharanipragada et al. [1] in order to

prevent formation of a separate Fe2O3 phase.

Table 2.1.1: Overview of synthesized support materials.

Abbreviation Support

0Fe MgAl2O4

0.5Fe MgFe0.007Al1.993O4

1.0Fe MgFe0.013Al1.987O4

2.5Fe MgFe0.04Al1.96O4



10Fe MgFe0.18Al1.82O4

15Fe MgFe0.26Al1.74O4

Al2O3 Al2O3

2.1.2 Active phase preparation

The incipient wetness (co)-impregnation method was applied for the preparation of the

synthesized catalysts [2], named as “as-prepared”. A catalyst library including monometallic,

bimetallic and tri-metallic samples was prepared (Table 2.2) using as precursors an aqueous

solution of the corresponding nitrates (Ni(NO3)2·6H2O (99.99+%, Sigma-Aldrich®),

Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®), Rh(NO3)3.xH2O (36% Rh basis, Sigma-Aldrich®),

Pd(NO3)2·2H2O nitrate (~40% Pd basis, Sigma-Aldrich®)). NH4OH (ACS reagent, 28.0-30.0%

NH3 basis) was added as precipitating agent to the precursors’ solution to adjust the pH to

10, at 333 K. The produced precipitate was filtered, dried in air at 393 K for 12 h and

subsequently calcined (typically at 1023 or 1073 K) for 4 h.

Table 2.2: Overview of synthesized catalysts.

Abbreviation Catalyst Used in

Chapter

Ni-αFe/MgAl2O4 (Monometallic and Bimetallic) 3 and 4

Ni-0Fe/MgAl 8wt.%Ni/MgAl2O4 ״

Ni-5Fe/MgAl 8wt.%Ni-5wt%Fe/MgAl2O4 ״


51

Ni-8Fe/MgAl 8wt.%Ni-8wt%Fe/MgAl2O4 ״

Ni-11Fe/MgAl 8wt.%Ni-11wt%/MgAl2O4 ״

Ni-Fe-βPd/MgAl2O4 (Bimetallic and Tri-metallic) 5

0-Pd 8wt.%Ni-4wt.%Fe/MgAl2O4 ״

0.1-Pd 8wt.%Ni-4wt.%Fe-0.1wt.%Pd/MgAl2O4 ״

0.2-Pd 8wt.%Ni-4wt.%Fe-0.2wt.%Pd/MgAl2O4 ״

0.4-Pd 8wt.%Ni-4wt.%Fe-0.4wt.% Pd/MgAl2O4 ״



Ni/zFe (Monometallic and Bimetallic) 6

Ni/0Fe 8wt%Ni/MgAl2O4 ״

Ni/0.5Fe 8wt%Ni/MgFe0.007Al1.993O4 ״






Ni/15Fe 8wt%Ni/MgFe0.26Al1.74O4 ״

zPd/0Fe (Monometallic) 5

mono-0.1Pd 0.1wt%Pd/ MgAl2O4 ״



Additional materials were synthesized either for better resolution when some

characterization techniques were used (i.e. XRD, EXAFS) or for comparison purposes (Table

2.3).

Table 2.3: Overview of samples synthesized for comparison purposes

Abbreviation Catalyst Used in Chapter

5wt.%Fe 5wt.%Fe/MgAl2O4 3

Ni-3Pd 10wt%Ni-3wt%Pd/MgAl2O4 5

Fe-3Pd 5wt%Fe-3wt%Pd/MgAl2O4 5


52

Ni-Fe-3Pd 10wt%Ni-5wt%-3wt%Pd/MgAl2O4 5

1.0Rh/0Fe 1.0wt%Rh/ MgAl2O4 6

2.2 Catalyst characterization methods

2.2.1 Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES)

ICP-AES measurements were carried out in order to obtain a quantitative elemental

analysis of each synthesized catalyst. The samples were mineralized either by an alkaline

fusion with a mix of Li-tetraborate and Li-metaborate or by acid fusion. A peristaltic pump

introduced the sample into a nebulizer and then in an Ar-plasma torch at ca. 8000 K (ICP-

AES, ICAP 6500, Thermo Scientific) after which ionization of the sample results in the

emission of light with characteristic peaks. The intensity of these characteristic peaks are a

measure for the concentration of the respective element in the sample, according to

Lambert-Beer’s law (Eq. 2.1).

𝐼𝑡 = 𝐼0𝑒−𝜇(𝐸)𝑟 ↔ ln (𝐼𝑡

𝐼0) = −𝜇(𝐸)𝑟 (2.1)

Where It is the measured intensity, I0 the measured reference intensity, μ(E) the absorption

coefficient [m-1] and r the path length of light [m].

The plasma flame was maintained by a strong electromagnetic field within a radio

frequency coil and continuously atomized the sample molecules, exciting the outer-shell

electrons of the separated atoms to a higher state. When falling back to the ground state,

those electrons emit characteristic radiation. This passes through a slit onto a mirror,

continuing to a grating which separates the radiation for the photomultipliers. The intensity

of the measured emission lines was compared with standards, thus allowing the calculation

of specific elements concentration [3].

2.2.2 Surface area and porosity

Surface area and porosity measurements were carried out on the catalyst samples. The

Brunauer-Emmett-Teller (BET) area (5 p/p0 points), Barrett-Joyner-Halenda (BJH) average

pore volume and pore diameter (15 p/p0 points), and Harkins and Jura pore size distribution


53

(ca. 40 pore width points) were measured in the Micromeritics Tristar-II Gemini using

nitrogen as adsorbate gas at 77 K. Prior to the measurements the samples were outgassed at

473 K for 2 h. The average pore volume and the average pore size were determined upon

desorption of nitrogen gas at 77 K by the BJH method. This method assumes the catalyst can

be modeled as a collection of parallel cylindrical pores. The Kelvin equation for capillary

condensation is used, assuming a hemispherical liquid-vapor meniscus and a well-defined

surface tension. The BJH theory uses a reference isotherm to account for thinning of the

adsorbed layer; the Kelvin equation is only applied to the ‘core’ fluid [4]. For the complete

pore diameter distribution, a Harkins and Jura adsorption model for slit pores was assumed.

For a more detailed description of the Harkins and Jura model, the reader is referred to

chapter 7 of Powder Surface Area and Porosity of S. Lowell and J.E. Shields [4].

2.2.3 Ex-situ X-ray diffraction (XRD)

The crystallographic phases of the materials in the “as prepared” state were confirmed by

ex-situ powder XRD measurements (Siemens Diffractometer Kristalloflex D500, Cu Kα

radiation). XRD is the most widely used method to identify the crystalline phases in a

material. This apparatus irradiates a monochromatic Cu Kα radiation of 0.154 nm on a 10-20

mg catalyst sample evenly spread out on a Si wafer. The XRD patterns were collected in a 2θ

range from 10o to 80o with a step of 0.02o and 30 s counting time per angle. XRD patterns of

known compounds are referenced by their corresponding number in the Powder Diffraction

File (PDF) database. By fitting a Gaussian function to a diffraction peak, the crystallite size

was determined from the peak width via the Scherrer equation (Eq. 2.2) [5], while the peak

position gave information about the lattice spacing based on the Bragg law of diffraction (Eq.

2.3) [6, 7].

𝑑𝑋𝑅𝐷 =𝐾𝜆

(𝛽 − 𝑏) ∙ 𝑐𝑜𝑠𝜃

(2.2)

𝑛𝜆 = 2𝑑𝑠𝑖𝑛𝜃 𝑤𝑖𝑡ℎ 𝑛 = 1,2,3, … (2.3)

Where dXRD is the crystallite size, K is a constant (0.9), λ is the incoming X-ray wavelength in

nm (0.154), β is the line broadening at half maximum intensity, b is in the instrumental

width, θ the incidence angle, n is the order where diffraction occurs, d is the interatomic

spacing. The instrumental width is calculated after a calibration curve using LaB6. It is


54

important to note that the Scherrer formula provides a lower bound on particle size; β is

usually influenced by the presence of defects and imperfections in the crystal lattice [7]. The

exposed fraction of the catalyst can be calculated with the following empirical equation:

𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑒𝑥𝑝𝑜𝑠𝑒𝑑 [%] =1

𝑑𝑋𝑅𝐷 [𝑛𝑚]∙ 100 (2.4)

This empirical formula assumes, for example, that a crystallite size of 2 nm corresponds to

an exposed fraction of 50%.

2.2.4 In-situ X-ray diffraction (XRD)

In-situ XRD measurements were performed in a reactor inside a Bruker-AXS D8 Discover

apparatus (Cu Kα radiation of 0.154 nm). The reactor had a Kapton foil window for X-ray

transmission. The setup was equipped with a linear detector covering a range of 20o in 2θ

with an angular resolution of 0.1o. The pattern acquisition time was 10 s. All temperatures

were measured with a K-type thermocouple and corrected afterwards according to a

calibration curve of the heating device, which is based on the eutectic systems Au–Si, Al–Si

and Ag–Si. For each sample, approximately 10-20 mg of catalyst were evenly spread on a

single crystal Si wafer. Interaction of the catalyst material with the Si wafer was never

observed. Prior to each experiment the reactor chamber was evacuated to a base pressure

of 4 Pa by a rotation pump. Gases were supplied to the reactor chamber with calibrated

mass-flow controllers. He (10 NmL/s) was flowing for 10 min before the flow was switched

to 5%H2/He or CO2 for Temperature Programmed Reduction (TPR) and Oxidation (TPO)

experiments (10 NmL/s, up to 1123 K, using a heating rate typically of 15-30 K/min),

respectively. After each in-situ measurement, a full scan pattern was obtained at room

temperature, in-situ, for better resolution, without removing the sample from the chamber.

2.2.5 X-ray absorption spectroscopy (XAS)

X-ray absorption near edge structure (XANES)

The experimental campaign was performed in the SOLEIL synchrotron, where the

electrons are centripetally accelerated in a magnetic field, resulting in polychromatic X-

radiation tangent on the storage ring with energies between 120 eV and 120 keV. The


55

incident beam intensity I0 attenuates according Lambert-Beer’s law integrated for a catalyst

sample of thickness r (Eq. 2.1), resulting in transmitted intensity It.

The absorption coefficient μ(E) decreases for increasing X-rays energies except when it is

equal to the binding energy Ebinding of a bound electron. In that case, the electron is knocked

out of its ground state. For any higher energy, the knocked out electron will have the kinetic

energy shown in Eq. 2.5.

𝐸𝑘 = ℎ𝜈 − 𝐸𝑏𝑖𝑛𝑑𝑖𝑛𝑔 (2.5)

This ‘absorption edge’ gives information about the electrons bound to a nucleus of the

catalyst/material, and therefore about its oxidation state. Due to the weak temperature

dependence and the short energy range of XANES, this technique is well suited to monitor

the in-situ catalytic reactions [8].

Extended X-ray absorption fine structure (EXAFS)

Each outgoing photo-electron of the bombarded anode material can be seen as a

spherical wave. This outgoing wave will be backscattered by neighboring atoms. The final

wave state function will be a result of interference of the outgoing wave state function and

the backscattered wave state function (Eq. 2.6).

𝜓𝑓𝑖𝑛𝑎𝑙 = 𝜓𝑜𝑢𝑡𝑔𝑜𝑖𝑛𝑔 + 𝜓𝑏𝑎𝑐𝑘𝑠𝑐𝑎𝑡𝑡𝑒𝑟𝑒𝑑 (2.6)

This results in a fine structure (EXAFS) in the X-ray absorption coefficient after the

absorption edge. The simplest theory to explain EXAFS is based on the single scattering

plane-wave approximation, viewing the electron wave as a plane wave where the electron is

assumed to scatter only once before it returns to the absorber atom. The oscillations are

then a summation over all coordination shells (Eq. 2.7).

𝜒(𝑘) = ∑ 𝐴𝑗(𝑘)

𝑆ℎ𝑒𝑙𝑙𝑠

𝑗=1

𝑠𝑖𝑛𝛷𝑗(𝑘) (2.7)

A coordination shell is defined by the identity of the scattering atom and the distance

from the absorber atom. The fact that χ(k) is a summation over all coordination shells

around a specific absorbing element illustrates that EXAFS is a bulk technique, which

provides information about the average coordination around the absorber atom.


56

In order to be able to model the EXAFS, model equations can be used for the amplitude

Aj(k) and the sine argument Φj(k). For more information on those model equations, the

reader is referred to the paper of Koningsberger [9]. By interpreting the EXAFS data using

those model equations, EXAFS can provide structural information such as the coordination

number of the central atom, distances between central and neighboring atoms and the

atomic number of neighboring atoms.

SOLEIL experimental setup and data analysis

In-situ and operando-QXAS experiments were carried out at the SOLEIL synchrotron ROCK

beamline (4-40 keV), which consists of a storage line providing the X-rays, an oscillating

monochromator, entrance slits, a first ion-chamber detecting the incoming beam I0, the

sample, a second ion chamber detecting the transmitted beam It, a reference metal foil

(either Fe or Ni) and a third ion chamber measuring a reference intensity.

The entrance slits reduced the radiation beam from the synchrotron to the desired spot

size. The monochromator oscillated over both Fe-K and Ni-K edges (7112 eV and 8331 eV)

using a macro for fast switching of edges.

The detection of X-rays by the ion-chambers is based on ionization of inert gases by

photons. The as-prepared material, 50% diluted by Boron nitride, was inserted into a 2mm

quartz capillary reactor and fixed by quartz wool plugs. The capillary reactor was

implemented in a frame which was connected to gas feed lines through Swagelok fittings

while the inlet gas flow rates were maintained by means of calibrated mass flow controllers.

Reference material pellets (FeO) were measured ex-situ. The transmission measurements

were coupled with kinetic measurements based on gas-phase compositions obtained from

MS during DRM. An external calibrated heat gun was used in order to reach the reaction

temperature of 1023 K, while a total flow between 10-60 mL/min was employed. Gas rig,

MS, capillary frame and heat blower were used from SOLEIL synchrotron.

In-situ XANES was performed during H2 temperature programmed reduction (TPR) (0.2

NmL/s of 5%H2/He using a heating ramp of 10 K/min) in a capillary reactor with an internal

diameter of 1mm in order to investigate the structural evolution of Ni/5.0Fe and Ni-

5Fe/MgAl. The capillary reactor was implemented in a frame which was connected to gas

feed lines [10] through Swagelok fittings while the inlet gas flow rates were maintained by


57

means of calibrated Brooks mass flow controllers. After reduction, in-situ XANES was carried

out as the sample was subjected to re-oxidation by CO2 (named “re-oxidized”), without

removing it from the reactor.

2.2.6 Electron microscopy techniques

Electron microscopy is one of the most important tools in order to visualize the

morphology of heterogeneous catalysts from micrometers down to atomic scale [11]. When

the electron beam irradiates a specimen surface, the interaction between the incident

electrons and atoms that compose the specimen generates various signals i.e. secondary

electrons, backscattered electrons, Auger electrons, characteristic x-rays,

cathodoluminescence, absorption current, transmitted electrons and other signals (Figure

2.1). Morphological examination can be carried out by exploiting these various signals.

Figure 2.1: Signals generated from an incident beam on a specimen. Reproduced from [12]

Scanning electron microscopy (SEM)- Energy Dispersive X-ray spectroscopy (EDX)

Scanning electron microscopy (SEM) produces an image by scanning it with a focused

electron beam. The difference between SEM and TEM is that the sample used in SEM does

not need to be very thin, since transmission is not used [13].

SEM (FEI Quanta 200F) was performed to evaluate the surface morphology of the studied

materials. An acceleration voltage of 25 or 12.5 kV was chosen for the secondary electron

SEM images. EDX spectroscopy (EDAX inc.) at 17.5 kV acceleration voltage provided a first


58

compositional analysis, typically in a 1-μm-thick volume. These measurements were carried

out at Department of Solid State Sciences, Ghent University by Olivier Janssens.

Transmission electron microscopy (TEM)

High angle annular dark field Scanning Transmission Electron Microscopy (HAADF-STEM),

Energy Dispersive X-ray (EDX) spectroscopy and Electron Energy-Loss Spectroscopy (EELS)

experiments were performed with a FEI Titan “cubed” electron microscope equipped with

an aberration corrector for the probe-forming lens, a high resolution EELS spectrometer

(Gatan GIF Quantum) and a wide solid angle “super-X” EDX detector, at 120 kV and 300 kV

acceleration voltage. The STEM convergence semi-angle used was ~22 mrad, while the inner

collection semi-angle for HAADF-STEM imaging at 300kV was ~50 mrad and at 120kV ~85

mrad. The measurements were carried out at the laboratory Electron Microscopy for

Materials Science (EMAT), University of Antwerp, by Dr. Maria Meledina and the results have

been documented in Chapter 6.

Further in Chapter 3, 4 and 5 the TEM measurements were performed using a JEOL JEM-

2200FS: Cs-corrected microscope at Department of Materials, Textiles and Chemical

Engineering (MaTCh), Ghent University, operated at 200 kV, equipped with a Schottky-type

field-emission gun (FEG), EDX JEOL JED-2300D and JEOL in-column omega filter (EELS).

Specimens were prepared by immersion of a lacey carbon film on copper support grid and

particles sticking to the grid were investigated. Elemental mapping was performed by EDX. A

beryllium specimen retainer was used to remove secondary X-ray fluorescence. These

measurements have been performed by Dr. Vitaliy Bliznuk (Department of Materials Science

and Engineering, Ghent University) and Ir. Lukas Buelens (Laboratory for Chemical

Technology, Ghent University).

2.2.7 X-ray Photoelectron Spectroscopy

XPS measurements were recorded with a S-Probe XPS spectrometer (VG, Surface Science

Instruments), equipped with a monochromatized 450W Al Kα source. The base pressure of

the analysis chamber was below 2∙10−7 Pa. Spectra were recorded with 200W source power.

The analyser axis made an angle of 45◦ with the specimen surface. Wide scan spectra were

measured with a pass energy of 157 eV and a 0.2 eV step, while core levels were recorded


59

with a step of 0.05 eV and a pass energy of 107.8 eV. Energy calibration to Al2p or C1s gave

the same result, so the main intensity of C1s at 284.6 eV was used for alignment.

2.2.8 RAMAN spectroscopy

Raman spectroscopy is a scattering technique. It is based on the Raman Effect, i.e. the

frequency of a small fraction of scattered radiation is different from frequency of

monochromatic incident radiation [14]. The laser light interacts with molecular vibrations,

phonons or other excitations in the system, resulting in the energy of the laser photons

being shifted up or down. The shift in energy gives information about the vibrational modes

in the system. Infrared spectroscopy yields similar, but complementary, information. Raman

analysis of the samples was performed on a RXN1 Raman spectrometer (Kaiser Optical

Systems) fitted with a 532 nm wavelength laser operating at 40 mW.

2.3 Activity and stability tests

Activity and stability tests were carried out in a plug flow reactor (i.d. 9 mm and length

460 mm) placed in an electric furnace. Quartz reactors were used for the experiments at

atmospheric pressure while a coated (by amorphous silicon) incoloy alloy 800HT reactor was

used for the high pressure experiments. The activity of the incoloy alloy reactor was always

tested in a blank experiment, at the reaction conditions, before each measurement. Typically

8-20 mg of catalyst (pellet size of ~50 μm) was homogeneously diluted with 50-125 μm-

diameter α-Al2O3 and placed onto a quartz wool plug. The catalyst bed temperature was

measured by K-type thermocouples touching the inside and outside of the reactor at the

height of the catalyst bed.

The setup can be divided broadly into three sections: feed, reactor and analysis sections

(Figure 2.2). A detailed flow diagram along with the list of symbols can be found in Appendix

B. The advantage of the setup is that it allows to choose between various feed gases or

liquids (that will be vaporized) and an inert flow through the reactor. Thus, it provides

flexibility to perform reactions under various feed specifications without changing the

reactor conditions of temperature and pressure. The main advantage of this setup is fast

switching between different feed streams through a fast switching four way valve, enabling


60

the users to simulate a “step response” or an “alternate pulse” experiment in the desired

feed conditions.

Figure 2.2: The overview of the step response setup with the feed, reactor and analytical sections.

Incoming gas flow rates were controlled by Bronkhorst mass flow controllers (MFCs).

Calibration curves and conversion factors for the mass flow controllers were used to convert

the MFC set-point to the actual volumetric flow. Prior to each activity measurement, an

activation step for the catalyst was taking place. This included the catalyst reduction under a

gas flow of diluted hydrogen (typically 1 NmL/s of 10-20vol% H2/Ar, using a heating ramp of

15-20 K/min up to 1123 K and a dwell time of 30 min). Afterwards the flow was switched to

inert gas (1 NmL/s of Ar or He) for 5-10 min. Subsequently, the catalyst activity during

reforming reactions (dry reforming, bi-reforming and steam reforming of methane) was

Feed

Analytical: MS

Reactor


61

measured at different conditions of temperature, pressure and feed composition. The outlet

gas stream was monitored online using a ThermoStar Pfeiffer mass spectrometer (MS). H2,

He, CH4, H2O, CO, Ar and CO2 were respectively measured at an atomic mass of 2, 4, 16, 18,

28, 40 and 44. A correction was applied to remove contributions from unavoidable

interference with fragmentation peaks of other gases. For example, CO2 (44 amu) fragments

within the MS chamber partially into CO (28 amu) and O (16 amu). Therefore, a correction

was applied to remove these contributions: the CO signal was subtracted by 7% of the CO2

signal while the CH4 signal was subtracted by 12% of the CO2 signal. A carbon balance with a

deviation of 4-8% was obtained from the performed experiments.

The MS calibration factors were calculated prior to each independent experiment

according to equation (Eq. 2.8):

𝐶𝐹𝑖 =

𝑆𝑖,𝑅𝑇

𝐹𝑖,𝑅𝑇

𝑆𝑖.𝑠.

𝐹𝑖.𝑠.

𝑓𝑜𝑟 𝑖 = 𝐶𝐻4, 𝐶𝑂2 𝑒𝑡𝑐. (2.8)

Where Si,RT and Si.s. are the references of MS signal at room temperature, using an inert bed,

for compound i and internal standard (Ar or He), respectively, Fi,RT and Fi.s. are the

volumetric flow rates at room temperature for compound i and internal standard,

respectively.

The following expressions are used to determine the activity of different catalysts. The

percent conversion for a reactant is calculated as:

𝑋𝑖 = 𝐹𝑖

0−𝐹𝑖

𝐹𝑖0 · 100% (2.9)

where 𝐹𝑖0 and 𝐹𝑖 are the inlet and outlet molar flow rates of gas (mol·s-1).

The CH4 consumption rate (mmol·s-1·gmetals-1, Eq. 2) was calculated from the difference

between the inlet and outlet molar flow rates, as measured relative to an internal standard

(Ar), while the space time yield (STY, mmol·s-1·gmetals-1, Eq. 3) was calculated for the products:

CH4 consumption rate = |𝐹𝑖

0−𝐹𝑖|

𝑚𝑚𝑒𝑡𝑎𝑙𝑠 (2.10)

STYi = 𝐹𝑖

𝑚𝑚𝑒𝑡𝑎𝑙𝑠 (2.11)


62

where 𝐹𝑖0and 𝐹𝑖 are the inlet and outlet molar flow rates of gas i (mol·s-1), and mmetals the

amount of metals (g) present in the catalyst after ICP analysis.

To ensure the plug flow regime, the absence of pressure drop as well as the absence of

mass transfer limitations the diagnostic criteria presented below were used. However, radial

heat transfer limitations remained an issue during the performed activity tests showing

radial temperature gradient in the range of 20-29 K. This temperature gradient was taken

into account for the calculation of the equilibrium conversion. More details about it can be

found below.

2.3.1 Absence of pressure drop

When the fluids flowing through the reactor are incompressible, gravitational forces can

be neglected and the Fanning definition of viscous losses can be implemented. Integrating

the momentum balance along the entire reactor length then leads to Eq. 2.12.

∆𝑝

𝐿= 0.5 ∙ 𝑓𝑓 ∙ 𝑎𝑣 ∙ 𝜌⟨𝑢⟩2 (2.12)

Δp is the pressure drop along the length L of the reactor, <u> is the local flow velocity, ρ is

the fluid density, and A is the reactor cross-sectional surface area. av is thus the available

pellet surface area divided by its total pellet volume (Eq. 2.13):

𝑎𝑣 =4

𝑑𝑝 (2.13)

In literature, generally a modified friction factor fm is reported, as shown in Eq. 2.14 [15].

𝑓𝑚 = 2 ∙ 𝑓𝑓 (2.14)

Several correlations can be found in literature for fm, used in Eq. 2.14, when assuming a

packed bed with a particle size distribution. For each of these correlations, the conditions of

validity are verified. Subsequently, the log-average of Δp is taken for all these correlations.

The criterion for absence of pressure drop is then given by Eq. 2.15 for a first order reaction.

𝛥𝑝 < 0.2 ∙ 𝑝𝑡𝑜𝑡 (2.15)

Where ptot is the total pressure in the reactor.


63

2.3.2 Ideal plug flow

Axial effective diffusion

Axial dispersion can be assumed if Eq. 2.16 is valid [16, 17].

𝐿

𝑑𝑝>

8

𝐵𝑜𝑛 ∙ ln (

1

1 − 𝑋𝐶𝐻4) (2.16)

XCH4 is the conversion of CH4. In order to calculate Bo = u·L/DCH4,ax, the diffusion coefficient

DCH4,ax is calculated as a log average from a set of correlations from literature within their

validity range [18-20]. u is the flow velocity.

Radial diffusion

A simple criterion for verification of the radial dispersion in a packed bed was used (Eq. 2.17)

[17].

𝑑𝑡

𝑑𝑝> 8 (2.17)

Where dt is the tube diameter. The criterion predicts when the wall effects will cause a radial

velocity profile, or in other words when there will be bypass of gas through the voids near

the reactor wall. A more sophisticated criterion that takes into account the influence of CH4

conversion is shown in Eq.16. Pe’ is the modified Péclet number (Eq. 2.18):

𝑃𝑒 ∙ (2 · 𝑑𝑡

𝐿)

2

=𝑢 · 𝐿

𝐷𝑒𝑟∙ (

2 · 𝑑𝑡

𝐿)

2

(2.18)

Where Der is the effective radial diffusivity, a and b are constants (Eq. 2.19).

ln (1

1 − 𝑋𝐶𝐻4) < 𝑎 +

𝑏

𝑃𝑒′ (2.19)

Dilution

The ratio of inert material to the total volume is denoted as b [m³inert·m-3

inert+cat]. It is

demanded that the maximum accepted deviation Δ, shown in Eq. 2.20, is 0.05. This is

equivalent Eq. 2.21, where xdil is the fraction diluted material [21].

𝛥 = (𝑏

1 − 𝑏)

𝑥𝑑𝑖𝑙 ∙ 𝑑𝑝

2 ∙ 𝐿 (2.20)


64

𝑏 <1

1 + 10 ∙ 𝑥𝑑𝑖𝑙 ∙ (𝑑𝑝/𝐿) (2.21)

Note that for higher amounts of dilution material, the probability for zones containing

relatively more dilution than catalyst will also be higher and this can negatively affect the

conversion.

2.3.3 Mass and heat transport limitations

Dimensionless criteria were applied for evaluating the significance of external and

internal heat and mass transfer limitations.

External heat transport limitations

From a balance over the film around the pellet, the maximal temperature difference can be

calculated using Eq. 2.22 [22].

𝛥𝑇(𝑓𝑖𝑙𝑚) =𝑅𝑣,𝐶𝐻4

𝑜𝑏𝑠 ∙ |∆𝑟𝐻| ∙ 𝑑𝑝

6 ∙ 𝛼𝑝<

0.05 ∙ 𝑅 ∙ 𝑇𝑓2

𝐸𝑎 (2.22)

Where ΔrH is the reaction enthalpy [kJ/mol], 𝑅𝑣,𝐶𝐻4𝑜𝑏𝑠 is the observed volumetric reaction

rate of CH4 [molCH4/(m3cat·s], αp is the heat transfer coefficient between pellet and gas in the

bulk, usually calculated from Nusselt number [W/(m2·K)], dp is the pellet size [m], Tf is the

temperature of the fluid in the bed [K] and Ea is the (apparent) activation energy [kJ/mol].

External mass transport limitations

The Carberry number (Eq. 2.23) was used to verify that the resistance due to external

mass transfer is less than 5% of the resistance due to chemical reaction [23].

𝐶𝑎 =𝑅𝑣,𝐶𝐻4

𝑜𝑏𝑠

𝑘𝑔 ∙ a𝑣 ∙ 𝐶𝐶𝐻4,𝑏=

𝐶𝐶𝐻4,𝑏 − 𝐶𝐶𝐻4,𝑠

𝐶𝐶𝐻4,𝑏<

0.05

𝑛 (2.23)

Where CCH4,b and CCH4,s are the bulk and external surface CH4 concentrations [mol/m3], av

is the specific external surface area of one pellet (=6/dp) [m2/m3], kg is the mass transfer

coefficient between pellet and the gas phase (also referred as km) and n is the reaction

order.


65

Radial and axial heat transfer limitations

Radial heat transfer limitations within bed can be neglected if the following criterion is

satisfied (Eq. 2.24) [22]:

𝛥𝑇 =𝑅𝑣,𝐴

𝑜𝑏𝑠 ∙ |∆𝑟𝐻| ∙ (1 − 𝜀𝑏) ∙ (1 − 𝑏) ∙ 𝑑𝑡2

32 ∙ 𝜆𝑒𝑟<

0.05 ∙ 𝑅 ∙ 𝑇𝑤2

𝐸𝑎 (2.24)

Where εb is the bed porosity, b is the fraction of inert material in the bed, Tw is the wall

temperature [K] and dt is the internal tube diameter [m]. There are numerous empirical

correlations available for the effective radial conductivity λer [W/(m·K)], consisting of the

contributions λer,0 + λer,conv to describe the radial heat transfer in the plug flow reactor. Axial

isothermicity will be assumed a priori, since this maximally equal to the corresponding radial

temperature gradient.

Internal mass transfer limitations

For internal mass transfer limitations, the Weisz-Prater criterion is used. In comparison to

the Thiele modulus φ, the Weisz modulus Φ is based on the measured production rate,

allowing direct calculation based on observable quantities only. This is shown by Eq. 2.25 for

first order kinetics [24].

𝛷 = 𝜂 ∙ 𝜑2 =𝑅𝑣,𝐶𝐻4

𝑜𝑏𝑠 ∙ (𝑑𝑝

6 )²

𝐷𝐶𝐻4,𝑒𝑓𝑓 ∙ 𝐶𝐶𝐻4,𝑠≪ 1 (2.25)

Where η is the effectiveness factor, 𝑅𝑣,𝐶𝐻4𝑜𝑏𝑠 is the observed volumetric reaction rate of CH4

[molCH4/(m3cat·s], CCH4,s is the external surface CH4 concentration [mol/m3], dp is the pellet

size [m] and DCH4,eff is the effective diffusivity of methane in the pellet [m2/s] . The Eq. 2.25

can be easily proven equivalent to the Thiele criterion for internal mass transfer limitations.

Internal heat transport limitations

It is assumed that a simple parabolic temperature profile is present throughout spherically-

shaped pellet. The effect on the internal temperature gradient (inside the pellet) on the net

production is smaller than 5%, if the following criterion (Eq. 2.26) holds:


66

Rv,CH4obs ∙ |∆rH| ∙ dp

2

60 ∙ λp<

0.05 ∙ R ∙ Tf2

Ea (2.26)

2.4 Temporal Analysis of Products

The Temporal Analysis of Products (TAP) reactor system is an important tool for

investigating heterogeneous reactions, particularly reactions on industrial catalysts.

Originally created by John T. Gleaves in 1988 [25] and later modified by him in 1997 [26].

Today, a third generation of TAP reactor is available, Gleaves et al. [27] investigated the

advancement in TAP reactor design from the TAP-1 system to the TAP-3 system to highlight

its relevance as a valuable tool for elucidating mechanistic and kinetic aspects of adsorption,

diffusion, and reaction in gas–solid systems.

A TAP pulse response experiment consists of injecting a very small amount of gas,

typically nanomoles per pulse, into a tubular fixed bed reactor that is kept under vacuum.

The time-dependent exit flow rate of each gas is detected by a mass spectrometer. The high

time resolution of the TAP technique allows detection of short-(millisecond time scale)

and/or long-lived (>1 s) reaction intermediates, which helps to formulate the mechanism of

reaction [28].

Two advantages can be attributed to the operation in high vacuum. The first is that

external mass transfer limitations are completely absent. The second is that gas transport

through the catalyst bed is governed by Knudsen diffusion. In this case, the diffusivities of

the individual component of a gas mixture are independent of pressure, concentration, or

the composition of the gas mixture. Moreover, performing transient response experiments

in the limit of Knudsen diffusion minimizes collisions between gas-phase molecules so that

the transient response of product molecules in the TAP system is a measure of gas–solid

interactions [28].

A basic TAP setup, illustrated in Figure 2.3, has some characteristics common to an

molecular beam scattering (MBS) experiment, and others common to conventional

microreactor experiments. The key components are a reaction zone or the microreactor, a

fast-pulse gas feed system, a mass spectrometer detector, and a high-throughput ultrahigh

vacuum system. The reaction zone, which holds the catalyst sample, is a temperature


67

controlled cylindrical tube usually made of stainless steel, inconel, or quartz. One end of the

tube receives input from the feed system, and the opposite end is open to vacuum. During

an experiment the reaction zone and catalyst sample are continuously evacuated. The

reaction zone can hold practical catalysts, which are commonly studied in conventional

microreactor experiments, as well as single particles, single crystals, or model catalysts,

which are studied in MBS experiments [27].

Figure 2.3: Schematical representation of TAP system. Obtained from [27].

TAP pulse-response experiments extract kinetic information in a similar manner as

molecular beam scattering experiments. Injection of a narrow gas pulse into the reaction

zone initiates an experiment. The gas molecules travel through the reaction zone where they

encounter the catalyst and can react to form product molecules. Molecules that exit the

reaction zone are monitored by the mass spectrometer positioned in the outlet. An

experiment ends when the flow of reactant and product molecules is no longer detected by

the mass spectrometer. The observed characteristic feature in a TAP experiment is the time-

dependent gas flow F(t) [mol·s-1] or [molecules·s-1] that escapes from the exit of the

microreactor. The flow dependencies have integral characteristics that are related to the

moments of the flow-time. Kinetic and mechanistic information are obtained by analyzing

these dependencies, and gas transport is used as a measuring stick to determine the rates of


68

chemical transformations. In addition to kinetic and mechanistic information, TAP

experiments have been used to study transport processes in porous materials, such as

zeolites, and oxygen diffusion in metal and metal oxide catalysts. Some typical pulse-

response experiment of inert, reactants and products are depicted in Figure 2.4. The

temperature dependence of TAP transient response curves provides intrinsic rate

parameters that can be associated with the surface composition of a catalyst. The number of

active sites can be determined in TAP titration experiments [27].

Figure 2.4: Analysis of a TAP experiment: inert (internal standard), reactant and product pulse responses.


69

2.5 References

[1] N.V.R.A. Dharanipragada, L.C. Buelens, H. Poelman, E. De Grave, V.V. Galvita, G.B. Marin, Mg-Fe-Al-O for advanced CO2 to CO conversion: carbon monoxide yield vs. oxygen storage capacity, J. Mater. Chem. A, 3 (2015) 16251-16262. [2] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst characterization, Pure & Appl. Chem., (1995). [3] Phillips Innovation Services: Technical note 12. Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES), (2013). [4] S.S. Lowell, J.E., Powder Surface Area and Porosity, 1991. [5] Scherrer, Nachr. Ges. Wiss. Göttingen, 1918. [6] Niemantsverdiet J.W., Spectroscopy in Catalysis, 2007. [7] W.J.C. Niemantsverdriet, I., Concepts of Modern Catalysis and Kinetics, Wiley-VCH, Weinheim, 2003. [8] M.G. Fernandez, In-situ X-ray absorption study of hydrocarbon oxidation over metal oxide catalysts, (2009). [9] D.C.M. Koningsberger, B.L.; van Dorssen, G.E.; Ramaker, D.E., XAFS spectroscopy; fundamental principles and data analysis, Topics in Catalysis, 10 (2000) 143-155. [10] V. Martis, A.M. Beale, D. Detollenaere, D. Banerjee, M. Moroni, F. Gosselin, W. Bras, A high-pressure and controlled-flow gas system for catalysis research, J. Synchrotron Radiat., 21 (2014) 462-463. [11] A.K. Datye, Electron microscopy of catalysts: recent achievements and future prospects, J. Catal., 216 (2003) 144-154. [12] F.H. Ribeiro, G.A. Somorjai, G. Wedler, B.C. Gates, C.T. Campbell, C. Xu, D.W. Goodman, A. Zecchina, D. Scarano, S. Bordiga, A. Cimino, F.S. Stone, Y. Iwasawa, M.A. Barteau, J.M. Vohs, J.P. Vigneron, Model Systems, Handbook of Heterogeneous Catalysis, Wiley-VCH Verlag GmbH2008, pp. 771-908. [13] A. Ponce, S. Mejia-Rosales, M. Jose-Yacaman, Scanning transmission electron microscopy methods for the analysis of nanoparticles, Methods in molecular biology (Clifton, N.J.), 906 (2012) 453-471. [14] G.S. Bumbrah, R.M. Sharma, Raman spectroscopy – Basic principle, instrumentation and selected applications for the characterization of drugs of abuse, Egyptian Journal of Forensic Sciences, 6 (2016) 209-215. [15] M.T. Maestri, E.; Berger, R.J.; Kapteijn, F.; Moulijn, J.A., EUROKIN: Overview of requirements for measurement of intrinsic kinetics in and overview of correlations for characteristics of the G-S and L-S fixed-bed reactor, (2013). [16] H. Gierman, Design of laboratory hydrotreating reactors, Appl. Catal., 43 (1988) 277-286. [17] D.E. Mears, The role of axial dispersion in trickle-flow laboratory reactors, Chem. Eng. Sci., 26 (1971) 1361-1366. [18] N. Wakao, S. Kaguei, T. Funazkri, Effect of fluid dispersion coefficients on particle-to-fluid heat transfer coefficients in packed beds, Chem. Eng. Sci., 34 (1979) 325-336. [19] M. Punčochář, J. Drahoš, The tortuosity concept in fixed and fluidized bed, Chem. Eng. Sci., 48 (1993) 2173-2175. [20] K.B. Bischoff, A note on gas dispersion in packed beds, Chem. Eng. Sci., 24 (1969) 607. [21] R.J. Berger, J. Pérez-Ramırez, F. Kapteijn, J.A. Moulijn, Catalyst performance testing: bed dilution revisited, Chem. Eng. Sci., 57 (2002) 4921-4932. [22] D.E. Mears, Diagnostic criteria for heat transport limitations in fixed bed reactors, J. Catal., 20 (1971) 127-131. [23] J.J. Carberry, D. White, On the role of transport phenomena in catalytic reactor behavior, Ind. Eng. Chem., 61 (1969) 27-35.


70

[24] G.F. Froment, K. Bischoff, J. De Wilde, Chemical Reactor Analysis and Design, 3rd ed., J. Wiley & Sons2011. [25] J. Gleaves, J. Ebner, T. Kuechler, Temporal analysis of products (TAP)—a unique catalyst evaluation system with submillisecond time resolution, Catalysis Reviews Science and Engineering, 30 (1988) 49-116. [26] J.T. Gleaves, G.S. Yablonskii, P. Phanawadee, Y. Schuurman, TAP-2: An interrogative kinetics approach, Applied Catalysis A: General, 160 (1997) 55-88. [27] J.T. Gleaves, G. Yablonsky, X. Zheng, R. Fushimi, P.L. Mills, Temporal analysis of products (TAP)—Recent advances in technology for kinetic analysis of multi-component catalysts, J. Mol. Catal. A: Chem., 315 (2010) 108-134. [28] G.S. Yablonsky, M. Olea, G.B. Marin, Temporal analysis of products: basic principles, applications, and theory, J. Catal., 216 (2003) 120-134.

Chapter 3: Enhanced carbon-resistant dry reforming Ni-Fe catalyst: role of Fe

71

Chapter 3

Enhanced carbon-resistant dry reforming Ni-Fe catalyst: role of Fe

Fe2O3 can be a suitable promoter for Ni based reforming catalysts as it combines good

redox properties, like CeO2, with Ni-Fe alloy formation upon intimate interaction between Fe

and Ni after reduction. To evaluate the role of Fe on Ni/MgAl2O4, a series of bimetallic Ni-

Fe/MgAl2O4 catalysts, with Fe:Ni ratios between 0 and 1.5, were synthesized. Methane dry

reforming from 923 to 1073 K, atmospheric pressure and a CH4:CO2 ratio of 1 was selected

as a reforming reaction in order to clarify whether and how the Ni-Fe alloy contributes to the

elimination of carbon deposition. The questions raised here, are addressed by performing in-

situ X-ray diffraction (XRD) characterization of Ni-Fe/MgAl2O4 catalysts during temperature-

programmed reduction (TPR), oxidation and dry reforming. Alternate pulse experiments are

also applied by switching the flow between CO2 and CH4 over Ni/MgAl2O4, Fe/ MgAl2O4 and

Ni-Fe / MgAl2O4.

The results of this chapter are published as:

S.A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, Enhanced Carbon-Resistant Dry Reforming Fe-

Ni Catalyst: Role of Fe, ACS Catal., (2015) 3028-3039


72

3.1 Introduction

Methane reforming processes such as steam reforming (SRM), partial oxidation (POM),

autothermal reforming (ARM) and dry reforming (DRM) have been investigated intensively

[1]. The final H2:CO product ratio, energetics and oxidant used, differ in these processes [2].

The H2:CO ratio from DRM is more favorable for Fischer-Tropsch and methanol synthesis

than from classical steam reforming. DRM has the lowest operating cost among these

processes and offers the additional advantage of converting CO2 into valuable chemicals:

𝐶𝐻4 + 𝐶𝑂2 ↔ 2𝐶𝑂 + 2𝐻2 ∆H0298K= 247 kJ·mol-1 (3.1)

Side reactions of importance include the water gas shift:

𝐻2𝑂 + 𝐶𝑂 ↔ 𝐻2 + 𝐶𝑂2 ∆H0298K= -41 kJ·mol-1 (3.2)

However, dry reforming technologies have the disadvantage of rapid catalyst

deactivation due to carbon deposition and sintering of both the support and the active metal

particles, at high temperatures. The coke originates mainly from two reactions [3, 4]:

𝐶𝐻4 → 𝐶 + 2𝐻2 ∆H0298K= 75 kJ·mol-1 (3.3)

2𝐶𝑂 ↔ 𝐶 + 𝐶𝑂2 ∆H0298K= -172 kJ·mol-1 (3.4)

Nickel and noble metals such as Pt, Ru and Rh are the most investigated catalysts for dry

reforming. The aforementioned noble metals show high activity and resistance towards

carbon formation[3-11]. Despite their excellent performance, the high cost and limited

availability restrict their application in industry. Nickel-based catalysts can offer an

alternative but they are sensitive to deactivation by sintering and carbon encapsulation of Ni

under reforming conditions[12-14]. Hence, improvements with respect to both activity and

deactivation of Ni catalysts are required. Therefore much research effort has focused on the

effect of supports, promoters and other metals on Ni-based catalysts [15-18].

In many cases, Al2O3 has been used as a catalyst support [14, 19]. In addition, various

other support materials such as MgO, ZrO2, La2O3, TiO2, SiO2, SiO2- Al2O3 have been tested

[19-28]. There is an agreement in the literature that the mechanism of DRM is strongly

dependent on support materials. For catalysts supported on inert materials like SiO2 [29], the


73

mechanism follows a mono-functional pathway, where both reactants are activated by the

metal alone. Once carbon formation occurs by dehydrogenation of methane, subsequent

activation of CO2 and reaction with carbon is limited, leading to catalyst deactivation. On an

acidic [10] (Al2O3) or basic[30, 31] (La2O3, CeO2, MgO,…) support, a bi-functional mechanism

takes place: CH4 is activated on the metal while CO2 is activated on the support. Zhang and

Verykios [30] showed that Ni/La2O3 activity increases after 2-5 h of reaction. The studies

revealed a large CO2 pool, stored in the form of La2O2CO3. Slagtern and co-workers [32]

proposed a bi-functional mechanism for Ni/La2O3: methane is activated on the Ni particles

and carbon dioxide interacts with La2O3 forming carbonates which oxidize carbon deposited

on nickel at the Ni-La2O3 interface. Ceria is another example of active support. Bobin and co-

workers[33] showed that DRM on metal nanocrystalline doped CeZrO2 oxides demonstrate a

high activity and coking resistance by acting as oxygen storage materials. They found that

the DRM mechanism on metal nanocrystalline doped CeZrO2 oxides follows a simple redox

scheme: CO2 dissociated on reduced oxide supports while methane on metal sites. Other

than being directly involved in the reactant activation process on its acidic or basic sites, the

supports also affect the metal particle size or metal dispersion. Suitable supports have to be

resistant to high temperature and should be able to maintain the metal dispersion during

operation.

Suitable supports have to be resistant to high temperature and should be able to

maintain the metal dispersion during operation [34]. The addition of MgO to Al2O3 prevents

the interaction between active metals (Ni and Fe) and the alumina forming NiAl2O4 and

FeAl2O4 spinel phases that are inactive [12]. Guo and co-workers[34] compared Ni/γ-Al2O3,

Ni/MgO-γ-Al2O3 and Ni/MgAl2O4 during methane dry reforming. They found that Ni/MgAl2O4

exhibit higher activity and better stability compared to the other samples. The MgAl2O4

spinel can effectively suppress the phase transformation to form NiAl2O4 and can stabilize Ni

crystallites. Penkova and co-workers [35] found that incorporating a basic element (Mg) into

the alumina support, alters the surface structure, leading to a modification of the acid-base

properties of the support. The resulting magnesium aluminate spinel (MgAl2O4) combines

desirable properties for use in ceramics, due to its high melting temperature (2408 K),

mechanical strength and good chemical stability[34]. Hence, in this work MgAl2O4 is

examined as a Ni support because of its low acidity, compared to Al2O3, and sintering


74

resistance due to strong interaction of metal oxides with MgAl2O4, resulting in high metal

dispersion after reduction [36-38].

The accumulation of carbon on the catalyst surface depends on the oxidation rate of

dehydrogenated CHx species, originating from CH4 decomposition and thus on the surface

oxygen species availability. In view of increasing the available O species, the addition of

oxides to Ni has been examined. A possible oxide is CeO2 because it is known for its ability to

easily release oxygen and quickly re-oxidize in the presence of oxidative agents [33, 39].

CeO2 indeed enhances the catalytic performance by forming a Ni-CeO2 nanocomposite [39].

The good catalytic activity and high resistance toward carbon formation were assigned to

the large interface between metallic Ni and the CeO2 surface and to the metal/ceria

interactions which strongly affect their redox and catalytic properties [40].

An additional way of improving catalytic performance is by alloy formation between Ni

and other metals[41, 42]. It has been reported that the addition of noble metals (Ru, Rh, Pd,

Pt) to Ni-based catalysts results in alloy formation after reduction, improving the catalyst

activity and stability[43-46]. Apart from noble metals, a synergistic effect of alloy formation

between Ni and Co has been reported in tar steam reforming [47]. In this respect, Fe2O3 can

also make a suitable promoter as it combines good redox properties, like CeO2, with N alloy

formation upon intimate interaction between Fe and Ni. An improved performance in steam

reforming activity of Ni-Fe/Al2O3 was attributed to the formation of a Ni-Fe alloy so that Fe

acts as a co-catalyst [42]. Moreover, under industrial CO methanation reaction conditions,

Ni-Fe bimetallic catalysts on γ-Al2O3 exhibited higher catalytic activity than the monometallic

catalysts [48]. These results make iron oxide an interesting material for improvement of the

Ni/MgAl2O4 activity and resistance toward carbon deposition. Ashok and Kawi studied

toluene steam reforming over a Ni/Fe2O3-Al2O3 catalyst examining the effect of iron-alumina

calcination temperature on steam reforming [14]. They found that Ni-Fe alloys were formed

and remained stable throughout the reforming reaction.


75

3.2 Experimental methods

For more details about the characterization techniques that have been used in this

chapter the reader is referred to the Chapter 2.

3.2.1 Catalyst preparation

Support preparation

MgAl2O4 support material was prepared by co-precipitation from an aqueous solution of

Mg(NO3)2·6H2O (99%, Sigma-Aldrich®) and Al(NO3)3·9H2O (98.5%, Sigma-Aldrich®) (molar

ratio Mg:Al=1:2). A precipitating agent, NH4OH (ACS reagent, 28.0-30.0% NH3 basis) was

added to adjust the pH to 10, at 333 K. The produced precipitate was filtered, dried at 393 K

for 12h and subsequently calcined in air at 1023 K for 4h.

Catalyst preparation

Four Ni-Fe catalysts were prepared by incipient wetness impregnation on the support

(MgAl2O4) using an aqueous solution of corresponding nitrates Ni(NO3)2·6H2O (99.99+%,

Sigma-Aldrich®) and Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®)[49]. The catalysts were dried

at 393 K for 12h and subsequently calcined in air at 1023 K for 4 h. The Ni loading was kept

stable at 8wt% while the Fe varied from 0 to 11 wt%. Samples are labeled as “Ni-XFe”, where

their X the wt% of Fe (Table 3.1). Also, a 5wt.%Fe sample was synthesized on the same

support for comparison purposes.

3.2.2 Catalyst characterization

The Brunauer-Emmett-Teller (BET) surface area of each sample was determined by N2

adsorption at 77 K (five point BET method using Gemini Micromeritics) after outgassing the

sample at 473 K for 2h. The crystallographic phases of the materials as prepared were

confirmed by ex-situ XRD measurements (Siemens Diffractometer Kristalloflex D500, Cu Kα

radiation). The powder patterns were collected in a 2θ range from 10o to 80o with a step of

0.02o and 30 s counting time per angle. XRD patterns of known compounds are referenced

by their corresponding number in the Powder Diffraction File database. By fitting a Gaussian

function to a diffraction peak, the crystallite size was determined from the peak width via

http://en.wikipedia.org/wiki/Water_of_crystallization


76

the Scherrer equation [50], while the peak position gives information about the lattice

spacing based on the Bragg law of diffraction[51].

The bulk chemical composition of support and as-prepared catalysts were determined by

means of inductively coupled plasma atomic emission spectroscopy (ICP-AES, ICAP 6500,

Thermo Scientific). The samples were mineralized by alkaline fusion with a mix of Li-

tetraborate and Li-metaborate.

Scanning electron microscopy (SEM; FEI Quanta 200F) was performed to evaluate the

surface morphology of the materials. An acceleration voltage of 25 or 12.5 kV was chosen for

the secondary electron SEM images. Energy Dispersive X-ray (EDX) spectroscopy (EDAX inc.)

at 17.5 kV acceleration voltage provided a first compositional analysis, typically in a 1-μm-

thick volume.

High-resolution transmission electron microscopy (HRTEM) was used for structural

analysis, while EDX yielded local chemical analysis. These techniques were implemented

using a JEOL JEM-2200FS, Cs-corrected microscope operated at 200kV, which was equipped

with a Schottky-type field-emission gun (FEG) and EDX JEOL JED-2300D. All samples were

deposited by immersion onto a lacey carbon film on a copper support grid.

3.2.3 In-situ time resolved XRD

In-situ XRD measurements were performed in a home-built reaction chamber housed

inside a Bruker-AXS D8 Discover apparatus (Cu Kα radiation of 0.154 nm). The reactor

chamber had a Kapton foil window for X-ray transmission. The setup was equipped with a

linear detector covering a range of 20o in 2θ with an angular resolution of 0.1o. The pattern

acquisition time was approximately 10 s. All temperatures were measured with a K-type

thermocouple and corrected afterwards to a calibration curve of the heating device, which is

based on the eutectic systems Au–Si, Al–Si and Ag–Si.For each sample, approximately 10 mg

of powdered sample was evenly spread on a single crystal Si wafer. Interaction of the

catalyst material with the Si wafer was never observed. Before each TPR/TPO experiment,

the reactor chamber was evacuated to a base pressure of 4 Pa by a rotation pump. Gases

were supplied to the reactor chamber from a rig with calibrated mass-flow meters. He (1


77

mL/s) was flowing for 15 minutes before the flow switched to 1 mL/s of 10% H2/He or CO2

for TPR and TPO respectively.

The evolution of the catalyst structure during TPR, TPO and DRM reaction was

investigated. For TPR and TPO experiments, the sample was heated from room temperature

to 1123 K at a heating rate of 30 K/min in a flowing gas stream (1mL/s of 10% H2/He or CO2

respectively). For DRM reaction experiments the sample was first reduced (1 mL/s of 10%

H2/He) at 1123 K while the reaction temperature was 973 K. A full XRD scan (10o to 65o with

step of 0.02o) was taken at room temperature before and after each TPR, TPO and DRM

reaction experiment. Samples were cooled in helium flow to room temperature after each

experiment.

It should be noted that the peaks in the in-situ XRD patterns appeared at slightly shifted

angular positions compared to both full scans and tabulated values due to temperature-

induced lattice expansion and different sample heights. These shifts in peak positions which

are not related to underlying physicochemical processes, were taken into account during

peak assignment.

3.2.4 Catalytic activity

Activity measurements were performed at atmospheric pressure in a quartz U-tube

reactor with an internal diameter of 9 mm, which was housed inside an electric furnace

(AutoChem II 2920, Micromeritics equipped with a TCD detector). 10-20 mg of sample with

particle size of 30 μm, diluted with inert Al2O3 (ratio catalyst/inert =1/60), was packed

between quartz wool plugs. The temperature of the catalyst bed was measured with K-type

thermocouples touching the outside and inside of the reactor at the position of the catalyst

bed. The inlet gas flow rates were always maintained by means of calibrated Brooks mass

flow controllers. The outlet gas stream was monitored online using a calibrated OmniStar

Pfeiffer mass spectrometer (MS). MS signals were recorded for all major fragments. For

quantification of each component, the MS is focused to different amu signals, the selection

of which was based on the analysis of the mass spectra of the individual components. H2 was

monitored at 2, He at 4, CH4 at 16, H2O at 18, CO at 28, Ar at 40 and CO2 at 44. When there

was an unavoidable interference with fragmentation peaks of other gases, a correction was

applied to remove their contributions, e.g., CO is monitored at 28 amu, subtracting the


78

contribution of CO2, i.e., not more than 10% of the peak at 28 amu. A carbon balance with a

maximum deviation of 5% was obtained. Dimensionless criteria were applied for evaluating

the significance of external and internal heat and mass transfer limitations. The absence of

external and internal mass transfer limitations was verified using Carberry number [52] and

Weisz-Prater criterion [53], while for heat transport limitations the diagnostic criteria

reported by Mears [54] were applied.

The activity was measured from 823 K to 1073 K. First the sample was reduced in 1 mL/s

flow of 5%H2/Ar at 1123 K for 30 minutes and then the flow was switched to He for 10

minutes. After this step a mixture of 50%CH4-50%Ar and CO2 (volume ratio 2:1, volume ratio

CH4:CO2=1:1, Ar: internal standard) started flowing inside the reactor for the methane dry

reforming reaction. Produced CO, H2 and unconverted CH4 and CO2 were detected at the

outlet. For each catalyst sample, activity was measured at 5 different temperatures. At each

temperature the reaction took place for 6 minutes, when the rate of CH4 and CO2

consumption was stable.

The activity was also measured as a function of time-on-stream. The experiment was

carried out at the same temperature, pressure, dilution ratio between catalyst and inert,

flow rates of reactants (CH4:CO2= 1:1) and conversion. The same conversion (XCH4=51%) for

all the investigated samples was achieved by varying the amount of catalyst.



𝑋𝑖 = 𝐹𝑖,𝑜−𝐹𝑖

𝐹𝑖,0 (3.5)

The CH4 consumption rate (mmol·s-1·gNi-1, Eq 3.6) was calculated from the difference in

the inlet and outlet molar flow rates, divided by the amount of active metals, as measured

relative to an internal standard (Ar), while the space time yield (STY, mmol·s-1·gNi-1, Eq. 3.7)

was calculated for the products:


0−𝐹𝑖|

𝑚𝑁𝑖 (3.6)

STYi = 𝐹𝑖

𝑚𝑁𝑖 3.7


79

where 𝐹𝑖0and 𝐹𝑖 are the inlet and outlet molar flow rates of gas i (mol·s-1), and mNi the

amount of Ni (g) present in the catalyst after ICP analysis.

Regeneration cycles were performed combining periods of methane dry reforming with

periods of catalyst oxidation with CO2 for carbon removal and finally reduction for alloy

formation [55]. Burning carbon by oxygen or air is a strongly exothermic reaction and hence

would lead to catalyst bed temperature increase. In each methane dry reforming period, the

methane consumption rate was calculated after TOS=30 minutes. The standard error was

calculated out of three independent experiments.

Carbon formation over the studied catalysts was investigated using methane

decomposition considering that this is the main source of carbon formation at 1023 K. The

reaction took place in a quartz tube reactor at 1023 K for six minutes. The inlet 50%CH4-

50%Ar gas flow rate was 1 mL/s. The deposited carbon was then oxidized by CO2, while the

produced CO and unconverted CO2 were detected in the outlet. Then the flow was switched

to O2 in order to burn the remaining carbon, if any.

3.2.5 Alternate Pulse experiment

An isothermal pulse experiment was performed using a quartz tube reactor at 1023 K.

The reactor was filled with catalyst and inert alumina as diluent, in a 1:60 ratio. As a first

step, a pulse of CO2 was admitted during 1 minute, while monitoring the CO production by

online MS. Then, He started to flow inside the reactor in order to remove the remaining CO2.

As a final step, a CH4 pulse followed for 1 minute under the same conditions, with

monitoring of the CO production. The 5%Fe/MgAl2O4 sample was also tested during

alternate pulse experiment.

3.3 Results


The Brunauer-Emmett-Teller (BET) surface area, metal content and average diameter of

Ni crystallites are reported in Table 3.1 for the samples used in this study. The supported Ni-

based catalysts have lower surface areas than the bare support. This is mainly due to pore

blockage [14]. The surface area of the support and the pure Ni supported-sample were


80

above previously reported values [4, 34, 56]. It is observed that the BET surface area

decreases as metal content increases, but it remains stable for the last two samples (Ni-

8Fe/MgAl and Ni-11Fe/MgAl). The crystallite sizes calculated from XRD are in line with the

average particle sizes as observed in HRTEM.

Table 3.1: Catalyst and support properties

Abbreviation

Catalyst

Fe:Ni

(mol/mol)

Metal loading

(wt.%) BET

(m2/gcat)

Metal

fraction

exposeda

(%)

Crystallite

size of active

phase

Ni and Ni-Fe

alloy (nm)b

Ni Fe

MgAl MgAl2O4 -

- - 100.3 ±

15.2

- -

Ni-0Fe/MgAl 8wt.%Ni/MgAl2O4 0

7.85 - 81.9 ±

9.8

9.5 10.5

Ni-5Fe/MgAl 8wt.%Ni-

5wt.%Fe/MgAl2O4 0.7

7.50 4.95 84.7 ±

5.8

7.0 14.3


8wt.%Fe/MgAl2O4 1.1

7.20 7.30 47.6 ±

11.4

7.5 13.4


11wt.%Fe/MgAl2O4 1.6

7.20 10.90 68.4 ±

6.7

5.1 19.5

aFraction exposed of Ni sites was calculated as (1/dNi)·100, where dNi is crystallite diameter in nm

bThe crystallite size of Ni and Ni-Fe alloy were calculated from full scan XRD patterns obtained after

H2-reduction

The crystalline phases of as-prepared, reduced and reoxidized catalysts were confirmed

by ex situ powder XRD. Figure 3.1(A) shows an example of pure support, as-prepared and

reduced Fe- promoted sample (Ni-5Fe/MgAl) while Figure 3.1 (B) illustrates the full scan XRD

patterns of Ni-5Fe/MgAl after CO2 oxidation. MgAl2O4 (31.7o,37o,45o, 55.5o, 59o, 65o, Powder

Diffraction File (PDF) card number: 00-021-1152) remained stable during reduction and

oxidation for all samples (full scan XRD patterns for the other studied samples are not

shown). Fe2O3 (Maghemite), NiO and NiAl2O4 peaks (30.2o, 35.6o, 43.3o, 57.3o, 62.9o, PDF:

00-039-1346, 37.3o, 43.3o and 62.9o, PDF: 01-089-5881 and 37.0o, 45o, 59.7o and 65.5o, PDF:

00-010-0339 respectively), some of which are overlapping, were observed for the as-

prepared sample. It cannot be excluded that the peaks (30.2o, 35.6o, 43.3o, 57.3o, 62.9o)


81

correspond to NiFe2O4 instead of Fe2O3 as they cannot be distinguished due to overlapping.

Upon reduction, NiO and Fe2O3 diffractions disappeared, while NiO completely integrated in

the spinel structure forming NiAl2O4 was not reduced. This is in accordance with Guo and co-

workers[34] who also observed the formation of NiAl2O4 crystalline phase. In addition,

diffractions are present at angles lower than metallic Ni. The shift is attributed to formation

of a Ni-Fe alloy (44o and 51.5o) and this was also observed in previous reports[14, 42].

According to the Ni-Fe phase diagram [57] at least one regular Ni-rich alloy with Ni3Fe

composition is known. Other Ni-Fe alloy structures with composition NiFe, Ni3Fe2 and Ni2Fe

have been also reported [56]. The XRD pattern following the CO2-TPO (Figure 3.1(B)) shows

that Ni-Fe alloy was decomposed to Ni (44.5o) and Fe3O4 (30.1o,36o,43.5o,57o,63o), while the

NiAl2O4 and MgAl2O4 substrate diffractions remained stable. For the Ni-8Fe/MgAl and Ni-

11Fe/MgAl, the XRD patterns showed an extra peak, after H2-TPR, originating from metallic

Fe (45o) which was not engaged in alloy formation (not shown).

The elemental distribution of Ni-5Fe/MgAl is indicated in Figure 3.2 using energy-

dispersive X-ray spectroscopy (EDX)-STEM mapping. Both Fe (red) and Ni (green) elements

were distributed uniformly in the sample after reduction, implying the alloy formation. In

contrast, after CO2 oxidation Ni and Fe particles were segregated as Fe is oxidized to Fe3O4

and the Ni-Fe alloy is decomposed.


82

Figure 3.1. Full XRD scans of (A) MgAl2O4, as-prepared and reduced Ni-5Fe/MgAl (1mL/s of 10%H2/He

mixture at a total pressure of 101.3 kPa and 1123K). The upper-right inset shows the highlighted rectangular

area in higher resolution. (B) Ni-5Fe/MgAl after CO2-oxidation (1mL/s of CO2, at a total pressure of 101.3 kPa

and 1123K).

Figure 3.2: EDX element mapping of Ni-5Fe/MgAl. (A) after H2-reduction (1mL/s of 5%H2/Ar mixture at a total

pressure of 101.3 kPa and 1123K). (B) after CO2 oxidation (1mL/s of CO2 at a total pressure of 101.3 kPa and

1123K). Red and green colors correspond to Fe and Ni elements respectively.

3.3.2 In-situ XRD time resolved measurements

H2-TPR.

The Ni-Fe alloy formation during the H2-TPR process was investigated using in-situ XRD

measurements for all studied samples. The results for the Ni-5Fe/MgAl sample are

presented in Figure 3.3 (A). Diffraction peaks associated to Fe2O3 were not detected by in-


83

situ XRD due to the low concentration and overlapping with MgAl2O4 peaks. During

reduction, NiO peaks disappear above 973 K. The Ni related diffraction shifts to an angle of

44o, lower than for metallic Ni, which hence corresponds to a Ni-Fe alloy peak [14, 42, 58,

59]. The kinetics of NiO reduction and alloy formation are depicted in Figure 3.3 (B) as

integral intensity variation of the diffraction peaks with temperature. At 973 K the intensity

in the angle region 43.7o-44.2o increases because of the appearance of the Ni-Fe alloy

diffraction.

Figure 3.3. In-situ XRD during H2-TPR. (A) 2D XRD pattern for Ni-5Fe/MgAl. Heating rate: 30 K/min, maximum

temperature 1123 K, flow rate: 1 mL/s, 10%H2/He. (B) Integral intensity variation of (A) for diffraction areas

35.5o-36.5

o (NiO), 43.7

o-44.2

o (Ni-Fe alloy).

(A)


84

The intensity of characteristic peaks associated with crystalline MgAl did not change

throughout H2-TPR up to the final temperature of 1123K, implying that MgAl was not

reduced in this temperature range. The small shift of angular positions compared to full scan

XRD patterns at room temperature can be ascribed to temperature-induced lattice

expansion and different sample height.

CO2-TPO

To test the stability of the Ni-Fe alloy in the presence of CO2, CO2-TPO was performed

immediately after cool down following H2-TPR (Figure 3.4(A)). The oxidation reaction of the

Ni-Fe alloy in Ni-5Fe/MgAl resulted in alloy decomposition above 900 K into Ni and Fe3O4.

The intermediate phase of FeO was not observed. It has been reported that the formation of

FeO strongly depends on reaction temperature and gas composition [60]. Further oxidation

of Fe3O4 to Fe2O3 cannot be achieved by applying gaseous CO2. The kinetics of the alloy

decomposition are illustrated in Figure 3.4 (B) as time-dependent integral intensities of the

Ni-Fe alloy and Fe3O4 angular regions. The Ni peak intensity is not depicted because of

overlap with a MgAl2O4 peak.

Fe3O4

MgAl2O4MgAl2O4

Ni-Fe alloy

MgAl2O4

Fe3O4

Ni

MgAl2O4 MgAl2O4(A)


85

Figure 3.4. In-situ XRD during CO2-TPO. (A) 2D XRD pattern for Ni-5Fe/MgAl. Heating rate: 30 K/min,

maximum temperature 1123 K, flow rate: 1 mL/s, 100% CO2. (B) Integral intensity variation of (A) for

diffraction areas 35.4o-36.4

o (Fe3O4) and 43.7

o-44.2

o (Ni-Fe alloy).

A graphical illustration of the Ni-Fe alloy formation and decomposition is depicted in

Figure 3.5. The alloy was decomposed during CO2 oxidation between 850 K and 1123 K

yielding two separate phases of Ni and Fe3O4 (see EDX elemental mapping image Figure 3.2).

Metallic Ni remained stable under CO2 flow and was not oxidized to NiO. A subsequent H2

reduction step led again to Ni-Fe alloy formation for all the studied samples (not shown).

Figure 3.5. Schematic diagram of Ni-Fe alloy formation, during H2-reduction, and decomposition, during CO2

oxidation.

3.3.3 Activity tests

The effect of Ni-Fe alloying on the catalytic properties of the Ni-Fe/MgAl2O4 was

examined by a set of activity measurements under methane dry reforming reaction

Fe3O4

NiNiO

Fe2O3

H2

Reduction

Ni

Fe

CO2

H2

MgAl2O4 MgAl2O4 MgAl2O4


86

conditions. Figure 3.6 shows the variation of CH4:CO2 consumption rates with temperature.

Each temperature was maintained for 6 minutes until the measured temperature inside the

catalyst bed was stable. The ratio of CH4:CO2 consumption rates was above unity for almost

all catalyst samples, implying that in the first few minutes of reaction, CH4 consumption was

higher than CO2. Based on DRM and Reverse Water Gas Shift (RWGS) reactions, it was

expected that the ratio of CH4:CO2 consumption rates would remain below 1. There are two

relative rates that contribute to this ratio. The first depends on the methane decomposition

rate to surface carbon and H2 (MD reaction in Table 1.2). The second depends on surface

carbon oxidation by oxygen which is derived from CO2, at the applied operating temperature

and reactant partial pressure. Initially the rate of methane decomposition is higher than

surface carbon oxidation, but it decreased after 10 to 15 minutes, due to deactivation of Ni

surface sites, which are more active for methane cracking [61, 62]. This was ascribed to a

lower CO2 consumption, comparing to Fe-promoted samples, as the methane conversion

was the same for all the studied catalysts.

Figure 3.7 illustrates the dependence of the ratio of CH4:CO2 consumption rates on time-

on-stream during isothermal dry reforming at 1023 K for all catalysts. Ni-0Fe/MgAl showed

the highest ratio.

Figure 3.6. Ratio of CH4:CO2 consumption rates as a function of temperature for methane dry reforming (6

min at each temperature, Wcat:FCH4= 0.40-0.71 kgcat·s·mol-1

): ◊: Ni-0Fe/MgAl, □: Ni-5Fe/MgAl, ∆: Ni-8Fe/MgAl,

x: Ni-11Fe/MgAl.

0.5

0.7

0.9

1.1

1.3

1.5

1.7

1.9

2.1

800 850 900 950 1000 1050 1100

Rat

io C

H4:C

O2

con

sum

pti

on

rat

es

Temperature (K)


87

Figure 3.7. Ratio of CH4:CO2 consumption rates as a function of Time-On-Stream (TOS, min) for methane dry

reforming at 1023K (Wcat:F0

CH4= 0.40-0.71 kgcat·s·mol-1

), CH4:CO2=1:1, total pressure of 101.3 kPa: ◊: Ni-

0Fe/MgAl, □: Ni-5Fe/MgAl, ∆: Ni-8Fe/MgAl, x: Ni-11Fe/MgAl.

The addition of Fe increase the CO2 conversion. This means that CO2 selectively reacts

with methane or that CO2 is consumed according to the water-gas-shift reaction (WGS

reaction in Table 1.2).

The CO and H2 STY results of a series of isothermal activity measurements in a fixed bed

reactor are depicted in Figure 3.8 (Ni-0Fe/MgAl and Ni-5Fe/MgAl are shown). CO production

was higher for all Fe-promoted samples than for Ni-0Fe/MgAl. Specifically, the Ni-5Fe/MgAl

catalyst showed the higher CO production and the CO:H2 ratio was close to unity (Figure 3.9)

making it suitable for the Fischer-Tropsch process and methanol synthesis.

0.5

0.7

0.9

1.1

1.3

1.5

1.7

1.9

0 50 100 150 200 250

Rat

io C

H4:

CO

2

con

sum

pti

on

rat

es

TOS (min)

(A)


88

Figure 3.8. Space Time Yield for H2 and CO (STY, mmol·s-1

·gNi-1

) with Time-On-Stream (TOS, min) for methane

dry reforming at 1023K (Wcat:FCH4= 0.40-0.71 kgcat·s·mol-1

), CH4:CO2=1:1 and total pressure of 101.3 kPa. (A):

Ni-0Fe/MgAl (B): Ni-5Fe/MgAl

Figure 3.9. CO:H2 ratio with Time-On-Stream (min) for all studied catalysts. ◊: Ni-0Fe/MgAl, □: Ni-5Fe/MgAl,

∆: Ni-8Fe/MgAl, x: Ni-11Fe/MgAl (error bars not shown).

It is also observed that Ni-5Fe/MgAl has an optimal composition as it results in higher CO

and H2 production and less deactivated, from 73.8 to 50.0 mmolCO·s-1·gNi-1 and from 97.1 to

37.0 mmolH2·s-1·gNi-1. The addition of Fe promotes methane dry reforming in the range of

molar ratio Fe:Ni≤0.7. More Fe addition results in catalyst activity decrease. These results

(B)

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

0 50 100 150 200 250

Rat

io C

O:H

2

TOS, min


89

are in accordance with Wang and co-workers [42] who found that the addition of Fe can

have either a promoting or a suppressing effect on catalytic steam reforming of tars.

Carbon formation on Fe-promoted and non-promoted samples was compared by SEM

micrographs and EDX analysis of spent catalysts. Figure 3.10(A) shows carbon filaments

grown on Ni-0Fe/MgAl [63] after 4h on-the-stream under methane dry reforming at 1023 K.

The corresponding EDX spectrum, see Figure 3.10(B), confirms the high concentration of

carbon in this sample. On the other hand, a negligible amount of carbon was deposited on

Ni-8Fe/MgAl (Figure 3.10(C)) as can be verified by the respective EDX spectrum, see Figure

3.10(D).

Figure 3.10. SEM micrographs and EDX analysis of spent catalysts. (A) Ni-0Fe/MgAl SEM image. (B) Ni-

0Fe/MgAl EDX. (C) Ni-8Fe/MgAl SEM image. (D) Ni-8Fe/MgAl EDX. Temperature 1023K, CH4:CO2=1:1,

reaction time 4h.

(A) (B)

(C) (D)


90

The high amount of carbon deposited on used Ni-0Fe/MgAl agrees with the higher ratio

of CH4:CO2 consumption rates than for Fe-promoted samples (see Figure 3.8 and 3.9),

observed during methane dry reforming [14, 42]. Higher ratio means that more CH4 was

converted than CO2 and as a result more carbon is formed (Figure 3.10). Fe addition

increases CO2 conversion but also suppresses carbon deposition.

3.3.4 Regeneration cycles

As seen in Figure 3.8, the activity of the studied catalysts decreases with time-on-

stream. To evaluate the catalyst regeneration ability, three catalytic cycles were performed

that combined periods of methane dry reforming with periods of catalyst oxidation with CO2

for carbon removal and finally reduction for alloy formation. In each DRM period, the

methane consumption rate was calculated after TOS=30 minutes (Figure 3.11). The duration

of the DRM in the first cycle was four hours while the duration in the second and third cycle

was reduced to one hour. The CO2 oxidation step was always 20 minutes at 1023 K and the

subsequent H2 reduction step also 20 minutes at 1123 K.

Figure 3.11. CH4 consumption rate (mmol·s-1

·gNi-1

) during three catalytic cycles 1st

cycle DRM TOS=4h, 2nd and

3rd cycle TOS=1h. CO2 oxidation and H2 reduction: each 20 min. The CH4 consumption rate for each cycle was

calculated after TOS=30 minutes.

Figure 3.11 shows all catalysts can be regenerated. The pure Ni sample remains at a

slightly lower level of activity even after the second regeneration (CH4 consumption rate in

0

5

10

15

20

25

30

35

CH

4co

nsu

mp

tio

n r

ate

(mm

ol·

s-1·g

Ni-1

)

Ni-0Fe/MgAl Ni-5Fe/MgAl Ni-8Fe/MgAl Ni-11Fe/MgAl


91

3rd cycle). The Fe-promoted samples showed the same activity after the first and second

regeneration cycle. The higher activity that can be observed from Figure 3.11 after the

second regeneration cycle is attributed to the experimental error [55]. Furthermore, a higher

amount of carbon was deposited on Ni-0Fe/MgAl (4·102 molc·kgcat-1) than on Fe-promoted

samples (5.0, 0.9 and 0.5 molc·kgcat-1 for Ni-5Fe/MgAl, Ni-8Fe/MgAl and Ni-11Fe/MgAl

respectively) during DRM of the 1st cycle. This is in agreement with the SEM-EDX

characterization of spent catalysts (Figure 3.10(A-D)) where more carbon deposition was

observed on Ni-0Fe/MgAl than on the Fe-promoted sample.

3.3.5 Effect of alloy on carbon formation

The contribution of the alloy formation to the elimination of carbon deposition was

investigated under methane decomposition reaction conditions (see §3.1). The carbon

accumulated according to methane decomposition reaction (see Table 1.2, Chapter 1) [64],

was subsequently oxidized by CO2 according to the Boudouard reaction(see Table 1.2,

Chapter 1). In order to examine whether the filamentous carbon formed, actually interacted

with CO2 to be burnt, the flow was switched afterwards to O2 to burn off any remaining

carbon. All deposited carbon was however removed during CO2 oxidation as no CO

production was observed during subsequent O2 oxidation.

Figure 3.12 illustrates the produced CO by CO2 oxidation of deposited carbon. The area

under each curve corresponds to the produced CO (mol) and this value was normalized per

gram of metal loading (Ni and Fe). The Ni-8Fe/MgAl sample accumulated more carbon as

0.07 mol CO/gNi+Fe were produced after carbon removal, while on Ni-5Fe/MgAl and Ni-

0Fe/MgAl less carbon was deposited (production of 0.06 mol CO/gNi+Fe and 0.04 mol

CO/gNi+Fe respectively). The reference 5Fe sample produced 0.04 mol CO/gNi+Fe (not shown),

implying that Fe can be an active site for the methane decomposition reaction [65].


92

Figure 3.12. CO Space Time Yield (STY, mmolCO·s-1

·gNi+Fe-1

) upon CO2 oxidation of deposited carbon for

different catalysts. The carbon was formed after pulsing methane for 6 min at 1023 K for all pre reduced

samples (1123 K, 1mL/s 5%H2/He). Total amount of CO produced: (1) 0.04 mol CO/gNi+Fe. (2) 0.06 mol

CO/gNi+Fe. (3) 0.07 mol CO/gNi+Fe.

As the total metal loading increases in the mixed Ni-Fe samples, more carbon is formed.

This is in accordance with Shah and co-workers [63] and Baker [66] also showed that

bimetallic catalysts are more active for methane decomposition than the monometallic

catalysts. This implies that the reduced amount of carbon observed in SEM-EDX after

methane dry reforming over Fe-promoted catalysts, was not due to the alloy formation

between Ni and Fe. Rather, another phenomenon is responsible for this observation.

3.3.6 In-situ XRD time resolved during DRM

To understand the reason for the lower amount of carbon deposited on Fe modified

samples under dry reforming reaction conditions, despite the higher amount deposited

during CH4 decomposition, in-situ XRD experiments were performed on all studied catalysts.

Initially, all samples were reduced during H2-TPR at 1023 K, at heating rate of 30 K/min, flow

rate of 1 mL/s 10%H2/He, in order to form the active Ni-Fe alloy (period 1 in Figure 3.13).

Then, a flow of CH4 and CO2 was introduced to the reactor (periods 2, 3 and 4 in Figure 3.13),

stepwise increasing the CO2 partial pressure and investigating the effect on alloy

decomposition. When the CO2 partial pressure was 6-times higher than the CH4 partial

0

5

10

15

20

25

30

0 1 2 3 4 5 6 7 8

STY C

O(m

mo

l·s-1

·gN

i+Fe

-1)

TOS (min)

Ni-0Fe/MgAlNi-5Fe/MgAl

Ni-8Fe/MgAl

(1)(2)

(3)


93

pressure, a FeO peak was observed in in-situ XRD, implying that Fe was extracted from the

alloy. Likely, FeO is formed in the presence of CO2 even at lower partial pressures but it is

then not detected by in-situ XRD due to its low concentration. Samples with higher Fe

loading than Ni-5Fe/MgAl exhibited a FeO peak even at higher CH4:CO2 ratios.

Figure 3.13. Time resolved in-situ XRD patterns of Ni-5Fe/MgAl sample. (1) Ni-Fe alloy formation in 10%H2-

TPR to 1023 K, heating rate: 30 K/min. Methane dry reforming: (2) CH4:CO2=1:2, (3) CH4:CO2=1:3, (4)

CH4:CO2=1:6, at 1023 K and total pressure of 101.3 kPa.

According to Galvita and co-workers [67] carbon deposition is not favored on iron oxides,

in the presence of an oxidizing agent, for temperatures between 1023 K and 1173 K. Hence,

it is likely that most of the carbon will be formed during dry reforming on the Ni surface. In

this mechanism, methane is decomposed over a Ni particle to form hydrogen and carbon. In

the presence of CO2 metallic Fe is segregated from the alloy and FeOx is formed.

Subsequently, oxygen from the FeOx lattice is transferred to nearby Ni atom oxidizing the

formed carbon and producing CO. This is in accordance with Galvita and co-workers [64]

who performed reduction/oxidation cycles over Ni/CeO2-Fe2O3. The initial reduction step of

the as-prepared catalyst in a gas stream of CH4+CO2 resulted in Ni-Fe alloy formation which

was stable under inert environment. After CO2 oxidation they observed that surface carbon

was further oxidized by iron oxide lattice oxygen since under inert environment, the iron

oxide was further reduced to metallic Fe.


94

3.3.7 Alternate Pulse experiment

To understand the mechanism for CH4 and CO2 activation on Ni-Fe catalysts, an alternate

pulse experiment was performed. CO2 and CH4 were sequentially pulsed into the reactor at

1023 K over a reduced catalyst while the outlet flow rate of CO production was measured.

Figure 3.14(A) and (B) depict that CO was produced on the Ni-8Fe/MgAl and 5Fe catalyst

during the CO2 pulse. Normalized per gram of Fe, Ni-8Fe/MgAl and 5Fe produced 0.15 and

0.11 molCO/gFe, respectively. However, there was no CO production on Ni-0Fe/MgAl sample

during a CO2 pulse (not shown), which is in accordance with Figure 3.4(A) where it was

shown that Ni is not oxidized under CO2 flow. Subsequently, He was introduced into the

reactor to remove the remaining CO2. During the subsequent CH4 pulse, CO was formed only

over Ni-8Fe/MgAl catalyst (0.13 molCO/gFe), while no CO production was observed over a

pure 5Fe sample (Figure 3.14(B)) implying that FeOx is not active for the methane

decomposition reaction. The behavior of CO2 oxidation and CH4 reduction observed in

Figure(A) is consistent with Mars–van Krevelen (MvK) mechanism.

During the first pulse, Fe is oxidized by CO2 to form FeOx and CO is produced. During the

reduction step (second pulse) methane is activated and decomposed to carbon and H2 only

on the Ni surface due to low activity of FeOx in methane activation (Figure 3.14(B)). Lattice

oxygen atoms from the surface and bulk of FeOx are consumed by carbon deposited on

nearby Ni. The carbon balance during CH4 pulse was calculated and the estimated carbon

remaining on the catalyst surface was 24% of the carbon feed. This can be attributed either

to the fact that the amount of pulsed CO2 was not high enough to remove all carbon or to

different types or location of carbon [68]. This carbon can be burnt only by oxygen at higher

than 1023 K.

The opposite sequence of pulses was also investigated (see § 3.3.5) where the CH4 pulse

was firstly sent, followed by the CO2 pulse. CO production was observed for all the examined

samples, during the CO2 pulse, due to oxidation of the deposited carbon.

In general, a dry reforming reaction is typically accompanied by the simultaneous

occurrence of the reverse water-gas shift reaction (CO2 + H2 CO + H2O). To study the

interaction between the hydrogen, being the product of methane dry reforming, and CO2, a

CO2 and H2 alternate pulse experiment was performed using the same volumetric flow rates

http://www.sciencedirect.com/science/article/pii/S0021951711001825#f0005

http://www.sciencedirect.com/science/article/pii/S0021951711001825#f0005


95

as in the CH4:CO2 alternate pulse experiment. It followed that the reverse water-gas shift

reaction over Fe-promoted catalysts proceeds via a simple redox mechanism: CO2 oxidizes

the Fe surface while H2 reduces it.

Figure 3.14: CO production during CO2 and CH4 alternate pulse experiment over (A) Ni-8Fe/MgAl; (B) 5%Fe.

Period I: 1mL/s of CO2, 1min; Period II: 1mL/s of He, 2min; Period III: 1mL/s of CH4, 1min at 1023 K and total

pressure of 101.3 kPa. molCO/gmetal produced: (1) 0.15, (2) 0.13 and (3) 0.11 (4) 0.0.

The results obtained from pulse experiments on the catalyst can give us some insights

about the major mechanistic aspects of the dry reforming reaction. It is suggested that both

methane dry reforming and reverse water-gas shift reactions can proceed via a MvK

(1) (2)

(A)I II III

(3) (4)

IIII II (B)


96

mechanism over Ni-8Fe/MgAl. The constituent formal reactions of methane dry reforming

can be described by the following equations:

However, the carbon oxidation from surface oxygen produced by CO2 dissociation over

Ni cannot be excluded [69]. In the Ni-Fe catalytic system, CH4 is activated by Ni sites and

dissociates to H2 and chemisorbed carbon (Eq. 3.8) whereas CO2 is dissociatively adsorbed

on Fe sites producing surface CO and Fe oxide (Eq. 3.9). In contrast to the non-promoted

sample, Ni-0Fe/MgAl, the lattice oxygen from iron oxide participates in the reaction

mechanism, oxidizing the carbon deposited on the nearby Ni atoms, thereby forming CO and

carbon free Ni sites (Eq. 3.10) [64]. The FeOx lattice oxygen can also react with produced H2

resulting in H2O formation (Eq. 3.11). Eq. 3.8 and 3.10 are the steps for the reverse WGS

reaction that arises simultaneously with the DRM reaction.

3.4 Conclusions

The evolution of the crystallographic structure of bimetallic Ni-Fe/MgAl2O4 catalysts with

varying Fe:Ni ratios was investigated using time-resolved in-situ XRD. During H2-TPR, Fe2O3

and NiO were reduced above 973K to form a Ni-Fe alloy, which constitutes the active phase

for the methane dry reforming reaction. This alloy remained stable in a flowing gas stream of

CO2 during re-oxidation, up to 900 K, but was decomposed to metallic Ni and Fe3O4 above

this temperature.

The effect of Fe addition to the activity of a Ni/MgAl2O4 catalyst, under methane dry

reforming reaction conditions was found to depend on the employed Fe:Ni ratio. Activity

measurements indicated that Ni-5Fe/MgAl presented a high selectivity towards CO and a

CH4 CNi + 2H2 Eq. (10)Ni

Fe + xCO2 FeOx + xCO Eq. (11)

FeOx + CNi xCO + Ni + Fe Eq. (12)

FeOx + H2 Fe + H2O Eq. (13)

(3.8)

(3.9)

(3.10)

(3.11)


97

CO:H2 ratio close to unity (1.3). Furthermore, the implementation of regeneration cycles

showed that the catalyst activity can be restored.

Alternate CH4 and CO2 pulse experiments and in-situ XRD over Fe/MgAl2O4, Ni/MgAl2O4

and Ni-Fe/MgAl2O4 catalysts allowed to propose a mechanism of methane dry reforming

over the Ni-Fe catalysts. The process of dry reforming on Ni-Fe could be described by the

Mars-Van Krevelen mechanism where CO2 oxidizes Fe to FeOx, and CH4 is activated on Ni

sites to form H2 and surface carbon. The latter was re-oxidized by lattice oxygen from FeOx,

producing CO. Lower amounts of accumulated carbon were observed after DRM on Fe-

promoted samples compared to the pure Ni/MgAl2O4 sample.


98

3.5 References

[1] P. Littlewood, X. Xie, M. Bernicke, A. Thomas, R. Schomäcker, Ni0.05Mn0.95O catalysts for the dry reforming of methane, Catal. Today, 242, Part A (2015) 111-118. [2] D. Pakhare, V. Schwartz, V. Abdelsayed, D. Haynes, D. Shekhawat, J. Poston, J. Spivey, Kinetic and mechanistic study of dry (CO2) reforming of methane over Rh-substituted La2Zr2O7 pyrochlores, J. Catal., 316 (2014) 78-92. [3] I. Luisetto, S. Tuti, E. Di Bartolomeo, Co and Ni supported on CeO2 as selective bimetallic catalyst for dry reforming of methane, Int. J. Hydrogen. Energ., 37 (2012) 15992-15999. [4] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, Ni-based catalysts for reforming of methane with CO2, Int. J. Hydrogen Energ., 37 (2012) 15966-15975. [5] B. Nematollahi, M. Rezaei, M. Khajenoori, Combined dry reforming and partial oxidation of methane to synthesis gas on noble metal catalysts, Int. J. Hydrogen Energ., 36 (2011) 2969-2978. [6] M. Rezaei, S.M. Alavi, S. Sahebdelfar, Z.-F. Yan, Syngas Production by Methane Reforming with Carbon Dioxide on Noble Metal Catalysts, J. Nat. Gas Chem., 15 (2006) 327-334. [7] M.C.J. Bradford, M.A. Vannice, CO2 Reforming of CH4, Catal. Rev. - Sci. Eng., 41 (1999) 1-42. [8] J. Guo, C. Xie, K. Lee, N. Guo, J.T. Miller, M.J. Janik, C. Song, Improving the Carbon Resistance of Ni-Based Steam Reforming Catalyst by Alloying with Rh: A Computational Study Coupled with Reforming Experiments and EXAFS Characterization, ACS Catal., 1 (2011) 574-582. [9] V.A. Sadykov, E.L. Gubanova, N.N. Sazonova, S.A. Pokrovskaya, N.A. Chumakova, N.V. Mezentseva, A.S. Bobin, R.V. Gulyaev, A.V. Ishchenko, T.A. Krieger, C. Mirodatos, Dry reforming of methane over Pt/PrCeZrO catalyst: Kinetic and mechanistic features by transient studies and their modeling, Catal. Today, 171 (2011) 140-149. [10] V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, Transient Kinetic Study of the Oxidation and Hydrogenation of Carbon Species Formed during CH4/He, CO2/He, and CH4/CO2 reactions over Rh/Al2O3 Catalyst, J. Catal., 161 (1996) 31-42. [11] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, Reforming of Methane with Carbon Dioxide to Synthesis Gas over Supported Rhodium Catalysts: II. A Steady-State Tracing Analysis: Mechanistic Aspects of the Carbon and Oxygen Reaction Pathways to Form CO, J. Catal., 158 (1996) 64-75. [12] J. Ashok, S. Kawi, Steam reforming of toluene as a biomass tar model compound over CeO2 promoted Ni/CaO–Al2O3 catalytic systems, Int. J. Hydrogen Energy, 38 (2013) 13938-13949. [13] J. Ashok, M. Subrahmanyam, A. Venugopal, Hydrotalcite structure derived Ni–Cu–Al catalysts for the production of H2 by CH4 decomposition, Int. J. Hydrogen Energy, 33 (2008) 2704-2713. [14] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catal., 4 (2014) 289-301. [15] K. Nakamura, T. Miyazawa, T. Sakurai, T. Miyao, S. Naito, N. Begum, K. Kunimori, K. Tomishige, Promoting effect of MgO addition to Pt/Ni/CeO2/Al2O3 in the steam gasification of biomass, Appl. Catal., B, 86 (2009) 36-44. [16] Z. Alipour, M. Rezaei, F. Meshkani, Effect of alkaline earth promoters (MgO, CaO, and BaO) on the activity and coke formation of Ni catalysts supported on nanocrystalline Al2O3 in dry reforming of methane, J. Ind. Eng. Chem., 20 (2014) 2858-2863. [17] H.M. Swaan, V.C.H. Kroll, G.A. Martin, C. Mirodatos, Deactivation of supported nickel catalysts during the reforming of methane by carbon dioxide, Catal. Today, 21 (1994) 571-578. [18] A.S. Bobin, V.A. Sadykov, V.A. Rogov, N.V. Mezentseva, G.M. Alikina, E.M. Sadovskaya, T.S. Glazneva, N.N. Sazonova, M.Y. Smirnova, S.A. Veniaminov, C. Mirodatos, V. Galvita, G.B. Marin, Mechanism of CH4 Dry Reforming on Nanocrystalline Doped Ceria-Zirconia with Supported Pt, Ru, Ni, and Ni–Ru, Top. Catal., 56 (2013) 958-968. [19] L. Xu, H. Song, L. Chou, Carbon dioxide reforming of methane over ordered mesoporous NiO–MgO–Al2O3 composite oxides, Appl. Catal., B, 108–109 (2011) 177-190.


99

[20] Z. Xu, Y. Li, J. Zhang, L. Chang, R. Zhou, Z. Duan, Bound-state Ni species — a superior form in Ni-based catalyst for CH4/CO2 reforming, Appl. Catal. A, 210 (2001) 45-53. [21] K.-M. Kang, H.-W. Kim, I.-W. Shim, H.-Y. Kwak, Catalytic test of supported Ni catalysts with core/shell structure for dry reforming of methane, Fuel Process. Technol., 92 (2011) 1236-1243. [22] H. Arandiyan, J. Li, L. Ma, S.M. Hashemnejad, M.Z. Mirzaei, J. Chen, H. Chang, C. Liu, C. Wang, L. Chen, Methane reforming to syngas over LaNixFe1−xO3 mixed-oxide perovskites in the presence of CO2 and O2, J. Ind. Eng. Chem., 18 (2012) 2103-2114. [23] S. He, H. Wu, W. Yu, L. Mo, H. Lou, X. Zheng, Combination of CO2 reforming and partial oxidation of methane to produce syngas over Ni/SiO2 and Ni–Al2O3/SiO2 catalysts with different precursors, Int. J. Hydrogen Energy, 34 (2009) 839-843. [24] Q.-H. Zhang, Y. Li, B.-Q. Xu, Reforming of methane and coalbed methane over nanocomposite Ni/ZrO2 catalyst, Catal. Today, 98 (2004) 601-605. [25] S. Sokolov, E.V. Kondratenko, M.-M. Pohl, U. Rodemerck, Effect of calcination conditions on time on-stream performance of Ni/La2O3–ZrO2 in low-temperature dry reforming of methane, Int. J. Hydrogen Energ., 38 (2013) 16121-16132. [26] X. Xie, T. Otremba, P. Littlewood, R. Schomäcker, A. Thomas, One-Pot Synthesis of Supported, Nanocrystalline Nickel Manganese Oxide for Dry Reforming of Methane, ACS Catal., 3 (2012) 224-229. [27] M.M. Nair, S. Kaliaguine, F. Kleitz, Nanocast LaNiO3 Perovskites as Precursors for the Preparation of Coke-Resistant Dry Reforming Catalysts, ACS Catal., 4 (2014) 3837-3846. [28] V.V. Gal'vita, V.D. Belyaev, V.N. Parmon, V.A. Sobyanin, Conversion of methane to synthesis gas over Pt electrode in a cell with solid oxide electrolyte, Catal. Lett., 39 (1996) 209-211. [29] V.C.H. Kroll, H.M. Swaan, S. Lacombe, C. Mirodatos, Methane Reforming Reaction with Carbon Dioxide over Ni/SiO2 Catalyst: II. A Mechanistic Study, J. Catal., 164 (1996) 387-398. [30] Z. Zhang, X.E. Verykios, Carbon dioxide reforming of methane to synthesis gas over Ni/La2O3 catalysts, Appl. Catal. A 138 (1996) 109-133. [31] D. Pakhare, J. Spivey, A review of dry (CO2) reforming of methane over noble metal catalysts, Chem. Soc. Rev., 43 (2014) 7813-7837. [32] A. Slagtern, Y. Schuurman, C. Leclercq, X. Verykios, C. Mirodatos, Specific Features Concerning the Mechanism of Methane Reforming by Carbon Dioxide over Ni/La2O3 Catalyst, J. Catal., 172 (1997) 118-126. [33] A.S. Bobin, V.A. Sadykov, V.A. Rogov, N.V. Mezentseva, G.M. Alikina, E.M. Sadovskaya, T.S. Glazneva, N.N. Sazonova, M.Y. Smirnova, S.A. Veniaminov, C. Mirodatos, V. Galvita, G.B. Marin, Mechanism of CH4 Dry Reforming on Nanocrystalline Doped Ceria-Zirconia with Supported Pt, Ru, Ni, and Ni–Ru, Top. Catal., 56 (2013) 958-968. [34] J. Guo, H. Lou, H. Zhao, D. Chai, X. Zheng, Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels, Appl. Catal. A 273 (2004) 75-82. [35] A. Penkova, L. Bobadilla, S. Ivanova, M.I. Domínguez, F. Romero-Sarria, A.C. Roger, M.A. Centeno, J.A. Odriozola, Hydrogen production by methanol steam reforming on NiSn/MgO–Al2O3 catalysts: The role of MgO addition, Appl. Catal. A 392 (2011) 184-191. [36] N. Hadian, M. Rezaei, Z. Mosayebi, F. Meshkani, CO2 reforming of methane over nickel catalysts supported on nanocrystalline MgAl2O4 with high surface area, J. Nat. Gas Chem., 21 (2012) 200-206. [37] J. Guo, H. Lou, H. Zhao, X. Wang, X. Zheng, Novel synthesis of high surface area MgAl2O4 spinel as catalyst support, Mater. Lett., 58 (2004) 1920-1923. [38] E.N. Alvar, M. Rezaei, Mesoporous nanocrystalline MgAl2O4 spinel and its applications as support for Ni catalyst in dry reforming, Scr. Mater., 61 (2009) 212-215. [39] A. Kambolis, H. Matralis, A. Trovarelli, C. Papadopoulou, Ni/CeO2-ZrO2 catalysts for the dry reforming of methane, Appl. Catal. A 377 (2010) 16-26. [40] T. Miyazawa, T. Kimura, J. Nishikawa, S. Kado, K. Kunimori, K. Tomishige, Catalytic performance of supported Ni catalysts in partial oxidation and steam reforming of tar derived from the pyrolysis of wood biomass, Catal. Today, 115 (2006) 254-262.


100

[41] M. Koike, C. Ishikawa, D. Li, L. Wang, Y. Nakagawa, K. Tomishige, Catalytic performance of manganese-promoted nickel catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Fuel, 103 (2013) 122-129. [42] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Catalytic performance and characterization of Ni-Fe catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Appl. Catal. A 392 (2011) 248-255. [43] D. Li, I. Atake, T. Shishido, Y. Oumi, T. Sano, K. Takehira, Self-regenerative activity of Ni/Mg(Al)O catalysts with trace Ru during daily start-up and shut-down operation of CH4 steam reforming, J. Catal., 250 (2007) 299-312. [44] K. Nagaoka, A. Jentys, J.A. Lercher, Methane autothermal reforming with and without ethane over mono- and bimetal catalysts prepared from hydrotalcite precursors, J. Catal., 229 (2005) 185-196. [45] C. Crisafulli, S. Scirè, R. Maggiore, S. Minicò, S. Galvagno, CO2 reforming of methane over Ni–Ru and Ni–Pd bimetallic catalysts, Catal. Lett., 59 (1999) 21-26. [46] D. Li, Y. Nakagawa, K. Tomishige, Methane reforming to synthesis gas over Ni catalysts modified with noble metals, Appl. Catal. A 408 (2011) 1-24. [47] L. Wang, D. Li, M. Koike, H. Watanabe, Y. Xu, Y. Nakagawa, K. Tomishige, Catalytic performance and characterization of Ni–Co catalysts for the steam reforming of biomass tar to synthesis gas, Fuel, 112 (2013) 654-661. [48] D. Tian, Z. Liu, D. Li, H. Shi, W. Pan, Y. Cheng, Bimetallic Ni–Fe total-methanation catalyst for the production of substitute natural gas under high pressure, Fuel, 104 (2013) 224-229. [49] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst characterization, Pure & Appl. Chem., (1995). [50] Scherrer, Nachr. Ges. Wiss. Göttingen, 1918. [51] Niemantsverdiet J.W., Spectroscopy in Catalysis, 2007. [52] U. Olsbye, O. Moen, Å. Slagtern, I.M. Dahl, An investigation of the coking properties of fixed and fluid bed reactors during methane-to-synthesis gas reactions, Appl. Catal. A, 228 (2002) 289-303. [53] G.F. Froment, K. Bischoff, J. De Wilde, Chemical Reactor Analysis and Design, 3rd ed., J. Wiley & Sons2011. [54] D.E. Mears, Diagnostic criteria for heat transport limitations in fixed bed reactors, J. Catal., 20 (1971) 127-131. [55] V.V. Galvita, H. Poelman, G.B. Marin, Combined chemical looping for energy storage and conversion, J. Power Sources, 286 (2015) 362-370. [56] A.L. Kustov, A.M. Frey, K.E. Larsen, T. Johannessen, J.K. Nørskov, C.H. Christensen, CO methanation over supported bimetallic Ni–Fe catalysts: From computational studies towards catalyst optimization, Appl. Catal. A 320 (2007) 98-104. [57] A. Intenational, Handbook of Binary Alloy Phase Diagrams, in: t. edition (Ed.), 1996. [58] S.B. Simonsen, D. Chakraborty, I. Chorkendorff, S. Dahl, Alloyed Ni-Fe nanoparticles as catalysts for NH3 decomposition, Appl. Catal. A 447–448 (2012) 22-31. [59] K. Gheisari, S. Javadpour, J.T. Oh, M. Ghaffari, The effect of milling speed on the structural properties of mechanically alloyed Fe–45%Ni powders, J. Alloys Compd., 472 (2009) 416-420. [60] V.V. Galvita, H. Poelman, V. Bliznuk, C. Detavernier, G.B. Marin, CeO2-Modified Fe2O3 for CO2 Utilization via Chemical Looping, Ind. Eng. Chem. Res., 52 (2013) 8416-8426. [61] D.W. Goodman, Structure/reactivity relationships for alkane dissociation and Hydrogenolysis using single crystal kinetics, Catal. Today, 12 (1992) 189-199. [62] T.P. Beebe, D.W. Goodman, B.D. Kay, J.T. Yates, Kinetics of the activated dissociative adsorption of methane on the low index planes of nickel single crystal surfaces, J. Chem. Phys., 87 (1987) 2305-2315. [63] N. Shah, D. Panjala, G.P. Huffman, Hydrogen Production by Catalytic Decomposition of Methane, Energy Fuels, 15 (2001) 1528-1534. [64] V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Catalyst-assisted chemical looping for CO2 conversion to CO, Appl. Catal., B, 164 (2015) 184-191.


101

[65] V.I. Alexiadis, N. Boukos, X.E. Verykios, Influence of the composition of Fe2O3/Al2O3 catalysts on the rate of production and quality of carbon nanotubes, Mater. Chem. Phys., 128 (2011) 96-108. [66] R.T.K. Baker, Coking problems associated with hydrocarbon conversion processes, Preprint Paper Am. Chem. Soc. Div. Fuel Chem., 41 (1996) 521-524. [67] V. Galvita, T. Schröder, B. Munder, K. Sundmacher, Production of hydrogen with low COx-content for PEM fuel cells by cyclic water gas shift reactor, Int. J. Hydrogen Energy, 33 (2008) 1354-1360. [68] A. Horváth, G. Stefler, O. Geszti, A. Kienneman, A. Pietraszek, L. Guczi, Methane dry reforming with CO2 on CeZr-oxide supported Ni, NiRh and NiCo catalysts prepared by sol–gel technique: Relationship between activity and coke formation, Catal. Today, 169 (2011) 102-111. [69] J. Wei, E. Iglesia, Isotopic and kinetic assessment of the mechanism of reactions of CH4 with CO2 or H2O to form synthesis gas and carbon on nickel catalysts, J. Catal., 224 (2004) 370-383.

Chapter 4: Carbon gasification from Ni-Fe catalysts after methane reforming reactions

103

Chapter 4

Carbon gasification from Ni-Fe catalysts after

methane reforming reactions

Despite all the different ways to reduce carbon deposition, carbon accumulation during

reforming reactions remains an issue. Eventually, catalyst regeneration is required, by

removing all carbon species by gasification. The catalyst regeneration mechanisms from a Ni-

Fe catalyst were studied using CO2 and O2 as oxidizing agents. Methane dry reforming at

1023 K, atmospheric pressure and a CH4:CO2 molar ratio of 1.1 was used as a reforming

reaction for ageing the catalysts through carbon deposition. The aged/deactivated and

regenerated catalysts were characterized using X-ray photoelectron spectroscopy (XPS),

Raman spectroscopy and energy-dispersive X-ray spectroscopy (EDX)-STEM mapping. The

catalyst regeneration was studied by CO2 and O2 temperature programmed oxidation (TPO)

and by operando time-resolved X-ray diffraction (XRD). A transient response technique,

Temporal Analysis of Products (TAP) , was applied to investigate the isothermal carbon

species gasification.


S.A. Theofanidis, R. Batchu, V.V. Galvita, H. Poelman, G.B. Marin, Carbon gasification from Fe–Ni catalysts after methane dry reforming, Appl. Catal., B, 185 (2016) 42-55.


104

4.1 Introduction

Methane dry reforming (DRM) has been a subject of several studies for a long time[1].

The H2:CO ratio from DRM is more favorable for Fischer-Tropsch and methanol synthesis

than the ratio obtained from classical steam reforming [2]. Moreover, DRM has the lowest

operating cost among these processes and offers the additional advantage of converting CO2

into valuable chemicals:

𝐶𝐻4 + 𝐶𝑂2 ↔ 2𝐶𝑂 + 2𝐻2 ∆H0298K= 247 kJ·mol-1 (4.1)

Side reactions of importance include the reverse water gas shift:

𝐻2 + 𝐶𝑂2 ↔ 𝐻2𝑂 + 𝐶𝑂 ∆H0298K= 41 kJ·mol-1 (4.2)

However, dry reforming technologies have the inherent disadvantage of rapid catalyst

deactivation due to carbon deposition, i.e. coke [3]. Carbon deposition can originate either

from the methane decomposition reaction (Eq. 4.4) or from CO disproportionation [4] (Eq.

4.3):

2𝐶𝑂 ↔ 𝐶 + 𝐶𝑂2 ∆H0298K = -172 kJ·mol-1 (4.3)

𝐶𝐻4 → 𝐶 + 2𝐻2 ∆H0298K = 75 kJ·mol-1 (4.4)

Carbon formation on metals from hydrocarbons decomposition is a complex process which

includes surface catalysis and solid state reactions. Lobo and Trimm in 1971 [5-7] suggested

a mechanism for steady-state carbon deposition on Ni catalysts. They suggested that the

carbon atoms migrate through the Ni surface, upon which the hydrocarbon decomposition

takes place, towards the active growth regions (carbides) where the process of carbon

growth (nucleation) takes place [6]. The carbide phase decomposes at some stage to give

graphite. Molecular mass and chemical structure of these carbon species may vary

depending on the reaction type, conditions and catalyst [3, 8, 9]. Similarly, the carbon

species deposited during DRM vary in morphology and may be carbidic, amorphous,

graphene-like, graphitic, filamentous [4, 10-13].

Several ways have been examined in order to inhibit or control the deactivation

originating from carbon species deposition. Higher dispersion of the active metals on the


105

support surface [14], increase of catalyst basicity to achieve a higher activation rate of mildly

acidic CO2 [15, 16] and addition of materials that offer an oxygen reservoir through their

redox behavior [17-20] have been investigated. Sadykov et al. [21] used CeO2 as an oxygen

reservoir material for DRM and showed that the oxygen mobility of CeO2 can be increased

by incorporation of rare earth metals (La, Gd, Pr) as dopants. They found that catalytic

activity is correlated with oxygen near the surface and/or bulk mobility for elimination of

deposited carbon species. Theofanidis et al. [22] on the other hand used Fe2O3 as promoter

for a Ni/MgAl2O4 catalyst because of its good redox properties [23]. They found lower

amounts of deposited carbon after methane dry reforming on Fe-modified samples in

comparison with pure Ni/MgAl2O4. This was attributed to FeOx formation during DRM

reaction and subsequent oxidation of carbon by lattice oxygen [24-26]. Zhang and Verykios

[16] showed that Ni/La2O3 activity increases after 2-5 h time-on-stream (TOS). This study

revealed a CO2 pool, stored in the form of La2O2CO3.

However, despite all the different ways to reduce carbon deposition, carbon

accumulation during reforming reactions remains an issue. Eventually, catalyst regeneration

is required, by removing all carbon species by gasification [27, 28]. There are many reports in

literature proposing different mechanisms for catalytic carbon gasification. Some of the

mechanisms that are mentioned are: 1) carbon bulk diffusion, where carbon is transported

through the metal particle to the region where the gasification reaction takes place [6, 29,

30], 2) oxygen spillover where metal-activated oxygen may migrate over a considerable

distance over the support towards the deposited carbon species and oxidizing them [31, 32]

and 3) the redox mechanism, where the catalyst provides oxygen towards carbon (reduction

step) and is itself oxidized by the gas phase (oxidation step) [33-35]. Figueiredo and Trimm in

1975 [29] studied the gasification of carbon deposits on Ni foil and supported Ni catalysts.

They found a zero reaction order for carbon gasification by steam, concluding that the

reaction is controlled by the diffusion of carbon through Ni. Machida et al. [36] used CeO2 as

an active catalyst for soot oxidation concluding to two possible reaction pathways: i) soot

oxidation by adsorbed superoxide species (O2-) at the three-phase boundary (catalyst,

carbon, gas) and ii) soot oxidation by active lattice oxygen at the CeO2/ soot interface.

The rate of gasification depends on the structure of the carbon [37], its location [11] and

on the nature of the catalysts present [10, 38, 39]. One important method for carbon species


106

characterization is temperature programmed (TP) techniques [40, 41]. However, isothermal

studies are also required in order to understand the kinetics [28]. Oxygen is one of the gases

most often used [42] to infer type and location of carbon species on the catalyst while there

are also studies on carbon gasification by CO2, H2O and H2 [4, 29, 31, 43]. Among the latter,

the mechanism according to which carbon species are oxidized by CO2, as well as, the

differences between CO2 and O2 carbon species removal await clarification. In view of the

promising results on Ni-Fe/MgAl2O4 [22] regarding reduced carbon species deposition, this

material was used for further investigation of carbon removal. The questions raised are

addressed by performing X-ray diffraction (XRD) characterization of used Ni-Fe/MgAl2O4

catalysts, both in-situ during O2 temperature-programmed oxidation (TPO), as well as

operando by coupling in-situ XRD with MS. Further, for first time to our knowledge a

transient response technique, Temporal Analysis of Products (TAP) , has been used to

investigate the isothermal carbon species gasification process. The latter is recognized as an

important experimental method for heterogeneous catalytic reaction studies. A TAP pulse

response experiment consists of injecting a very small amount of gas, typically nanomoles

per pulse, into a tubular fixed bed reactor that is kept under vacuum. The time-dependent

exit flow rate of each gas is detected by a mass spectrometer. The high time resolution of

the TAP technique allows detection of short-( millisecond time scale) and/or long-lived (>1s)

reaction intermediates, which helps to formulate the mechanism of reaction [44, 45].

4.2 Experimental Methods

For more details about the characterization techniques that have been used in this

chapter the reader is referred to the Chapter 2.

4.2.1 Catalyst preparation

Support preparation

The support material of MgAl2O4 was prepared by co-precipitation from an aqueous

solution of Mg(NO3)2·6H2O (99%, Sigma-Aldrich®) and Al(NO3)3·9H2O (98.5%, Sigma-Aldrich®)

(molar ratio Mg:Al=1:2). A precipitating agent, NH4OH (ACS reagent, 28.0-30.0% NH3 basis)

was added to adjust the pH to 10, at 333 K. The formed precipitate was filtered, dried at 393

K for 12h and finally calcined in air at 1023 K for 4h.


107


8wt%Ni-5wt%Fe (named as “as-prepared Ni-Fe”) was prepared by incipient wetness

impregnation on the support (MgAl2O4) using an aqueous solution of corresponding nitrates

Ni(NO3)2·6H2O (99.99+%, Sigma-Aldrich®) and Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®)[46].

The catalysts were dried at 393 K for 12h and subsequently calcined in air at 1023 K for 4 h.


The BET surface area of the catalyst was 84.7 ± 5.8 m2/gcat. A more detailed

characterization of the catalyst has been reported elsewhere[22].The crystallographic phases

of the materials were confirmed by ex-situ XRD measurements (Siemens Diffractometer

Kristalloflex D5000, Cu Kα radiation). The powder patterns were collected in a 2θ range from

10o to 80o with a step of 0.02o and 30 s counting time per angle. XRD patterns of known

compounds are referenced by their corresponding number in the Powder Diffraction File

database.






XPS measurements were recorded with a S-Probe XPS spectrometer (VG, Surface Science

Instruments), equipped with a monochromatized 450W Al Kα source. The base pressure of

the analysis chamber was below 2∙10−7 Pa. Spectra were recorded with 200W source power.

The analyser axis made an angle of 45◦ with the specimen surface. Wide scan spectra were

measured with a pass energy of 157 eV and a 0.2 eV step, while core levels were recorded

with a step of 0.05 eV and a pass energy of 107.8 eV. Energy calibration to Al2p or C1s gave

the same result, so the main intensity of C1s at 284.6 eV was used for alignment.

Raman analysis of the samples was performed on a RXN1 Raman spectrometer (Kaiser

Optical Systems) fitted with a 532 nm laser operating at 40 mW.



108

4.2.3 Ageing of catalyst during DRM

The ageing of the catalyst was performed at atmospheric pressure in a quartz reactor

with an internal diameter of 9 mm, which was housed inside an electric furnace. The

temperature of the catalyst bed was measured with K-type thermocouples touching the

outside and inside of the reactor at the position of the catalyst bed. Based on previous

experiments, this treatment allowed to fully reduce the material (see §3.3.1). 70 mg of

sample (Ni-5Fe/MgAl, see §3.2) with particle size of 30 μm, diluted with inert Al2O3 for

improved heat conductivity (ratio catalyst/inert =1/40), was packed between quartz wool

plugs resulting in approximately 1.3·10-2 m of catalyst bed length. Control characterization

experiments were performed on this material (referred to as “reduced Ni-Fe”). After this

step a mixture of CH4, CO2 and He (total flow of 80 mL/min, volumetric ratio

CH4:CO2:He=1.1:1:1, He internal standard) started flowing through the reactor for carbon

species deposition on the catalyst during DRM reaction at 1023 K and atmospheric pressure

for 1h, yielding a “used Ni-Fe catalyst”. Produced CO, H2 and unconverted CH4 and CO2 were

detected at the outlet using a calibrated OmniStar Pfeiffer mass spectrometer (MS). MS

signals were recorded for all major fragments. For more details about the quantification

methodology of reactants and products the reader is referred to §2.3.

4.2.4 Carbon species temperature programmed oxidation (TPO)

The TPO measurements were performed subsequent to the ageing experiment at

atmospheric pressure after cooling down to room temperature. In a typical TPO experiment,

the used catalyst was heated from ambient temperature to 1123 K under CO2 or 10%O2/He

stream. The heating rate used for CO2- and O2-TPO was 10 K/min and the total gas flow rate

was 1 NmL/s. The TPO experiment was performed over the used Ni-Fe catalyst (Ni-

5Fe/MgAl, see §3.3.1) as well as over Ni-0Fe/MgAl (see §3.3.1) and commercial graphite

(Sigma-Aldrich®, powder, <20 μm), as reference materials for comparison purposes.

4.2.5 Operando XRD

The evolution of the structure during CO2-TPO was investigated with time-resolved XRD.




109



were measured with a K-type thermocouple and corrected afterwards to a calibration curve

of the heating device, which is based on the eutectic systems Au–Si, Al–Si and Ag–Si. For

each sample, approximately 10 mg of powdered sample was evenly spread on a single

crystal Si wafer. Interaction of the catalyst material with the Si wafer was never observed.

Gases were supplied to the reactor chamber with calibrated mass-flow controllers. The

sample was heated from room temperature to 1123 K at a heating rate of 20 K/min in a

flowing gas stream (10mL/s of CO2 or 10%O2/He). The outlet gas stream (CO, CO2 and He)

was monitored online using a calibrated OmniStar Pfeiffer mass spectrometer (MS).

A full XRD scan (10o to 65o with step of 0.02o) was taken at room temperature before and

after the TPO experiment. Samples were cooled in helium flow to room temperature after

each experiment.

4.2.6 Isothermal Temporal Analysis of Products (TAP) experiments for carbon

gasification

Transient measurements were performed in a TAP-3E reactor (Mithra Technologies, St.

Louis, USA) equipped with an Extrel Quadrupole Mass Spectrometer (QMS). The details of

TAP experiments, can be found in [44]. For the experiments, 20 mg (250<d<500 μm catalyst

fraction) of the calcined catalyst was placed in a quartz microreactor (I.D = 4mm and ~2 mm

bed length), which was located between two inert beds of quartz particles with the same

sieved fraction. The temperature of the catalyst was measured by a K-type thermocouple

housed inside the catalytic zone. Prior to the experiments, the catalyst was reduced at 1023K

in a flow of 10%H2/He at atmospheric pressure. A series of CH4 pulses (~10-7 mol/pulse) were

used isothermally for carbon deposition on the catalyst, yielding an “aged Ni-Fe catalyst”,

and the amu signals of 2 and 16 were monitored by QMS. For every 1 second, data were

recorded with millisecond time resolution in each pulse.

A sequence of isothermal O2 or CO2 pulses were admitted into the microreactor for

carbon gasification and CO2, CO responses were monitored at amu signals of 44 and 28

respectively. The latter were recorded with millisecond time scale resolution for 20 seconds


110

collection time. The burn-off sequence continued up to the completion of carbon

gasification with CO2 and CO intensity responses equal to zero.

4.3 Results

4.3.1 Catalyst Characterization

The crystalline phases of the Ni-Fe catalyst after DRM and after CO2-TPO for carbon

species removal are presented in Figure 4.1. Identification of the XRD patterns is performed

using the invariable peaks of the support as standard and interpreting the remainder

diffractions in terms of position and relative intensity. The used catalyst shows peaks of

MgAl2O4 (31.7o,37o,45o, 55.5o, 59o, 65o, Powder Diffraction File (PDF) card number: 00-021-

1152), Ni-Fe alloy peaks (44o and 51.5o, PDF card number: 00-038-0419) and graphite

diffraction peaks (26.6o, 43.4o, 46.3o,56.7o PDF card number: 01-075-2078) . Nickel or iron

carbide diffraction peaks were not detected. The XRD pattern following the CO2-TPO shows

that the Ni-Fe alloy was decomposed to Ni (44.5o and 51.8o) and Fe3O4

(30.1o,36o,43.5o,57o,63o). The NiAl2O4 crystalline phase that was present in the reduced

sample (not shown) [22] disappeared after DRM but a small diffraction peak appeared after

CO2 oxidation. MgAl2O4 diffraction peaks remained stable after CO2-oxidation at 1123 K

while the graphite peak was still present but with reduced intensity. This implies that the

carbon species cannot be completely removed by CO2. The XRD pattern after O2-TPO

showed that the graphite peak completely disappeared and that the catalyst structure was

now similar to the structure of the fresh catalyst with separate Fe2O3 (maghemite) and NiO

supported on MgAl2O4.

The elemental distribution of Ni-Fe is indicated in Figure 4.2 using energy-dispersive X-ray

spectroscopy (EDX)-STEM mapping. Carbon species (red) cover the Ni (blue) and Fe (green)

particles that are in alloy formation (light blue) after DRM (Figure 4.2(A)). In contrast, after

CO2 oxidation Ni and Fe particles are segregated as Fe is oxidized to Fe3O4 and the Ni-Fe alloy

is decomposed. As it is reported elsewhere [22] a subsequent H2 reduction step can lead

again to Ni-Fe alloy formation restoring the initial activity.


111

Figure 4.1: Full XRD scans of used Ni-5Fe/MgAl (DRM for 1 h, 1023 K, CH4:CO2:He= 1.1:1:1, total pressure of

101.3 kPa), after CO2-TPO (maximum temperature 1123 K, flow rate 10 mL/s) and after O2-TPO (maximum

temperature 1123 K, flow rate 10 mL/s).

Figure 4.2: EDX element mapping of Ni-5Fe/MgAl. A) after DRM (1023 K, CH4:CO2:He= 1.1:1:1, total pressure

of 101.3 kPa, reaction time 1 h). (B) after CO2 oxidation (1mL/s of CO2 at a total pressure of 101.3 kPa and

1123K). Red, green and blue colors correspond to carbon, Fe and Ni elements respectively.

Furthermore, it is observed that the carbon species in contact with the active metal phase

are mainly removed. Figure 4.2(B) shows that after CO2 oxidation there are carbon species

located at a distance from Ni and Fe. This implies that gas phase CO2 does not interact

directly with the deposited carbon species, up to 1123 K.

22 27 32 37 42 47 52 57 62 67 72

Inte

nsi

ty a

.u.

2θ

θ Fe3O4 ◊ Graphite

• MgAl2O4 Δ Niο NiAl2O4 + Fe2O3

□ Ni-Fe alloy * NiO

•

•

•

• •

οοθ

θθ

□

□

◊

Δ

Δ

θ

◊

◊

After CO2-TPO

After O2-TPO

Spent catalyst

(A) (B)


112

4.3.2 Carbon species temperature programmed oxidation

CO2- and O2-TPO profiles are presented in Figure 4.3 comparing used Ni-Fe with the

reference materials of Ni/MgAl2O4 and graphite, for the same heating rate of 10 K/min.

Carbon oxides intensities as a function of temperature are illustrated. The CO signal for the

used Ni sample (Figure 4.3(A)) showed one maximum at 890 K and no multiple-peak profile.

However, one maximum at 1020 K and one shoulder at 890 K were observed during CO2-TPO

over used Ni-Fe. The shoulder at lower temperature can be attributed to carbon species

oxidation by Ni surface oxygen, originated from CO2 dissociation, as it coincides with the

peak maximum of the TPO profile obtained over used Ni. Metallic Ni remains stable under

CO2 and is not oxidized to NiO up to 1123 K [22, 23]. Furthermore, the main peak maximum

for used Ni-Fe can be assigned to carbon species oxidation by Fe oxide. This implies that the

mechanism of carbon species removal over Ni is different from that over Ni-Fe catalyst. Pure

graphite was also investigated during CO2-TPO but no produced CO was detected up to 1173

K.

Figure 4.3(B) displays the CO2 intensity upon O2-TPO with respect to temperature for used

Ni, Ni-Fe and pure graphite. The deposited carbon species can either be oxidized by lattice

oxygen, by surface oxygen originated from O2 dissociation or by direct interaction with gas

phase O2. The CO2 signal now showed one maximum for the studied samples. However, the

peak maximum obtained by oxidation of carbon species over Ni and Ni-Fe is shifted to lower

temperatures than that of pure graphite oxidation, implying that the deposited carbon

species are mainly oxidized by lattice or surface oxygen. However, as the temperature

increases, direct interaction of carbon species with gas phase O2 also takes place,

contributing to the gasification process. The lower peak maximum of 850 K for Ni vs 920 K

for Ni-Fe, implies that the carbon removal mechanism is somewhat different over both

samples.


113

Figure 4.3: (A) CO2-TPO profile: CO intensity as a function of temperature for used (DRM for 1 h, 1023 K,

CH4:CO2:He= 1.1:1:1, total pressure of 101.3 kPa) Ni-0Fe/MgAl, Ni-5Fe/MgAl and graphite heating rate of 10

K/min, flow rate of 1 mL/s of CO2. (B) O2-TPO profile: CO2 intensity as a function of temperature for used Ni-

0Fe/MgAl, Ni-5Fe/MgAl and graphite, heating rate of 10 K/min, flow rate of 1 mL/s of 10%O2/He. Black line:

Ni-5Fe/MgAl, grey line: Ni-0Fe/MgAl and dashed line: graphite

Figure 4.3 shows that a higher amount of carbon was deposited on Ni-5Fe/MgAl than on

the Ni catalyst. This is attributed to the CH4-rich environment in which DRM takes place

during catalyst ageing (volumetric CH4:CO2:He= 1.1:1:1). Bimetallic catalysts are more active

for methane decomposition reaction than monometallic catalysts [47, 48]. Theofanidis and


114

co-workers [22] showed that during DRM with CH4:CO2= 1:1, Fe addition suppresses carbon

deposition on Ni as oxygen from the FeOx lattice is transferred to nearby Ni atom, producing

CO.

4.3.3 Carbon characterization

Raman spectroscopy is widely used in order to investigate the structure and the crystallite

size of carbon species [49]. It provides information about the electronic properties and can

detect the presence of ordered carbon species [4]. The Raman spectrum of a single crystal

graphene sample only shows the G band at approximately 1581 cm-1 Raman shift. The

commercial graphite, that was used as a reference, shows two peaks at 1555 and 1580 cm-1

Raman shift (blue line in Figure 4). However, in case of imperfect, polycrystalline graphite

and other carbonaceous materials [50], additional bands are detected at 1355 cm-1 (D band)

and 1620 cm-1 (D’ band). The ratio of areas ID/IG has been correlated to the inverse crystallite

size of graphite [51]. Figure 4.4 shows the Raman spectra for the graphite, the used Ni-

5Fe/MgAl catalyst after 1h TOS during DRM at 1023 K, the used Ni-5Fe/MgAl catalyst after

CO2-TPO to 950 K (removal of shoulder peak on Figure 4.3A) and the used Ni-5Fe/MgAl

catalyst after CO2-TPO to 1123 K. The analysis of used Ni-5Fe/MgAl (black line in Figure 4.4)

confirmed the existence of two types of carbon species structures. The G band of single

crystal graphene shifted from 1581 cm-1 to 1584 cm-1 implying the presence of graphitic-like

carbon species in the catalyst (more graphene layers). According to literature, the G Raman

peak changes in position, shape and intensity as a function of the number of graphene layers

[52]. The D and D’ bands at 1350 and 1619 cm-1 were also observed and can be attributed to

a defective and disordered structure [50, 52]. The disordered carbon species structure,

following from the D band, can be amorphous. Amorphous and graphitic-like carbon were

also observed by Guo and co-workers [4], who performed Raman spectroscopy over

Ni/MgAl2O4 after coking via CH4 temperature programmed decomposition. The Raman

spectrum of Ni-5Fe/MgAl after CO2-TPO at 950 K (grey line in Figure 4.4) showed the same

peaks as the used Ni-5Fe/MgAl catalyst (black line in Figure 4.4) that was not treated by

oxidation, implying the existence of the same types of carbon. This enhances the

aforementioned statement that the shoulder peak in the CO2-TPO profile of used Ni-

5Fe/MgAl (Figure 4.3A) is ascribed to carbon species oxidation by Ni surface oxygen,

originating from CO2 dissociation. Finally, the same type of carbon is observed on the Ni-


115

5Fe/MgAl catalyst after CO2 treatment at 1123 K, implying that only carbon close to active

metals can be oxidized by CO2. This is in agreement with the XRD full scan patterns (Figure

4.1) and EDX element mapping of Ni-5Fe/MgAl (Figure 4.2) where carbon was still observed

after CO2 oxidation.

Figure 4.4: Raman spectrum of the used Ni-5Fe/MgAl sample (DRM for 1 h, 1023 K, CH4:CO2= 1.1, total

pressure of 101.3 kPa). Blue line: graphite, black line: used Ni-5Fe/MgAl catalyst, grey line: used Ni-

5Fe/MgAl catalyst after CO2-TPO up to 950 K (removal of shoulder peak on Figure 2A), purple line: Ni-

5Fe/MgAl catalyst after CO2-TPO up to 1123 K.

Figure 4.5(A) shows a TEM image of a used Ni-5Fe/MgAl catalyst. The presence of

filamentous carbon with Ni-5Fe/MgAl alloy particles on top is observed, which can be

verified by the EDX mapping (Figure 4.5(B-D)).


116

Figure 4.5: (A): HRTEM image of a used Ni-5Fe/MgAl catalyst (after DRM at 1023 K, CH4:CO2:He= 1.1:1:1,

total pressure of 101.3 kPa, reaction time 1 h). EDX element mapping of (B): carbon, (C): Ni and (D): Fe.

XPS was performed on the material in two different states: reduced and used Ni-

5Fe/MgAl . The XPS wide scans showed photoemission lines for C, O, Mg, Al, Ni and Fe. The

surface concentrations of Ni and Fe in the fresh sample amounted to 4.5 and 1.0 at%,

respectively. The wide scan after reaction in CH4:CO2 with 1:1 ratio stood out as its survey

was dominated by a very strong C1s intensity. Next to this photoemission peak, only O1s

was detected, while other elements were not observable. The latter evidenced the presence

of a large amount of C at the surface, covering all of the catalyst surface with considerable

thickness, thereby suppressing photoelectron signals from other surface elements.

Figure 4.6 shows the C1s photoemission signal for these two states of the material,

calibrated to 284.6 eV. For the reduced state, a shoulder is apparent at binding energy 288.3

eV, next to the main C intensity. This C1s signal is representative of adventitious carbon

contamination and consists of a main component at 284.6 eV for C-C bonds, and smaller

contributions at 286 eV for O-C and 288.3 eV for C=O.

Fe-Ni alloy particle

Carbon filaments

(B) (C) (D)

(A)


117

Figure 4.6: XPS detail windows of the C1s photoline for Ni-5Fe/MgAl : a) reduced b) after DRM reaction in

CH4:CO2:He= 1:1:1.

The C1s signal after reaction with CH4:CO2 in a 1:1 ratio differs substantially. First, the

main C1s contribution is much more intense than in the previously discussed state.

Secondly, it is asymmetric to higher binding energy and has an extra feature at 291 eV. Such

C1s photoline is consistent with graphite or graphitic compounds with shake-up satellite ~6

eV higher in binding energy due to π π* transitions [53, 54].

Summarizing, the existence of two types of carbon species, amorphous and graphitic-like

i.e. filaments , can be concluded from both Raman spectroscopy, HRTEM and XPS. Nickel and

iron carbide were not detected from XPS which is in agreement with Figure 1 where no

carbide diffraction peaks were observed during XRD full scans. However, Galvita at al. [23]

observed Fe3C formation during CH4 reduction of Ni/CeO2-Fe2O3. During DRM Fe partially

segregates from Ni-Fe alloy, forming FeOx due to CO2 presence. The FeOx formation removes

carbon species as these are oxidized by FeOx lattice oxygen producing CO. Hence, it is likely

that most of the carbon species will be formed during dry reforming on the Ni surface [22].


118

4.3.4 Mechanism of carbon removal by CO2.

An operando experiment was performed over used Ni-5Fe/MgAl during CO2-TPO coupling

in-situ time resolved XRD with MS. In-situ XRD (Figure 4.7 (A)) showed that during oxidation,

the Ni-Fe alloy peak (43.7o- 44.2o) shifted to higher 2θ values (~44.5o) corresponding to

metallic Ni. In addition, two peaks at diffraction areas 35.4o-36.4o and 42.8o-43.5o appeared

implying the formation of Fe3O4. This is in agreement with previous results [22] where Ni-Fe

alloy was found to decompose during CO2 oxidation yielding two separate phases of Ni and

Fe3O4. The intensity of the graphite diffraction in the angular region 25.8o-26.8o decreased at

1050 K because of its removal by CO2. However, Figure 4.7(A) shows that graphite was not

completely removed by CO2 which is in agreement with Figure 4.2(B) where carbon species

were still observed on the support after CO2-TPO. Figure 4.7(B) shows the CO intensity

variation with temperature obtained by MS corresponding to CO produced by carbon

species removal. Another operando CO2-TPO experiment (not shown) over reduced Ni-

5Fe/MgAl catalyst was performed in order to separate the CO production originating from Fe

oxidation and from carbon species removal by CO2. No CO intensity was detected by MS

because of low CO concentration whereas Fe3O4 formation was observed in the 2D in-situ

XRD patterns. This implies that CO intensity during the operando CO2-TPO over the used Ni-

5Fe/MgAl catalyst (Figure 4.7(B)) corresponds to carbon species removal by CO2. The kinetics

of alloy decomposition and graphite oxidation are illustrated in Figure 4.7(C) as time-

dependent integral intensities of the Ni-Fe alloy, Ni, Fe3O4 and graphite angular regions.


119

Figure 4.7: In-situ XRD coupled with MS during CO2-TPO (heating rate 20 K/min, maximum temperature 1123

K, flow rate 10 mL/s) of used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4:CO2:He= 1.1:1:1, total pressure

of 101.3 kPa):(A) 2D XRD pattern; (B) CO produced during carbon species removal as a function of

temperature; (C) Integral intensity variation of (A) for diffraction areas 25.8-26.8o (Graphite), 35.4

o-36.4

o

(Fe3O4) and 43.7o-44.2

o (Ni-Fe alloy).

The intensity of characteristic peaks associated with crystalline MgAl2O4 did not change

throughout CO2-TPO up to the final temperature of 1123K. The small shift of angular

positions compared to full scan XRD patterns at room temperature can be ascribed to

temperature-induced lattice expansion and different sample height. The peak maximum of

CO in Figure 4.7(B) at 1050 K is in agreement with the temperature at which the graphite


120

peak intensity (25.8o-26.8o) decreases in Figure 4.7(C). The formation of Fe3O4 was observed

only at 1050 K (Figure 4.7(C)), i.e. the same temperature at which the graphite diffraction

peak was removed, while it was expected to happen at lower temperature. Figure 4.8(A)

shows a selected angular region of 35o-40o from Figure 4.7(A), for used Ni-5Fe/MgAl , where

the most intense peak of Fe3O4 (35.5o) is located. Figure 4.8(B) shows the results obtained by

in-situ XRD for the same selected angular region (35o-40o) during CO2-TPO of the reduced Ni-

5Fe/MgAl catalyst. The start of Fe3O4 formation during CO2-TPO of the used Ni-5Fe/MgAl

catalyst is detected at higher temperature, 1050 K, than the 850 K, for CO2-TPO of the

reduced catalyst. This is in agreement with Galvita et al. [55], who performed CO2-TPO over

reduced Ni/CeO2-Fe2O3 and reported reoxidation of Fe to Fe3O4 at 750 K. It implies that in

case of CO2-TPO of used catalyst two reactions occur involving Fe, namely oxidation of Fe by

CO2 and reduction of FeOx by carbon species respectively, up to 1050 K. At this temperature

Fe3O4 is formed because the carbon species that were in contact with the catalyst particles

have been removed.

Figure 4.8: 2D in-situ XRD pattern during CO2-TPO (heating rate 20 K/min, maximum temperature 1123 K,

flow rate 10 mL/s). (A) Used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4:CO2= 1.1, total pressure of 101.3

kPa). Selected angular region of 35o-40

o from Figure 6(A). (B) Reduced Ni-5Fe/MgAl catalyst (heating rate: 30

K/min, maximum temperature 1123 K, flow rate: 1 mL/s, 10%H2/He). Selected angular region of 35o-40

o.


121

The results obtained from the operando XRD experiment are consistent with a

mechanism consisting of two parallel processes: 1) dissociation of CO2 over Ni followed by

the oxidation of carbon species by surface oxygen; 2) Fe oxidation by CO2 and subsequent

carbon species oxidation by Fe oxide lattice oxygen (Fe oxide reduction step). In contrast,

CO2 does not interact directly from gas phase with carbon species that are located far from

the active sites.

A graphical illustration of carbon species removal over the Ni-5Fe/MgAl catalyst is

depicted in Figure 4.9. In the latter, the carbon representation is schematic and does not

reflect the real structure as determined from the carbon characterization section (see

§4.3.3).

Figure 4.9: Schematic representation of carbon species removal by CO2 over Ni-5Fe/MgAl catalyst. Cs:

deposited carbon. Os: surface oxygen, OL: lattice oxygen. Cm: carbon deposited on metals, Cs: carbon

deposited far from metals, Os: surface oxygen, OL: lattice oxygen. The carbon illustration is not corresponding

to the real carbon structure.

4.3.5 In-situ time resolved XRD during O2-TPO

To examine the differences in mechanism between carbon species oxidation by CO2 and

O2, an O2-TPO experiment over used Ni-5Fe/MgAl by means of in-situ XRD was performed

(Figure 4.10). During oxidation the Ni-Fe alloy peak (43.7o- 44.2o) shifted to lower 2θ values

(~43.4o) corresponding to NiO (Figure 4.10(A)). In addition, a peak in diffraction region 35.7o-

36.4o appeared implying the formation of Fe2O3. The graphite diffraction intensity of 25.8o-

26.8o disappeared at 950 K because of its oxidation by O2. This is in agreement with Figure

4.3(B) where it was shown that during O2-TPO, carbon species were completely burnt at 960

K. In contrast to CO2 oxidation (Figure 4.7(A)), where graphite was not completely removed

MgAl2O4

Ni Fe3O4Cs

Os

Cm Cm

FeOL


122

by CO2, O2 can oxidize graphite completely at lower temperature. The kinetics of alloy

decomposition and graphite oxidation are illustrated in Figure 4.10(B) as time-dependent

integral intensities of the graphite, Ni-Fe alloy and Fe2O3 angular regions.

Figure 4.10: In-situ XRD during O2-TPO of a used Ni-5Fe/MgAl catalyst (DRM for 1 h, 1023 K, CH4:CO2= 1.1,

total pressure of 101.3 kPa). (A) 2D XRD pattern for Ni-5Fe/MgAl . Heating rate: 20 K/min, maximum

temperature 1123 K, flow rate: 10 mL/s, air. (B) Integral intensity variation of (A) for diffraction areas 25.8o-

26.8o (graphite), 35.7

o-36.4

o (Fe2O3) and 43.7

o -44.2

o (Ni-Fe alloy).

The mechanism of carbon species gasification during O2-TPO is different than carbon

species removal during CO2-TPO (Figure 4.10). Fe starts to segregate from the Ni-Fe alloy at


123

780 K, while the decomposition is complete at 850 K and thereafter the graphite oxidation

starts. The carbon species oxidation could be ascribed to several processes: lattice oxygen

transfer from nickel and iron oxides, oxygen spillover originating from O2 dissociation,

particle migration to carbon that is deposited far from active metals and direct interaction of

gas phase O2 with the carbon. The oxidation of Fe and Ni to Fe2O3 and NiO respectively, was

only observed in XRD when graphite was completely removed.

4.3.6 O2-TPO over different catalyst bed configurations

To further investigate the contribution of oxygen spillover mechanism to the carbon

gasification process another experiment was performed involving O2- and CO2- TPO over

three different catalyst bed configurations: a mechanical mixture of graphite and Ni-

5Fe/MgAl catalyst, a two and a three layers catalyst bed. A CO2-TPO of all three

aforementioned catalyst bed configurations showed no CO production up to 1173 K (not

shown) implying that the carbon species which are located far from the active metals cannot

be removed by CO2. Figure 4.11(A) displays the CO2 intensity as a function of temperature

during O2-TPO of the mechanical mixture and pure graphite. The maximum of CO2

production for the case of the mechanical mixture is shifted to lower temperature (920 K)

than for pure graphite (1100 K). Complete oxidation of the graphite content of the

mechanical mixture is observed at 1040 K. Compared to the mechanical mixture, graphite

oxidation from the used Ni-5Fe/MgAl sample occurred at close to the same temperature

(see Figure 4.10(B)) implying that in carbon oxidation behavior, the used catalyst resembles

a mechanical mixture of Ni-5Fe/MgAl catalyst and graphite.

O2-TPO was also performed over a two- and three-layers catalyst bed (Figure 4.11(B)).

The two-layers catalyst bed included one layer of Ni-5Fe/MgAl catalyst and one layer of

graphite (dot-line) while the three-layers catalyst bed included Ni-5Fe/MgAl catalyst,

MgAl2O4 and graphite (dash-dot dot line). The peak maximum during O2-TPO over the two-

layers reactor bed occurred at lower temperature (1030 K) compared to the three-layers

reactor bed (1100 K). This was attributed to the fact that there was contact at the interface

between the catalyst and graphite, allowing graphite oxidation through lattice oxygen of

nickel and iron oxides.


124

Figure 4.11: O2-TPO profile: CO2 intensity as a function of temperature for (A) mechanical mixture of graphite

and Ni-5Fe/MgAl catalyst (solid line) and for graphite only (dashed line), (B): two and three layers catalyst

bed configuration, including one layer of Ni-5Fe/MgAl catalyst and one layer of graphite (dot line) and a Ni-

5Fe/MgAl catalyst, MgAl2O4 support and graphite (dash-dot dot line), respectively. Heating rate of 10

K/min, flow rate of 1 mL/s of 10%O2/He.

The peak maximum during O2-TPO over the three-layers bed lies at the same temperature

as for pure graphite (1100 K) indicating that oxygen surface species do not migrate through

Graphite

Fe-Ni catalyst

(A)

Graphite

Fe-Ni catalyst

MgAl2O4

(1) (2) (B)


125

the layer of MgAl2O4 in between Ni-5Fe/MgAl catalyst and graphite. Therefore, there is no

evidence of oxygen spillover to graphite and its contribution to the carbon gasification

process is considered negligible. Furthermore, the better the contact between the catalyst

and the carbon, the lower the temperature for carbon oxidation (Figure 4.11).

4.3.7 Temporal Analysis of Products (TAP) experiments

TAP reactor experiments were applied in order to investigate the isothermal carbon

gasification process by CO2 and O2 at 993 K. As a first step, the catalyst was aged by a

sequence of 400 CH4 pulses. Then, pulses of the oxidizing gas (CO2 or O2) were admitted to

the reactor. Figure 4.12 shows the 2D view of CO2 responses as a function of pulse number

over Ni and Ni-5Fe/MgAl . No CO formation was observed during O2 pulses. Changing of

color from blue to dark red corresponds to an increase in CO2 produced during pulsing.

Figure 4.12(A) showed a gradual intensity increase of the CO2 response during the first five

pulses, while shifting the peak maximum to higher times. In order to visualize the individual

response peak shapes, Figure 4.12(B) displays CO2 responses after the 1st, 4th and 7th O2

pulse over aged Ni. After the 1st pulse, the CO2 peak maximum grew into a sharp peak

located around time equal to 0.5 s, while at the 7th pulse a more broad peak is located at 6s.

Throughout these pulses, conversion of O2 was 100% (see Figure 4.13B). The low CO2

response intensity at the start of the TAP experiment thus indicates O2 is being consumed in

a process, adsorption or metal oxidation, other than carbon gasification. The gradual

increase in CO2 intensity then shows gasification is becoming more important. The peak

maximum shift on the other hand, points to an additional process of carbon gasification,

requiring more time for CO2 to appear. Most of the carbon is burnt-off after the 7th O2 pulse.

For aged Ni-5Fe/MgAl , the onset of CO2 response is longer compared to Ni (Figure 4.12(C)).

The CO2 intensity steadily increased and shifted to higher peak maximum times from 0.1 s

for pulse 1, 0.7 s at pulse 15, to 1.2 s for the 29th pulse, further to 1.8 s at the 40th and back

to 0.12 s at the 50th pulse. Up to the 45th pulse, the O2 conversion remained at 100% (see

Figure 4.13C). Qualitatively similar results were obtained for carbon gasification by O2 pulses

at lower temperatures of 873 K and 923 K, which can be ascribed to the strong exothermicity

of the carbon gasification reaction, yielding an increase in the local temperature.


126

Figure 4.12: CO2 response during O2 pulses at TAP reactor at 993 K. (A): 2D view for Ni, (B): CO2 molar flow

rate produced during selected O2 pulses over Ni, (C): 2D view for Ni-5Fe/MgAl , (D): CO2 molar flow rate

produced during selected O2 pulses over Ni-5Fe/MgAl . Ni and Ni-5Fe/MgAl aged by a sequence of 400 CH4

pulses.

The onset in CO2 response intensity observed for both Ni and Ni-5Fe/MgAl catalysts

implies that the initial O2 pulses are not involved in carbon gasification. From XRD during O2-

TPO (Figure 4.10) it was observed that alloy decomposition and metal oxidation took place

before graphite’s burn-off. Hence, from the present TAP experiments it is concluded that

metals oxidation occurred first and only thereafter the carbon gasification process started.

0

4

8

12

16

20

Tim

e, s

4 62 8 10 12 14 16 18 20 22

Pulse number

2

4

6

8

10

12

x 10-7

mo

ls-1

0 2 4 6 8 100

1x10-6

2x10-6

3x10-6

Mo

lar

flo

w r

ate

, m

ol C

O2 s

-1

Time, s

1st

4th

7th

mo

ls-1

(A) (B)

Tim

e, s

Pulse number

(C)(D)


127

Figure 4.13: 2D projected view of O2 response from TAP during O2 pulses at 993 K over: (A) reduced Ni, (B)

aged Ni and (C) aged Ni-Fe catalyst.

The shift in time of the CO2 peak maximum indicates a change in process of CO2

formation. Possible processes are: a) oxygen spillover from the active metals to carbon on

the support, b) migration of active particles towards deposited carbon, c) carbon bulk

diffusion through the metal particle to the region where the gasification reaction takes

place.

The observed time shift is related to surface transport, implying a diffusion process. The

diffusion coefficient can be assessed by applying the Einstein approximation equation (Eq.

5):

D= x2/t (5),

where x is the mean traveled distance of the species (m) and D is the diffusion coefficient

(m2·s−1). The characteristic distance, which needs to be bridged by species to reach reactive


128

sites can be estimated from TEM of a TAP aged catalyst (Figure 4.14) and is approximately

equal to 8-10 nm. This distance can either apply to O spillover, moving from the active metal

to the edge of the carbon deposit, to carbon transfer from the edge towards the active

metal, or to metal transfer. For carbon gasification by O2 over the catalysts at 993 K, the

diffusion coefficient calculated according to Eq. 5 was approximately 10-17 m2·s−1.

Oxygen spillover provides active oxygen species on the metal surface, which then

transfer towards carbon over the support. For such oxygen, the surface diffusion coefficient

on MgO amounts to approximately 10-16 m2·s−1, i.e. close to the calculated value [56]. To

further evaluate oxygen spillover as possible process, another TAP experiment of O2 pulses

was performed over an oxidized Ni-5Fe/MgAl catalyst (not shown). The O2 peak that was

measured in the outlet of the reactor was sharp and short in time, 0.3 s while the amount of

O2 that was measured in the outlet corresponded to 95% of the inlet pulse, implying the

formation of a negligible amount of surface oxygen. This observation enhances the

conclusion that was made about the negligible contribution of oxygen spillover mechanism

to the carbon gasification process (see section 4.3.6).

The mobility of metal particles on graphite under O2 and CO2 was evidenced in 1964

[57]. McKee showed mobility of catalytic particles with channeling on graphite by hot stage

optical microscopy [58]. Recently, the Environmental TEM (ETEM) technology was applied

for in-situ imaging during the carbon oxidation process on transition metal particles. Booth

et al. [59] and Gardini et al. [60] have demonstrated that such particles may engage in

transport through the layers of carbon deposits during carbon gasification by O2. The

mobility temperature of various metal and metal oxide particles dispersed on graphite was

investigated by Baker [61]. For 10nm particle size of several metals and oxides, he found that

the mobility temperature coincided with the Tammann temperature of these materials,

which was expressed as 0.52·Bulk melting point(K). In case of NiO and Fe2O3, the Tammann

temperature is equal to 1290 K and 1080 K, respectively, close to the temperature at which

the TAP experiments were performed (993 K).

Carbon gasification by O2 is an exothermic process (ΔH=-390 kJ/mol), which will result in

a local temperature increase. Gasification occurring on the active metal surface will make

the metal particle hotter than its surroundings, which will increase its mobility and hence


129

allow it to migrate. Hence, carbon species which are located away from the metal particles

can still be oxidized after the latter migrate.

Figure 4.14: HRTEM image of aged Ni catalyst after 400 CH4 pulses in TAP.

As for carbon transfer, it is well documented that the diffusion coefficient of carbon inside

of Ni or Fe is equal to approximately 10-13 m2·s−1 [62] which value is significant higher,

implying that it is a faster process than the calculated value. Hence, carbon transfer cannot

account for the time shifts observed.

The above isothermal TAP experiments were repeated with CO2 pulses for carbon

oxidation. CO was monitored as a carbon gasification product. No onset occurred for either

catalyst, and the responses were all narrow and remained stable in time position. The

amount of CO produced gradually decreased with pulse number as carbon was removed

from the surface (see Figure 4.15). Subsequent pulses of O2 however gave more CO2

response, indicating that not all carbon was gasified by CO2.

Ni

Carbon


130

Figure 4.15: 2D TAP spectrum: CO response during CO2 pulses at 993 K over (A): Ni, (B):Ni-Fe catalyst.

4.3.8 Carbon gasification mechanism by O2

Based on above-mentioned results from different techniques, a concluding mechanism

for O2 mediated carbon gasification is proposed. The oxidation of carbon species by O2 can

be described as follows: 1) oxidation of metal particles Ni and Fe; 2) surface carbon

gasification through lattice oxygen, from the active metal oxides, 3) particle migration to the

contact points between catalyst and carbon species, enhanced by high local temperature

arising from 2) and 4) subsequent oxidation of distant carbon through lattice oxygen of

nickel and iron oxides. The effect of oxygen spillover to the support towards distant carbon is

considered to be negligible.

A graphical illustration of the O2 carbon species oxidation over Ni-5Fe/MgAl catalyst is

depicted in Figure 4.16, including the contribution for local and distant carbon gasification.

Tim

e, s

Pulse number

Tim

e, s

Pulse number


131

During O2 oxidation, Fe2O3 and NiO are formed before carbon species oxidation, as it was

concluded from TAP experiments (Figure 4.12). Figure 4.16 (I) shows the local carbon

gasification process over the Ni-5Fe/MgAl catalyst. Figure 4.16 (II) illustrates the mechanism

of particles diffusion to carbon species that are located far from the active metals [59, 60].

Their migration is made possible due to the local temperature increase, originating from

surface carbon oxidation. Then, the distant carbon species are oxidized through the lattice

oxygen of the nickel and iron oxides. Alternatively, the mechanism of carbon bulk diffusion

through the metal particles, resulting in dissolved carbon, introduced by Figueiredo and

Trimm [29], cannot be excluded. However, our observation is that initially metal oxides are

formed and afterwards the carbon gasification process starts.

Figure 4.16: Schematic representation of carbon species oxidation by O2 over Ni-5Fe/MgAl catalyst. Two

different mechanisms are illustrated: (I) carbon gasification on the active metal (II) particles migration

followed by carbon gasification through lattice oxygen. Cm: carbon deposited on metals, Cs: carbon deposited

far from metals, Os: surface oxygen, OL: lattice oxygen. The carbon illustration does not represent the actual

carbon structure.

Fe

Fe2O3NiO

Ni

MgAl2O4

Cs Cs

(I)

Particles migration

Particles migration

CmOLOL

Fe2O3NiO

MgAl2O4

CsCs

(II)

OL OL


132

4.4 Conclusions

Carbon removal is examined from 8wt%Ni-5wt%Fe catalysts on a MgAl2O4 support after

methane dry reforming at 1023 K atmospheric pressure and a CH4:CO2 ratio of 1.1. The

existence of two different carbon species structures, graphitic and amorphous, was

determined by Raman spectroscopy, XPS and TEM.

CO2-regeneration resulted in the removal of carbon on the active metals of the catalysts.

However, EDX-STEM mapping showed the persistence of carbon species located far from the

catalyst active metals, implying the absence of direct interaction between carbon species

and CO2 from the gas phase.

Operando XRD and isothermal TAP experiments allowed to identify the major

mechanistic aspects of carbon species removal by CO2 over used Ni-5Fe/MgAl catalyst. The

process could be described by two parallel contributions. One contains the dissociation of

CO2 over Ni and subsequent oxidation of carbon species by the surface oxygen. The second

consists of the Fe oxidation by CO2 followed by carbon species oxidation by Fe oxide lattice

oxygen, i.e. Fe oxide reduction.

The mechanism of carbon species oxidation by O2 was examined using time-resolved in-

situ XRD, TPO of different catalyst bed configurations and isothermal TAP experiments. The

gasification of carbon species can be described by two processes: 1) oxidation of surface

carbon and 2) particles migration to the carbon species that are deposited far from active

metals and subsequent oxidation through lattice oxygen of the iron and/or nickel oxides. The

contribution of oxygen spillover in carbon gasification was considered to be negligible.


133

4.5 References

[1] L. Guczi, G. Stefler, O. Geszti, I. Sajó, Z. Pászti, A. Tompos, Z. Schay, Methane dry reforming with CO2: A study on surface carbon species, Appl. Catal., A, 375 (2010) 236-246. [2] D. Pakhare, V. Schwartz, V. Abdelsayed, D. Haynes, D. Shekhawat, J. Poston, J. Spivey, Kinetic and mechanistic study of dry (CO2) reforming of methane over Rh-substituted La2Zr2O7 pyrochlores, J. Catal., 316 (2014) 78-92. [3] I. Luisetto, S. Tuti, E. Di Bartolomeo, Co and Ni supported on CeO2 as selective bimetallic catalyst for dry reforming of methane, Int. J. Hydrogen. Energ., 37 (2012) 15992-15999. [4] J. Guo, H. Lou, X. Zheng, The deposition of coke from methane on a Ni/MgAl2O4 catalyst, Carbon, 45 (2007) 1314-1321. [5] L.S. Lobo, Carbon formation from hydrocarbons on metals Department of Chemical Engineering and Chemical Technology, Imperial College of Science and Technology, 1971. [6] L.S. Lobo, J.L. Figueiredo, C.A. Bernardo, Carbon formation and gasification on metals. Bulk diffusion mechanism: A reassessment, Catal. Today, 178 (2011) 110-116. [7] L.S. Lobo, D.L. Trimm, Complex Temperature Dependencies of the Rate of Carbon Deposition on Nickel, Nature Phys. Sci., 234 (1971) 15-16. [8] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, Ni-based catalysts for reforming of methane with CO2, Int. J. Hydrogen Energ., 37 (2012) 15966-15975. [9] C.H. Bartholomew, Mechanisms of catalyst deactivation, Appl. Catal., A, 212 (2001) 17-60. [10] T.H. Gardner, J.J. Spivey, E.L. Kugler, D. Pakhare, CH4–CO2 reforming over Ni-substituted barium hexaaluminate catalysts, Appl. Catal. A, 455 (2013) 129-136. [11] D. Pakhare, J. Spivey, A review of dry (CO2) reforming of methane over noble metal catalysts, Chem. Soc. Rev., 43 (2014) 7813-7837. [12] A. Horváth, G. Stefler, O. Geszti, A. Kienneman, A. Pietraszek, L. Guczi, Methane dry reforming with CO2 on CeZr-oxide supported Ni, NiRh and NiCo catalysts prepared by sol–gel technique: Relationship between activity and coke formation, Catal. Today, 169 (2011) 102-111. [13] C.E. Daza, S. Moreno, R. Molina, Co-precipitated Ni–Mg–Al catalysts containing Ce for CO2 reforming of methane, Int. J. Hydrogen Energy, 36 (2011) 3886-3894. [14] J.R. Rostrupnielsen, J.H.B. Hansen, CO2-Reforming of Methane over Transition Metals, J Catal., 144 (1993) 38-49. [15] A. Penkova, L. Bobadilla, S. Ivanova, M.I. Domínguez, F. Romero-Sarria, A.C. Roger, M.A. Centeno, J.A. Odriozola, Hydrogen production by methanol steam reforming on NiSn/MgO–Al2O3 catalysts: The role of MgO addition, Appl. Catal. A 392 (2011) 184-191. [16] Z. Zhang, X.E. Verykios, Carbon dioxide reforming of methane to synthesis gas over Ni/La2O3 catalysts, Appl. Catal. A 138 (1996) 109-133. [17] A.S. Bobin, V.A. Sadykov, V.A. Rogov, N.V. Mezentseva, G.M. Alikina, E.M. Sadovskaya, T.S. Glazneva, N.N. Sazonova, M.Y. Smirnova, S.A. Veniaminov, C. Mirodatos, V. Galvita, G.B. Marin, Mechanism of CH4 Dry Reforming on Nanocrystalline Doped Ceria-Zirconia with Supported Pt, Ru, Ni, and Ni–Ru, Top. Catal., 56 (2013) 958-968. [18] A. Kambolis, H. Matralis, A. Trovarelli, C. Papadopoulou, Ni/CeO2-ZrO2 catalysts for the dry reforming of methane, Appl. Catal. A 377 (2010) 16-26. [19] V.V. Galvita, H. Poelman, V. Bliznuk, C. Detavernier, G.B. Marin, CeO2-Modified Fe2O3 for CO2 Utilization via Chemical Looping, Ind. Eng. Chem. Res., 52 (2013) 8416-8426. [20] H. Ay, D. Üner, Dry reforming of methane over CeO2 supported Ni, Co and Ni–Co catalysts, Appl. Catal., B, 179 (2015) 128-138. [21] V. Sadykov, V. Muzykantov, A. Bobin, N. Mezentseva, G. Alikina, N. Sazonova, E. Sadovskaya, L. Gubanova, A. Lukashevich, C. Mirodatos, Oxygen mobility of Pt-promoted doped CeO2–ZrO2 solid solutions: Characterization and effect on catalytic performance in syngas generation by fuels oxidation/reforming, Catal. Today, 157 (2010) 55-60.


134

[22] S.A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, Enhanced Carbon-Resistant Dry Reforming Fe-Ni Catalyst: Role of Fe, ACS Catal., (2015) 3028-3039. [23] V.V. Galvita, H. Poelman, G.B. Marin, Combined chemical looping for energy storage and conversion, J. Power Sources, 286 (2015) 362-370. [24] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catal., 4 (2014) 289-301. [25] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Catalytic performance and characterization of Ni-Fe catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Appl. Catal. A 392 (2011) 248-255. [26] A. Djaidja, H. Messaoudi, D. Kaddeche, A. Barama, Study of Ni–M/MgO and Ni–M–Mg/Al (M=Fe or Cu) catalysts in the CH4–CO2 and CH4–H2O reforming, Int. J. Hydrogen Energy, 40 (2015) 4989-4995. [27] L.S. Lobo, Intrinsic kinetics in carbon gasification: Understanding linearity, “nanoworms” and alloy catalysts, Applied Catalysis B: Environmental, 148–149 (2014) 136-143. [28] L.S. Lobo, Catalytic Carbon Gasification: Review of Observed Kinetics and Proposed Mechanisms or Models - Highlighting Carbon Bulk Diffusion, Catalysis Reviews, 55 (2013) 210-254. [29] J.L. Figueiredo, D.L. Trimm, Gasification of carbon deposits on nickel catalysts, J. Catal., 40 (1975) 154-159. [30] C.A. Bernardo, D.L. Trimm, The kinetics of gasification of carbon deposited on nickel catalysts, Carbon, 17 (1979) 115-120. [31] A. Tomita, Y. Tamai, Hydrogenation of carbons catalyzed by transition metals, J. Catal., 27 (1972) 293-300. [32] E. Baumgarten, A. Schuck, Oxygen spillover and its possible role in coke burning, Appl. Catal., 37 (1988) 247-257. [33] F. Kapteijn, J.A. Moulijn, Kinetics of Catalysed and Uncatalysed Coal Gasification, in: J. Figueiredo, J. Moulijn (Eds.) Carbon and Coal Gasification, Springer Netherlands1986, pp. 291-360. [34] D. Fino, N. Russo, C. Badini, G. Saracco, V. Specchia, Effect of active species mobility on soot-combustion over Cs-V catalysts, AICHE J., 49 (2003) 2173-2180. [35] B.R. Stanmore, V. Tschamber, J.F. Brilhac, Oxidation of carbon by NOx, with particular reference to NO2 and N2O, Fuel, 87 (2008) 131-146. [36] M. Machida, Y. Murata, K. Kishikawa, D. Zhang, K. Ikeue, On the Reasons for High Activity of CeO2 Catalyst for Soot Oxidation, Chem. Mater., 20 (2008) 4489-4494. [37] N. Zong, Y. Liu, Learning about the mechanism of carbon gasification by CO2 from DSC and TG data, Thermochim. Acta, 527 (2012) 22-26. [38] W. Gac, A. Denis, T. Borowiecki, L. Kępiński, Methane decomposition over Ni–MgO–Al2O3 catalysts, Appl. Catal., A, 357 (2009) 236-243. [39] A.S.A. Al–Fatish, A.A. Ibrahim, A.H. Fakeeha, M.A. Soliman, M.R.H. Siddiqui, A.E. Abasaeed, Coke formation during CO2 reforming of CH4 over alumina-supported nickel catalysts, Appl. Catal. A, 364 (2009) 150-155. [40] C.A. Querini, S.C. Fung, Coke characterization by temperature programmed techniques, Catal. Today, 37 (1997) 277-283. [41] K.M. Hardiman, C.G. Cooper, A.A. Adesina, R. Lange, Post-mortem characterization of coke-induced deactivated alumina-supported Co–Ni catalysts, Chem. Eng. Sci., 61 (2006) 2565-2573. [42] J.M. Kanervo, A.O.I. Krause, J.R. Aittamaa, P.H. Hagelberg, K.J.T. Lipiäinen, I.H. Eilos, J.S. Hiltunen, V.M. Niemi, Kinetics of the regeneration of a cracking catalyst derived from TPO measurements, Chem. Eng. Sci., 56 (2001) 1221-1227. [43] A. Tomita, Catalysis of Carbon–Gas Reactions, Catal. Surv. Jpn., 5 (2001) 17-24. [44] J.T. Gleaves, G. Yablonsky, X. Zheng, R. Fushimi, P.L. Mills, Temporal analysis of products (TAP)—Recent advances in technology for kinetic analysis of multi-component catalysts, J. Mol. Catal. A: Chem., 315 (2010) 108-134. [45] U. Menon, V.V. Galvita, G.B. Marin, Reaction network for the total oxidation of toluene over CuO–CeO2/Al2O3, J. Catal., 283 (2011) 1-9.


135

[46] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst characterization, Pure & Appl. Chem., (1995). [47] R.T.K. Baker, Coking problems associated with hydrocarbon conversion processes, Preprint Paper Am. Chem. Soc. Div. Fuel Chem., 41 (1996) 521-524. [48] N. Shah, D. Panjala, G.P. Huffman, Hydrogen Production by Catalytic Decomposition of Methane, Energy Fuels, 15 (2001) 1528-1534. [49] D. Espinat, H. Dexpert, E. Freund, G. Martino, M. Couzi, P. Lespade, F. Cruege, Characterization of the coke formed on reforming catalysts by laser raman spectroscopy, Appl. Catal., 16 (1985) 343-354. [50] H. Darmstadt, L. Sümmchen, J.M. Ting, U. Roland, S. Kaliaguine, C. Roy, Effects of surface treatment on the bulk chemistry and structure of vapor grown carbon fibers, Carbon, 35 (1997) 1581-1585. [51] T. Jawhari, A. Roid, J. Casado, Raman spectroscopic characterization of some commercially available carbon black materials, Carbon, 33 (1995) 1561-1565. [52] A.C. Ferrari, Raman spectroscopy of graphene and graphite: Disorder, electron–phonon coupling, doping and nonadiabatic effects, Solid State Commun., 143 (2007) 47-57. [53] H. Estrade-Szwarckopf, XPS photoemission in carbonaceous materials: A “defect” peak beside the graphitic asymmetric peak, Carbon, 42 (2004) 1713-1721. [54] A.V. Shchukarev, D.V. Korolkov, XPS Study of group IA carbonates, Cent. Eur. J. Chem., 2 (2004) 347-362. [55] V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Catalyst-assisted chemical looping for CO2 conversion to CO, Appl. Catal., B, 164 (2015) 184-191. [56] D. Martin, D. Duprez, Mobility of Surface Species on Oxides. 1. Isotopic Exchange of 18O2 with 16O of SiO2, Al2O3, ZrO2, MgO, CeO2, and CeO2-Al2O3. Activation by Noble Metals. Correlation with Oxide Basicity, J. Phys. Chem., 100 (1996) 9429-9438. [57] J.M. Thomas, P.L. Walker, Mobility of Metal Particles on a Graphite Substrate, J. Chem. Phys., 41 (1964) 587-588. [58] D.W. McKee, The copper-catalyzed oxidation of graphite, Carbon, 8 (1970) 131-139. [59] T.J. Booth, F. Pizzocchero, H. Andersen, T.W. Hansen, J.B. Wagner, J.R. Jinschek, R.E. Dunin-Borkowski, O. Hansen, P. Bøggild, Discrete Dynamics of Nanoparticle Channelling in Suspended Graphene, Nano Lett., 11 (2011) 2689-2692. [60] D. Gardini, J.M. Christensen, C.D. Damsgaard, A.D. Jensen, J.B. Wagner, Visualizing the mobility of silver during catalytic soot oxidation, Appl. Catal., B, 183 (2016) 28-36. [61] R.T.K. Baker, The relationship between particle motion on a graphite surface and Tammann temperature, J. Catal., 78 (1982) 473-476. [62] J.J. Lander, H.E. Kern, A.L. Beach, Solubility and Diffusion Coefficient of Carbon in Nickel: Reaction Rates of Nickel‐Carbon Alloys with Barium Oxide, J. Appl. Phys., 23 (1952) 1305-1309.

Chapter 5: Controlling the stability of a Ni-Fe reforming catalyst: structural organization of the active components

137

Chapter 5

Controlling the stability of a Ni-Fe reforming

catalyst: structural organization of the active

components

Ni-Fe catalysts present high activity in methane reforming reactions, in absence of

impurities i.e. H2S, with high carbon resistance, but suffer from deactivation during long

time-on-stream (TOS> 4h) via sintering and Fe segregation. Enhanced control of the stability

and activity of Ni-Fe/MgAl2O4 can be achieved by means of noble metal addition. Pd was

selected, systematically investigating the effect of its loading, while the Fe:Ni ratio kept

constant. The evolution of the catalyst structure during H2 Temperature Programmed

Reduction (TPR) and CO2 Temperature Programmed Oxidation (TPO) was investigated using

time-resolved in-situ X-ray diffraction (XRD). Methane dry reforming, as well as steam-dry

reforming, using a close to bi-reforming CH4:H2O:CO2 ratio, of a gas mixture which simulates

the clean effluent of a biomass gasifier (after removal of tars and sulfur compounds) were

used to investigate the effect of the Ni:Pd ratio upon catalyst activity and stability. Finally the

experimental results were coupled with DFT calculations.


Theofanidis, S. A.; Galvita, V. V.; Sabbe, M.; Poelman, H.; Detavernier, C.; Marin, G. B., Controlling the stability of a Fe–Ni reforming catalyst: Structural organization of the active components. Appl. Catal., B 2017, 209, 405-416.


138

5.1 Introduction

Global climate changes impose the need to investigate new routes for utilization of

available resources such as natural gas, to produce fuels and chemicals [1]. Methane

originates from petroleum reserves and landfill gas [2], and its conversion to higher value

products will become increasingly important for the foreseeable future.

Synthesis of chemicals can be carried out via conversion of methane, either directly or

indirectly. A direct process consists of one-step conversion of methane to e.g. halocarbons,

hydrocyanic acid, acetylene, carbon black, aromatics, methanol or carbon disulfide.

However, these processes suffer from low net yields of the desired products. This is due to

the high C-H bond dissociation energy (436 kJ/mol) of the first C-H bond, making it virtually

impossible to limit bond breaking to a single C-H bond[3]. Therefore, all commercial

processes use methane indirectly as a feedstock for higher value products, by reforming it

into syngas as an intermediate resource[4].

The most widely studied technologies for methane reforming to syngas are: steam

reforming (SRM)[5, 6], dry reforming (DRM)[7], partial oxidation (POM)[8, 9] and

autothermal reforming (ARM)[10]. The oxidant used, the final H2:CO ratio, the kinetics and

energetics differ in these processes[11, 12]. Ross and co-workers concluded that among

these processes DRM has a 20% lower operating cost[13] while producing high purity syngas,

with low CO2 content and H2:CO ≤1[14]:

𝐶𝐻4 + 𝐶𝑂2 2𝐶𝑂 + 2𝐻2 ∆H0298K = 247 kJ·mol-1 (1)

Methane bi-reforming (BIM) on the other hand combines steam and dry reforming in

one step, utilizing CH4 and CO2, the two most important greenhouse gases, along with H2O

to produce syngas[4, 15, 16]. BIM offers the advantage of a 2:1 H2:CO ratio, named as

metgas in order to distinguish it from syngas mixtures of different H2:CO ratio, which is

suitable for methanol synthesis[16, 17]:

3𝐶𝐻4 + 𝐶𝑂2 + 2𝐻2𝑂 4𝐶𝑂 + 8𝐻2 ∆H0298K = 220 kJ·mol-1 (2)

Side reactions of importance include the reverse water gas shift (RWGS):

𝐻2 + 𝐶𝑂2 𝐻2𝑂 + 𝐶𝑂 ∆H0298K = 41 kJ·mol-1 (3)


139

There are certain disadvantages associated with DRM and BIM, such as high

endothermicity and catalyst deactivation due to carbon deposition and sintering. The carbon

deposition originates mainly from two reactions[18-20]:

2𝐶𝑂 𝐶 + 𝐶𝑂2 ∆H0298K = -172 kJ·mol-1 (4)

𝐶𝐻4 𝐶 + 2𝐻2 ∆H0298K = 75 kJ·mol-1 (5)

Noble metal catalysts like Pt, Rh, Pd and Ru outweigh these disadvantages: they show

high activity for methane reforming and resistance towards carbon formation, but then

again they are expensive[21-25]. Nickel-based catalysts have been investigated as an

alternative but these are prone to deactivation due to carbon deposition, especially at

CO2/CH4 ratios close to unity[26-28]. However, the alloy formation between Ni and other

metals does lead to improved catalytic activity [29]. Addition of noble metals (Pd, Rh, Pt, Ru)

to Ni-based catalysts has been investigated and alloy formation has been reported after the

reduction treatment, enhancing the catalyst activity and stability [30, 31]. In addition, such

bimetallic catalysts often show improved dispersion, resistance against sintering and higher

intrinsic reactivity compared to the original monometallic catalyst [32-35]. Damyanova and

co-workers [36] investigated DRM over a series of PdNi catalysts. The higher activity

observed for bimetallic Pd-Ni in comparison with monometallic catalysts was attributed to a

synergetic effect between Pd and Ni resulting in higher metallic surface area and higher

reducibility. Ferrandon and co-workers[37] studied steam and autothermal reforming of n-

butane over Ni-Rh catalysts. They concluded that a low Rh loading in the Ni catalysts

performed better than monometallic catalysts due to Ni-Rh alloy formation. Similarly,

Steinhauer and co-workers[38] investigated DRM over bimetallic Ni-Pd catalysts on different

supports. They also concluded that the activity of the bimetallic catalysts was higher than

that of monometallics.

It has been reported that the addition of Fe as non-noble metal promoter to Ni catalysts

[7] also has a beneficial impact, up to a certain Fe:Ni ratio, by suppressing carbon deposition.

Fe oxide species increase the catalyst oxygen mobility, which helps to reduce carbon

formation. Ashok and Kawi[39] studied toluene steam reforming over an Ni-Fe catalyst

supported on Fe2O3-Al2O3. They concluded on the existence of a strong metal-support

interaction which eliminated the sintering and thus resulted in stable catalyst performance.


140

Koike et al. [40] studied the steam reforming of phenol over Ni-Fe/MgAlOx hydrotalcite

catalysts and found higher activity and stability compared to the monometallic catalysts,

since Fe enabled the adsorption of phenol. Fe might also improve the reducibility of Ni

according to Djaidja et al. [41] who synthesized Ni-Fe/MgO and (Ni-Fe-Mg)2Al catalysts, for

which the reducibility increased due to the presence of Fe. The activity results showed high

performance and a good resistance against carbon formation.

Despite the positive findings with Ni-Fe/MgAl2O4[7], for long time-on-stream (TOS > 4 h)

the catalyst proved not stable during DRM at high temperature and the origin of this

deactivation awaits clarification. To increase the catalyst stability, in the present work a

noble metal (Pd) was added. Materials with constant Fe:Ni ratio and different Pd loadings,

supported on MgAl2O4 (Ni-Fe-Pd/MgAl2O4), were used to investigate the effect of the Ni:Pd

ratio upon catalyst activity and stability during DRM. The questions raised are addressed by

performing X-ray diffraction (XRD) characterization of trimetallic Ni-Fe-Pd/MgAl2O4 catalysts,

both ex-situ and in-situ during H2 temperature-programmed reduction (TPR) and CO2

temperature-programmed oxidation (TPO). The best candidate was also tested for activity

during steam-dry reforming, using a close to bi-reforming CH4:H2O:CO2 ratio, of a gas

mixture which simulates the clean effluent of a biomass gasifier (after removal of tars and

sulfur compounds).


5.2.1 Catalyst Preparation

Support preparation

MgAl2O4 was used as a support material and prepared by co-precipitation from an aqueous

solution of Mg(NO3)2·6H2O (99%, Sigma-Aldrich®) and Al(NO3)3·9H2O (98.5%, Sigma-Aldrich®)

(molar ratio Mg:Al=1:2). NH4OH (ACS reagent, 28.0-30.0% NH3 basis) was added as a

precipitating agent to adjust the pH to 10, at 333 K. The precipitate was filtered, dried at 393

K for 12 h and subsequently calcined in air at 1023 K for 4 h.


141


Six Ni-Fe-Pd catalysts were prepared by incipient wetness impregnation on the support

(MgAl2O4) using an aqueous solution of corresponding nitrates Ni(NO3)2·6H2O (99.99+%,

Sigma-Aldrich®), Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®) and Pd(NO3)2·2H2O nitrate (~40%

Pd basis, Sigma-Aldrich®) [42]. The catalysts were dried at 393 K for 12 h and subsequently

calcined in air at 1023 K for 4 h. The molar ratio Fe:Ni was approximately stable at 0.5, while

varying the loading of Pd from 0, 0.1, 0.2, 0.4, 0.8 to 1.6wt%. Samples are labeled using their

Pd content x as x-Pd as shown in Table 1. In addition, monometallic Ni (10wt%) and Pd

(0.1wt%. 0.2wt% and 0.5wt%) samples were synthesized on the same support, named as

“Ni” and “mono-zPd” (z equal to 0.1, 0.2 and 0.5wt%), for comparison purposes.

Effect of catalyst preparation method

In view of easier characterization, Ni-Fe-Pd catalysts on the same MgAl2O4 support, with

Fe:Ni molar ratio of 0.5 and higher Pd loadings (3wt%) were prepared. Two different

methods were examined, incipient wetness co-impregnation and sequential impregnation,

in order to compare the alloy that is formed after reduction treatment. During the co-

impregnation, all the aforementioned precursors were mixed together (named as “Ni-Fe-

3Pd”), while in case of the sequential impregnation Ni and Fe precursors were mixed first.

The samples were dried at 393 K for 12 h and subsequently calcined in air at 1023 K for 4 h.

Then Pd was added by incipient wetness impregnation on bimetallic Ni-Fe (named as “Ni-

Fe+3Pd”). The sample was again dried at 393 K for 12 h and subsequently calcined in air at

1023 K for 4 h. For comparison purposes, bimetallic samples, 10wt%Ni-3wt%Pd and 5wt%Fe-

3wt%Pd (named as “Ni-3Pd” and “Fe-3Pd”), were equally prepared.



adsorption at 77 K (five point BET method using Tristar Micromeritics) after outgassing the

sample at 473 K for 2 h. The crystallographic phases of the materials as prepared were

confirmed by ex-situ XRD measurements (Siemens Diffractometer Kristalloflex D500, Cu Kα

radiation). The powder patterns were collected in a 2θ range from 10o to 80o with a step of

0.02o and 30 s counting time per angle. XRD patterns of known compounds are referenced


142

by their corresponding number in the Powder Diffraction File database. By fitting a Gaussian

function to a diffraction peak, the crystallite size was determined from the peak width via

the Scherrer equation [43], while the peak position gave information about the lattice

spacing based on the Bragg law of diffraction [44].

The bulk chemical composition of support and as-prepared catalysts was determined by


Thermo Scientific). The samples were mineralized by acid fusion.











were measured with a K-type thermocouple and corrected afterwards according to a


and Ag–Si. For each sample, approximately 10 mg of catalyst were evenly spread on a single

crystal Si wafer. Interaction of the catalyst material with the Si wafer was never observed.

Prior to each experiment the reactor chamber was evacuated to a base pressure of 4 Pa by a

rotation pump. Gases were supplied to the reactor chamber with calibrated mass-flow

controllers. He (10 NmL/s) was flowing for 10 min before the flow was switched to 5%H2/He

or CO2 for TPR and TPO experiments (10 NmL/s, up to 1123 K, heating rate of 30 K/min,

dwell time at 1123 K was 45 minutes), respectively.

It should be noted that the peaks in the in-situ XRD patterns appeared at slightly shifted

angular positions compared to both full scans and tabulated values due to temperature-

induced lattice expansion and different sample heights. These shifts in peak positions which


143

are not related to underlying physicochemical processes, were taken into account during

peak assignment.


All activity measurements were performed at 1023 K and 101.3 kPa in a quartz reactor

with an internal diameter of 9 mm, which was housed inside an electric furnace. The

temperature of the catalyst bed was measured with K-type thermocouples touching the

outside and inside of the reactor at the position of the catalyst bed. The inlet gas flow rates

were always maintained by means of calibrated Bronkhorst mass flow controllers. The

sample with particle size fraction of 50-125 μm was diluted with inert Al2O3 for improved

heat conductivity (ratio catalyst:inert ~1:60) and packed between quartz wool plugs. Prior to

each experiment, the as-prepared sample was reduced in a 1 NmL/s flow of 10vol%H2/He at

1123 K (heating rate of 20 K/min) for 30 minutes (named as “reduced”) and then the flow

was switched to 1 NmL/s of He for 20 minutes. A mixture of CH4, CO2 and He (total flow of

237 NmL/min, 37vol% CH4/ 37vol% CO2/ 26vol% He, He as internal standard) was used as

feed for DRM. The produced CO, H2 and unconverted CH4 and CO2 were detected at the

outlet using a calibrated OmniStar Pfeiffer mass spectrometer (MS). MS signals were

recorded for all major fragments. For quantification of reactants and products, the MS was

focused to different amu signals. H2 was monitored at 2, He at 4, CH4 at 16, H2O at 18, C2H4

at 27, CO at 28 and CO2 at 44 amu. A correction was applied to remove contributions from

unavoidable interference with fragmentation peaks of other gases. To evaluate the

significance of external and internal mass transfer limitations, the Carberry number [45] and

Weisz-Prater criterion [46] were applied, while for heat transport limitations the Mears

criteria [47] were employed. A radial temperature gradient of ~22 K was calculated. This

temperature gradient was taken into account for the calculation of the equilibrium

conversion.

The activity was measured as a function of time in a short term experiment (time-on-

stream (TOS) = 4 h) during DRM at 1023 K and 101.3 kPa. The same conversion range (XCH4=

50±5%) was achieved for all of the investigated samples by varying the amount of catalyst

(Wmetals:FCH4= 0.018-0.025 kgmetals·s·mol-1CH4). With the better performing catalyst candidates,

stability tests were carried out over longer TOS (24 h).


144

The activity was also investigated as a function of space time during isothermal bi-

reforming experiments (TOS = 1 h), using a gas mixture, which simulates the clean effluent of

a biomass gasifier (12.4 vol% CO, 9.8% H2, 6.2% CO2, 13.5% H2O, 6.2% CH4, 2.9% C2H4 and

49.0% He). It should be noted the absence of mass and heat transport limitations during

these experiments at low space time values.



𝑋𝑖 = 𝐹𝑖

0−𝐹𝑖

𝐹𝑖0 · 100% (6)

where Fin,i and Fout,i are the inlet and outlet molar flow rates of reactant i.

The CH4 consumption rate (mmol·s-1·kgNi-1, Eq. 7) was calculated from the difference in

the inlet and outlet molar flow rates, as measured relative to an internal standard (He).


0−𝐹𝑖|

𝑚𝑚𝑒𝑡𝑎𝑙𝑠 (7)

where Fi, mol·s-1, is the molar flow rate of component i and mmetals, the amount of Ni and Pd

(g) present in the catalyst.

To determine the amount of carbon species deposited after the stability tests, TPO

measurements were performed in the same experimental setup at atmospheric pressure.

Once the reactor was cooled down to room temperature, the catalyst bed was subjected to

a 1 NmL/s flow of 10% O2/He. With a heating rate of 15 K/min, the temperature was raised

to 1173 K. Assuming that all carbon underwent complete combustion in the presence of

oxygen, the CO2 peaks measured by MS allow calculation of the amount of carbon present

on the used catalyst. TPO was carried out on all studied samples after the stability tests.

To evaluate the catalyst regeneration ability, regeneration cycles were performed after

the stability tests, combining periods of oxidation by 10vol%O2/He, reduction by

10vol%H2/He and DRM at 1023 K.


145

5.2.5 Catalyst models and DFT calculations

The effect of Pd on the Ni-Fe alloy behavior in different environments was modeled using

DFT calculations. Two scenarios were examined, a Ni-Fe alloy without Pd and a Ni-Fe-Pd

alloy. In the studied catalysts, the molar Fe:Ni ratio is around 0.5, regardless of the presence

of Pd. In the overall trimetallic alloy composition, there is at most 1 Pd per 8 Ni atoms, yet

the experimentally observed Pd-rich shell will possess a higher Pd:Ni ratio.

It is generally accepted that Ni3Fe is a stable intermetallic compound with fcc structure at

temperatures below 500 K [48]. At higher temperatures, the Ni3Fe alloy is known to be in an

fcc γ-phase, though there is some discussion whether it is γ-Fe, γ-Ni, or a combination of

both mentioned γ-phases[48]. The Fe:Ni molar ratio of 0.33 in Ni3Fe is smaller than the

nominal ratio of 0.50 in the present samples, still it can serve as Ni-Fe alloy approximation.

Then, if one Ni is substituted by 1 Pd atom, the resulting Ni2FePd structure has the ratio of

0.50. Since Ni-Pd and Ni-Pt are known to form an fcc solid solution for the whole range of

compositions [49, 50], we can assume that Ni2FePd most likely will possess an fcc structure

as well.

Therefore, the Ni3Fe structure is taken to model the Pd-free phase, and the Ni2FePd alloy

to model the Pd-rich alloy surface phase. Both are modeled as close-packed fcc structures. In

the Ni3Fe bulk unit cell, the 3 Ni atoms occupy the face-centered positions of the bulk unit

cell, while the Fe atom is located at the corner of the cell. For the Ni2FePd bulk unit cell, one

of the Ni atoms in this structure is replaced by a Pd atom.

For both fcc lattices, only the close-packed (111) surface is studied since it is typically the

lowest surface energy plane and hence the most abundant surface on most particles, and

the subject of interest is only to calculate the qualitative segregation behavior of iron. The

(111) surface is modeled by a 4-layered slab with 4 atoms per layer, constructed from the

optimized bulk fcc unit cell (see Supporting Information Secion S1.1).

The thermodynamic driving force for segregation of iron to the surface is represented by

the segregation energy ΔΕseg, defined as the difference between the energy of the

segregated slab, Eslab,seg, and the energy of the non-segregated slab, Eslab, expressed per

number of migrating solute atoms nM (which is 1 for the considered surface unit cell) [51]:


146

M

.

n

EEE

slabsegslab

seg

(8)

The segregation behavior of the alloys can be dependent on the adsorbate coverage. To

this end, the minimum energy configurations for several adsorbate coverages can be

calculated. In line with earlier work [51, 52], the segregation energy in the presence of an

adsorbate, ads

segE , is calculated as

M solutes

,,

n

EEE

segnonadsslabsegadsslabads

seg

(9)

i.e., the difference between the energy of the segregated adsorbate-covered slab

Eslab+ads,seg and the energy of the non- segregated adsorbate-coverage slab Eslab+ads,non-seg

divided by the number of migrated solute atoms, nsolutesM.

All DFT calculations have been carried out using the Perdew-Burke-Ernzerhof (PBE) GGA

functional [53], with the Vienna Ab initio Simulation Package (VASP 5.3). [54-57] Details of

the computational method can be found in Supporting Information Section S1.1, including

details of the surface unit cells, the geometries, and the actual composition of the

considered adsorbate overlayers.

5.3 Results


The metal content, Brunauer-Emmett-Teller (BET) surface area, Ni:Pd and Fe:Ni ratio are

reported in Table 5.1 for the support material and the studied catalysts. The surface area for

the MgAl2O4 support has the highest value while for the catalysts it remains stable at

approximately 70 m2/gcat.

The crystalline phases of “as-prepared”, “reduced” and “reoxidized” catalysts were

determined by ex-situ powder XRD. Catalysts with higher metal concentration, Ni-Fe-3Pd,

were used in order to obtain more clear diffraction patterns. Figure 5.1 (A) displays the full

scan XRD patterns of pure support, as-prepared, reduced and reoxidized Ni-Fe-3Pd catalyst.


147

The reduced states of all the possible bimetallic combinations among Ni, Fe and Pd (Ni-3Pd

and Fe-3Pd) supported on MgAl are illustrated at Figure 5.1 (B).

Table 5.1: Catalyst and support properties

Abbreviation

Catalyst

Metal loadinga

(wt.%)

Fe:Ni

(mol/mol)

Ni:Pd

(mol/mol) BET

b

(m2/gcat)

Ni Fe Pd - -

MgAl MgAl2O4 - - - - - 100.3 ± 5.0

0-Pd 9%Ni-4%Fe/MgAl2O4 8.4 4.30 - 0.54 - 78.0 ± 9.9

0.1-Pd 9%Ni-4%Fe-0.1%Pd/MgAl2O4 9.30 4.10 0.10 0.46 171 67.7 ± 2.6

0.2-Pd 9%Ni-4%Fe-0.2%Pd/MgAl2O4 8.80 3.80 0.20 0.45 75 71.0 ± 1.9

0.4-Pd 9%Ni-4%Fe-0.4% Pd/MgAl2O4 8.90 3.80 0.37 0.45 44 70.3 ± 2.1

0.8-Pd 9%Ni-4%Fe-0.8% Pd/MgAl2O4 7.85 3.30 0.78 0.50 21 61.3 ± 6.0

1.6-Pd 9%Ni-4%Fe-1.6% Pd/MgAl2O4 7.38 3.20 1.58 0.51 8 69.4 ± 1.6

a: ICP of as-prepared samples; b: as-prepared

MgAl2O4 (31.3o,37o,45o, 55.5o, 59o, 65o, Powder Diffraction File (PDF) card number: 00-

021-1152) remained stable during reduction and oxidation for all samples. The “as-

prepared” sample contained diffraction peaks of Fe2O3 (maghemite), NiO, PdO and NiAl2O4

(30.2o, 35.6o, 43.3o, 57.3o and 62.9o, PDF: 00-039-1346; 37.3o, 43.3o and 62.9o, PDF: 01-089-

5881; 33.8o, 41.9o 54.7o and 60.2o, PDF: 00-041-1107 and 37.0o, 45o, 59.7o and 65.5o, PDF:

00-010-0339), some of which are overlapping. The latter phase is in accordance with

Theofanidis and co-workers [7] and Guo and co-worders [58], who found the formation of

NiAl2O4 in the as-prepared sample. NiFe2O4 diffraction peaks (30.2o, 35.7o, 43.3o, 57.3o,

62.9o, PDF: 00-010-0325) overlap with those of Fe2O3 and therefore its presence cannot be

excluded. Upon reduction, the oxides diffractions (NiO, PdO and Fe2O3) disappeared, while

the spinel NiAl2O4 phase was not decomposed. In addition, diffractions were present at 2θ

angles (42o and 48.8o) lower than those of Ni-Fe alloy (44.2o and 51.5o, PDF: 00-038-0419) [7,

39, 59-61]. In order to assign those shifted peaks to a crystallite structure, different

bimetallic combinations among Ni, Fe and Pd, supported on the same MgAl, were examined

in reduced state (Figure 5.1B). The peaks observed at 42o and 48.8o in the pattern of reduced

Ni-Fe-3Pd are higher than those of Fe-Pd and Ni-Pd alloy (41.2o and 47.3o, PDF: 00-002-1441

and 41.9o and 48.8o, PDF:03-065-5788), observed in the patterns of reduced bimetallic Fe-

3Pd and Ni-3Pd, respectively. Therefore, the diffraction peaks at 42o and 48.8o were


148

attributed to the formation of a trimetallic alloy phase of Ni, Fe and Pd. In the XRD pattern

following the CO2-TPO (Figure 5.1A), the aforementioned trimetallic alloy phase was

decomposed to Fe3O4 (30.1o,35.5o,43o,57o,62.5o, PDF:03-065-3107) and Ni-Pd alloy, while

NiAl2O4 and MgAl support diffractions remained unchanged.

Figure 5.1: Full XRD scans of (A) Ni-Fe-3Pd supported on MgAl2O4, as-prepared, reduced and

oxidized (1 NmL/s of 10%H2/He or CO2 at a total pressure of 101.3 kPa and 1123 K). (B) Ni-Fe-3Pd

reduced state along with possible bimetallic combinations (Fe-3Pd, Ni-3Pd) supported on MgAl (1

NmL/s of 10%H2/He, at a total pressure of 101.3 kPa and 1123 K).

(A)

(B)


149

The elemental distribution of 1.6-Pd is indicated in Figure 5.2, using energy-dispersive X-

ray spectroscopy (EDX)-STEM mapping. In the as-prepared sample Ni (green), Fe (red) and Pd

(blue) form clusters as oxides, while upon reduction the elements are redistributed, resulting

in the formation of a trimetallic alloy in the outer shell. Based upon the element loadings,

this implies that the core of the alloy will be close to bimetallic Ni-Fe, while the surface will

be truly trimetallic Ni-Fe-Pd.

Figure 5.2: EDX element mapping of 1.6-Pd. (A): as-prepared (B) reduced (1 NmL/s of 5%H2/He

mixture at a total pressure of 101.3 kPa and 1123 K). Red, green and blue colors correspond to Fe,

Ni and Pd elements, respectively.

5.3.2 In-situ XRD time resolved measurements

H2-TPR

The trimetallic Ni-Fe-Pd alloy formation during the H2-TPR process was investigated using

in-situ XRD measurements. A catalyst with higher concentration on metals, Ni-Fe-3Pd was

used in order to obtain better signal resolution. The results are presented in a 2D in-situ XRD

pattern (Figure 5.3). Diffraction peaks associated to Fe2O3 were not detected by in-situ XRD

due to the low concentration and their overlapping with MgAl2O4 peaks. During reduction,

PdO peaks disappeared at 400 K and NiO peaks above 800 K. The NiAl2O4 spinel phase

remained stable upon reduction in accordance with the ex-situ XRD full scans (Figure 5.1).

(B)(A)

Ni-Fe alloy

Ni-Fe-Pdarea

PdO

Fe2O3 Fe2O3

PdO

NiO


150

Figure 5.3: 2D in-situ XRD pattern during H2-TPR for Ni-Fe-3Pd. Heating rate: 30 K/min, maximum

temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He.

The characteristic peaks associated with the MgAl support did not change throughout H2-

TPR up to the final temperature of 1123 K for all studied samples, implying that MgAl was

not reduced in this temperature range. The Pd related diffraction shifted from 40.1o to an

angle of 42.0o, above 820 K, higher than that for Ni-Pd alloy (41.9o), which was hence

attributed to a trimetallic Ni-Fe-Pd alloy diffraction peak.

A graphical illustration of the Ni-Fe-Pd phase formation after catalyst reduction is

depicted in Figure 5.4. The alloy phase consists of a close to bimetallic core, surrounded by a

thin trimetallic layer, as it was observed from EDX (Figure 5.2B).

Figure 5.4: Schematical representation of the alloy formation during H2-reduction up to 1123 K.

CO2-TPO

The stability of the trimetallic Ni-Fe-Pd alloy active phase in oxidizing conditions was

investigated during CO2-TPO immediately after cooling down following H2-TPR (Figure 5.5).

The diffraction peak assigned to the Ni-Fe-Pd alloy phase shifted to lower 2θ values above

Fe2o3 H2-reduction

As-prepared Reduced

PdONiO

Ni-Fe

Ni-Fe-Pd


151

950 K, implying the alloy decomposition to Ni-Pd alloy and Fe3O4, as it can be seen from the

highlighted area in Figure 5.5. The same result is expected in case steam would be used as an

oxidizing gas. The Ni-Pd alloy remained stable up to 1123 K along with the NiAl2O4 and

MgAl2O4 phases.

Figure 5.5: 2D in-situ XRD pattern during CO2-TPO for Ni-Fe-3Pd. Heating rate: 30 K/min, maximum

temperature 1123 K, flow rate: 1 NmL/s, CO2. The highlighted rectangular area shows the integral

intensity variation of trimetallic Ni-Fe-Pd alloy for diffraction angles 41.7-42.2o as a function of

temperature.

5.3.3 Effect of catalyst preparation on alloy formation

The effect of catalyst synthesis method on the formation of the trimetallic Ni-Fe-Pd alloy

phase was assessed by comparing the catalyst’s crystallite phases formed after co-

impregnation and sequential impregnation. Figure 5.6 shows the crystalline phases of as-

prepared and reduced catalysts prepared by co-impregnation and sequential impregnation.

The diffraction peaks of support MgAl2O4 (31.3o,37o,45o, 55.5o, 59o, 65o, Powder Diffraction

File (PDF) card number: 00-021-1152) are the same in the as-prepared and reduced state of

the samples prepared by both methods. The patterns of the as-prepared catalysts are similar

for both synthesis methods, containing the diffraction peaks of Fe2O3 (Maghemite), NiO,

NiAl2O4 and PdO phases. However, differences are apparent in the patterns of reduced

catalysts. Two diffraction peaks (42o and 48.8o) are observed in the case of the co-

impregnated catalyst, that were previously assigned to a trimetallic Ni-Fe-Pd alloy phase (see

Figure 5.1). On the other hand, two diffraction peaks at higher angles (44.2o and 51.5o) and

two at lower angles (40.1o and 46.7o) are observed in the case of the catalyst prepared by

sequential impregnation. These are attributed to Ni-Fe alloy (PDF: 00-038-0419) and metallic

Ni-Fe-Pd alloy


152

Pd (PDF: 00-046-1043), respectively, implying the crucial role of the preparation method on

the trimetallic alloy phase formation.

Figure 5.6: Full XRD scans of Ni-Fe-3Pd and Ni-Fe+3Pd supported on MgAl2O4, as-prepared and

reduced (1 NmL/s of 10%H2/He at a total pressure of 101.3 kPa and 1123K).

The evolution of the catalyst structure, prepared by sequential impregnation, during H2-

TPR was examined by time-resolved in-situ XRD (Figure 5.7). Diffraction peaks associated to

Fe2O3, NiO and PdO were detected in the as-prepared sample. During reduction, PdO peaks

disappeared at 400 K, while the Pd related diffraction peak at 40.1o remained stable up to

1123 K, implying the absence of a trimetallic alloy phase (Figure 5.3). The NiO diffraction

peaks disappeared above 973 K and the Ni related peak shifted to a lower value (44.2o),

which was previously assigned to a Ni-Fe alloy phase [7, 39, 59-61].

25 30 35 40 45 50 55 60 65 70

Inte

nsi

ty a

.u.

2θ

● MgAl2O4 ¤ Ni-Fe alloy γ Fe2O3

■ PdO NiO + NiAl2O4

Ni ‡ Pd ♦ Ni-Fe-Pd

Support

As-prepared co-impregnation

As-preparedsequential impregnation

Reducedco-impregnation

Reducedsequential impregnation


153

Figure 5.7: 2D in-situ XRD pattern during H2-TPR for Ni-Fe+3Pd. Heating rate: 30 K/min, maximum

temperature 1123 K, flow rate: 1 NmL/s, 10%H2/He.

5.3.4 Activity and stability tests

Methane dry reforming (DRM)

The catalysts prepared by co-impregnation, where the trimetallic active phase was

formed after reduction by H2, were compared with Ni and 0-Pd samples in activity

measurements during isothermal DRM at 1023 K, 101.3 kPa and CH4:CO2 = 1:1 in order to

investigate the effect of Ni-Fe-Pd alloying and Ni:Pd ratio on the catalytic properties of all

studied samples. The CH4 consumption rate of a series of isothermal activity measurements,

after TOS = 4 h, are depicted in Figure 5.8.

PdO

MgAl2O4

MgAl2O4

Fe2O3

MgAl2O4Pd

NiO

MgAl2O4

Ni-Fe alloy

Pd

Ni-Fe alloy


154

Figure 5.8: CH4 consumption rate (mmol·s-1·g-1metals) of samples after 4 h time-on-stream (TOS),

under DRM at 1023 K (Wmetals:F0

CH4 = 0.018-0.025 kgmetals·s·molCH4-1, total pressure of 101.3 kPa and

CH4:CO2 = 1:1). XCH4 = 50 ± 5% for all studied samples. The x-Pd catalysts are trimetallic Ni-Fe-Pd

samples with constant Fe:Ni ratio, containing different wt% of Pd (see Table 5.1). Error bars were

calculated after three independent experiments representing standard deviation (68% confidence

interval).

The CH4 space time yield for 0.2-Pd was slightly higher than the one for monometallic Ni

and 0-Pd. The addition of 0.2wt%Pd (molar ratio Ni:Pd equal to 75) improved, but not

significantly, the catalytic activity, while further raising the Pd concentration (or decreasing

the Ni:Pd molar ratio) did not improve the activity. This is in accordance with Ferrandon and

co-workers [37], who studied steam and autothermal reforming of n-butane using another

noble metal, Rh, along with Ni. They found that a Rh/Ni atomic ratio close to 1/100 had the

best performance, while further increase of Rh concentration did not significantly improve

the catalyst activity. In literature, it is mentioned that the addition of noble metals improves

the dispersion and enhances the resistance against sintering [62, 63]. So, the better catalyst

performance could originate from an enhanced dispersion of Ni particles due to the addition

of Pd.

To assess the effect of Pd addition on NiO dispersion, five different batches of 0-Pd, 0.1-

Pd and 0.2-Pd were prepared and their crystallite structure was examined using ex-situ XRD.

0

5

10

15

20

25

30

35

Ni 0-Pd 0.1-Pd 0.2-Pd 0.4-Pd 0.8-Pd 1.6-Pd

CH

4 c

on

sum

pti

on

rat

e

(mm

ol·

s-1·g

me

tals

-1)


155

The NiO diffraction peaks were fitted with a Gaussian function and the crystallite size was

estimated using the Scherrer equation[43] (Figure 5.9).

Figure 5.9: Calculated NiO crystallite size (nm) from different batches of 0-Pd (●), 0.1-Pd (■) and

0.2-Pd (▲) as determined from ex-situ XRD full scans.

The dispersion of NiO is slightly affected by Pd addition as the average crystallite size of

0.2-Pd is 8.5 ± 0.5 nm, compared to 10.0 ± 1.7 nm and 11.3 ± 1.3 nm, for 0.1-Pd and 0-Pd,

respectively. This means that the activity increase observed in case of 0.2-Pd, compared to 0-

Pd (Figure 5.8), could indeed originate from higher dispersion of Ni.

In order to differentiate the contribution of the different elements to the catalyst

activity, another test was performed over monometallic Pd catalysts, supported on MgAl2O4,

under DRM at 1023 K (Wcat:FCH4= 0.26 kgcat·s·molCH4-1, total pressure of 101.3 kPa and

CH4:CO2 = 1:1). Figure 5.10 shows the CH4 conversion as a function of TOS for the studied

samples.

0

2

4

6

8

10

12

14

16

0 1 2 3 4 5

NiO

cry

stal

lite

size

(n

m)

Sample number


156

Figure 5.10: Methane conversion as a function of time-on-stream (TOS) over monometallic Pd

catalysts supported on MgAl2O4 under DRM at 1023 K (Wcat:F0

CH4 = 0.26 kgcat·s·molCH4-1, total

pressure of 101.3 kPa and CH4:CO2 = 1:1). x: mono-0.1Pd, *: mono-0.2Pd, +: mono-0.5Pd.

The mono-0.2Pd catalyst reached an average CH4 conversion of 12%, implying that Pd is

active for DRM. The aforementioned result suggests that the origin of the improved activity

of 0.2-Pd compared to 0-Pd (Figure 5.8) includes, apart from the higher dispersion of Ni as

shown in Figure 5.9, the creation of extra sites upon Pd addition.

Stability tests

The stability of the catalysts was investigated during DRM for a TOS of 24 h. The best

candidate among the studied catalysts, 0.2-Pd, was tested along with 0-Pd, 0.1-Pd and Ni for

comparison purposes. CO and H2 (Figure 5.11) as products along with unreacted CH4 and CO2

were observed in the reactor outlet for every studied sample. Helium was also detected as it

was used as internal standard. Figure 5.11 illustrates the CH4 and CO2 consumption rates for

all tested samples along with CO:H2 ratio.

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250

CH

4C

on

vers

ion

TOS (min)


157

(A)

(B)

(C)


158

Figure 5.11: Stability tests during DRM at 1023 K (total pressure of 101.3 kPa and CH4:CO2 = 1:1).

(A): CH4 consumption rate (mmolCH4·s-1·g-1

metals); (B): CO2 consumption rate (mmolCO2·s-1·g-1

metals); (C):

CO:H2 ratio. ♦: Ni (Wmetals:F0

CH4 = 0.022 kgmetals·s·mol-1CH4), XCH4 from 67% to 53% ●: 0-Pd (Wmetals:F

0CH4 =

0.025 kgmetals·s·mol-1CH4), XCH4: from 62% to 24% ■: 0.1-Pd (Wmetals:F

0CH4 = 0.018 kgmetals·s·mol-1

CH4), XCH4

from 64% to 36%; ▲: 0.2-Pd (Wmetals:F0

CH4 = 0.024 kgmetals·s·mol-1CH4), XCH4 from 58% to 44%. Error

bars were calculated after three independent experiments representing standard deviation (68%

confidence interval).

The pure Ni catalyst was stable during the 24 h TOS as it showed a decrease of 21% and

10% on CH4 and CO2 consumption rates, respectively. However, a carbon amount of 14.7

molC/kgcat was deposited on this pure Ni sample during 24 h TOS of DRM at 1023 K,

confirmed by O2-TPO over the used sample.

During the first 2 h TOS, 0-Pd showed the highest CH4 and CO2 space time yield. This is in

agreement with Theofanidis and co-workers [7], who studied different Ni-Fe catalysts (Fe:Ni

weight ratio from 0.7 to 1.6) supported on MgAl2O4 during DRM at 1023 K. They found that

the sample with a Fe:Ni weight ratio of 0.7 had a higher CO and H2 production and the CO:H2

ratio was close to 1.3 (see Chapter 3). However, after a total TOS of 24 h the 0-Pd had

deactivated considerably with an activity drop of 62% and 57% for CH4 and CO2, respectively.

Carbon deposition was examined as a possible cause of deactivation using O2-TPO, however,

no CO2 production was detected, implying that carbon formation is not involved in the

observed deactivation. This result verifies the enhanced carbon resistance of Ni-Fe catalysts

that was reported elsewhere [7, 64]. The CO:H2 ratio increased from 1.3 after 1 h TOS to 2.5

after 24 h TOS (Figure 5.11C), implying a modification in the nature of active sites during the

reaction.

The 0.1-Pd catalyst showed improved stability compared to 0-Pd, since the CH4 and CO2

space time yields decreased by 44% and 42%, respectively. However, its activity was not

improved compared to 0-Pd. The best performance was observed for the 0.2-Pd catalyst,

which showed the highest activity after 5 h TOS and better stability, since the CH4 and CO2

consumption rates were reduced by only 23% and 20%, respectively. The produced CO:H2

ratio was close to 1.5 (Figure 5.11C), while no deposited carbon was observed over 0.2-Pd

catalyst after 21 h TOS. Previously, it has been reported that bimetallic Ni-Pd catalysts

(without Fe) supported on mesoporous MCM-41 do suffer from carbon deposition after


159

DRM [36]. The absence of carbon on 0.2-Pd after long use confirms the positive effect of Fe

addition on suppression of carbon formation.

To evaluate the reversibility of the deactivation, a regeneration cycle was applied after

the stability tests for 0-Pd and 0.2-Pd catalysts. After regeneration, the catalyst activity was

examined in-situ for 2 h TOS without removing the catalyst from the reactor. Figure 5.12

displays the CH4 consumption rate at the beginning of the stability test presented in Figure

5.11 (initial), after TOS of 21 h (used) and after regeneration of 0-Pd and 0.2-Pd catalysts.

Figure 5.12: CH4 consumption rate (mmol·s-1·g-1metals) of used (after TOS=21h shown in Figure 5.11)

and regenerated 0-Pd (blue) and 0.2-Pd (green) samples under DRM at 1023 K. Catalyst

regeneration took place using 10vol%O2/He and 10vol%H2/He at 1023 K and immediately after the

reaction.

Figure 5.12 indicates that the regenerated 0.2-Pd catalyst did not recover its initial

activity, remaining at a CH4 consumption of 0.026 molCH4·s-1·g-1metals, i.e. the value after 21 h

TOS. The latter implies an irreversible deactivation mechanism, which is attributed to

sintering. On the other hand, 0-Pd recovered 76% of its initial activity after regeneration. So,

in 0-Pd two parallel processes contribute to the deactivation mechanism. One is originating

from sintering, and it is irreversible upon regeneration, while the other is attributed to Fe

segregation from the Ni-Fe alloy, which is reversible so that the activity can be restored by

Ni-Fe alloy reconstruction. Upon reuse in reaction, 0-Pd will again deactivate. However, in

TOS=21h RegeneratedInitial


160

0.2-Pd, the addition of Pd to Ni-Fe prevents the segregation of Fe from the Ni-Fe alloy,

leaving only sintering as reason for limited activity loss in Ni-Fe-Pd.

Hydrocarbons mixture bi-reforming

The activity of the best candidate, 0.2-Pd, was further tested during bi-reforming of a

hydrocarbon mixture (inset table of Figure 5.13), which simulates the clean effluent of a

biomass gasifier, at 1023 K and 101.3 kPa along with 0-Pd and pure Ni samples for

comparison. The CH4 and C2H4 conversions as a function of space time are illustrated in

Figure 5.13 along with the CH4 and C2H4 equilibrium conversions.

Figure 5.13: Isothermal conversion of CH4 and C2H4 as a function of space time during bi-reforming

at 1023 K and 101.3 kPa. The solid symbols represent the CH4 conversion while the open symbols

represent the C2H4 conversion. ♦: Ni, ●: 0-Pd, ▲: 0.2-Pd. C2H4 equilibrium; CH4

equilibrium. Inset table: composition of feed mixture.

At high space time value, all the samples approach the equilibrium C2H4 conversion,

while the CH4 conversion remains below equilibrium for all examined space time values. The

differences among the catalysts are more pronounced at low space times, below 0.2

kg·s/molHC, where the effect of C2H4 addition to the feed stream on CH4 conversion can be


161

seen. At 0.04 kg·s/molHC, the CH4 conversion of 0.2-Pd is 3 times higher than that of Ni, while

at 0.1 kg·s/molHC it is almost 2 times higher than that of 0-Pd, implying the improved

performance of 0.2-Pd during bi-reforming of a CH4 and C2H4 mixture. These results are in

agreement with the higher activity and stability of 0.2-Pd catalyst (Figures 5.8 and 5.11,

respectively) during DRM at 1023 K and imply that a catalyst suitable for DRM can be also

applied for BIM, in the presence of both CO2 and H2O.

5.4 Discussion

It is well known that Ni catalysts are prone to deactivation due to carbon deposition,

especially under reforming conditions using a CH4:CO2 mixture with ratio near unity, which is

considered thermodynamically to be a carbon-rich area [65]. The addition of Fe as a

promoter to Ni catalysts has proven to be beneficial by suppressing carbon formation [7],

however, after 5 h TOS the activity of the 0-Pd catalyst has dropped significantly (see Figure

5.11).

Since no carbon was deposited, the origin of deactivation can be sintering, an

irreversible contribution, or structural rearrangement of the surface Ni-Fe alloy, which is

reversible. The ratio between reducing and oxidizing gases (CO+H2:CO2+H2O), Rc, is

important for the iron/iron oxides system and thus for the stability of Fe containing alloys

[61, 66]. During DRM over a Ni-Fe catalyst, which proceeds following the Mars-van Krevelen

mechanism (see Chapter 3), CH4 and CO2 are present as reactants, where CO2 oxidizes Fe to

FeOx and CH4 is activated on Ni[7]. The aforementioned oxidation of Fe by CO2 during DRM

at high temperature leads to its partial segregation from the alloy, even at reducing

environment (Rc>1). This redistribution of elements could eventually result in Fe species

located on top of alloy particles. The latter follows from fast reduction of FeOx through

interaction with C, CHx and H species at the surface. This is in agreement with Wang and co-

workers, who found Fe species enriched the outer layers of alloy particles [59]. The observed

deactivation can then be attributed to the high surface Fe:Ni ratio, since Fe is less active in

DRM than Ni [7]. This is also in agreement with the increase of CO:H2 ratio, observed for 0-

Pd, as a function of TOS (Figure 5.11C). A similar mechanism of deactivation can be invoked

for any Fe containing alloy: it can decompose at high temperature under CO2 [7, 60, 61], as it

was observed in Figure 5.5 for Ni-Fe-3Pd, resulting in segregation of Fe from the alloy.


162

A graphical illustration of the proposed deactivation for 0-Pd during DRM is depicted in

Figure 5.14. The Ni and Fe elements were well mixed and uniformly distributed after

reduction, as it was observed elsewhere [7].

Figure 5.14: Schematical representation of the proposed deactivation for 0-Pd due to Ni-Fe surface

alloy reconstruction during DRM at high temperature (1023 K).

The addition of 0.2wt% Pd (molar ratio Ni:Pd equal to 75) stabilizes the alloy core, by

reconstructing its surface into a trimetallic Ni-Fe-Pd alloy. This outer layer is formed on top

of the mostly bimetallic Ni-Fe core and acts as a barrier for Fe segregation from the core

during reaction (see Figure 5.2 and 5.4). As a result, only Fe located within the trimetallic

alloy is accessible to CO2 during DRM and can therefore be extracted. This is in agreement

with Felicissimo and co-workers [67], who investigated a Pd-Fe bimetallic model catalyst.

They found that Pd forms a shell in parallel with an intermixing of Pd and Fe during the

deposition of Pd on Fe particles.

In order to understand the effect of conversion, and thus of Rc, on the catalyst stability,

another experiment was performed at high initial CH4 conversion (more reducing

atmosphere due to H2 and CO presence), XCH4~ 80% and Rc~16, during DRM at 1023 K over 0-

Pd and 0.2-Pd (Figure 5.15). The higher conversion was achieved by varying the space time.

Ni-Fe alloydecomposition

After reduction

Ni-Fe alloy

Reduction of FeOx during

DRMNi-Fe alloy

Fe layer

Ni-Fe alloy

FeOx


163

Figure 5.15: Stability tests during DRM at 1023 K (total pressure of 101.3 kPa and CH4:CO2=1:1) and

high initial CH4 conversion (~80%). ●: 0-Pd (Wmetals:F0

CH4= 0.037 kgmetals·s·mol-1CH4); ▲: 0.2-Pd

(Wmetals:F0

CH4= 0.031 kgmetals·s·mol-1CH4).

A similar deactivation trend as in Figure 5.11A was observed for 0-Pd as the activity

decreased by 61%, while only 29% of activity drop was observed for 0.2-Pd. Hence, although

at higher conversion a more reducing environment is present (Rc = 16), the latter does not

improve the stability. This implies that the environment during DRM, at high or low

conversion, does not play a role in the 0-Pd deactivation, supporting the statement that the

Ni-Fe catalyst follows the Mars van Krevelen mechanism [7], where part of Fe will be

oxidized by CO2 and thus segregated from the alloy structure.

This is in alignment with the result observed during an isothermal experiment using

time-resolved in-situ XRD, under different Rc values, by changing the H2:CO2 ratio (Figure

5.16) simulating the reducing/oxidizing gases during DRM at low conversions. Figure 5.16

shows the 2D in-situ XRD pattern during isothermal cycles of H2:CO2 from 1:1 to 1:3. Initially,

the sample was reduced during H2-TPR up to 1173 K, at a heating rate of 30 K/min and a flow

rate of 20 NmL/s 10vol%H2/N2 in order to form Ni-Fe alloy. Then, the temperature was

decreased to 1023 K for the isothermal cycles. Ni-Fe alloy is detected in the reduced sample,

remaining stable at a H2:CO2 ratio of 1 and 2. However, the alloy was decomposed when a

H2:CO2 ratio of 3 was applied.

0

5

10

15

20

25

30

35

40

45

0 2 4 6 8 10 12 14 16 18 20 22

CH

4co

nsu

mp

tio

n r

ate

(mm

ol·

s-1·g

me

tals

-1)

TOS (h)


164

Figure 5.16: Time-resolved in-situ XRD pattern during isothermal treatment of 0-Pd under different

reducing/oxidizing ratios (Rc) at 1023 K. The upper right inset shows the intensity variation as a

function of time-on-stream (TOS) for the selected diffraction angles 43.2o-44o (Ni-Fe alloy).

The upper right inset shows the integral intensity variation as a function of time during

H2:CO2=1:3, condition where the formation of FeO is observed. The selected diffraction area

corresponds to the Ni-Fe alloy, showing the decomposition of the alloy into Ni when the

partial pressure of CO2 was 3 times higher than the H2 partial pressure. The Ni-Fe alloy was

formed again during the next pure H2 step. FeO probably formed already at higher H2:CO2

ratios, due to partial segregation of Fe from the alloy particles, but it was not clearly

detected because of low concentration. This is in agreement with the results presented in

Figure 3.13 (Chapter 3), where an in-situ XRD experiment with different ratios of CH4:CO2

during DRM was performed. It was observed that FeO was formed at CH4:CO2=1:6, but were

unable to detect FeO at lower CO2 partial pressures due to low concentration.

To evaluate whether the thermodynamic tendency of Fe to move towards the alloy

surface is reduced in the presence of Pd, the iron segregation energies were calculated using

Density Functional Theory for the simplified models Ni3Fe(111), as Ni-Fe alloy, and Ni2FePd

(111), representing the Pd-rich shell of the alloy particle. For both model systems, an iron

atom from the second layer was exchanged with an atom from the surface layer, i.e. Ni for

the first, and Ni or Pd for the second alloy, representative of Fe migration to the surface (see

Appendix B). Under vacuum conditions, in the absence of adsorbates on the metal,

segregation of Fe to the surface is not favorable, particularly not in exchange for a Pd atom,

with segregation energies up to +100 kJ mol-1 (see Table 5.2). The presence of a monolayer

H2:CO2=1:1

H2:CO2=1:3

H2:CO2=1:2H

e

H2

He

H2

H2

FeONi-Fe alloy

Isothermal at 1023 K

MgAl2O4

MgAl2O4 MgAl2O4

MgAl2O4

Ni

H2:CO2=1:3H2 H2

Ni-Fe alloy

Ni

Ni-Fe alloy


165

of hydrogen or CO does not change this, with the only difference that exchanging a

subsurface Fe for a surface Pd becomes more likely, to almost energy-neutral for CO

monolayer coverage. The presence of an oxygen monolayer, which can form upon CO2

dissociation, turns this picture upside-down: it entails a strong tendency to extract a Fe from

the bulk to the surface, with segregation energies from -76 to -218 kJ mol-1. However, as

under DRM conditions the formation of a monolayer of oxygen is highly unlikely, more

realistic coverages, containing a mixture of adsorbed CO, O, CH and H, were investigated.

From the moment the fractional oxygen coverage falls to 0.25 or below, with CO, CH and H

making up the remaining coverage, the segregation of Fe to the surface is almost energy-

neutral, except for the exchange with surface Pd, for which the segregation energy still

amounts to +38 to +143 kJ mol-1.

Table 5.2: Segregation energies without (ΔEseg, kJ/mol) and with adsorbates (ads

segE ) for the

exchange of Fe in the subsurface layer of a (111) surface of Ni3Fe or Ni2FePd, with Ni or Pd from the

surface layer, for various coverages, representative for DRM. All coverages refer to the species

adsorbed on the fcc sites of a periodically repeated unit cell with 4 surface atoms.

ΔEseg (kJ/mol) Ni3Fe Ni2PdFe

adsorbate overlayer Fe Ni Fe Ni Fe Pd

0% (vacuum) +55 +53 +104

100% H +49 +52 +43

100% CO +29 +42 +3

100% O −94 −76 −218

50% CO, 50% O −25 −8 −92

50% CO, 25% O, 25% H +1 +11 +38

25% of CH, CO, O, H +2 +9 +143

The DFT calculations therefore indicate that (i) the segregation behavior of Fe is a very

strong function of the adsorbate layer present, and (ii) the presence of Pd in a Ni-Fe alloy

will reduce the tendency of Fe to segregate to the surface for coverages that are close to

what can be expected during DRM conditions, which can explain the improved deactivation

behavior of the Ni-Fe-Pd catalysts.


166

All of the above leads to the statement that the 0-Pd catalyst mainly deactivates due to

extraction of Fe from the Ni-Fe alloy during DRM. The addition of 0.2wt% of Pd to Ni-Fe

forms a trimetallic layer at the surface of the particles, leading to a more bimetallic Ni-Fe

core with a trimetallic Ni-Fe-Pd shell. The latter suppresses segregation of Fe from within the

core to the surface. Figure 5.17 illustrates in a schematical representation the structure of

the Pd-modified samples under reforming conditions.

Figure 5.17: Schematical representation of the proposed structure of Pd-modified samples (i.e 0.2-

Pd) during DRM at high temperature (1023 K).

5.5 Conclusions

Although a Ni-Fe catalyst presents high activity in DRM, with high carbon resistance, it

does suffer from deactivation via sintering and Fe segregation. Therefore, enhanced control

of the stability and activity of Ni-Fe/MgAl2O4 was examined by means of Pd addition. During

reduction of a co-impregnated sample, a core shell alloy forms at the surface, consisting of a

Ni-Fe core and a Ni-Fe-Pd shell. The alloy is decomposed during CO2-TPO into a Ni-Pd alloy

and Fe3O4, above 850 K.

The effect of Ni:Pd ratio on the catalytic properties of Ni-Fe/MgAl2O4 was investigated

systematically by varying the Pd concentration, during DRM at 1023 K and 101.3 kPa. The

addition of Pd stabilizes the Ni-Fe alloy by means of a thin Ni-Fe-Pd surface layer. The latter

acts as a barrier for Fe segregation from the core during DRM. Segregation of Fe from the

trimetallic shell will still occur but to a lesser extent as the Fe concentration is lower.

Moreover, DFT calculations show that the presence of Pd in a Ni-Fe alloy reduces the

tendency of Fe to segregate to the surface for coverages that are relevant to DRM

conditions. A Ni:Pd molar ratio equal to 75 was found to be the optimal promotion as it

DRM conditions

Reduced

Ni-Fe

Ni-Fe-Pd

Ni-Fe

Fe or FeOx

Ni-Pd


167

increased both the activity and stability of the catalyst. The origin of the improved catalytic

activity derives from higher Ni dispersion along with the creation of extra active sites, due to

Pd addition. The best candidate was examined under bi-reforming (CO2 + H2O) at 1023 K and

101.3 kPa of a gas mixture, which simulates the clean effluent of a biomass gasifier (absence

of tars and sulfur compounds), confirming that the catalyst is not only active for DRM but

also for other reforming reactions, containing steam in the feed mixture.


168

5.6 References

[1] M. Maestri, D.G. Vlachos, A. Beretta, G. Groppi, E. Tronconi, A C1 microkinetic model for methane conversion to syngas on Rh/Al2O3, AICHE J., 55 (2009) 993-1008. [2] M.P. Kohn, M.J. Castaldi, R.J. Farrauto, Auto-thermal and dry reforming of landfill gas over a Rh/γAl2O3 monolith catalyst, Appl. Catal., B, 94 (2010) 125-133. [3] A. Caballero, P.J. Perez, Methane as raw material in synthetic chemistry: the final frontier, Chem. Soc. Rev., 42 (2013) 8809-8820. [4] N. Kumar, M. Shojaee, J.J. Spivey, Catalytic bi-reforming of methane: from greenhouse gases to syngas, Curr. Opin. Chem. Eng., 9 (2015) 8-15. [5] S.D. Angeli, L. Turchetti, G. Monteleone, A.A. Lemonidou, Catalyst development for steam reforming of methane and model biogas at low temperature, Appl. Catal., B, 181 (2016) 34-46. [6] E.T. Kho, J. Scott, R. Amal, Ni/TiO2 for low temperature steam reforming of methane, Chem. Eng. Sci., 140 (2016) 161-170. [7] S.A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, Enhanced Carbon-Resistant Dry Reforming Fe-Ni Catalyst: Role of Fe, ACS Catal., (2015) 3028-3039. [8] M.D. Salazar-Villalpando, D.A. Berry, T.H. Gardner, Partial oxidation of methane over Rh/supported-ceria catalysts: Effect of catalyst reducibility and redox cycles, Int. J. Hydrogen Energy, 33 (2008) 2695-2703. [9] A. Scarabello, D. Dalle Nogare, P. Canu, R. Lanza, Partial oxidation of methane on Rh/ZrO2 and Rh/Ce–ZrO2 on monoliths: Catalyst restructuring at reaction conditions, Appl. Catal., B, 174–175 (2015) 308-322. [10] Y. Yan, Z. Zhang, L. Zhang, X. Wang, K. Liu, Z. Yang, Investigation of autothermal reforming of methane for hydrogen production in a spiral multi-cylinder micro-reactor used for mobile fuel cell, Int. J. Hydrogen Energy, 40 (2015) 1886-1893. [11] A. M. Gaddalla, M. E. Sommer, Carbon dioxide reforming of methane on nickel catalysts, Chem. Eng. Sci., 44 (1989) 2825-2829. [12] D. Pakhare, J. Spivey, A review of dry (CO2) reforming of methane over noble metal catalysts, Chem. Soc. Rev., 43 (2014) 7813-7837. [13] J.R.H. Ross, Natural gas reforming and CO2 mitigation, Catal. Today, 100 (2005) 151-158. [14] J. Galuszka, R.N. Pandey, S. Ahmed, Methane conversion to syngas in a palladium membrane reactor, Catal. Today, 46 (1998) 83-89. [15] G.A. Olah, G.K.S. Prakash, A. Goeppert, Anthropogenic Chemical Carbon Cycle for a Sustainable Future, J. Am. Chem. Soc., 133 (2011) 12881-12898. [16] G.A. Olah, A. Goeppert, M. Czaun, G.K.S. Prakash, Bi-reforming of Methane from Any Source with Steam and Carbon Dioxide Exclusively to Metgas (CO–2H2) for Methanol and Hydrocarbon Synthesis, J. Am. Chem. Soc., 135 (2013) 648-650. [17] G.A. Olah, A. Goeppert, M. Czaun, T. Mathew, R.B. May, G.K.S. Prakash, Single Step Bi-reforming and Oxidative Bi-reforming of Methane (Natural Gas) with Steam and Carbon Dioxide to Metgas (CO-2H2) for Methanol Synthesis: Self-Sufficient Effective and Exclusive Oxygenation of Methane to Methanol with Oxygen, J. Am. Chem. Soc., 137 (2015) 8720-8729. [18] I. Luisetto, S. Tuti, E. Di Bartolomeo, Co and Ni supported on CeO2 as selective bimetallic catalyst for dry reforming of methane, Int. J. Hydrogen. Energ., 37 (2012) 15992-15999. [19] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, Ni-based catalysts for reforming of methane with CO2, Int. J. Hydrogen Energ., 37 (2012) 15966-15975. [20] S.A. Theofanidis, R. Batchu, V.V. Galvita, H. Poelman, G.B. Marin, Carbon gasification from Fe–Ni catalysts after methane dry reforming, Appl. Catal., B, 185 (2016) 42-55. [21] V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, Transient Kinetic Study of the Oxidation and Hydrogenation of Carbon Species Formed during CH4/He, CO2/He, and CH4/CO2 reactions over Rh/Al2O3 Catalyst, J. Catal., 161 (1996) 31-42.


169

[22] V.A. Sadykov, E.L. Gubanova, N.N. Sazonova, S.A. Pokrovskaya, N.A. Chumakova, N.V. Mezentseva, A.S. Bobin, R.V. Gulyaev, A.V. Ishchenko, T.A. Krieger, C. Mirodatos, Dry reforming of methane over Pt/PrCeZrO catalyst: Kinetic and mechanistic features by transient studies and their modeling, Catal. Today, 171 (2011) 140-149. [23] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, Reforming of Methane with Carbon Dioxide to Synthesis Gas over Supported Rhodium Catalysts: II. A Steady-State Tracing Analysis: Mechanistic Aspects of the Carbon and Oxygen Reaction Pathways to Form CO, J. Catal., 158 (1996) 64-75. [24] M.C.J. Bradford, M.A. Vannice, CO2 Reforming of CH4, Catal. Rev. - Sci. Eng., 41 (1999) 1-42. [25] J. Guo, C. Xie, K. Lee, N. Guo, J.T. Miller, M.J. Janik, C. Song, Improving the Carbon Resistance of Ni-Based Steam Reforming Catalyst by Alloying with Rh: A Computational Study Coupled with Reforming Experiments and EXAFS Characterization, ACS Catal., 1 (2011) 574-582. [26] D.L. Trimm, Coke formation and minimisation during steam reforming reactions, Catal. Today, 37 (1997) 233-238. [27] D.L. Trimm, Catalysts for the control of coking during steam reforming, Catal. Today, 49 (1999) 3-10. [28] J. Ashok, S. Kawi, Steam reforming of toluene as a biomass tar model compound over CeO2 promoted Ni/CaO–Al2O3 catalytic systems, Int. J. Hydrogen Energy, 38 (2013) 13938-13949. [29] L. Wang, D. Li, M. Koike, H. Watanabe, Y. Xu, Y. Nakagawa, K. Tomishige, Catalytic performance and characterization of Ni–Co catalysts for the steam reforming of biomass tar to synthesis gas, Fuel, 112 (2013) 654-661. [30] D. Li, I. Atake, T. Shishido, Y. Oumi, T. Sano, K. Takehira, Self-regenerative activity of Ni/Mg(Al)O catalysts with trace Ru during daily start-up and shut-down operation of CH4 steam reforming, J. Catal., 250 (2007) 299-312. [31] K. Nagaoka, A. Jentys, J.A. Lercher, Methane autothermal reforming with and without ethane over mono- and bimetal catalysts prepared from hydrotalcite precursors, J. Catal., 229 (2005) 185-196. [32] T. Huang, W. Huang, J. Huang, P. Ji, Methane reforming reaction with carbon dioxide over SBA-15 supported Ni–Mo bimetallic catalysts, Fuel Process. Technol., 92 (2011) 1868-1875. [33] Z. Hou, T. Yashima, Small Amounts of Rh-Promoted Ni Catalysts for Methane Reforming with CO2, Catal. Lett., 89 193-197. [34] B. Pawelec, S. Damyanova, K. Arishtirova, J.L.G. Fierro, L. Petrov, Structural and surface features of PtNi catalysts for reforming of methane with CO2, Appl. Catal., A, 323 (2007) 188-201. [35] K. Nagaoka, K. Takanabe, K.-i. Aika, Modification of Co/TiO2 for dry reforming of methane at 2 MPa by Pt, Ru or Ni, Appl. Catal., A, 268 (2004) 151-158. [36] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, C. Sener, T. Dogu, MCM-41 supported PdNi catalysts for dry reforming of methane, Appl. Catal., B, 92 (2009) 250-261. [37] M. Ferrandon, A.J. Kropf, T. Krause, Bimetallic Ni-Rh catalysts with low amounts of Rh for the steam and autothermal reforming of n-butane for fuel cell applications, Appl. Catal., A, 379 (2010) 121-128. [38] B. Steinhauer, M.R. Kasireddy, J. Radnik, A. Martin, Development of Ni-Pd bimetallic catalysts for the utilization of carbon dioxide and methane by dry reforming, Appl. Catal., A, 366 (2009) 333-341. [39] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catal., 4 (2014) 289-301. [40] M. Koike, D. Li, H. Watanabe, Y. Nakagawa, K. Tomishige, Comparative study on steam reforming of model aromatic compounds of biomass tar over Ni and Ni–Fe alloy nanoparticles, Appl. Catal., A, 506 (2015) 151-162. [41] A. Djaidja, H. Messaoudi, D. Kaddeche, A. Barama, Study of Ni–M/MgO and Ni–M–Mg/Al (M=Fe or Cu) catalysts in the CH4–CO2 and CH4–H2O reforming, Int. J. Hydrogen Energy, 40 (2015) 4989-4995. [42] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst characterization, Pure & Appl. Chem., (1995).


170

[43] Scherrer, Nachr. Ges. Wiss. Göttingen, 1918. [44] Niemantsverdiet J.W., Spectroscopy in Catalysis, 2007. [45] J.J. Carberry, D. White, On the role of transport phenomena in catalytic reactor behavior, Ind. Eng. Chem., 61 (1969) 27-35. [46] G.F. Froment, K. Bischoff, J. De Wilde, Chemical Reactor Analysis and Design, 3rd ed., J. Wiley & Sons2011. [47] D.E. Mears, Diagnostic criteria for heat transport limitations in fixed bed reactors, J. Catal., 20 (1971) 127-131. [48] G.I. Silman, Compilative Fe-Ni phase diagram with author's correction, Met. Sci. Heat Treat., 54 (2012) 105-112. [49] C.E. Dahmani, M.C. Cadeville, J.M. Sanchez, J.I. Moranlopez, Ni-Pt Phase -diagram - experiment and theory, Phys. Rev. Lett., 55 (1985) 1208-1211. [50] R.H. Davies, A.T. Dinsdale, J.A. Gisby, J.A.J. Robinson, S.M. Martin, MTDATA - thermodynamic and phase equilibrium software from the national physical laboratory, Calphad, 26 (2002) 229-271. [51] M.K. Sabbe, L. Lain, M.F. Reyniers, G.B. Marin, Benzene adsorption on binary Pt3M alloys and surface alloys: a DFT study, PCCP, 15 (2013) 12197-12214. [52] Y.G. Ma, P.B. Balbuena, Surface segregation in bimetallic Pt3M (M = Fe, Co, Ni) alloys with adsorbed oxygen, Surf. Sci., 603 (2009) 349-353. [53] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. Lett., 77 (1996) 3865. [54] G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Comput. Mat. Sci., 6 (1996) 15. [55] G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B, 54 (1996) 11169. [56] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals, Phys. Rev. B, 47 (1993) 558. [57] G. Kresse, J. Hafner, Ab initio molecular-dynamics simulation of the liquid-metal-amorphous-semiconductor transition in germanium, Phys. Rev. B, 49 (1994) 14251. [58] J. Guo, H. Lou, H. Zhao, D. Chai, X. Zheng, Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels, Appl. Catal. A 273 (2004) 75-82. [59] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Catalytic performance and characterization of Ni-Fe catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Appl. Catal. A 392 (2011) 248-255. [60] V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Catalyst-assisted chemical looping for CO2 conversion to CO, Appl. Catal., B, 164 (2015) 184-191. [61] J. Hu, L. Buelens, S.-A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, CO2 conversion to CO by auto-thermal catalyst-assisted chemical looping, J. CO2 Util., 16 (2016) 8-16. [62] J.C.S. Wu, H.-C. Chou, Bimetallic Rh–Ni/BN catalyst for methane reforming with CO2, Chem. Eng. J., 148 (2009) 539-545. [63] Z. Hou, P. Chen, H. Fang, X. Zheng, T. Yashima, Production of synthesis gas via methane reforming with CO2 on noble metals and small amount of noble-(Rh-) promoted Ni catalysts, Int. J. Hydrogen Energy, 31 (2006) 555-561. [64] S.M. Kim, P.M. Abdala, T. Margossian, D. Hosseini, L. Foppa, A. Armutlulu, W. van Beek, A. Comas-Vives, C. Copéret, C. Müller, Cooperativity and Dynamics Increase the Performance of NiFe Dry Reforming Catalysts, J. Am. Chem. Soc., (2017). [65] V. Galvita, K. Sundmacher, Cyclic water gas shift reactor (CWGS) for carbon monoxide removal from hydrogen feed gas for PEM fuel cells, Chem. Eng. J., 134 (2007) 168-174. [66] L.C. Buelens, V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Super-dry reforming of methane intensifies CO2 utilization via Le Chatelier’s principle, Science, 354 (2016) 449-452. [67] M.P. Felicissimo, O.N. Martyanov, T. Risse, H.-J. Freund, Characterization of a Pd–Fe bimetallic model catalyst, Surf. Sci., 601 (2007) 2105-2116.

Chapter 6: Controlling the stability of a Ni-Fe reforming catalyst: down tuning the Fe concentration

171

Chapter 6

Controlling the stability of a Ni-Fe reforming catalyst: down tuning the Fe concentration

Replacing noble metals in heterogeneous catalysis remains a challenge. Viable

substitutes should combine their excellent stability and carbon resistance with high

abundancy. In Chapter 5, noble metals were used in order to stabilize the Ni-Fe alloy during

reforming conditions. However, this Chapter introduces a novel MgFexAl2-xO4 support for Ni-

Fe, offering excellent stability and carbon control. Fe was incorporated into the magnesium

aluminate spinel, forming a support with redox functionality and high thermal stability. By

carefully down tuning the amount of Fe in MgFexAl2-xO4, an active alloy was formed with a

Fe:Ni molar ratio ~1:10, which fully mitigates carbon formation, while keeping Ni highly

accessible. This Ni-Fe/MgFexAl2-xO4 makes a combination of stable alloy and active support, a

worthy substitute for noble metals.


172

6.1 Introduction

Syngas production from methane has been widely investigated by various reforming

technologies like steam reforming (SMR), dry reforming (DRM), bi-reforming (BIM) [1, 2],

partial oxidation (POM) and autothermal reforming (ARM) [3] over different series of

supported catalysts [4]. The used oxidant, energetics and the final H2:CO ratio differ in these

processes [5].

There is no industrial technology developed for DRM yet because the selection and

development of a reforming catalyst remains an important challenge [6]. Only a combination

of steam and dry reforming plant (bi-reforming) has been recently demonstrated by the

Japan Oil, Gas and Metals National Corporation [7]. Nickel-based are preferred over noble

metal catalysts due to their high availability and lower cost. However, they are sensitive to

deactivation by carbon deposition and sintering [8-10]. Hence, much research effort has

focused on improving catalyst activity and stability by the addition of promoters [11, 12],

synthesis of bimetallic catalysts [13-15] and changing the nature of the support [6, 16]. In

that respect, Fe-modified Ni catalysts restrain carbon accumulation by forming Ni-Fe alloys

[10, 13, 17, 18], but despite years of research still do not provide the stability offered by

noble metals: during methane reforming Fe segregates from the alloy [17, 19] and blocks the

Ni sites, leading to deactivation.

Support plays a crucial role in the activity and stability of a catalyst and therefore it has

been widely investigated for the aforementioned reforming reactions. Al2O3 has been widely

used as a catalyst support [10, 20] while various other support materials such as MgO, ZrO2,

La2O3, TiO2, SiO2, SiO2- Al2O3 have been also examined [20-29]. The required properties for a

support material are: 1) high temperature resistance, 2) ability to maintain the dispersion of

active metals during reaction conditions and 3) oxygen storage capacity [30].

The present study suggests that MgFexAl2-xO4 can be used for reforming reactions as a

cheaper alternative support, compared to i.e. CeO2 or CeZrO2, with redox functionality,

suppressing significantly the carbon deposition during reforming reactions at high

temperatures (1023 K).


173


6.2.1 Support and catalyst preparation

Support preparation

MgFexAl2-xO4 support materials (where x = 0, 0.007, 0.013 ,0.04 ,0.09, 0.13, 0.18 and

0.26) were prepared by co-precipitation from an aqueous solution of Mg(NO3)2·6H2O (99%,

Sigma-Aldrich®), Al(NO3)3·9H2O (98.5%, Sigma-Aldrich®) and Fe(NO3)3·9H2O (99.99+%, Sigma-

Aldrich®). A precipitating agent, NH4OH (ACS reagent, 28.0-30.0% NH3 basis) was added to

adjust the pH to 10, at 333 K. The produced precipitate was filtered, dried at 393 K for 12 h

and subsequently calcined in air at 1073 K for 4 h. The calcined support samples are labeled

using their weight percent “z”, as zFe as shown in Table 6.1.


8wt%Ni catalysts were prepared by incipient wetness impregnation on the support

materials (MgFexAl2-xO4) using an aqueous solution of nitrate Ni(NO3)2·6H2O (99.99+%,

Sigma-Aldrich®®) [31]. The catalysts were dried at 393 K for 12 h and subsequently calcined

in air at 1073 K for 4 h. The calcined samples, named “as-prepared”, are labeled as Ni/zFe

where z is the Fe weight percent of the support (Table 6.3) In addition, a bimetallic 8wt%Ni-

4wt%Fe catalyst was prepared by incipient wetness impregnation on the MgAl2O4 support,

named as “Ni-4.0Fe” as reference material for comparison purposes, using an aqueous

solution of corresponding nitrate Ni(NO3)2·6H2O (99.99+%, Sigma-Aldrich®®) and

Fe(NO3)3·9H2O (99.99+%, Sigma-Aldrich®®). Finally, a monometallic noble metal 1wt%Rh

catalyst supported on MgAl2O4 was prepared as a noble metal reference catalyst, using an

aqueous solution of Rh(NO3)3.xH2O (36% Rh basis, Sigma-Aldrich®®).

6.2.2 Support and catalyst characterization


adsorption at 77 K (five point BET method using Tristar Micromeritics) after outgassing the

sample at 473 K for 2 h. The crystallographic phases of the as-prepared, reduced and re-

oxidized materials were determined by ex-situ XRD measurements (Siemens Diffractometer

Kristalloflex D5000, Cu Kα radiation). The powder patterns were collected in a 2θ range from



174

10o to 80o with a step of 0.02o and 30 s counting time per angle. XRD patterns of known

compounds are referenced by their corresponding number in the Powder Diffraction File

database. By fitting a Gaussian function to a diffraction peak, the crystallite size was

determined from the peak width via the Scherrer equation [32], while the peak position gave

information about the lattice spacing based on the Bragg law of diffraction [33].

The bulk chemical composition of support and as-prepared catalysts was determined by


Thermo Scientific). The samples were mineralized by acid fusion.

The Fe-K edge XANES (X-ray absorption near edge structure) and EXAFS (Extended X-ray

absorption fine structure) spectra of the support (MgFexAl2-xO4) were collected at the Dutch-

Belgian Beam Line (DUBBLE, BM26A) at the European Synchrotron research Facility (ESRF).

The measurements were carried out in transmission mode using Ar/He filled ionization

chambers at ambient temperature and pressure. The energy of the X-ray beam was tuned

using a double crystal monochromator operating in fixed exit mode using a Si (111) crystal

pair.

In-situ XANES was performed during H2 temperature programmed reduction (TPR) (0.2

NmL/s of 5%H2/He using a heating ramp of 10 K/min) in a capillary reactor with an internal

diameter of 1mm. The capillary reactor was implemented in a frame which was connected

to gas feed lines [34] through Swagelok fittings, while the inlet gas flow rates were

maintained by means of calibrated Brooks mass flow controllers. After reduction, in-situ

XANES was carried out as the sample was subjected to re-oxidation by CO2 (named “re-

oxidized”), without removing it from the reactor.

Theoretical simulations of XANES spectra were performed with the FDMNES [35] software

package. The XANES spectra, three scans per sample, were energy-calibrated, averaged and

analyzed using the multiple scattering theory based on the muffin-tin approximation of the

potential well. The muffin-tin radii were tuned to have a 10% overlap between the different

spherical potentials and the Hedin–Lundqvist exchange potential was applied. Further, a

core-hole broadening of 1.2 eV was used, and the FWHM of the Gaussian used for the

energy resolution set to 2 eV. The XANES spectrum was simulated, considering all atoms

surrounding the Fe absorber within a 7 Å radius sphere. The same approach was used for the


175

as-prepared/non-distorted and reduced/distorted Fe environment. The approximation of

non-excited absorbing atoms was used which better reproduces the experimental data. We

have not considered the effect of structural disorder in the XANES simulations.

The linear combination fitting range was limited to a [-30eV, + 30eV] range around the

edge µ(E) inflection point to limit the metal amount estimation based on pre-edge features.

For the fitting of the Ni-K and Fe-K edges, the standards used were metallic Ni and Fe,

respectively, along with the “as-prepared” material.

EXAFS spectra of the as-prepared materials were measured on pellets diluted with a

suitable concentration of inert diluent BN to avoid self-absorption. These spectra were

measured in a closed cycle He-cryostat (Oxford Instruments) at 80 K to minimize the thermal

noise contributions from the Debye Waller factors. The EXAFS measurements, following the

H2-TPR (named “reduced”) and CO2-TPO treatment (named “re-oxidized”) were carried out

at room temperature. Three scans per material were measured, energy calibrated, averaged

and then analyzed using the GNXAS methodology [36]. In this approach, the local atomic

arrangement around the absorbing atom is decomposed into model atomic configurations

containing 2, ..., n atoms. The theoretical EXAFS signal χ(k) is given by the sum of the n-body

contributions γ2, η3, ..., which take into account all the possible single and multiple

scattering paths between the n atoms. The fitting of χ(k) to the experimental EXAFS signal

allows to refine the relevant structural parameters of the different coordination shells; the

suitability of the model is evaluated by comparison of the Fourier transformed (FT)

experimental EXAFS signal with the FT of the calculated χ(k) function. The global fit

parameters that were allowed to vary during the fitting procedure were the distance R(Å),

Debye-Waller factor (σ2) and the angles of the η3 contributions, which were defined

according to the spinel structure used in the data analysis. The edge energy E0 was defined

at 7112 eV according to the value of the Fe foil.

For the analysis of the fresh and re-oxidized samples, one γ2 term, representing the six

fold Fe-O distance (R1) at ~1.97 Å in the octahedral spinel site, as well as two three-body

configurations η3 involving the Fe-O-M (M=Al,Mg) triangular arrangement (Figure 6.1) and

their multiple scattering effects were taken into account. In this case the coordination

numbers were released as free parameters. The obtained values are reported in Table 6.2.


176

Figure 6.1: Structure showing the γ2 two- and η3 three-body configurations with angles at 116°

(planar configuration) and 95° (perpendicular configuration) used in the modeled EXAFS signal. Mg

is located in tetrahedral sites while Fe or Al in octahedral. Blue color corresponds to oxygen, red to

Fe and light blue to Al atoms.

For the reduced sample, two different Fe-O distances can be detected in the EXAFS data

which can be assigned to a nearly unmodified Fe3+ (non-distorted) and a distorted Fe2+ local

environment. The existence of two different local Fe environments is in good agreement

with XANES. Therefore, the Fe local environment has been modeled considering two γ2

terms, corresponding to Fe-O distances of ~1.92 Å and ~2.10 Å for non-distorted and

distorted Fe sites respectively, to account for partial reduction. Accordingly, four η3 three-

body configurations, involving the Fe-O-M (M=Al,Mg) were used. During the reduction

treatment, Fe3+ reduces to Fe2+ which has a larger ionic radius [37]. Thus, an increase of the

average Fe-O distance is expected upon reduction.

In order to carry out a quantitative assessment of the two phases, we constrained the

total coordination number of the first shell to its fully coordinated value ± 1 (coordination

number Fe-O equal to 6) and parameterized the respective contributions as a function of the

total Fe amount (parameter β, Table 6.2). The main reason to fix the coordination numbers

is that the error in this parameter is higher than the number of possible oxygen vacancies

confined to the Fe neighborhood, which can be present in the reduced structure. Because

the disorder term of the first shell is correlated to the ones of the higher shells, whose

95o

116o


177

distance is correlated to the first shell, we fixed the Debye Waller factor to the value as

obtained from the single shell analysis.

The suitability of the model was evaluated by comparing the Fourier transform (FT) of

experiment and model data to refine parameters such as coordination numbers, bond

distances and Debye-Waller factors from the two-body and three-body configurations, which

were defined according to the crystallographic structures used in the model.

6.2.3 Operando quick-XAS (QXAS)

An operando-QXAS study was performed during H2-TPR. QXAS was measured in

transmission at the ROCK [38] beam line of the SOLEIL synchrotron using one oscillating

monochromator over both Fe-K and Ni-K edges (7112 and 8331 eV), by means of a macro for

fast switching between edges. Approximately 5mg of as-prepared material, 50% diluted by

boron nitride, was inserted into a 2 mm quartz capillary reactor and fixed by quartz wool

plugs. The capillary reactor was implemented in a frame which was connected to gas feed

lines through Swagelok fittings while the inlet gas flow rates were maintained by means of

calibrated Brooks mass flow controllers. The transmission measurements were coupled with

kinetic measurements based on gas-phase compositions obtained from MS. An external

calibrated heat gun was used in order to reach the reaction temperature of 1023 K, while a

total flow between 10-60 NmL/min was employed. Gas rig, MS, capillary frame and heat

blower were used from the SOLEIL synchrotron. ). For more details the reader is referred to

§2.2.5.





with an angular resolution of 0.1o. Patterns were recorded in 10 s acquisition time. All

temperatures were measured with a K-type thermocouple and corrected afterwards to a


and Ag–Si. For each sample, approximately 10 mg of powdered sample was evenly spread on

a single crystal Si wafer. No interaction of the catalyst material with the Si wafer was


178

observed. Before each experiment the reactor chamber was evacuated to a base pressure of

4 Pa by a rotation pump. Gases were supplied to the reactor chamber with calibrated mass-

flow controllers. He (10 NmL/s) was flowing for 10 min before the flow switched to 10 NmL/s

of 5%H2/He or CO2 for TPR and TPO experiments (up to 1123 K, heating rate of 20 K/min),

respectively.

Peaks in the in-situ XRD patterns appeared at slightly shifted angular positions compared

to both full scans and tabulated values due to temperature-induced lattice expansion and

different sample heights. These shifts in peak positions, which are not related to underlying

physicochemical processes, were taken into account during peak assignment.

6.2.5 Isothermal Temporal Analysis of Products (TAP) experiments

Transient measurements were performed in a TAP-3E reactor (Mithra Technologies, St.

Louis, USA) equipped with an Extrel Quadrupole Mass Spectrometer (QMS). The details of

TAP experiments can be found in [39]. For the experiments, 20 mg (250<d<500 μm catalyst

fraction) of the calcined catalyst was placed in a quartz microreactor (I.D. = 4 mm and ~2

mm bed length), which was located between two inert beds of quartz particles with the

same sieved fraction. The temperature of the catalyst was measured by a K-type

thermocouple housed inside the catalytic zone. Prior to the experiments, the catalyst was

reduced at 1173 K reduced in a 1 NmL/s flow of 30%H2/Ar at 1123 K (heating rate of 20

K/min) and atmospheric pressure. A series of CH4 pulses (~10-7 mol/pulse) were used

isothermally at 993 K. For every 1 second, data were recorded with millisecond time

resolution in each pulse. CO2, CO, H2, H2O, CH4 and Ar (internal standard) responses were

monitored at amu signals of 44, 28, 2, 18, 16 and 40, respectively.


Activity and stability measurements were performed at 1023 K and 111.3 kPa in a quartz

reactor with an internal diameter of 9 mm, which was housed inside an electric furnace. A

coated (by amorphous silicon) incoloy alloy 800-HT reactor of the same dimensions was used

for the high pressure experiments (911.7 kPa). The activity of the coated incoloy alloy

reactor was always tested in a blank experiment, at the reaction conditions, before each

measurement. The temperature of the catalyst bed was measured with K-type


179

thermocouples touching the outside and inside of the reactor at the position of the catalyst

bed. The inlet gas flow rates were always maintained by means of calibrated Bronkhorst

mass flow controllers. The sample with particle size fraction of ~50 μm was diluted with inert

Al2O3 for improved heat conductivity (ratio catalyst/inert ~1/90) and packed between quartz

wool plugs resulting in approximately 10-2 m of catalyst bed length. Prior to each

experiment, the as-prepared sample was reduced in a 1 NmL/s flow of 30%H2/Ar at 1123 K

(heating rate of 20 K/min) for 20 minutes (named “reduced”) and then the flow was

switched to 1 NmL/s of Ar for 10 minutes. A mixture of CH4, CO2 and Ar (total flow of 105

NmL/min, volumetric ratio CH4:CO2:Ar = 1:1:0.5, Ar internal standard) was used for DRM and

a mixture of CH4, CO2, H2O and Ar (total flow of 300 NmL/min, 15% CH4, 11% CO2, 4% H2O

and 70% Ar, Ar internal standard) for BIM, while the produced CO, H2 and unconverted CH4

and CO2 were detected at the outlet using a calibrated OmniStar Pfeiffer mass spectrometer

(MS). MS signals were recorded for all major fragments. For quantification of reactants and

products, the MS was focused to different amu signals. H2 was monitored at 2, CH4 at 16,

H2O at 18, CO at 28, Ar at 40 and CO2 at 44 amu. A correction was applied to remove

contributions from unavoidable interference with fragmentation peaks of other gases. A

carbon balance with a maximum deviation of 6% was obtained. To evaluate the significance

of external and internal mass transfer limitations, the criteria of Carberry number [40] and

Weisz-Prater [41] were applied, while for heat transport limitations the diagnostic criteria

reported by Mears [42] were employed. A radial temperature gradient of ~29 K was

calculated. This temperature gradient was taken into account for the calculation of the

equilibrium conversion.

The activity was also measured as a function of time in a short term experiment (time-on-

stream (TOS) = 2 h) during DRM at 1023 K and 111.3 kPa, using a mixture of CH4, CO2 and Ar

(total flow of 105 NmL/min, volumetric ratio CH4:CO2:Ar = 1:1:0.5, Ar internal standard). The

same conversion range (XCH4 ~ 70%) was achieved for all of the investigated samples, by

varying the amount of catalyst (WNi:F0

CH4 = 0.024-0.08 kgNi·s·molCH4-1). It always remained

below the equilibrium conversion (Xeq,CH4 ~ 80% at 1023 K and 111.3 kPa considering the

above mentioned radial temperature gradient).

The best candidates of the activity tests were further tested for longer TOS (stability)

during DRM at 1023 K and 111.3 kPa. The space time values (WNi:F0

CH4) of the samples used


180

during the stability tests are in the range [0.023 - 0.025] kgNi·s·mol-1CH4 for Ni/0.5Fe, Ni/1.0Fe

and Ni/2.5Fe (initial XCH4~ 70%). One stability test was also performed for Ni/0.5Fe at lower

initial CH4 conversion value (~54%), using a WNi:F0

CH4 of 0.011 kgNi·s·mol-1CH4.

Two samples were examined under BIM at 1023 K and 911.7 kPa. The best candidate of

the activity and stability tests was compared with the Ni/0Fe. The space time values

(WNi:F0

CH4) of the samples used during the activity tests at 911.7 kPa were in the range

[0.003 - 0.004] kgNi·s·mol-1CH4 for Ni/1.0Fe and Ni/0Fe.



𝑋𝑖 = 𝐹𝑖

0−𝐹𝑖

𝐹𝑖0 · 100% (1)

where 𝐹𝑖0 and 𝐹𝑖 are the inlet and outlet molar flow rates of gas (mol·s-1).

The CH4 consumption rate (mmol·s-1·gNi-1, Eq. 2) was calculated from the difference

between the inlet and outlet molar flow rates, as measured relative to an internal standard

(Ar), while the space time yield (STY, mmol·s-1·gNi-1, Eq. 3) was calculated for the products:


0−𝐹𝑖|

𝑚𝑁𝑖 (2)

STYi = 𝐹𝑖

𝑚𝑁𝑖 (3)

where 𝐹𝑖0and 𝐹𝑖 are the inlet and outlet molar flow rates of gas i (mol·s-1), and mNi the

amount of Ni (kg) present in the catalyst.

6.2.7 Mechanical mixture

A mechanical mixture of graphite and 10Fe support (100mg graphite - 300 mg support)

was prepared in order to investigate the support oxygen transfer properties under a

temperature programmed experiment in an inert environment (1 NmL·s-1 Ar flow). The

temperature was increased from room temperature up to 1173 K, using a heating rate of 10

K·min-1. A CO or CO2 signal in the outlet of the reactor then indicates graphite oxidation.


181

Another mechanical mixture of 7.5Fe support and Ni/0Fe was used in a reduction

experiment, in order to investigate the Fe reduction and migration process towards the

surface Ni. The mechanical mixture was reduced the under 10 NmL·s-1 of 5%H2/He mixture

at a total pressure of 101.3 kPa and 1073 K, using in-situ XRD.

6.3 Results

6.3.1 Support Characterization

The metal content and the textural properties are reported in Table 6.1 for the support

materials and in Table 6.3 for the studied catalysts. The 0Fe support showed the highest

surface area and the largest pore volume, while the addition of Fe resulted in surface area

decrease for the support materials. No significant changes were observed in the pore

volume by increasing the Fe content. The surface area tabulated values are in accordance

with literature reports [43].

Table 6.1: Support properties. Metal loading was determined by means of inductively coupled

plasma atomic emission spectroscopy (ICP-AES).

Abbreviation

“zFe” Catalyst

Metal loading

(wt.%) BET

(m2/gcat)

Pore volume

(cm³/gcat) Mg Al Fe

0Fe MgAl2O4 14.35 32.91 - 81.0 ± 4.1 0.23 ± 0.03

0.5Fe MgFe0.007Al1.993O4 15.90 36.00 0.51 78.1 ± 1.9 0.21 ± 0.05

1.0Fe MgFe0.013Al1.987O4 15.50 35.80 1.01 79.3 ± 0.8 0.20 ± 0.02

2.5Fe MgFe0.04Al1.96O4 14.12 30.85 2.87 77.1 ± 1.9 0.18 ± 0.01

5.0Fe MgFe0.09Al1.91O4 12.76 29.29 5.76 63.7 ± 3.5 0.17 ± 0.01

7.5Fe MgFe0.13Al1.87O4 12.66 28.22 8.88 57.9 ± 2.1 0.20 ± 0.01

10Fe MgFe0.18Al1.82O4 11.90 26.80 12.18 50.1 ± 2.2 0.18 ± 0.01

15Fe MgFe0.26Al1.74O4 11.78 26.50 20.20 31.1 ± 0.3 0.13 ± 0.00

Figure 6.2 displays the pore size distribution of the support materials that have been

examined. The addition of Fe in the support material results in an increase of the pore size,


182

as the peak maximum shift to higher pore width values. The same trend with no significant

changes were observed after Ni impregnation on the corresponding support materials.

Figure 6.2: Pore size distribution calculated by Jura and Harkins model (pressure range 0.25-0.98)

for different MgFexAl2-xO4 support materials.

A mechanical mixture of MgFe0.18Al1.82O4 support with commercial graphite (weight ratio

3:1) showed clear redox activity during temperature programmed treatment in an inert

environment (Ar) (Figure 6.3a). The oxidation of graphite by lattice oxygen gave rise to CO

formation, starting at 1000 K and up to 1173 K. In this process 5 mmolO·kgcat-1 were

consumed. The lattice oxygen can easily be regenerated by CO2, H2O or O2, making MgFexAl2-

xO4 an active support, with redox properties based on oxygen mobility, as exemplified by

Temporal Analysis of Products (TAP) experiments. Figure 6.3b shows the catalyst bed

configuration during these experiments. CO2 was produced during CH4 pulses over an “as-

prepared” MgFexAl2-xO4 support (Figure 6.3c), implying that CH4 was decomposed on

MgFexAl2-xO4, in some Fe sites (CH4(g) → C (s) + 2H2(g)). Then, the deposited carbon was

oxidized by lattice oxygen, producing CO2. 15.6 mmolO·kgsupport-1 were consumed during the

CH4 pulses, corresponding to 30% of the maximum oxygen storage capacity (for complete

reduction of Fe3+ to Fe2+).


183

Figure 6.3: Support oxygen mobility. (a): Catalyst bed configuration with mechanical mixture of

MgFe0.18Al1.82O4 and graphite. inset: graphite oxidation by lattice oxygen. (b): Catalyst bed

configuration during pulses in TAP reactor, (c): CO2 molar flow rate during CH4 pulses; inset: CO2

production as a function of pulse number.

In the as-prepared state, the support lattice contains randomly distributed iron as shown

by High Angle Annular Dark Field Scanning Transmission Electron Microscopy (HAADF-STEM)

combined with Energy-Dispersive X-ray spectroscopy (EDX) (Figure 6.4a-d). The structure of

MgFexAl2-xO4 was examined during reduction/oxidation by means of in-situ X-ray Absorption

Near Edge Structure (XANES) and Extended X-ray Absorption Fine Structure (EXAFS). XANES

spectra (Figure 6.4e) give insight in electronic changes around the absorber atom and its

local symmetry [44-46], while EXAFS allows for detailed identification of the neighbouring

shells. The XANES region of MgFexAl2-xO4 shows reduction through a shift of edge energy

a b

c


184

position towards lower values. Reduction to reference FeO (red line) and metallic Fe (black

line) is not observed implying that Fe is not segregated from the support lattice. The EXAFS

analysis of as-prepared MgFexAl2-xO4 identified an AB2O4 spinel structure with Mg2+ in the

tetrahedral A site along with Fe3+ and Al3+ in the octahedral B sites, showing Fe incorporation

into MgFexAl2-xO4 (Figure 6.4f).

Figure 6.4: Support characterization. a, HAADF-STEM image of as-prepared MgFe0.09Al1.81O4 along

with (b-d): EDX elemental maps. (e): XANES spectra of support MgFe0.09Al1.91O4 at the Fe-K edge

during H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10 K/min). (f): local environment of

randomly dispersed Fe inside the spinel framework. Brown corresponds to Mg, light blue to Al,

dark blue to oxygen and red to Fe atoms. (g): experimental and simulated XANES spectra for

MgFe0.09Al1.81O4 after H2-TPR with (h): contributions from two different Fe sites, non-distorted

(top), and reduced and distorted (bottom). ( ) simulated, ( ) experimental , ( )

distorted octahedron, ( ) non-distorted octahedron.

h

a

f

eb

c

g

d

Simulated Experimental


185

During temperature-programmed reduction (TPR) in hydrogen, Fe3+ partially transforms

to Fe2+, while remaining in the lattice (Figure 6.5). Reduction induces a distortion of the

octahedron around Fe2+, leading to a change in Fe-O bond distance (Table 6.2).

Figure 6.5: (a): FT of the fresh and reduced sample. The arrow highlights the shift in R occurring in

the reduced sample. The shift indicates the formation of Fe2+ in the support. (b): Left figure shows

the signals used in the fitting of the reduced sample. The γ and η signals represent contributions

from the two-bodies and the three-bodies configuration, respectively. The signals in blue

correspond to the distorted Fe local environment; the signals drawn in black are related to the

non-distorted environment. The EXAFS data, the calculated fit and the residual are reported below.

The right figure shows the EXAFS data of the fresh and reduced sample, with their respective fit.

The XANES spectrum after reduction (Figure 6.4g) was simulated using two octahedral Fe

sites: reduced Fe with strong distortions and adjusted Fe-O distance, and non-distorted and

a

bγ²1

γ²2

η³1

η³2

η³3

η³4

fit

residual

reduced

fresh

fresh

reduced


186

fully oxidized Fe, i.e. identical to the as-prepared sample. Approximately 60% ± 10% of Fe

appeared in a distorted environment (Table 6.2 and Figure 6.5). Re-oxidation using CO2

restores the as-prepared state of fully oxidized spinel. Adding Fe to MgAl2O4 thus makes a

uniquely stable (Figure 6.6) and active support, MgFexAl2-xO4, with 0.007 ≤ x ≤ 0.26, that can

also serve as catalyst with redox capacity.

Table 6.2: The results from the EXAFS model data of the as prepared, reduced and re-oxidized

sample 5.0Fe. N is the coordination number, σ² the Debye Waller factor, R the radial distance from

the central Fe absorber, θ the angle in the Fe-O-M three-body configuration. In reduced state, two

Fe contributions are considered: 1: non-distorted, 2: reduced/distorted octahedron. β represents

the corresponding fractions. Bold values were fitted, non-bold values were calculated.

As-prepared

Reduced Re-oxidized

Non-distorted Distorted

β (%) 100 40 60 100

N1 6.0 ± 0.5 2.8 ± 0.5 3.0 ± 0.5 5.9 ± 0.9

σ21 0.0037 ± 0.0007 0.0190 ± 0.0008 0.0190 ± 0.0008 0.0180 ± 0.0007

R1Fe-O (Å) 1.97 ± 0.02 1.92 ± 0.02 2.08 ±0.02 1.96 ± 0.02

N2 5.3 ± 0.5 2.4 ± 0.5 3.2 ± 0.5 5.1 ± 0.5

σ22 0.0031 ± 0.0007 0.0030 ± 0.001 0.0032 ± 0.0008 0.0032 ± 0.0008

R2Fe-O-Al (Å)

θ

2.95 ± 0.05

95ο ± 5

2.90 ± 0.05

98ο ± 5

2.94 ± 0.05

95ο ± 5

2.89 ± 0.05

95ο ± 5

N3 4.4 ± 0.5 2.2 ± 0.5 1.0 ± 0.5 4.0 ± 0.5

σ23 0.0051 ± 0.0007 0.0052 ± 0.0008 0.0052 ± 0.0008 0.0052 ± 0.0008

R3Fe-O-Mg (Å)

θ

3.41 ± 0.05

116ο ± 5

3.42 ± 0.05

118ο ± 5

3.50 ± 0.05

116ο ± 5

3.43 ± 0.05

116ο ± 5


187

Figure 6.6: CO yield (mol·kgsupport-1) of MgFe0.09Al1.91O4 support as a function of redox cycles at

750°C. The CO yield remained stable during 60 cycles illustrating the stability of the MgFexAl2-xO4

support under reducing/oxidizing environment. Each cycle lasted for 16 min and it was composed

of 4 min of reduction (5%H2/Ar), 4 min of inert gas (He), 4 min of oxidation (CO2) and 4 min of inert

gas (He). All the gas flow rates were 1.1 NmL/s.

6.3.2 Catalyst Characterization

Using these new supports, 8wt%Ni-based catalysts were synthesized by incipient

wetness impregnation. Table 6.3 displays their properties. The Ni catalysts have lower

surface area and pore volume than the corresponding supports mainly due to surface

covering and pore blockage, similar to what has been reported about the effect of different

additives on the support properties [13, 47, 48]. Shi and Zang [47] investigated the role of

MgO over Pd supported on γ-Al2O3, varying its loading from 1 to 10 wt%. Up to 1wt%, there

was no noticeable change in the textural properties of γ-Al2O3. By increasing the loading of

MgO from 3 to 10 wt% the surface area and the pore volume gradually decreased. In the

same line, a similar trend of surface area and pore volume decrease was also observed by

Damyanova and co-workers who added phosphorous on alumina support used for bimetallic

NiPd [48].


188

Table 6.3: Catalyst properties. Metal loading was determined by means of inductively coupled

plasma atomic emission spectroscopy (ICP-AES).

Abbreviation

“Ni/zFe” Catalyst

Metal loading (wt.%) BET (m2/gcat)

Pore volume

(cm³/gcat) Ni Fe

Ni/0Fe 8%Ni/MgAl2O4 8.51 - 62.8 ± 2.9 0.17 ± 0.01

Ni/0.5Fe 8%Ni/MgFe0.007Al1.993O4 9.10 0.48 55.0 ± 4.0 0.16 ± 0.05

Ni/1.0Fe 8%Ni/MgFe0.013Al1.987O4 9.08 0.92 63.0 ± 3.0 0.19 ± 0.04

Ni/2.5Fe 8%Ni/MgFe0.04Al1.96O4 8.30 2.52 54.4 ± 2.3 0.14 ± 0.01

Ni/5.0Fe 8%Ni/MgFe0.09Al1.81O4 8.29 5.13 45.1 ± 2.7 0.13 ± 0.01

Ni/7.5Fe 8%Ni/MgFe0.13Al1.87O4 8.05 7.95 41.1 ± 2.2 0.15 ± 0.01

Ni/10Fe 8%Ni/MgFe0.18Al1.82O4 7.99 10.94 35.4 ± 1.3 0.14 ± 0.00

Ni/15Fe 8%Ni/MgFe0.26Al1.74O4 9.25 17.40 21.8 ± 1.7 0.10 ± 0.01

Their crystallographic structure was characterized as-prepared, reduced and CO2-re-

oxidized, using XRD. Figure 6.7 displays the full scan XRD patterns of pure support, as-

prepared, reduced and reoxidized catalyst (Ni/7.5Fe is displayed for better signal quality).

MgFexAl2-xO4 spinel (31.3o,37o,45o, 55.5o, 59o, 65o, Powder Diffraction File (PDF) card

number: 01-071-1238) was the predominant phase for the as-prepared catalysts, while NiO

peaks (37.3o, 43.3o, 63o, PDF: 01-089-5881) disappeared after reduction. Reduction, as a pre-

treatment, is crucial for catalyst activation. The diffraction pattern of the reduced catalyst

showed peaks at lower angles than metallic Ni. This shift can be attributed to Ni-Fe alloy

formation as it was already observed in previous reports [10, 13, 15, 49], implying that

metallic Fe can migrate from the support towards Ni and form an alloy. Upon temperature

programmed CO2 re-oxidation of this reduced catalyst, the same crystalline phases were

observed in XRD as for the as-prepared sample (Figure 6.7). This was unexpected since

oxidation of Ni by CO2 at these conditions is not favourable, as established previously [13]

for Ni/MgAl2O4. CO2 decomposes the alloy and simultaneously oxidizes the support, which in

turn oxidizes Ni using lattice oxygen. Hence, the redox activity of MgFexAl2-xO4 enables

unsuspected Ni re-oxidation.


189

Figure 6.7: Full XRD scans of 7.5Fe support and fresh, reduced (10 NmL/s of 10%H2/He mixture at a

total pressure of 101.3 kPa and 1123K) and re-oxidized (10 NmL/s of CO2 mixture at a total

pressure of 101.3 kPa and 1123K) Ni/7.5Fe.

6.3.3 Temperature Programmed Reduction (TPR)

In-situ XRD Time-resolved measurements

The Ni-Fe alloy formation, via Fe migration from the support during H2-TPR was

investigated for all studied samples using in-situ XRD. Figure 6.8 displays the 2D in-situ XRD

pattern for the Ni/7.5Fe sample. A 2θ diffraction angles window was selected in order to

follow the evolution of MgFexAl2-xO4 support and NiO peaks. During reduction, NiO peak

disappears above 980 K, while the Ni-related diffraction peak shifts to a 2θ angle of 44o,

lower than metallic Ni, implying the formation of Ni-Fe alloy [10, 13, 17, 49]. The intensity of

characteristic peaks associated with MgFexAl2-xO4 decrease as a function of temperature but

remain present up to 1123 K, implying that Fe can be partially, but not completely,

segregated from MgFexAl2-xO4 lattice. This is in accordance with Dharanipragada and co-

workers [43] who studied the reduction of Mg-Al-Fe-O and concluded that the spinel is only

partially reduced.

25 30 35 40 45 50 55 60 65 70

Inte

nsi

ty (

a.u

.)

2θ ( )

7.5Fe

Ni/7.5Fe

after H2-TPR

Ni/7.5Fe

as-prepared

MgFeAlOx NiO − Ni-Fe

Ni/7.5Fe

after CO2-TPO


190

Figure 6.8: Time-resolved in-situ XRD patterns of Ni/7.5Fe during H2-TPR. Conditions: maximum

temperature 1123K , flow rate of 10 NmL/s, 5%H2/He, heating rate 20K/min.

The small shift of angular positions compared to full scan XRD patterns at room

temperature can be ascribed to temperature-induced lattice expansion and different sample

height.

In-situ XAS

To further understand the mechanism under which Ni enhances the reducibility of Fe

from the support lattice in order to form alloy along with Ni, during the reduction process,

an in-situ XAS experiment was performed on the Ni/5.0Fe. The XANES spectra were recorded

quasi-simultaneously at the Fe-K and Ni-K edges. XANES spectra at both Fe-K and Ni-K edges

(Figure 6.9) indicated that the addition of Ni onto MgFexAl2-xO4 causes deeper reduction of

Fe compared to the pure support (Figure 6.4e), in agreement with the in-situ XRD

experiments (Figure 6.7).

Figure 6.9: XANES evolution during H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10

K/min) of Ni/5.0Fe at the (B) Fe-K edge and (C) Ni-K edge.

MgFexAl2-xO4

NiO

Ni-Fe alloy

Ni-Fe alloy formation at 980 K

MgFexAl2-xO4

MgFexAl2-xO4

a b


191

Fe reduction occurs via hydrogen spillover [50, 51] from the Ni surface towards MgFexAl2-

xO4, as established in a reduction experiment (Figure 6.10) with a mechanical mixture of

Ni/MgAl2O4 (Ni/0Fe) and MgFe0.13Al1.87O4 (Ni/7.5Fe). The pattern of reduced Ni/7.5Fe was

added to Figure 6.10a for comparison purposes. The diffraction pattern of the reduced

mechanical mixture shows formation of metallic Ni along with metallic Fe. Metallic Fe

originates from hydrogen spillover, which was activated over metallic Ni particles.

Segregation of Fe during reduction of the pure 7.5Fe support was never observed previously

(Figure 6.4e).

Figure 6.10: (a): Full XRD scans of a mechanical mixture between 7.5Fe support and Ni/0Fe before

and after an isothermal in-situ treatment (named as “fresh” and “reduced”, respectively) under 10

NmL/s of 5%H2/He mixture at a total pressure of 101.3 kPa and 1073 K. b, HR-STEM of the

mechanical mixture after in-situ XRD along with c, EDX elemental map. Green color corresponds to

Ni and red to Fe. Yellow dots at the interface indicate initiated alloying between Ni and Fe.

a

b c


192

The deeper reduction of Fe in Ni/MgFexAl2-xO4 is crucial for Fe to become mobile. So,

hydrogen spillover induces deep reduction, steering the migration of Fe to Ni and formation

of a Ni-Fe alloy.

Reduction and alloying do not consume all Fe from the support (Figure 6.9a). Linear

Combination Fitting of the reduced Fe-K XANES spectrum indicated that approximately 50%

of Fe stayed incorporated in the support. This is beneficial as the support preserves oxygen

mobility, and therefore remains reducible. Only 59% of Ni reached the metallic state after

reduction (Figure 6.9b), while no NiAl2O4 spinel structure was detected, indicating a strong

metal-support interaction (SMSI) [52, 53]. SMSI comprises a number of effects occurring at

the interface between active metal and its support. One such effect is enhancing the metal

dispersion [54, 55], which is illustrated by the active phase crystallite size determined from

XRD (8.8 ± 0.6 nm for Ni/7.5Fe, compared to 10.5 ± 1.0 nm for Ni/0Fe). SMSI can also inhibit

the formation of carbon (Figure 6.13) [55, 56]. Further, SMSI in Ni/MgFexAl2-xO4 ensures that

Ni at the interface with the support remains oxidized, as thin NiOx layer, even under

reducing conditions [57], [58]. A schematic representation of Ni-Fe alloy formation upon Fe

migration from the support during H2-TPR is illustrated in Figure 6.11.

Figure 6.11: a,b: HAADF-STEM image of the mapped area of reduced Ni/1.0Fe together with the

EDX elemental map. Schematic representation of Ni-Fe alloy formation upon Fe migration from the

support during H2-TPR. Green represents NiO, grey Ni, black Ni-Fe alloy while the arrows indicate

the migration of Fe from the support towards Ni.

For comparison, a catalyst prepared by co-impregnation of Ni and Fe on MgAl2O4 (Ni-

4.0Fe/MgAl2O4), i.e. without incorporated Fe, was examined during reduction using in-situ

a b

c


193

XANES. The Fe-K edge XANES spectrum (Figure 6.12a) shows reduction towards metallic Fe,

deeper reduction compared to Ni/5.0Fe (Figure 6.8), which is also shown by the shift in the

pre-edge peak. Furthermore, deeper reduction of Ni was achieved, with 82% reaching the

metallic state (Figure 6.11), while the remainder formed a NiAl2O4 spinel phase (see Figure

3.1, § 3.1) [13]. SMSI have been previously reported for Ni impregnated on spinel supports

like Ca-MgAl2O4 [59].

Figure 6.12: XANES spectra of Ni-4.0Fe supported on MgAl2O4 at (a) Fe-K and (b) Ni-K edge during

H2-TPR (0.2 NmL/s of 5%H2/He using a heating ramp of 10 K/min).

As it was previously mentioned, SMSI can inhibit carbon formation. A schematic

representation of carbon formation over benchmark Ni catalyst supported on MgAl2O4 and

over a novel Ni catalyst supported on MgFexAl2-xO4 is illustrated in Figure 6.13. The process

includes hydrocarbon decomposition upon Ni sites during steady-state methane reforming

(Figure 6.13 a and c). In case of Ni/MgAl2O4, the deposited carbon atoms transport towards

the Ni-support interface. The accumulation of carbon extracts Ni particles from the support

causing the observed elongation and keeping the Ni sites active [56]. The mechanism of

carbon formation over the Ni/MgFexAl2-xO4 catalyst is now affected by the MgFexAl2-xO4

support. Carbon is formed on Ni particles, originating mainly from the methane

decomposition reaction. The FeOx islands, that are formed under reaction conditions,

actively participate in the reaction scheme by oxidizing the carbon deposited, through lattice

oxygen, producing CO. In addition, the support provides lattice oxygen to the carbon species

that have diffused towards the interface between support and active particle. Lattice oxygen

which is consumed, is then regenerated by gas phase CO2 producing CO. Direct interaction of

CO2 with carbon was not observed.


194

Figure 6.13: Schematic representation of carbon formation over a, b: benchmark Ni catalyst

supported on MgAl2O4 [56] and c, d: novel Ni catalyst supported on MgFexAl2-xO4. OL: lattice

oxygen.

6.3.4 Activity and stability tests

All catalysts were examined during methane dry reforming (DRM) in a fixed-bed reactor

at 1023 K, 111.3 kPa, molar ratio CH4:CO2 = 1 and at the same conversion (XCH4 = 70 ± 5%).

The CO and H2 space time yields (STYs) after a 2 h time-on-stream (TOS) are shown in Figure

6.14a. The best performing catalysts were further examined over a longer period under the

same conditions (Figure 6.14b) as well as at higher pressure (911.7 kPa, Figure 6.15) under

BIM. Ni/2.5Fe exhibited over 50% activity loss after 55 h TOS. Carbon formation was not

causing the observed deactivation, as none was detected. Similarly, the Ni/1.0Fe sample did

not show carbon deposition after 65 h TOS, whereas 0.4·102 molc·kgcat-1 had been deposited

on the Ni/0Fe sample after only 12 h TOS at 111.3 kPa (Figure 6.14c). Co-impregnation of

1.0wt% Fe with the same amount of Ni on a MgAl2O4 support also led to carbon formation,

with 0.4·102 molc·kgcat-1 deposited after 40 h TOS. Clearly, suppression of carbon formation

requires Fe to be part of the support lattice. As it was expected, by increasing the reaction

C

CH4

FeOx

MgFexAl2-xO4

C

CH4

Ni

MgAl2O4

C

CH4

Ni

MgAl2O4

a b

C

FeOx

CO

C C

CO COCC

MgFexAl2-xO4

OLOL

c d

NiO NiO

C

CC

C

CC


195

pressure more carbon is deposited. However, three times less carbon was deposited on

Ni/1.0Fe compared to Ni/0Fe (0.62·102 to 1.84·102 molc·kgcat-1, respectively). The carbon

formation can be further eliminated by increasing the partial pressure of the oxidizing gases,

CO2 or H2O. From the economic point of view, less CO2 or H2O will be required for Ni/1.0Fe

than for Ni/0Fe in order to reach the carbon-free region.

Figure 6.14: Activity and stability during methane reforming at 1023 K. a, STY for CO and H2 after 2

h TOS under DRM at 111.3 kPa for Ni/zFe (Table 6.3). b, CH4 consumption rate during DRM at 111.3

kPa: ■: Ni/2.5Fe, ♦: Ni/1.0Fe, x: Ni/0.5Fe at WNi:F0

CH4 = 0.024 kgNi∙s∙molCH4-1, x: Ni/0.5Fe at WNi:F

0CH4

= 0.011 kgNi∙s∙molCH4-1. Error bars: twice the standard deviation from three independent

experiments. c, Comparison of deposited carbon (molC·kgNi-1) during methane steam-dry reforming

(15%CH4, 11% CO2, 4% H2O and 70% Ar) at 111.3 and 911.7 kPa. ♦: Ni/1.0Fe, +: Ni/0Fe. Dashed lines

are guides to the eye. d, Deactivation due to Fe segregation from the Ni-Fe alloy for high (Fe:Ni >

1:3), low Fe content (Fe:Ni ≤ 1:9).

The catalyst activity and stability are sensitive to the Fe content of the support (Figure

6.14a-b). Under reforming conditions the Ni-Fe alloy rearranges (Figure 6.14d), according to

the Fe amount. This alloy restructuring process cannot be avoided and its rate is controlled

by the ratio between reducing and oxidizing gases [19, 60]. For high Fe content, this causes

deactivation as large patches of Fe cover Ni. Previously, noble metals were used to stabilize

a

d

b

c


196

the Ni-Fe alloy during reforming (see Chapter 5) [19]. However, we succeeded in obtaining

high stability without using noble metals. By including only a small amount of Fe inside the

magnesium aluminate support, the Ni-Fe alloy composition is such that it rearranges into an

active phase (Figure 6.14d) achieving high catalyst activity and stability, together with

efficient carbon removal. CH4 reforming on Ni-Fe/MgFexAl2-xO4 proceeds with CH4 reacting

on Ni sites to form H2 and surface carbon. CO2 can also be activated on Ni, but is more likely

to oxidize Fe to FeOx on the alloy surface patches, and Fe2+ to Fe3+ in the support. All surface

carbon is removed by interaction with lattice oxygen, producing CO. As such, oxidized iron

eliminates carbon accumulation, operating as scavenger.

Figure 6.15: CH4 consumption rate (mol·s-1·kgNi-1, closed symbols) and conversion (%, open

symbols) during steam-dry reforming, using a feed composition of 15% CH4, 11% CO2, 4% H2O and

70% Ar, at 1023 K and 911.7 kPa. ♦: 8wt%Ni/MgFe0.013Al1.987O4 at WNi:F0

CH4 = 0.00342 kgNi·s·mol-1CH4.

Looking to the future, iron and its oxides have just started to become a hot topic and

findings are already most rewarding [61-65]. To date noble metal catalysts still have a

tenfold higher activity (1wt%Rh/MgAl2O4 reached a CH4 consumption rate of 315 mol·s-

1·kgRh-1 at 1023 K and 111.3 kPa), but they are also a thousand times more expensive. We

found that a unique support containing a small amount of Fe creates a catalyst with

unmatched carbon resistance and stability in reforming of methane under various conditions

(Figures 6.14 and 6.15). Combined with good redox properties and thermal stability,


197

MgFexAl2-xO4 constitutes an outstanding support for Ni, providing catalysts that can truly

challenge noble metals.

6.4 Conclusions

The replacement of noble metals by viable substitutes in heterogeneous catalysis remains

a challenge. Ni-Fe catalyst have shown excellent resistance in carbon deposition but, on the

other hand, this property was not combined with high stability under reforming reactions. A

novel material was synthesized by incorporating Fe in the magnesium aluminate lattice

resulting in a MgFexAl2-xO4 spinel support. The incorporation of Fe occurs in the octahedral

sites of the spinel lattice, replacing Al and providing redox functionality in the support, while

at the same time preserves the high thermal stability that magnesium aluminate offers. The

addition of Ni on MgFexAl2-xO4 steers the formation of a surface Ni-Fe alloy via migration of

Fe due to hydrogen spillover, during the reduction process.

The activity and stability of the Ni/ MgFexAl2-xO4 were tested under various conditions and

strongly depend on the amount of incorporated Fe. By carefully down tuning the amount of

Fe in MgFexAl2-xO4, a surface active alloy Ni-Fe alloy is obtained (Fe:Ni molar ratio ~1:10),

which, in combination with MgFexAl2-xO4 support, fully mitigate carbon formation during

reforming reactions, offering excellent stability and carbon control. This Ni-Fe/MgFexAl2-xO4

makes a worthy substitute for noble metals, despite the tenfold higher activity that they

have, as they are also a thousand times more expensive.


198

6.5 References

[1] G.A. Olah, A. Goeppert, M. Czaun, T. Mathew, R.B. May, G.K.S. Prakash, Single Step Bi-reforming and Oxidative Bi-reforming of Methane (Natural Gas) with Steam and Carbon Dioxide to Metgas (CO-2H2) for Methanol Synthesis: Self-Sufficient Effective and Exclusive Oxygenation of Methane to Methanol with Oxygen, J. Am. Chem. Soc., 137 (2015) 8720-8729. [2] N. Kumar, M. Shojaee, J.J. Spivey, Catalytic bi-reforming of methane: from greenhouse gases to syngas, Curr. Opin. Chem. Eng., 9 (2015) 8-15. [3] P. Littlewood, X. Xie, M. Bernicke, A. Thomas, R. Schomäcker, Ni0.05Mn0.95O catalysts for the dry reforming of methane, Catal. Today, 242, Part A (2015) 111-118. [4] M.C.J. Bradford, M.A. Vannice, CO2 Reforming of CH4 over Supported Ru Catalysts, J. Catal., 183 (1999) 69-75. [5] D. Pakhare, V. Schwartz, V. Abdelsayed, D. Haynes, D. Shekhawat, J. Poston, J. Spivey, Kinetic and mechanistic study of dry (CO2) reforming of methane over Rh-substituted La2Zr2O7 pyrochlores, J. Catal., 316 (2014) 78-92. [6] E.N. Alvar, M. Rezaei, Mesoporous nanocrystalline MgAl2O4 spinel and its applications as support for Ni catalyst in dry reforming, Scr. Mater., 61 (2009) 212-215. [7] O. Muraza, A. Galadima, A review on coke management during dry reforming of methane, Int. J. Energy Res, 39 (2015) 1196-1216. [8] J. Ashok, S. Kawi, Steam reforming of toluene as a biomass tar model compound over CeO2 promoted Ni/CaO–Al2O3 catalytic systems, Int. J. Hydrogen Energy, 38 (2013) 13938-13949. [9] J. Ashok, M. Subrahmanyam, A. Venugopal, Hydrotalcite structure derived Ni–Cu–Al catalysts for the production of H2 by CH4 decomposition, Int. J. Hydrogen Energy, 33 (2008) 2704-2713. [10] J. Ashok, S. Kawi, Nickel–Iron Alloy Supported over Iron–Alumina Catalysts for Steam Reforming of Biomass Tar Model Compound, ACS Catal., 4 (2014) 289-301. [11] D. Li, Y. Nakagawa, K. Tomishige, Methane reforming to synthesis gas over Ni catalysts modified with noble metals, Appl. Catal. A 408 (2011) 1-24. [12] K. Nagaoka, A. Jentys, J.A. Lercher, Methane autothermal reforming with and without ethane over mono- and bimetal catalysts prepared from hydrotalcite precursors, J. Catal., 229 (2005) 185-196. [13] S.A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, Enhanced Carbon-Resistant Dry Reforming Fe-Ni Catalyst: Role of Fe, ACS Catal., (2015) 3028-3039. [14] L. Wang, D. Li, M. Koike, H. Watanabe, Y. Xu, Y. Nakagawa, K. Tomishige, Catalytic performance and characterization of Ni–Co catalysts for the steam reforming of biomass tar to synthesis gas, Fuel, 112 (2013) 654-661. [15] L. Wang, D. Li, M. Koike, S. Koso, Y. Nakagawa, Y. Xu, K. Tomishige, Catalytic performance and characterization of Ni-Fe catalysts for the steam reforming of tar from biomass pyrolysis to synthesis gas, Appl. Catal. A 392 (2011) 248-255. [16] Z. Alipour, M. Rezaei, F. Meshkani, Effect of alkaline earth promoters (MgO, CaO, and BaO) on the activity and coke formation of Ni catalysts supported on nanocrystalline Al2O3 in dry reforming of methane, J. Ind. Eng. Chem., 20 (2014) 2858-2863. [17] S.A. Theofanidis, R. Batchu, V.V. Galvita, H. Poelman, G.B. Marin, Carbon gasification from Fe–Ni catalysts after methane dry reforming, Appl. Catal., B, 185 (2016) 42-55. [18] S.M. Kim, P.M. Abdala, T. Margossian, D. Hosseini, L. Foppa, A. Armutlulu, W. van Beek, A. Comas-Vives, C. Copéret, C. Müller, Cooperativity and Dynamics Increase the Performance of NiFe Dry Reforming Catalysts, J. Am. Chem. Soc., (2017). [19] S.A. Theofanidis, V.V. Galvita, M. Sabbe, H. Poelman, C. Detavernier, G.B. Marin, Controlling the stability of a Fe–Ni reforming catalyst: Structural organization of the active components, Appl. Catal., B, 209 (2017) 405-416. [20] L. Xu, H. Song, L. Chou, Carbon dioxide reforming of methane over ordered mesoporous NiO–MgO–Al2O3 composite oxides, Appl. Catal., B, 108–109 (2011) 177-190.


199

[21] Z. Xu, Y. Li, J. Zhang, L. Chang, R. Zhou, Z. Duan, Bound-state Ni species — a superior form in Ni-based catalyst for CH4/CO2 reforming, Appl. Catal. A, 210 (2001) 45-53. [22] K.-M. Kang, H.-W. Kim, I.-W. Shim, H.-Y. Kwak, Catalytic test of supported Ni catalysts with core/shell structure for dry reforming of methane, Fuel Process. Technol., 92 (2011) 1236-1243. [23] H. Arandiyan, J. Li, L. Ma, S.M. Hashemnejad, M.Z. Mirzaei, J. Chen, H. Chang, C. Liu, C. Wang, L. Chen, Methane reforming to syngas over LaNixFe1−xO3 mixed-oxide perovskites in the presence of CO2 and O2, J. Ind. Eng. Chem., 18 (2012) 2103-2114. [24] S. He, H. Wu, W. Yu, L. Mo, H. Lou, X. Zheng, Combination of CO2 reforming and partial oxidation of methane to produce syngas over Ni/SiO2 and Ni–Al2O3/SiO2 catalysts with different precursors, Int. J. Hydrogen Energy, 34 (2009) 839-843. [25] Q.-H. Zhang, Y. Li, B.-Q. Xu, Reforming of methane and coalbed methane over nanocomposite Ni/ZrO2 catalyst, Catal. Today, 98 (2004) 601-605. [26] S. Sokolov, E.V. Kondratenko, M.-M. Pohl, U. Rodemerck, Effect of calcination conditions on time on-stream performance of Ni/La2O3–ZrO2 in low-temperature dry reforming of methane, Int. J. Hydrogen Energ., 38 (2013) 16121-16132. [27] X. Xie, T. Otremba, P. Littlewood, R. Schomäcker, A. Thomas, One-Pot Synthesis of Supported, Nanocrystalline Nickel Manganese Oxide for Dry Reforming of Methane, ACS Catal., 3 (2012) 224-229. [28] M.M. Nair, S. Kaliaguine, F. Kleitz, Nanocast LaNiO3 Perovskites as Precursors for the Preparation of Coke-Resistant Dry Reforming Catalysts, ACS Catal., 4 (2014) 3837-3846. [29] V.V. Gal'vita, V.D. Belyaev, V.N. Parmon, V.A. Sobyanin, Conversion of methane to synthesis gas over Pt electrode in a cell with solid oxide electrolyte, Catal. Lett., 39 (1996) 209-211. [30] J. Guo, H. Lou, H. Zhao, D. Chai, X. Zheng, Dry reforming of methane over nickel catalysts supported on magnesium aluminate spinels, Appl. Catal. A 273 (2004) 75-82. [31] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst characterization, Pure & Appl. Chem., (1995). [32] Scherrer, Nachr. Ges. Wiss. Göttingen, 1918. [33] Niemantsverdiet J.W., Spectroscopy in Catalysis, 2007. [34] V. Martis, A.M. Beale, D. Detollenaere, D. Banerjee, M. Moroni, F. Gosselin, W. Bras, A high-pressure and controlled-flow gas system for catalysis research, J. Synchrotron Radiat., 21 (2014) 462-463. [35] Y. Joly, X-ray absorption near-edge structure calculations beyond the muffin-tin approximation, Physical Review B, 63 (2001) 125120. [36] A. Filipponi, A. Di Cicco, C.R. Natoli, X-ray-absorption spectroscopy and n-body distribution functions in condensed matter. I. Theory, Phys. Rev. B, 52 (1995) 15122-15134. [37] R. Shannon, Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides, Acta Crystallogr., Sect. A, 32 (1976) 751-767. [38] V. Briois, C.L. Fontaine, S. Belin, L. Barthe, M. Th, V. Pinty, A. Carcy, R. Girardot, E. Fonda, ROCK: the new Quick-EXAFS beamline at SOLEIL, J. Phys.: Conf. Ser., 712 (2016) 012149. [39] J.T. Gleaves, G. Yablonsky, X. Zheng, R. Fushimi, P.L. Mills, Temporal analysis of products (TAP)—Recent advances in technology for kinetic analysis of multi-component catalysts, J. Mol. Catal. A: Chem., 315 (2010) 108-134. [40] J.J. Carberry, D. White, On the role of transport phenomena in catalytic reactor behavior, Ind. Eng. Chem., 61 (1969) 27-35. [41] G.F. Froment, K. Bischoff, J. De Wilde, Chemical Reactor Analysis and Design, 3rd ed., J. Wiley & Sons2011. [42] D.E. Mears, Diagnostic criteria for heat transport limitations in fixed bed reactors, J. Catal., 20 (1971) 127-131. [43] N.V.R.A. Dharanipragada, L.C. Buelens, H. Poelman, E. De Grave, V.V. Galvita, G.B. Marin, Mg-Fe-Al-O for advanced CO2 to CO conversion: carbon monoxide yield vs. oxygen storage capacity, J. Mater. Chem. A, 3 (2015) 16251-16262.


200

[44] M. Wilke, F. Farges, P.-E. Petit, G.E. Brown, F. Martin, Oxidation state and coordination of Fe in minerals: An Fe K-XANES spectroscopic study, Am. Mineral., 86 (2001) 714-730. [45] T.E. Westre, P. Kennepohl, J.G. DeWitt, B. Hedman, K.O. Hodgson, E.I. Solomon, A Multiplet Analysis of Fe K-Edge 1s → 3d Pre-Edge Features of Iron Complexes, J. Am. Chem. Soc., 119 (1997) 6297-6314. [46] I. Moog, C. Feral-Martin, M. Duttine, A. Wattiaux, C. Prestipino, S. Figueroa, J. Majimel, A. Demourgues, Local organization of Fe3+ into nano-CeO2 with controlled morphologies and its impact on reducibility properties, J. Mater. Chem. A, 2 (2014) 20402-20414. [47] C. Shi, P. Zhang, Role of MgO over γ-Al2O3-supported Pd catalysts for carbon dioxide reforming of methane, Appl. Catal., B, 170–171 (2015) 43-52. [48] S. Damyanova, B. Pawelec, K. Arishtirova, J.L.G. Fierro, Biogas reforming over bimetallic PdNi catalysts supported on phosphorus-modified alumina, International Journal of Hydrogen Energy, 36 (2011) 10635-10647. [49] V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Catalyst-assisted chemical looping for CO2 conversion to CO, Appl. Catal., B, 164 (2015) 184-191. [50] W. Karim, C. Spreafico, A. Kleibert, J. Gobrecht, J. VandeVondele, Y. Ekinci, J.A. van Bokhoven, Catalyst support effects on hydrogen spillover, Nature, 541 (2017) 68-71. [51] W.C. Conner, J.L. Falconer, Spillover in Heterogeneous Catalysis, Chem. Rev., 95 (1995) 759-788. [52] J.C. Matsubu, S. Zhang, L. DeRita, N.S. Marinkovic, J.G. Chen, G.W. Graham, X. Pan, P. Christopher, Adsorbate-mediated strong metal–support interactions in oxide-supported Rh catalysts, Nat Chem, 9 (2017) 120-127. [53] G.B. McVicker, J.J. Ziemiak, Chemisorption properties of platinum and iridium supported on TiO2

Al2O3 mixed-oxide carriers: Evidence for strong metal-support interaction formation, J. Catal., 95 (1985) 473-481. [54] G.J. Haddad, B. Chen, J.J.G. Goodwin, Characterization of La3+-Promoted Co/SiO2 catalysts, J. Catal., 160 (1996) 43-51. [55] X. Li, Y. Zhang, K.J. Smith, Metal–support interaction effects on the growth of filamentous carbon over Co/SiO2 catalysts, Appl. Catal., A, 264 (2004) 81-91. [56] S. Helveg, C. Lopez-Cartes, J. Sehested, P.L. Hansen, B.S. Clausen, J.R. Rostrup-Nielsen, F. Abild-Pedersen, J.K. Norskov, Atomic-scale imaging of carbon nanofibre growth, Nature, 427 (2004) 426-429. [57] N.J. Divins, I. Angurell, C. Escudero, V. Pérez-Dieste, J. Llorca, Influence of the support on surface rearrangements of bimetallic nanoparticles in real catalysts, Science, 346 (2014) 620-623. [58] X.Y. Shi, W. Zhang, C. Zhang, W.T. Zheng, H. Chen, J.G. Qi, Real-space observation of strong metal-support interaction: state-of-the-art and what's the next, J. Microsc., 262 (2016) 203-215. [59] K.Y. Koo, J.H. Lee, U.H. Jung, S.H. Kim, W.L. Yoon, Combined H2O and CO2 reforming of coke oven gas over Ca-promoted Ni/MgAl2O4 catalyst for direct reduced iron production, Fuel, 153 (2015) 303-309. [60] J. Hu, L. Buelens, S.-A. Theofanidis, V.V. Galvita, H. Poelman, G.B. Marin, CO2 conversion to CO by auto-thermal catalyst-assisted chemical looping, J. CO2 Util., 16 (2016) 8-16. [61] L.C. Buelens, V.V. Galvita, H. Poelman, C. Detavernier, G.B. Marin, Super-dry reforming of methane intensifies CO2 utilization via Le Chatelier’s principle, Science, 354 (2016) 449-452. [62] R.V. Jagadeesh, A.-E. Surkus, H. Junge, M.-M. Pohl, J. Radnik, J. Rabeah, H. Huan, V. Schünemann, A. Brückner, M. Beller, Nanoscale Fe2O3-Based Catalysts for Selective Hydrogenation of Nitroarenes to Anilines, Science, 342 (2013) 1073-1076. [63] E. Bykova, L. Dubrovinsky, N. Dubrovinskaia, M. Bykov, C. McCammon, S.V. Ovsyannikov, H.P. Liermann, I. Kupenko, A.I. Chumakov, R. Ruffer, M. Hanfland, V. Prakapenka, Structural complexity of simple Fe2O3 at high pressures and temperatures, Nat Commun, 7 (2016). [64] G. Chen, Y. Zhao, G. Fu, P.N. Duchesne, L. Gu, Y. Zheng, X. Weng, M. Chen, P. Zhang, C.-W. Pao, J.-F. Lee, N. Zheng, Interfacial Effects in Iron-Nickel Hydroxide–Platinum Nanoparticles Enhance Catalytic Oxidation, Science, 344 (2014) 495-499.


201

[65] M. Zhao, K. Yuan, Y. Wang, G. Li, J. Guo, L. Gu, W. Hu, H. Zhao, Z. Tang, Metal–organic frameworks as selectivity regulators for hydrogenation reactions, Nature, 539 (2016) 76-80.

Conclusions and perspectives

203

Chapter 7


In this work, different approaches were presented in order to control the deactivation of

Ni-based catalysts, and thus their stability, especially due to carbon formation, while

eliminating the use of noble metals in order to achieve this control. The efforts were focused

on catalyst optimization, containing different routes, from promoters addition to support

modifications. Fe, an economically viable and widely available element was chosen (1) as a

promoter for Ni-based catalysts and (2) to be incorporated in the magnesium aluminate

support lattice altering its properties. The location of Fe, either on the surface of the

MgAl2O4 support along with Ni, or incorporated in the support lattice, plays an important

role on catalyst stability and on mitigation of carbon deposition. An in-depth detailed

characterization and mechanistic insight study of methane reforming over Ni-Fe catalysts

was conducted in this work. The methodology consists of detailed in-situ and operando

material characterization in order to connect between physico-chemical properties and

catalyst performance.

Upon reduction, Ni and Fe interact forming an alloy which can be decomposed at

different temperatures depending on the applied oxidizing gas i.e. CO2, H2O, O2. The general

effect of Fe addition to a “state-of-the-art” Ni/MgAl2O4 catalyst, under methane reforming at

atmospheric pressure was found to depend on the employed Fe:Ni ratio. By increasing the

Fe:Ni ratio, the catalyst activity decreases, while pure Fe has negligible activity for methane

reforming at the investigated conditions. At the same time, the addition of Fe to Ni creates a

carbon-resistant catalyst. During reaction, Fe oxide species are formed increasing the

catalyst oxygen mobility, which helps to reduce carbon formation. Based on the

experimental observations the mechanism of methane reforming over Ni-Fe catalyst was

unraveled. It was found that the Mars-Van Krevelen mechanism (or redox mechanism)

prevails, where CO2 oxidizes Fe to FeOx, and CH4 is activated on Ni sites to form H2 and

surface carbon. The latter is re-oxidized by lattice oxygen from FeOx, producing CO.


204

Although iron-containing Ni catalysts restrain carbon accumulation, they still do not

provide the stability offered by noble metals: during methane reforming Fe segregates from

the alloy and blocks the Ni sites making them less accessible to reactants, leading to

deactivation. However, a viable catalyst should combine excellent stability and carbon

resistance using high abundance elements and eliminating the use of noble metals. In that

respect, optimization of the Ni-Fe catalyst was attempted by means of Pd addition at low

concentrations (≤ 2wt%). During reduction of a co-impregnated sample, a core shell alloy

forms at the surface, consisting of a Ni-Fe core and a Ni-Fe-Pd shell. A Ni:Pd molar ratio as

high as 75 was found to be the optimal as it increased both the activity and stability of the

catalyst. Further decrease of Ni:Pd ratio to 8 does not improve the activity of the catalyst,

while it increases 15 times the cost of it (Figure 7.1).

Figure 7.1: The effect of Ni:Pd ratio on the catalyst activity (CH4 consumption rate, mmol·s-1

·g-1

metals) and

price. Lines are guides to the eye.

A small Pd concentration stabilizes the Ni-Fe alloy by means of a thin tri-metallic Ni-Fe-Pd

surface layer. The origin of the improved catalytic activity derives from higher Ni dispersion

along with the creation of extra active sites, due to Pd addition. The latter acts as a barrier

for Fe segregation from the core during reforming reactions. Segregation of Fe from the

trimetallic shell will still occur but to a lesser extent as the Fe concentration is lower.


205

However, despite all the different ways to reduce carbon deposition, carbon

accumulation during reforming reactions remains an issue. Eventually, catalyst regeneration

is required, by removing all carbon species by gasification. An isothermal catalyst

“regeneration protocol” was developed for the Ni-Fe catalysts, when carbon deposition is

considered the main deactivation mechanism. The applied oxidizing gas plays an important

role on the restored activity. If air is applied directly on the spent catalyst, due to the

exothermicity of the reaction between carbon and O2, the nickel and iron oxide particles that

are formed, can migrate, collapse and then sinter. To avoid this phenomenon, CO2 or H2O

can be applied first, in order to remove the carbon species that are deposited close to the

active metals. The reaction of carbon with CO2 or H2O is endothermic and therefore the

migration of particles does not occur. Then, air can be used in order to oxidize the rest of

carbon deposits that are located far from the active metals.

The research of methane reforming to syngas has not yet reached its final destination.

Further insight on catalyst optimization can be acquired by introducing a new, low cost

support for reforming reactions. This unique support, that is presented in the framework of

this thesis, contains a small amount of Fe incorporated into the magnesium aluminate

lattice. A spinel structure AB2O4 with Mg2+ in the tetrahedral A site along with Fe3+ and Al3+ in

the octahedral B sites, was identified by detailed characterization techniques and modelling,

showing Fe incorporation into MgFexAl2-xO4, 0.007≤x≤0.26. The addition of Fe to magnesium

aluminate makes a uniquely stable and active support, with redox functionality based on

oxygen mobility. Combined with thermal stability, MgFexAl2-xO4 constitutes an outstanding

support for Ni, providing catalysts that can truly challenge noble metals, in terms of activity

and stability under various conditions of methane reforming.

As a follow up of this work, attention should be focused on the promising Ni catalyst

supported on MgFexAl2-xO4. The effect of temperature on Fe migration process, during

reduction, in order to form Ni-Fe alloy would provide useful information about the catalyst

dispersion and can eventually lead to the development of a pre-treating protocol. Also, the

reversibility of the catalyst structure under oxidizing agents like CO2, H2O and O2 would give

insights about the catalyst regeneration mechanism. In-situ XANES and EXAFS would be

useful tools to probe and identify the bulk structure of the catalyst, while XPS and EDX

mapping will help to differentiate between the bulk and surface iron. Finally the Ni


206

/MgFexAl2-xO4 catalyst can be examined during auto-thermal reforming conditions,

investigating the effect of oxygen concentration on its structure.

Further, the carbon-resistant Ni-Fe catalysts explored and developed in this work should

be tested using more complicated feedstocks. Given the transition of the research towards

renewable sources, biomass could be considered an interesting feedstock. Biomass is

heading for a great future due to its abundance and it can be converted by gasification,

resulting in a produced gas that contains significant amounts of light hydrocarbons i.e. CH4,

along with small quantities of higher hydrocarbons, tars and impurities, such as H2S, HCl, NH3

and alkali metals. From the aforementioned impurities, H2S is one of the strongest catalyst

deactivation agents as it chemisorbs on metal surfaces, with Ni being the most sensitive one.

Sulfur has a strong impact on the reforming activity of the catalysts since the deactivation is

severe even at small coverage, while at full coverage (θs), the catalyst is completely

deactivated. The rate of sulfur poisoning and hence sulfur resistance varies as a function of

catalyst composition and reaction conditions. Sulfur chemisorbs on Fe weaker than on Ni

and therefore their combination could result in a promising sulfur-resistant candidate

depending on the applied reaction conditions.

Appendix A

207

Appendix A

The step response set up consists of a fixed bed reactor and can be used for catalytic and

gas-solid reactions. The advantage of the setup is that it allows the users to choose between

a variety of feed lines consisting of various gases or liquids, that will be vaporized, and an

inert flow through the reactor. Thus, provides flexibility to perform reactions under various

feed specifications without changing the reactor conditions of temperature and pressure. In

addition switching between different feed mixtures is provided through a fast switching

valve. This enables the users to simulate a “step –response” or “alternate pulse” experiment

in the desired feed conditions, instead of a slow gradient in the feed composition and hence

it is called step response reactor. The reactor effluent can be analyzed using an online GC

and MS. Typically a catalyst/material mass of 0.01-1g can be loaded in the reactor.

The setup can be broadly, divided into three sections namely Feed, Reactor and Analysis.

The flow diagram is shown in Figure A.1. The various parameters in these sections such as

flow, temperature and pressure are PLC enabled

Appendix A

208

Figure A.1: The flow diagram of the step response setup with feed, reactor and analysis sections located at Laboratory for Chemical Technology (LCT),

Ghent University.

Appendix A

209

Figure A.2: The list of symbols used in the flow diagram in Figure A.1.

Appendix A

210

Feed section

The feed section (Figure A.2) can be categorized into gas and liquid vapor supply

sections, allowing to introduce various gas and liquid vapor feed streams into the

reactor. The inlet flow rates are always maintained by means of calibrated Brooks or

Bronkhorst mass flow controllers (MFCs).

Figure A.3: Overview of the feed selection section.

Gas feeding

This section enables the feeding of various gas mixtures including hydrocarbons (e.g.

CH4, C2H6, C3H8…..), oxidizing gases (e.g. CO2, O2) and other gases such as H2, He and Ar.

A summary of various possibilities of gas feeds are listed in Table B.1.

Liquid vapor feeding

The unit can dose liquid through a pressure vessel (Figure A.3) using a Coriolis MFC.

Further the liquid is vaporized in an evaporator (Max T ~ 180°C) downstream of the

Coriolis MFC. In addition, a saturator can be used when an additional vapor feed is

Appendix A

211

required, an inert gas (He or Ar) is used for that purpose. All the liquid feeding lines

towards the reactor inlet are heated. In the present work the vapor feed has been dosed

though the Coriolis MFC.

Table A.1: The overview of the mass flow controllers with their feed specifications.

S. No Mass flow controller Gas stream Flow range

(Nl/hr)

1 MFC-1 (Bronkhorst) Helium (He) / Argon (Ar) 0-60

2 MFC-2 (Bronkhorst) Hydrogen (H2) / Hydrocarbon 0-20

3 MFC-3 (Bronkhorst) Oxygen (O2) 0-20

4 MFC-4 (Brooks-Coriolis) Liquid feed 0-50*

5 MFC-5 (Bronkhorst) Helium (He) / Argon (Ar) 0-60

6 MFC-6 (Brooks) Helium (He) / Argon (Ar) 0-160

7 MFC-7 (Bronkhorst) CH4 (Internal standard) 0-12

8 MFC-8 (Brooks) Hydrocarbon 2 0-70

9 MFC-9 (Brooks) Diborane (B2H6) 0-20

10 MFC-10 (Brooks) Carbon dioxide (CO2) 0-60

*Flow rate units in g/h

Reactor section

Selection of reactor

Each reactor is approximately 470 mm long with an internal diameter of ~9 mm.

Depending on the reaction to be performed the user can choose between a quartz reactor or

a stainless steel reactor. The experiments mentioned in this thesis have been carried out in a

quartz and a coated stainless steel reactor for atmospheric and higher pressure experiments,

respectively. The bed temperature inside the reactor is measured by a K-type thermocouple

which can be placed inside the reactor (Figure A.4). The material inside the reactor can be

heated to the reaction temperature using three heating zones. The feed outlet of the reactor

is connected to a sampling port towards online analysis in the MS.

Appendix A

212

Figure A.4: The overview of the reactor with the catalyst bed.

Appendix A

213

Pressure

The pressure of the reactor is maintained by two back pressure regulators (BPRs, Figure A.5),

one for high pressure (max limit: 35 bar) and one for low pressure (max limit: 2 bar). For

quartz reactors a relief valve is provided to prevent overshoot of pressure with a limit of 2

bar.

Figure A.5: The back pressure regulators.

Analytical section

The reactor effluents were analyzed using an OmnistarTM (Pfeiffer Vacuum) mass

spectrometer (QMS 200) with capillary sampling.

Low pressure

High pressure

Appendix B

215

Appendix B

In this Appendix, the computational method used in Chapter 5, is discussed.

Appendix B

216

Computational method

All calculations have been carried out using the Perdew-Burke-Ernzerhof (PBE) GGA

functional [1], with the Vienna Ab initio Simulation Package (VASP 5.3) [2-5]. A plane-wave

energy cutoff of 400 eV was used. The projector augmented wave (PAW) method [6, 7]

describes the wave function close to the nuclei. Due to the presence of Fe and O atoms, all

calculations allow for spin polarization. A 551 Monkhorst-Pack [8] grid was used for

Brillouin-zone integration for the slab calculations. The Fermi level was smoothed using

Methfessel-Paxton 1st order electron smearing[9] with smearing width σ of 0.2 eV to

accelerate convergence. The electronic energy convergence criterion is 10-6 eV. This

approach proved appropriate to calculate segregation energies for four-layered fcc metal

slabs with unit cells of the studied size, i.e. containing 4 surface atoms [10]. The unit cell is

repeated in all directions to simulate the alloy surface. The top two layers were allowed to

fully relax in all calculations, while the two bottom layers were frozen at the optimized bulk

geometry. The vacuum layer between the slabs amounts to 10.6 Å. An intermediate artificial

dipole layer between the slabs is used to correct for the dipole interaction between them.

Convergence of the geometry optimization is assumed when all forces are below 0.05 eV/Å.

The bulk geometry, which is imposed on the two bottom layers, is obtained by

minimizing the electronic energy of the bulk fcc unit cell as a function of the cell volume,

using the same computational settings, but a finer 15x15x15 Monckhorst-Pack grid for the

integration of the Brillouin zone and an increased energy cutoff of 520 eV. In the Ni3Fe bulk

unit cell, the 3 Ni atoms occupy the face-centered positions of the bulk unit cell, while the Fe

atom is located at the corner of the cell. For the Ni2FePd bulk unit cell, one of the Ni atoms in

this structure is replaced by a Pd atom. The optimized bulk lattice parameters pertain to

354.3 pm for Ni3Fe and 367.2 pm for Ni2FePd.

Appendix B

217

Figure B.1: Ni3Fe (111) unit cell, without segregation (left) and with a Fe atom from the second

layer segregated to the surface layer in exchange for a Ni atom. Perspective view (top) and top

view (bottom) (Ni: green, Fe: red).

Appendix B

218

Figure B.2: Ni2FePd (111) unit cell, without segregation (left) and with a Fe atom from the second

layer segregated to the surface layer in exchange for a Ni atom (middle) or a Pd atom (right).

Perspective view (top) and top view (bottom) (Ni: green, Fe: red, Pd: blue).

Figure B.3: Oxygen monolayer coverage, adsorbed at the most stable fcc site for all studied

surfaces: Ni3Fe (top) and Ni2FePd (bottom). The same adsorption sites hold for a monolayer

coverage of H atoms and CO molecules (Ni: green, Fe: bright red, Pd: blue, O: small dark red atom).

Appendix B

219

Figure B.4: fcc adsorption sites with numbering, on all considered structures, Ni3Fe (top) and

Ni2FePd (bottom), non-segregated structure (left) and segregated structures in which an Fe is

exchanged with a surface Ni atom (middle) or a surface Pd atom (right) (Ni: green, Fe: bright red,

Pd: blue).

Table B.1: Location of the adsorbates for the mixed adsorbate layers on the fcc positions,

numbered according to Figure B.4.

fcc adsorption sites

Overlayer 1 2 3 4

50 % CO, 50%O a CO O O CO

50% CO, 25% O, 25% H CO CO O H

25% of CH, CO, O, H CO CH O H aOnly the Ni2FePd structure in which Fe is exchanged for Pd has a different

overlayer O-O-CO-CO for positions 1-2-3-4, because the Fe lines in the surface layer are oriented differently than in the other segregated structures.

Appendix B

220

References

[1] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. Lett., 77 (1996) 3865. [2] G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Comput. Mat. Sci., 6 (1996) 15. [3] G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B, 54 (1996) 11169. [4] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals, Phys. Rev. B, 47 (1993) 558. [5] G. Kresse, J. Hafner, Ab initio molecular-dynamics simulation of the liquid-metal-amorphous-semiconductor transition in germanium, Phys. Rev. B, 49 (1994) 14251. [6] P.E. Blöchl, Projector augmented-wave method, Phys. Rev. B, 50 (1994) 17953. [7] G. Kresse, D. Joubert, From ultrasoft pseudopotentials to the projector augmented-wave method, Phys. Rev. B, 59 (1999) 1758. [8] H.J. Monkhorst, J.D. Pack, Special Points for Brillouin-Zone Integrations, Phys. Rev. B: Condens. Matter, 13 (1976) 5188-5192. [9] M. Methfessel, A.T. Paxton, High-Precision Sampling for Brillouin-Zone Integration in Metals, Phys. Rev. B: Condens. Matter, 40 (1989) 3616-3621. [10] M.K. Sabbe, L. Lain, M.F. Reyniers, G.B. Marin, Benzene adsorption on binary Pt3M alloys and surface alloys: a DFT study, PCCP, 15 (2013) 12197-12214.

Appendix C

221

Appendix C

In this Appendix, an overview of the publications originated from this Thesis, is discussed.

Thesis publications

Chapter 3 ACS Catal., (2015) 3028-3039

Chapter 4 Appl. Catal., B, 185 (2016) 42-55

Chapter 5 Appl. Catal., B 2017, 209, 405-416

Other publications

J. CO2 Util., 16 (2016) 8-16

Oral presentations

Methane dry reforming of Ni-Fe catalysts:

effect of Fe content

IAP Annual Meeting 2014, Louvain-La-

Neuve, Belgium, September 19, 2014

Novel Fe-promoted MgAl2O4 support material

for control of carbon deposition during

reforming reactions

American Institute of Chemical Engineers

Annual Meeting (AIChE 2016), San

Francisco

Ni supported on MgFexAl2-xO4 as carbon

resistant catalyst for methane reforming:

structural and mechanistic study

North American Catalysis Society Meeting

(NAM25), 2017, Denver, Colorado

Poster presentations

Enhanced catalytic performance of Ni-Fe alloy

in methane dry reforming: role of Fe

Joint workshop of ENMIX, FASTCARD,

CASCATBEL, BIOGO, Stuttgart, Germany,

January 20-21, 2016

Appendix C

222

The mechanism of carbon gasification over a

Fe-Ni catalyst after methane dry reforming

International Congress on Catalysis (ICC

16), Beijing, China, July 3-8, 2016

Reforming Processes for Sustainable Syngas Production: · PDF file1.1 Catalysts ... B)...

Documents

Transcript of Reforming Processes for Sustainable Syngas Production: · PDF file1.1 Catalysts ... B)...