7/30/2019 1-s2.0-S0021951713000614-main

1/6

Structureactivity relationships for propane oxidative dehydrogenation by

anatase-supported vanadium oxide monomers and dimers

Lei Cheng a, Glen Allen Ferguson a, Stan A. Zygmunt c, Larry A. Curtiss a,b,

a Materials Science Division, Argonne National Laboratory, Argonne, IL 60439, United Statesb Center for Nanoscale Materials, Argonne National Laboratory, Argonne, IL 60439, United Statesc Department of Physics and Astronomy, Valparaiso University, Valparaiso, IN 46383, United States

a r t i c l e i n f o

Article history:

Received 5 October 2012Revised 6 December 2012Accepted 13 February 2013Available online 1 April 2013

Keywords:

Propane oxidative dehydrogenationSupported vanadium oxideDFTCAH activation

a b s t r a c t

To understand the importance of the effect of molecular structure on reactivity, we have studied theactivity of anatase TiO2 (001) supported VOx catalytic sites for propane oxidative dehydrogenation(ODH). First, possible structures of monomeric and dimeric VOx species on anatase (001) after VO4H3grafting and water elimination were determined. We then studied the conversion reaction of propaneto propanol by the supported VOx to elucidate the structurereactivity relationship. The coordinationnumber of the vanadium atom was the key structural parameter in predicting the catalytic activity. Thiskey structural difference alone resulted in an increase of up to 800 times in the reaction rate of CAH bondactivation (rate-determining for propane ODH) for the various vanadium oxide species at 600 K. Theseresults demonstrate the remarkable sensitivity of the catalytic site activity to its geometric structureand its implications for achieving optimal catalyst performance.

2013 Elsevier Inc. All rights reserved.

1. Introduction

Supported vanadium oxide catalysts have been used industri-ally for many years to promote selective oxidation reactions [1]and have recently attracted special attention due to their excep-tional activity and selectivity for the oxidative dehydrogenation(ODH) of light alkanes to alkenes. Dispersed vanadium oxide(VOx) can be prepared by anchoring VOx species atop a metal oxidesupport such as anatase (TiO2) at low vanadium oxide loading, andthese dispersed catalysts exhibit higher alkene selectivity thancrystalline V2O5 for ethane and propane ODH. However, due tothe difficultyof characterizing supported VOx structures in the sub-monolayer regime, the elucidation of the catalyst structureactiv-ity relationship, a long-standing goal of catalysis science, has notyet been achieved.

Many early attempts to characterize the surface structure ofsupported VOx species using techniques designed for bulk systemssuch as EXAFS, XANES, and 51V MAS-NMR led to conflicting results[27]. In contrast, Raman spectroscopy provides outstanding struc-tural information for metal oxide catalytic sites at a molecular level[816]. With decades of experimental studies [216] and more re-cently computational efforts [13,1720], it is generally acceptedthat a tetrahedral structure with one oxo group and three oxygen

atoms bridging to the support is the most likely structure for theVOx surface site. However, the catalytic surface is inhomogeneous.This distribution of surface structures results from variations incatalyst preparation conditions that produce different thermody-namically stable surface species. Since the species are stable byeither having a similar thermodynamic stability or by being kinet-ically trapped, they can coexist on the substrate. In a recent study[13], we reported three distinct vanadium oxide structures on theh-alumina surface using a deep-UV Raman spectroscopy supportedby DFT calculations. These structures result from different degreesof dehydration during the preparation process. Similar phenomenahave been observed on silica-supported vanadium oxide [8]. Moreimportantly, these different species were found to have differencesin reducibility [21]. Therefore, when we evaluate the catalyticactivities of these materials, it is important to distinguish differentreactivities of each individual structure and identify the most ac-tive species. This information is critical for the rational designand synthesis of catalysts with high activity and selectivity.

The CAH bond activation through hydrogen abstraction by theoxo group has been identified as the key step for alkane ODH to al-kene on V2O5 and supported vanadium oxide in several mechanis-tic studies [2226]. Cheng et al. examined propane ODH with aV4O10 cluster and found that the rate-determining step is hydrogenabstraction by the vanadium oxo group forming an isopropyl rad-ical [23]. A similar conclusion was reached by Rozanska et al. forpropane ODH on SiO2-supported vanadium oxide [22]. These stud-ies provided valuable information on the reaction mechanism, but

0021-9517/$ - see front matter 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.jcat.2013.02.012

Corresponding author at: Materials Science Division and Center for NanoscaleMaterials, Argonne National Laboratory, 9700 Cass Ave., Argonne, IL 60439, UnitedStates.

E-mail address: curtiss@anl.gov (L.A. Curtiss).

Journal of Catalysis 302 (2013) 3136

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j c a t

http://dx.doi.org/10.1016/j.jcat.2013.02.012mailto:curtiss@anl.govhttp://dx.doi.org/10.1016/j.jcat.2013.02.012http://www.sciencedirect.com/science/journal/00219517http://www.elsevier.com/locate/jcathttp://www.elsevier.com/locate/jcathttp://www.sciencedirect.com/science/journal/00219517http://dx.doi.org/10.1016/j.jcat.2013.02.012mailto:curtiss@anl.govhttp://dx.doi.org/10.1016/j.jcat.2013.02.012

7/30/2019 1-s2.0-S0021951713000614-main

2/6

did not consider that some hydroxyl groups may remain on thesupport surface or on the vanadium oxide [8,13] and play animportant role in changing the reactivity of the catalyst.

In this work, we present the activities of various possible mono-meric and dimeric VOx structures supported on an anatase (001)surface. These structures were constructed by grafting a VO4H3precursor molecule(s) on the hydroxylated anatase surface andeliminating one or more water molecules to find the possible sta-ble structures on the surface. The reaction energies and barriersfrom propane up to propanol formation on selected monomerand dimer structures were calculated to show the structurereactivity relationship of the catalyst toward propane ODH. Theexact minimum energy crossing point (MECP) between two elec-tronic states on the reaction path has also been located and, toour knowledge, is reported here for the first time. More impor-tantly, the calculations of reaction energies and barriers usingmore realistic monomer and dimer models provide insight intothe effect of the structure on catalyst reactivity.

2. Theoretical models and methods

2.1. Models

2.1.1. Gas-phase vanadium oxide models

The structure of the smallest models of active sites, VO4H3 andVO5H5, is shown in Fig. 1. VO4H3 and VO5H5 represent gas-phase

vanadia active sites with vanadiumin tetrahedral and square pyra-midal coordination spheres, respectively. In structure calculationsinvolving these two clusters, all atoms were allowed to relax. Wealso used a cluster, namely VO5H5_bulk, as the smallest model torepresent the crystalline V2O5 (01 0) surface. The structure ofVO5H5_bulk is very similar to VO5H5, except that its oxygen atomswere frozen at experimentally derived oxygen positions for V2O5bulk in all calculations, and all OAH bond lengths in the terminat-ing hydroxyl and aqua groups were also fixed at 0.96 , in accor-dance with our previous work [27]. All other atoms were allowedto relax in calculations with VO5H5_bulk. Key structural parame-ters of these three clusters are also shown in Fig. 1 to illustratethe differences.

2.1.2. Anatase-supported vanadium oxide models

Experimentally, vanadium oxide species are anchored atop sup-port surfaces using a variety of preparation methods[1]. When thisprocess is carried out in an aqueous environment, we expect theanatase surface to be hydroxylated. Computational studies haveshown [17,28], and our own work confirms, that water readily dis-sociates upon adsorption on the (00 1) anatase surface by breaking

one TiAO bond and forming two TiAOH bonds with enhanced sta-bility due to hydrogen bonding. A previous experimental study [29]also indicated more hydroxyl groups on the anatase support led tothe formation of more segregated VOx clusters instead of polymericchains. We used a hydrogen-terminated cluster H20O18Ti4, shown

VO4H3 VO5H5

(1.91) (1.91)

(1.75)(2.10)

1.57

1.82 1.82

1.782.27

1.57

1.77

1.771.77

1.56

VO5H5_bulk

Fig. 1. Structures of gas-phase vanadia species VO4H3, VO5H5 and VO5H5_bulk. Bond lengths in figure are in and numbers shown in parenthesis are fixed duringcalculations.

M-monodentateM-bidentate

M-bidentate_2

M-tridentate

M-molecular

M-dioxo

+H2O

H20O18Ti4

VO4H3

+

+H2O +2H2O

+3H2O

+3H2O

+3H2O

Fig. 2. Illustration of the grafting of a VO4H3 precursor on hydroxylated anatase (001) forming various supported structures.

32 L. Cheng et al. / Journal of Catalysis 302 (2013) 3136

7/30/2019 1-s2.0-S0021951713000614-main

3/6

in Fig. 2, to represent the hydroxylated (001) surface of anatase.Either during the VO4H3 precursorgrafting process or during subse-quent calcination of the catalyst, the hydroxyl groups on the sup-port are likely to condense as water and be driven off. It ispossible that some surface hydroxyl groups remain under reactionconditions. To explore this possibility, VOx/TiO2 structures wereconsidered with and without hydroxyl groups. Based on the workof Vittadini andSelloni [17], the grafting of VO

xunits on the surface

was envisioned as proceeding by the interaction of VO4H3 molecu-lar precursors with the hydroxyl groups on the TiO2 surface. Thisstep was followed by a series of water eliminations illustrated inFig. 2. In calculations of the supported clusters, constrained geom-etry optimizations were performed in which the terminal hydroxyland aqua groups were fixed. The terminal oxygen atoms were fro-zen at experimentally observed positions [30], and the H atomswere frozen at an hydroxyl distance of 0.96 along the directionof the titanium atom in the experimental structure.

2.2. Theoretical methods

Structures reported in this paper were optimized using the

B3LYP [31] hybrid density functional method with the 6-31G(d)basis set. For small cluster models VO4H3 and VO5H5, single pointcalculations using CCSD(T) [3235] and 6-31G(d) basis set combi-nations were also performed in order to assess and correct theB3LYP/6-31G(d) results. The GAUSSIAN 09 suite of programs [36]was used for these calculations.

We employed a fragment guess technique to find transitionstates on the open-shell singlet potential energy surface. The reac-tion path involves crossing of two states of different spin multiplic-ities. We used the method of Bearpark et al. [37,38] to locate theminimum energy crossing points (MECPs).

3. Results and discussion

3.1. Equilibrium structures and relative energetics of supportedmonomers and dimers

The first step in modeling propane ODH by supported VOx cat-alysts was to determine the geometries and relative energies ofmonomeric (M) and dimeric (D) VOx structures on the TiO2support.

The structures of six monomeric species (M-molecular,M-monodentate, M-bidentate, M-bidentate_2, M-dioxo and M-tri-dentate) and two dimeric species (D-molecular and D-bidentate)that can be formed during grafting of VO4H3 are shown in Figs. 2and 3, respectively. The electronic energies and Gibbs free energiesrelative to gas-phase VO4H3 and hydroxylated anatase (H20O18Ti4)at a water partial pressure of 0.001 atm (typical catalytic condi-

tion) and temperatures of 298, 600 and 800 K, respectively, are re-ported in Table 1. These energies are also plotted in a graph shownin Fig. S2. The relative stabilities of the species are temperaturedependent. As the temperature increases, the formation of M-molecular and M-mono species becomes less favorable (less exo-thermic), while the formation of other species becomes morefavorable (more exothermic or less endothermic). At 600 K, the for-mations of all structures are thermodynamically favorable (exo-thermic reaction) except for the M-dioxo species. Therefore, theM-dioxo structure is not likely to be formed in significant quanti-ties on the support. The energies reported in Table 1 are based so-lely on calculations of the stoichiometric formation reaction. Therelative population of these species can also be adjusted by otherfactors such as nature of precursors or water concentration.

The vanadium atoms in the M-monodentate, M-bidentate,M-bidentate_2, and D-bidentate structures have tetrahedral

configurations with an oxo group and are also bound to three otheroxygen atoms, in agreement with most supported V2O5 clusterresults in the literature [8,13,1720]. The coordination of the vana-

dium atoms in the M-molecular, M-tridentate, and D-molecular issomewhat unique: apart from the oxo group, vanadium is alsocoordinated with four, rather than three, other oxygen atoms,resulting in a distorted square pyramidal geometry. A square pyra-midal geometry is not uncommon for vanadium. In coordinationcomplexes, the square pyramidal structure is especially preferredwhen the metal has one ligand with a multiple bond, like inVV and VIV coordination complexes such as VOCl3acetonitrile ad-duct and VO(acac)2 (vanadyl acetylacetonate). The square pyrami-dal geometry was also observed recently in ab initio moleculardynamics simulations of anatase (001)-supported vanadia underhydration conditions [28] and was also found to be the most stablevanadium oxide monomeric configuration on anatase (001) sup-port in a periodic DFT calculation [39]. The V@O bond lengths in

all structures reported in Figs. 2 and 3 are very similar, being inthe range of 1.561.59 .

3.2. Reaction energies of propane to propanol by ODH on selected

monomeric and dimeric structures

Propanol formation from propane by supported vanadia in-volves two steps: propane CAH bond activation (or hydrogenabstraction) by the vanadyl group and OAC bond formation, asillustrated in Fig. 4. The relative reaction energies and barriersfor the two steps are reported in Table 2. The first hydrogenabstraction by a vanadyl group starts with a singlet ground stateI: C3H8 + O@V

5+ and ends with a diradical intermediate II:C3H

7AHOAV4. For all monomer structures reported in Table 2,

the diradical intermediates have two nearly degenerate electronicstates: the triplet (T) and the open-shell singlet (OS). For theM-molecular, M-tridentate, and M-bidentate clusters, the OS stateof II is slightly lower in energy than the T, while for the M-biden-tate_2, the OS state is slightly less stable than the T. The potentialenergy surfaces of the two states are highly parallel at this point ofthe reaction coordinate, and this is the point of the electronic curvecrossing from the initial singlet ground state potential energy sur-face (PES) to the triplet PES. After the hydrogen abstraction, propa-nol C3H7OHAV

3+ formation proceeds on the triplet PES because thetriplet propanol has significantly lower energy than the closed-shell singlet for all the calculated structures, consistent with othertheoretical work [22,23]. Fig. 4a illustrates the reaction energy pro-file of propane conversion to propanol on the M-molecular cluster,

with all transition states and intermediates, as well as the MECPs.The crossing prior to diradical (II) formation, noted as MECP1 in

D-molecular D-bidentate

H20O18Ti4

2VO4H3

+

+3H2O +5H2O

Fig. 3. Grafting of two VO4H3 precursors to form VOx dimers on anatase (0 01).

L. Cheng et al. / Journal of Catalysis 302 (2013) 3136 33

http://-/?-http://-/?-

7/30/2019 1-s2.0-S0021951713000614-main

4/6

Fig. 4, occurs at a much higher energy than both the TS1OS and theMECP2. At the crossing point geometry MECP1, the OS state energyis 55.3 kcal/mol, lower than the singlet/triplet energy at the same

crossing point geometry. This confirms that the open-shell singletpotential curve is lower in energy than the triplet and closed-shellsinglet at this point on the reaction coordinate, and the reactionproceeds through the TS1OS state instead of crossing to the tripletPES, which occurs later in the reaction coordinate. Mulliken spindensity analysis shows that in all the calculated TS1OS, II, and TS2structures, spin is localized on the C and V atoms; while for the

intermediate III structures, the spin is localized on the vanadium.The reduction in the support, a phenomenon that has been re-ported for reducible CeO2 support [40], was not observed on TiO2in the current study.

An alternative pathway for hydrogen abstraction by a VAOATibridging site is less favorable than the abstraction by the vanadylgroup for the systems in this study. One example is the reactionat the bridging oxygen of M-bidentate_2 cluster with activationbarrier and reaction energy of 55.2 and 47.5 kcal/mol, respectively.These energies are much higher than the vanadyl pathway with anactivation energy of 45.6 (TS1os) and a reaction energy of 42.2 kcal/mol (II).

The reaction mechanism on the two dimeric clusters is the sameas on the monomeric structures reported above. The reaction ener-gies and barriers of the two dimeric clusters are also both reportedin Table 2, and the reaction energy profile on the D-bidentate clus-ter is illustrated in Fig. 4b. In all the calculated TS1OS, II, and TS2structures, the spin is localized at the C and V atoms, while inthe triplet intermediate III structures, the spin is localized on thereacting vanadium atom of the D-bidentate cluster. On theD-molecular structure, the spin is distributed between the twovanadium atoms. Comparing reaction energies and barriers onthe two dimeric structures with the monomeric structures, we no-tice that these values of the D-bidentate are very similar to those ofthe monomeric structures M-bidentate and M-bidentate_2. ForD-molecular, the TS1os energy is similar to that of the M-molecular(37.1 vs. 39.8 kcal/mol), while the II, TS2, and III energies arenoticeably lower than those of the M-molecular.

In order to evaluate the accuracy of these B3LYP/6-31G(d) re-

sults, we calculated the propanol formation energies and hydrogenabstraction barriers on VO5H5 and VO4H3 clusters using accuratewavefunction-based model chemistries. As presented in the Sup-

Table 1

Electronic energies (Ee) and Gibbs free energies (G) in kcal/mol of monomeric structures in Fig. 2 relative to gas-phase VO4H3 and hydroxylated anatase

(H20O18Ti4) and dimeric structures in Fig. 3 relative to two gas-phase VO4H3 and hydroxylated anatase (H20O18Ti4). Energies in parenthesis are zero point

corrected.

Ee G (298.15K, 0.001 atm) G (600 K, 0.001 atm) G (800 K, 0.001 atm)

M-molecular + H2O 19.1 (18.7) 13.8 9.4 6.4M-mono + H2O 22.3 (21.8) 17.8 14.1 11.6M-bidentate + 2H2O 11.8 (13.5) 22.8 33.4 40.0

M-dioxo + 3H2O 61.87 (57.5) 32.8 5.1 12.7M-bidentate_2 + 3H2O 43.0 (38.6) 15.0 11.0 27.6M-tridentate + 3H2O 15.1 (11.7) 10.9 35.7 51.5D-molecular + 3H2O 20.6 (22.1) 55.8 30.8 33.9D-bidentate + 5H2O 3.6 (9.7) 12.4 79.6 103.3

Table 2

Relative energies (in kcal/mol) of intermediates and transition states on reaction pathway of propanol formation on monomeric and dimeric supported VO x structures. Numbers

in parenthesis include the zero point energy (ZPE) correction. Electronic configuration of each structure is also indicated: CS for closed-shell singlet, OS for open-shell singlet and T

for triplet.

Structure TS1OS II C3H

7AHOAVIV TS2 III (C3H7OHAV

III)

M-molecular 39.8 (34.6) 35.1(OS), 35.6(T) 47.4(T), 52.6(OS) 8.9(T), 38.6(CS)M-tridentate 40.1 (35.3) 37.4(OS), 39.1(T) 45.7(T), 50.5(OS) 1.7(T), 24.1(CS)M-bidentate 44.4 (39.8) 41.8(OS), 42.2(T) 53.3(T), 58.7(OS) 22.2(T), 48.5(CS)M-bidentate_2 45.6 (40.5) 41.9(T), 42.2(OS) 55.3(T), 60.5(OS) 23.0(T), 48.9(CS)D-molecular 37.1 (32.3) 29.8(T), 30.0(OS) 38.0(T), 42.7(OS) 5.7(T), 29.8(CS)

D-bidentate 43.9 (39.0) 41.2(OS), 41.8(T) 52.2(T), 57.7(OS) 18.8(T), 43.0(CS)

0.0, CS

41.2, OS

41.8

18.8

Triplet

43.0, CSTS1OS

52.2

57.753.5

61.1

44.6

C3H8 +O=V5+

MECP1 MECP2

TS2T

Singlet

43.9(39.0)

0.0, CS

35.1, OS

35.6

8.9

C3H7-HO-V

4+

Triplet

38.6, CS

TS1OS

47.4

52.657.047.6

C3H7OH-V3+

C3H8 +O=V5+

MECP1

MECP2

35.7 TS2T

Singlet

39.8(34.6)

(a)

(b)

C3H7-HO-V

4+C3H7OH-V

3+

TS2OS

TS2OS

Fig. 4. The potential energy (kcal/mol) diagram of the conversion of propane topropanol on (a) the M-molecular cluster, and (b) the D-bidentate cluster. Numbersin parenthesis include the zero point energy (ZPE) correction. On the singletpotential energy curve the open-shell (OS) and closed-shell (CS) electronicconfiguration are specified.

34 L. Cheng et al. / Journal of Catalysis 302 (2013) 3136

http://-/?-http://-/?-

7/30/2019 1-s2.0-S0021951713000614-main

5/6

porting information, we found that the B3LYP/6-31G(d) modelchemistry predicts energy differences between hydrogen abstrac-tion barriers of different species in close agreement with moreaccurate methods. The difference between the VO5H5 and VO4H3CAH activation barriers calculated by B3LYP/6-31G(d) is 0.2 kcal/mol lower than that calculated by CCSD(T)/6-31G(d).

3.3. Structurereactivity relationship of supported VOx clusters

To relate structure with reactivity of the monomeric clusters,we note that the active site VOx in both the M-bidentate and M-bidentate_2 structures has a tetrahedral coordination environ-ment, one oxo group, two bridging O atoms bonded to Ti atoms,and one terminal hydroxyl group. The only structural differencebetween the two is that the support of the M-bidentate structureis hydroxylated, while that of the M-bidentate_2 is not. The factthat the reaction energies and barriers of the two structures arevery similar, as shown in Table 2, indicates that the hydroxylgroups on the support surface have minimum effect on the reactiv-ity of the catalytic site. The reactivity of the catalyst does not sim-ply depend on the size of the active site either. The monomeric M-bidentate_2 shares a very similar reaction energy profile with thedimeric D-bidentate cluster, but differs significantly from someother monomeric structures such as M-molecular and M-tridentate.

The reactivity of the cluster does seem to be correlated to thecoordination environment of the active V site: the vanadiumatomsin the M-molecular and M-tridentate structures have square pyra-midal coordination environment, and they both correspond toslightly lower TS1OS and intermediate II energies, as well as signif-icantly lower TS2 and intermediate III energies, in comparison withthe tetrahedral M-bidentate and M-bidentate_2. Similar trendswere observed for the dimeric clusters as well: the V atom of theD-molecular cluster has a square pyramidal-like coordination,and it corresponds to a slightly lower TS1OS energy, as well as sig-nificantly lower intermediate II, TS2, and intermediate III energies

than the D-bidentate, in which the V has tetrahedral-like coordina-tion. This is because when a V5+ is reduced to V4+ or V3+, the elec-tron or electrons in d orbitals are more stabilized in a squarepyramidal structure. If we consider ligand field theory, when aV4+ or V3+ ion is placed in the surrounding oxygen ligand field,the degeneracy of the metal d orbitals is broken. Due to the sym-metry of the ligand field, the overlap between a square pyramidalfield and the V d orbitals (dxz and dyz) is more effective than thatwith the tetrahedral field. This leads to a larger decrease in orbitalenergies and thus more stabilization of the d electron(s) in thesquare pyramidal structure. This causes the M-molecular, M-tri-dentate and D-molecular clusters with the square pyramidal V tobe more facile to oxidation state change and more active towardreduction and dehydrogenation reactions.

3.4. Discussion

Despite the fact that TS2 structures are higher in energy thanTS1OS structures for all VOx structures studied, the initial CAH acti-vation is rate-limiting for the overall reaction of propane ODH bysupported vanadia, as evidenced by experimental studies [2426]and explained with a kinetic scheme based on computational re-sults for SiO2-supported VOx species by Rozanska et al. [22] Thehydrogen abstraction barriers calculated for various monomericand dimeric supported VOx structures are in the range of 3240 kcal/mol. These values are comparable with the previously re-ported experimental (27.5 5 kcal/mol)[26] and theoretical values(34.4 kcal/mol) [22]. However, the TS1OS energies of the most and

least active species reported in Table 2 differ by 8 kcal/mol, andthis difference (as well as the differences between TS1OS energies

of all other species) is the same when the entropic contributionsare included as shown in Fig. S3. We also expect the collision fre-quencies between propane and all sites are the same. Since theprefactors are almost the same between all species, the 8 kcal/mol difference in barriers leads to a difference of more than two or-ders of magnitude (150800 times) in reaction rate for these twospecies in the temperature range 600800 K. Therefore, structuraldifferences could actually lead to significant differences in the

turnover frequency of the catalytic site.We believe that effective stabilization of reduced V species (V4+

and V3+) is the key to the catalytic ODH reaction activity of thevanadium oxide species. Previous studies of CeO2-supported VOxhave shown that the origin of enhanced reactivity of the vanadia/ceria system in Marsvan Krevelen-type oxidation reactions isthe ability of ceria to stabilize reduced states by accommodatingelectrons in localized f-states [40,41]. While the ceria supportshowed a strong electronic effect that stabilizes reduced V, we findthat the square pyramidal structure stabilizes reduced V betterthan the tetrahedral structure, indicating a strong geometric/struc-tural effect.

This geometric effect on reactivity is reflected in the simplesmall gas-phase cluster as well: the propane hydrogen abstractionbarrier on the square pyramidal cluster VO5H5 is 9 kcal/mol lowerthan that of the tetrahedral cluster VO4H3 as reported in Table 3.This is consistent with the trend we have observed for the ana-tase-supported clusters, that is, the pyramidal coordinated vanadylgroup is more active for hydrogen abstraction than the tetrahedral.In addition, the relaxation of the ligands surrounding the vanadiumatom also plays a key factor in lowering the reaction barrier. Forexample, the barrier of the pyramidal cluster with terminatinghydroxyl and aqua ligands fixed at bulk V2O5 lattice distance(VO5H5_bulk in Table 3) corresponds to a barrier 9 kcal/mol higherthan fully relaxed VO5H5 cluster or supported square pyramidalclusters. In this sense, segregated VOx clusters on support surfacesare less restricted geometrically and can thus undergo greaterstructural relaxation, in comparison with crystalline V2O5. Thelow barrier of 19.5 kcal/mol, as reported by Cheng et al. [23] for

propane CAH activation by V2O5, may be partially due to the relax-ation of V4O10 clusters used in their calculation.

4. Conclusions

Density functional methods have been used to investigatestructures of anatase-supported monomeric and dimeric VOxstructures and their activity for propane oxidative dehydrogena-tion. We found that a variety of thermodynamically stable struc-tures can coexist on the support surface including ones with asquare pyramidal coordination environment and ones with tetra-hedral environment. The relative populations will depend on tem-perature, hydration condition, etc. These different structures show

significantly different reactivities for propane ODH: the supportedVOx structures with a square pyramidal coordination environment

Table 3

Reaction barriers (kcal/mol) of hydrogen abstraction reaction on

VO5H5 and VO4H3 clusters. Energies were calculated using B3LYP/

6-31G(d) optimized structures with all atoms allowed to relax for

the VO5H5 and VO4H3 clusters, and with the constraint that the

terminating hydroxyl and aqua groups were fixed at lattice V2O5bulk distances for the VO5H5_bulk. Numbers in parenthesis

include the zero point energy (ZPE) correction.

Structure TS1OS

VO4H3 47.8 (43.1)VO5H5 38.5 (34.5)VO5H5_bulk 47.5 (43.0)

L. Cheng et al. / Journal of Catalysis 302 (2013) 3136 35

http://-/?-http://-/?-http://-/?-http://-/?-

7/30/2019 1-s2.0-S0021951713000614-main

6/6

being much more active for CAH bond activation. The origin of thisdifference in activity is that the square pyramidal coordinationprovides more effective stabilization of reduced V species in thereaction intermediate structures than the tetrahedral one. Theseresults show that the coordination environment of the active vana-dium site is a key structural parameter for its activity. This struc-tural difference can result in an increase of 800 times in reactionrate of CAH activation at 600 K. Furthermore, the varied activitiesof catalytic structure coexisting on the support suggest that ad-vanced catalyst synthesis technique to control site specificity couldprovide greatly improved catalytic efficiency.

Acknowledgments

This research was supported by the U.S. Department of Energy,Office of Basic Energy Sciences, Division of Materials Sciences, Con-tract No. DE-AC-02-06CH11357. We also thank the Argonne Centerfor Nanoscale Materials for computing resources.

We are grateful for discussion with Dr. Paul Redfern, ArgonneNational Laboratory.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jcat.2013.02.012 .

References

[1] B.M. Weckhuysen, D.E. Keller, Catal. Today 78 (2003) 2546.[2] B.M. Weckhuysen, J.M. Jehng, I.E.J. Wachs, Phys. Chem. B 104 (2000) 7382

7387.[3] T. Tanaka, H. Yamashita, R. Tsuchitani, T. Funabiki, S.J. Yoshida, Chem. Soc.

Faraday Trans. I 84 (1988) 29872999.[4] M. Ruitenbeek, A.J. van Dillen, F.M.F. de Groot, I.E. Wachs, J.W. Geus, D.C.

Koningsberger, Top. Catal. 10 (2000) 241254.[5] U.G. Nielsen, N.Y. Topsoe, M. Brorson, J. Skibsted, H.J. Jakobsen, J. Am. Chem.

Soc. 126 (2004) 49264933.

[6] R. Kozlowski, R.F. Pettifer, J.M. Thomas, J. Phys. Chem. 87 (1983) 51765181.[7] Y. Izumi, F. Kiyotaki, H. Yoshitake, K. Aika, T. Sugihara, T. Tatsumi, Y. Tanizawa,T. Shido, Y. Iwasawa, Chem. Commun. (2002) 24022403.

[8] Z.L. Wu, S. Dai, S.H. Overbury, J. Phys. Chem. C 114 (2010) 412422.[9] G.T. Went, S.T. Oyama, A.T. Bell, J. Phys. Chem. 94 (1990) 42404246.

[10] M.A. Vuurman, I.E. Wachs, J. Phys. Chem. 96 (1992) 50085016.[11] N. Magg, B. Immaraporn, J.B. Giorgi, T. Schroeder, M. Baumer, J. Dobler, Z.L. Wu,

E. Kondratenko, M. Cherian, M. Baerns, P.C. Stair, J. Sauer, H.J. Freund, J. Catal.226 (2004) 88.

[12] E.L. Lee, I.E. Wachs, J. Phys. Chem. C 111 (2007) 1441014425.

[13] H.S. Kim, S.A. Zygmunt, P.C. Stair, P. Zapol, L.A. Curtiss, J. Phys. Chem. C 113(2009) 88368843.

[14] D.E. Keller, F.M.F. de Groot, D.C. Koningsberger, B.M. Weckhuysen, J. Phys.Chem. B 109 (2005) 1022310233.

[15] C. Cristiani, P. Forzatti, G. Busca, J. Catal. 116 (1989) 586589.[16] J.L. Bronkema, A.T. Bell, J. Phys. Chem. C 111 (2007) 420430.[17] A. Vittadini, A. Selloni, J. Phys. Chem. B 108 (2004) 73377343.[18] T.K. Todorova, M.V. Ganduglia-Pirovano, J. Sauer, J. Phys. Chem. C 111 (2007)

51415153.[19] T.K. Todorova, M.V. Ganduglia-Pirovano, J. Sauer, J. Phys. Chem. B 109 (2005)

2352323531.[20] V. Brazdova, M.V. Ganduglia-Pirovano, J. Sauer, J. Phys. Chem. B 109 (2005)

2353223542.[21] H. Kim, G.A. Ferguson, L. Cheng, S.A. Zygmunt, P.C. Stair, L.A. Curtiss, J. Phys.

Chem. C 116 (2012) 29272932.[22] X. Rozanska, R. Fortrie, J. Sauer, J. Phys. Chem. C 111 (2007) 60416050.[23] M.-J. Cheng, K. Chenoweth, J. Oxgaard, A. van Duin, W.A. Goddard III, J. Phys.

Chem. C 111 (2007) 51155127.[24] K.D. Chen, A. Khodakov, J. Yang, A.T. Bell, E. Iglesia, J. Catal. 186 (1999) 325

333.[25] K.D. Chen, E. Iglesia, A.T. Bell, J. Catal. 192 (2000) 197203.[26] M.D. Argyle, K.D. Chen, A.T. Bell, E. Iglesia, J. Catal. 208 (2002) 139149.[27] P.C. Redfern, P. Zapol, M. Sternberg, S.P. Adiga, S.A. Zygmunt, L.A. Curtiss, J.

Phys. Chem. B 110 (2006) 83638371.[28] A.E. Lewandowska, M. Calatayud, F. Tielens, M.A. Banares, J. Phys. Chem. C 115

(2011) 2413324142.[29] S.T. Choo, Y.G. Lee, I.S. Nam, S.W. Ham, J.B. Lee, Appl. Catal. A: Gen. 200 (2000)

177188.[30] M. Horn, C.F. Schwerdtfeger, E.P.Z. Meagher, Kristallogr. New Cryst. Struct.

136 (1972) 273281.[31] A.D. Becke, J. Chem. Phys. 98 (1993) 56485652.[32] G.E. Scuseria, H.F. Schaefer, J. Chem. Phys. 90 (1989) 37003703.[33] G.E. Scuseria, C.L. Janssen, H.F. Schaefer, J. Chem. Phys. 89 (1988) 73827387.[34] G.D. Purvis, R.J. Bartlett, J. Chem. Phys. 76 (1982) 19101918.[35] J.A. Pople, M. Headgordon, K. Raghavachari, J. Chem. Phys. 87 (1987) 5968

5975.[36] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,

G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M.Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.Honda, O. Kitao, H. Nakai, T. Vreven, J.A. MontgomeryJr., J.E. Peralta, F. Ogliaro,M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J.Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M.Cossi, N. Rega, N.J. Millam, M. Klene,J.E. Knox,J.B. Cross, V. Bakken, C. Adamo, J.

Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C.Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth,P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, . Farkas, J.B. Foresman,

J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian, Inc., Wallingford, CT, 2009.[37] J.N. Harvey, M. Aschi, H. Schwarz, W.Koch, Theor. Chem. Acc. 99 (1998) 9599.[38] M.J. Bearpark, M.A. Robb, H.B. Schlegel, Chem. Phys. Lett. 223 (1994) 269274.[39] Y.J. Du, Z.H. Li, K.N. Fan, Surf. Sci. 606 (2012) 956964.[40] M.V. Ganduglia-Pirovano, C. Popa, J. Sauer, H. Abbott, A. Uhl, M. Baron, D.

Stacchiola, O. Bondarchuk, S. Shaikhutdinov, H.J. Freund, J. Am. Chem. Soc. 132(2010) 23452349.

[41] C. Popa, M.V. Ganduglia-Pirovano, J. Sauer, J. Phys. Chem. C 115 (2011) 73997410.

36 L. Cheng et al. / Journal of Catalysis 302 (2013) 3136

http://dx.doi.org/10.1016/j.jcat.2013.02.012http://dx.doi.org/10.1016/j.jcat.2013.02.012