1-s2.0-S0360319915314518-main

download 1-s2.0-S0360319915314518-main

of 7

Transcript of 1-s2.0-S0360319915314518-main

  • 8/18/2019 1-s2.0-S0360319915314518-main

    1/7

    Metal substrate enhanced hydrogen production of 

    aluminum fed liquid phase GaeIn alloy inside

    aqueous solution

    Xiao-Hu Yang  a, Bin Yuan  a, Jing Liu  a,b,* 

    a Beijing Key Lab of Cryo-Biomedical Engineering and Key Lab of Cryogenics, Technical Institute of Physics and

    Chemistry, Chinese Academy of Sciences, Beijing 100190, Chinab Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing 100084, China

    a r t i c l e i n f o

    Article history:

    Received 13 November 2015

    Received in revised form

    2 March 2016

    Accepted 3 March 2016

    Available online 24 March 2016

    Keywords:

    Hydrogen production

    Liquid metalEnhancement

    Metal substrate

    Surface roughness

    a b s t r a c t

    Hydrogen is a kind of promising clean energy with huge potential values for human being.

    Developing a convenient real-time and on-demand hydrogen production and utilization

    method is critical for its wide practices. Liquid metal was recently found to be able to easily

    activate aluminum in aqueous solution to continuously and quickly generate hydrogen at

    room temperature, and thus provides a straightforward hydrogen production way. To

    further improve this technology, we are dedicated here to investigate the effects of the

    container substrate on such hydrogen generation process. An interesting phenomenon

    was found that metal substrate material would evidently enhance the hydrogen generation

    rate. For conceptual illustration purpose, comparative experiments were conducted to

    demonstrate the different hydrogen evolution modes between the present method andformer approach, and the enhancement mechanism lying behind was revealed. In addi-

    tion, the influence of the surface roughness of the substrate on the hydrogen production

    performance was also clarified. The present finding would help motivate an improved

    extremely simple, straightforward and low cost way for the liquid metal assisted

    aluminum hydrogen production.

    Copyright  ©  2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

    reserved.

    Introduction

    Hydrogen, a clean and carbon-emission-free energy with high

    gravimetric energy density, has attracted much attention over

    the past few decades due to the increasingly serious issues on

    fossil fuel shortage and air pollution. So far, there are already

    many hydrogen production approaches ever proposed, which

    mainly include: steam reforming of alcohols and coal gasifi-cation   [1e3], water electrolysis and photolysis   [4e6], metal

    induced hydrogen evolving chemical reactions   [7e9], etc.

    Among them, the first approach still dominants the current

    commercial market and owns advantages such as cost-

    effective, technical maturity, etc. However, viewing from the

    perspective of raw material, this method is essentially fossil

    energy consuming. Moreover, the storage and transportation

    *   Corresponding author. Technical Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, PR China.

    Tel.: þ86 10 82543765; fax:  þ86 10 82543767.

    E-mail address: [email protected] (J. Liu).

     Available online at www.sciencedirect.com

    ScienceDirect 

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

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 9

    http://dx.doi.org/10.1016/j.ijhydene.2016.03.020

    0360-3199/Copyright  ©  2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

    mailto:[email protected]://www.sciencedirect.com/science/journal/03603199http://www.elsevier.com/locate/hehttp://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://www.elsevier.com/locate/hehttp://www.sciencedirect.com/science/journal/03603199http://crossmark.crossref.org/dialog/?doi=10.1016/j.ijhydene.2016.03.020&domain=pdfmailto:[email protected]

  • 8/18/2019 1-s2.0-S0360319915314518-main

    2/7

    of hydrogen remains a big challenge due to its high flamma-

    bility and extremely low volumetric energy density, and thus

    caused technical difficulties encountered in compression and

    liquefaction.

    To accelerate the extensive use of hydrogen energy, it is of 

    great significance to develop real-time and on-demand

    hydrogen production and utilization method, which can

    dramatically reduce the storage and transportation cost. Suchdemand can be realized by hydrogen evolving chemical re-

    actions of reactive metals andhydrogen sources such as water

    and alcohol, etc. Among which, aluminumewater reaction is

    identified as the most promising candidate   [10,11]. Al is the

    largestabundantmetalonearthcrust,anditislowcostandhas

    stable storage quality. When Al is exposed in air, a denseoxide

    layer (Al2O3) is formed on its surface rapidly due to its active

    chemical property, which protects the Al from further oxida-

    tion. However, this passivation layer also hinders the hydro-

    lysis reaction of Al, which is unfortunately unfavorable for the

    hydrogen production. Such problem can be solved by alloying 

    Al with certain low melting point elements such as Ga, Zn and

    Bi, etc. which can activate Al and promote the hydrolysis re-action. By now, many important efforts have been made to

    investigate the preparation and hydrogen generation perfor-

    mance of such Al based alloys [9,12e17]. Gallium is considered

    to be an excellent activator of Al to generate hydrogen in

    normal conditions. Its activation effect is attributed to the Ga

    induced destructionof theoxidelayeron theAl surface andthe

    liquid metal embrittlement (LME) effect of Al caused by the

    penetration ofGa into thegrainboundaryof thepolycrystalline

    Al, which is well-known as the Rebinder effect [18e20].

    In our latest work, a new conceptual liquid phase hydrogen

    generation method was proposed [21], which is straightforward

    and simple, and can work without evident time and energy

    consuming alloying treatment process. In that work, a typicalroom temperature liquid metal, GaeIn alloy, was used for the

    activation of the reaction. The dynamic hydrogen generation

    phenomenon of aluminum fed liquid phase GaeIn alloy inside

    NaOH electrolyte, which is much different from the conven-

    tional hydrogen evolution process of solid Al rich alloys, was

    experimentally demonstrated there. As a continuing work to

    further improve the technology, this paper is dedicated to

    explore the influence of different container substrates on the

    hydrogen production events. We were surprised to find that,

    when put on a metal substrate, such as stainless steel, the

    hydrogen production process of Alewater reaction will be

    significantly enhanced compared with the former process

    which was carried out on a glass petri dish. More importantly,the hydrogen evolution rate in this work was promoted to 2e3

    times that of the previous onewhichis of practical significance.

    In addition, the liquid metal and the stainless steel substrate

    used here were not consumed during the hydrogen production

    process, just like catalysis. All these features are beneficial for

    the future extensive applications of such a straightforward, low

    cost and real-time liquid phase hydrogen production method.

    Principles and methods

    The Alewater reaction can occur both in deionized water and

    aqueous solution, such as NaCl or NaOH solution, etc. Since

    NaOH performances best, here, we choose 0.5 mol/L NaOH

    solution for conceptual demonstrations, and similar results

    can be obtained using deionized water or NaCl solution. It is

    worthy to note that, even there is no liquid metal, the

    Alewater reaction can also happen under assistance of al-

    kalis, because that the alkaline solution will destroy the

    passivation layer on the Al surface which hinders the reaction.

    However, when in deionized water or NaCl solution, nothing can react with the oxide film layer, thus it is critical for the

    reaction in this situation to deposit liquid metal for the acti-

    vation of Al. A series of reactions will occur in the Alewater

    reaction with assistance of alkalis, as given by:

    2Al þ 6H2O þ 2NaOH/2NaAlðOHÞ4 þ 3H2   (1)

    NaAlðOHÞ4/NaOH þ AlðOHÞ3   (2)

    2Al þ 6H2O/2AlðOHÞ3 þ 3H2   (3)

    where,   reactions (1) and (2)   indicate the intermediately

    involved hydrogen evolution processes and Eq. (3) is an overall

    expression of such processes. It can be seen that, essentially,

    only Al and water are consumed in the whole reaction. The

    hydrogen evolution procedure of aluminumewater reaction is

    dependent on many factors such as temperature, the con-

    centration of the aqueous solution, the composition, ratio and

    size of the alloy, etc. Here, our focus is on exploring the metal

    substrate enhanced hydrogen evolution reaction of Al fed

    liquid metal and the mechanism lying behind, thus the factors

    aforementioned will not be investigated here.

    A stainless steel dish was chosen as the conductive sub-

    strate, and was filled with 0.5 mol/L NaOH solution inside.

    1 mL liquid metal (GaIn10, 90% Gallium and 10% Indium by

    weight, with a melting point of about 16  C) was dropped into

    the solution, and a slide of Al foil (0.03 g) was immersed in the

    solution and contacted with the liquid metal. A soap film

    flowmeter was used to measure the real-time hydrogen yield

    and a camera (Sony Cyber-shot Dsc-W690, Japan) was

    configured to record the dynamic hydrogen generation pro-

    cess. Comparative experiments were conducted using a glass

    petri dish as substrate. All those experiments were carried out

    at room temperature (about 20  C).

    Results and discussions

    Hydrogen evolution mode

    Fig. 1  shows the different hydrogen evolution modes under

    four different conditions, including the liquid metal activated

    Alewater reaction on stainless steel substrate (Fig. 1(a))and on

    glass substrate (Fig. 1(b)), and Al foil without liquid metal in

    NaOH solution (Fig. 1(c)) and NaCl solution (Fig. 1(d)) for

    comparison. When a slice of Al foil was put in contact with the

    liquid metal droplet in aqueous solution, it was tightly

    attracted by the liquid metal and then embrittled into small

    grains, caused by the LME effect, and finally swallowed into

    the liquid metal droplet. The hydrogen evolution process of Al

    fed liquid metal can be divided into two main sub-processes,

    namely the embrittlement and swallow process at the

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 96194

    http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020

  • 8/18/2019 1-s2.0-S0360319915314518-main

    3/7

    beginning and the subsequent escape process. Here,   “escape”

    means that the embrittled tiny Al grains escape out from the

    Al-liquid metal amalgam to its surface. The differenthydrogen evolution modes induced by the two different sub-

    strates mainly occurred in the later process, namely the

    escape process, and our attention will be focused on that

    process.

    On the glass substrate, there were two hydrogen evolution

    forms. Some hydrogen was generated on the almost homo-

    genously scattered sites on the bottom of the liquid metal, and

    some hydrogen was generated on the side surface of the liquid

    metal through the aggregated Al grains, in Fig. 1(b). From the

    hydrogen evolution yield viewpoint, the second form was

    dominant. The generated tiny bubbles beneath the liquid

    metal would merge to form a bigger bubble. And finally the

    bigger bubble was released from the edge of the liquid metalbottom. At the same time, aqueous solution compensated into

    the bottom space, thus the reaction happened there was kept

    running. The aggregated Al grains on the side surface reacted

    violently with the surrounding solution and intensive tiny

    bubbles were generated there.

    When the substrate was replaced with stainless steel,

    different hydrogen evolution mode was observed, in Fig. 1(a).

    Intensive tiny hydrogen bubbles were ejected from the whole

    edge of the liquid metal bottom, and big bubbles covered on

    the whole upper surface of the liquid metal. The big bubbles

    desorbed from the liquid metal surface due to buoyancy force

    and then were quickly compensated by newly generated

    bubbles. In addition, vertical oscillation of the liquid metal

    was observed, which did not occur on the glass substrate. The

    oscillation was not very regular and the frequency was about

    2e

    5 Hz. This oscillation, in some degree, promoted thedesorption of the hydrogen bubbles and the compensation of 

    solution beneath the liquid metal, thus accelerated the

    hydrogen evolution process. In addition, the intensive

    hydrogen ejection and liquid metal oscillation enhanced the

    solution flow and take the reaction byproduct Al(OH)3 away.

    This is beneficial to the efficient and continuous hydrogen

    evolution process.

    As controls,   Fig. 1(c), (d) show the hydrogen evolution

    modes of Al foil without liquid metal in NaOH and NaCl so-

    lutions, respectively. In NaOH solution, tiny hydrogen bubbles

    were generated on the whole surface of the Al foil after its

    surface oxide layer was destroyed by NaOH. While in NaCl

    solution, no hydrogen bubble was generated due to the pro-tection of the surface passivation layer.

    Hydrogen evolution rate

    Comprehensive experiments on the hydrogen yields in about

    18 min under the aforementioned four conditions were per-

    formed with typical outputs presented in  Fig. 2. For compari-

    son purpose, the mass of Al foil used for all of the 4 conditions

    are kept the same, namely 0.03 g. In addition, the concentra-

    tion of the solution is a constant of 0.5 mol/L and the volume

    of liquid metal is a constant of 1 mL. The shadowed area on

    the left part of the picture presents the embrittlement and

    swallow process, while the rest presents the escape process. It

    Fig. 1 e  Hydrogen evolution modes under 4 different conditions. a) Al fed liquid metal (0.03 g Al, 1 mL GaIn10 ) in 0.5 mol/L

    NaOH solution, on stainless steel substrate; b) Al fed liquid metal in 0.5 mol/L NaOH solution, on glass substrate; c) Al foil

    without liquid metal in 0.5 mol/L NaOH solution, on stainless steel substrate; d) Al foil without liquid metal in 0.5 mol/L NaCl

    solution, on stainless steel substrate.

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 9   6195

    http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020

  • 8/18/2019 1-s2.0-S0360319915314518-main

    4/7

    can be seen that, the most intensive hydrogen evolution re-action happened in the embrittlement stage, and the

    hydrogen generation performance on both substrates in this

    processhad almost no difference. While in the escape process,

    the stainless steel substrate showed significant superiority,

    due to its long lasting intensive reaction, as shown in  Fig. 1(a).

    About 20 min later, the hydrogen evolution reaction on the

    stainless substrate was almost completed, while the reaction

    on the glass lasted about 1 h. It is clearthat, on a stainless steel

    substrate, the Al fed liquid metal displays the best hydrogen

    production performance.

    Enhancement mechanism discussion

    To clarify the reason why the hydrogen generation rate was

    enhanced on the stainless steel substrate, it is necessary at

    first to reveal the basic hydrogen evolution mechanism on the

    glass substrate. According to our former experimental results

    [22,23], after the embrittlement and swallow process, the

    embrittled tiny Al grains can be roughly considered uniformly

    dispersed in the whole Al-liquid metal amalgam, as shown in

    Fig. 3(a). As the density of Al is about two times lower than

    that of the liquid metal, some Al grains, which have larger

    volume and thus larger buoyancy, float up to the amalgam

    surface due to buoyancy force, while some very tiny grains

    will not float up due to the viscous drag. Since the Al grain is

    mechanically instable on the quasi-atomically smooth arcsurface of the liquid metal [24], it slips down to the side sur-

    face.The escaped Al grain is still tightlyabsorbed by theliquid

    metal and will react with the surrounding solution, and the

    surface tension of the amalgam there will become larger,

    which is caused by the electrochemical reaction induced

    charge variation of the surface electric double layer (EDL).

    More details can refer to our previous work  [25]. The Al grain

    site where has the largest surface tension pulls surrounding 

    surface liquid to flow there, which is called Marangoni flow,

    and can be easily observed in the experiment. To compensate

    the pulled away surface liquid, the inner liquid will flow out,

    and a flow loop is thus formed, as shown in  Fig. 3(b). The Al

    grains inside the amalgam are carried out by the dual driving 

    force from the flowing liquid metal and the buoyancy, and

    then aggregate at the side surface of the amalgam where the

    surface tension is the largest, as shown in the experimental

    result in Fig. 1(b) and in Fig. 3(b), respectively. The aggregated

    Al grains released hydrogen violently both through direct

    hydrolysis reaction with the surrounding solution and the Al/

    liquid metal bipolar electrochemical reaction. At the same

    time, some of the larger grains are embrittled into smallergrains and swallowed by the liquid metal once again, and then

    participate in the loop flow, until all of the Al grains are hy-

    drolyzed. It is worthy to note that, the situation shown in

    Fig. 3(a) in fact hardly exists, for that in the embrittlement

    process, Marangoni flow occurs, and the situation in Fig.3(b) is

    already present.

    It can be concluded from the aforementioned analysis that

    the buoyancy force and the Marangoni flow are the main

    reason that incessantly provides the Al grains for hydrogen

    evolution reaction, accelerate the Al grains supply rate will

    lead to the enhancement of the hydrogen generation rate.

    Bearing this in mind and based on the experimental obser-

    vation, it is presumed that, on the stainless steel substrate,tremendous Al grains were attracted and quickly aggregated

    at the bottom of the amalgam, thus violent hydrogen evolu-

    tion reaction happened there, as shown in   Fig. 3(c). This

    assumption can be easily verified by a simple and intuitive

    demonstrative experiment, in Fig. 3(d). When a stainless steel

    bar was inserted into the amalgam, rapidly, tremendous tiny

    Al grains aggregated near the bar, and significant hydrogen

    evolution reaction happened both on the bar and on the Al

    grains surface. The stainless steel induced attraction effect

    can be explained by the bipolar electrochemical reaction

    happened there. Al acted as anode:  Al þ 3OH/AlðOHÞ3 þ 3e,

    and the stainless steel bar acted as cathode:

    3H2O þ 3e/3=2H2 þ 3OH. The mutual attraction betweentwo opposite electrodes promotes the Al grains to move to the

    bar.

    The same analysis can be applied to the situation where

    the reaction happens on a stainless steel substrate. Different

    from the situation on the glass substrate, where the Al grains

    aggregate on certain side of the amalgam, the stainless steel

    substrate attracts all of the escaped Al grains to the bottom of 

    the amalgam. In this situation, the escaped Al grains are not

    passively draggedby the surface flow, but actively and quickly

    move towards the substrate. It will not retard the surface flow

    as that of the situation on the glass substrate, but in turn

    accelerate the surface flow. Since the surface liquid flows

    faster, the compensation from the inner liquid acceleratesaccordingly, and a more efficient Al grain supply loop flow is

    thus formed, and the enhancement effect is obtained.

    Effect of surface roughness

    Another factor that may cause varied hydrogen generation

    performances on both substrates is the surface roughness. To

    rule out this potential influence factor that may induce the

    different hydrogen revolution modes aforementioned, addi-

    tional experiments were conducted. The glass petri dish was

    scratched by a 200 mesh sand paper (M200, which means that

    the diameter of the sand is about 74   mm) to get a rougher

    surface, and similar hydrogen production experiments as

    Fig. 2  e  Hydrogen yield curves over time under 4 different 

    conditions.

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 96196

    http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020

  • 8/18/2019 1-s2.0-S0360319915314518-main

    5/7

    stated before were carried out. The result showed that, on the

    rough glass substrate, the hydrogen evolution mode was the

    same as that of the former smooth one, and the hydrogen

    generated on the side surface of the liquid metal was still

    dominant, almost no hydrogen generation enhancement ef-

    fect was achieved.

    Although the surface roughness of the substrate cannot

    induce the change of hydrogen evolution mode, it does can

    affect the hydrogen revolution process beneath the liquid

    metal, and this effect is not significant for the reaction on

    glass substrate. However, on stainless steel, where the

    hydrogen evolution reaction mainly occurs on the bottom of 

    the liquid metal, this effect cannot be neglected.  Fig. 4 shows

    the hydrogen yield curves on stainless substrate with

    different surface roughness. The substrates were scratched by

    M200 and M80 sand papers, respectively. In addition, a sub-

    strate without surface treatment was set as control group.

    Here, only the corresponding hydrogen yields in the escape

    process were measured and presented. It can be seen that, a

    rougher surface can promote the hydrogen evolution process,

    which may be because that such rough surface facilitated the

    compensation of solution beneath the liquid metal and pro-

    vided tremendous nucleation sites for the hydrogen evolution

    reaction.

    Different conductive substrates

    Further, copper and graphite plates, had also been investi-

    gated as substrates. When a drop of liquid metal was placed

    on the copper substrate, it dissolved some copper and its

    surface tension was deduced, thus caused the liquid metal to

    spread out and adhere on the copper substrate  [26]. Different

    from the hydrogen evolution mode on stainless steel sub-

    strate, hydrogen bubbles were mainly generated on the whole

    copper surface. However, the corrosion of copper by liquid

    metal will pollute this catalyst, especially in alkaline solution

    Fig. 3 e  Schematic diagrams for illustrating the different hydrogen evolution mechanisms on glass/stainless steel

    substrates. a) Section view of the Al-liquid metal amalgam after the embrittlement and swallow process; b) Marangoni flow

    inside and on the surface of the amalgam on glass substrate; c) On the stainless steel substrate; d) Stainless steel bar

    induced Al grains aggregation on the liquid metal surface near the bar and corresponding intensive hydrogen evolution

    reaction happened both on the Al grains surface and the bar surface.

    Fig. 4 e  Hydrogen yield curves on stainless substrate with

    different surface roughness. The substrates were scratched

     by M200 and M80 sand paper respectively, and a substrate

    without surface treatment was set as control.

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 9   6197

    http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020

  • 8/18/2019 1-s2.0-S0360319915314518-main

    6/7

    and high temperature situation, thus it is unsuitable to select

    copper as the substrate for future practices.

    As for graphite substrate, similar hydrogen evolution mode

    as that on stainless steel was observed, while inferior

    hydrogen generation performancewas observed. To conclude,

    some prerequisites must be satisfied when selecting the

    conductive substrate material. The substrate should be

    conductive. Further, it should have high chemical stabilityand do not react with the aqueous solution, and cannot be

    corroded by liquid metal. In addition, the substrate material

    should be readily available, easy to fabricate, and clearly low

    cost. Taking all those requirements into consideration,

    stainless steel is an excellent candidate.

    It is worthy to note that, the liquid metal GaIn10 is essen-

    tially used for destroying the oxide film layer on the Al surface

    and embrittling Al into tiny grains, thus activating the

    Alewater reaction and increasing the reaction area. The metal

    substrate used here is only for promoting the escape of the Al

    grains from the liquid metal-Al amalgam, and thus acceler-

    ating the hydrogen evolution process. Both the liquid metal

    and the metal substrate can be seen as the cathode of thebipolar electrochemical reaction, and are not consumed in the

    whole reaction processes, which can be verified by the

    composition analysis of the after-reaction aqueous solution

    examined by Inductively Coupled Plasma Emission Spec-

    trometer (ICP, Varian 710-ES,USA). 2.5 mg/mL Ga and 0.2 mg/mL

    Fe was measured, with total volume about 50 mL, which

    means that 1.8  mmol Ga and 0.18  mmol Fe were tested, far less

    than the amount of the reactant Al 1.11 mmol. The tested Ga

    and Fe are thought to be due to dissolution by the solution,

    similar to the situation in Tang 's work [27], thus it is believed

    that both the liquid metal and the substratedid not participate

    in the hydrogen generation reaction.

    Last but not least, at the end of the reaction happened onstainless steel substrate, brownish-black chemical substance

    was generated on the liquid metal surface, which might be

    gallium oxide. This oxide layer can be easily removed by

    transferring the liquid metal into NaOH solution on a glass/

    plastic substrate. To avoid such oxidation reaction, one can

    transfer the liquid metal out from the container with stainless

    steel substrate to one with glass substrate before the end of 

    the hydrogen evolution reaction, thus ensuring that the liquid

    metal is not polluted and can be used repeatedly, well

    resembling the same functions that “catalyst” can achieve. On

    graphite substrate, no such oxidation reaction will occur, thus

    a compromise must be made when choosing stainless steel or

    graphite as the substrate.

    Conclusion

    Liquid metal can easily destroy the oxide film layer of Al and

    embrittle it into tiny grains at room temperature, thus activate

    the Alewater reaction and increase the reaction area. In such

    a liquid phase hydrogen production method, the reaction

    process is much different from that of the conventional solid

    Al rich alloy hydrogen production method. To better utilize

    the strategy, conductive materials, typically stainless steel,

    was found to be able to evidently enhance the hydrogen

    evolution reaction of the Al fed liquid phase Gae

    In alloy inside

    aqueous solution. According to our experimental in-

    vestigations, such an enhancement effect is thought to be

    achieved by the attraction of the substrate and the Al grains,

    which accelerates the Marangoni flow of the amalgam and

    thus accelerates the Al grains supply rate. Stainless steel is a

    good choice to be the conductive substrate, due to its high

    chemical stability and low cost. Rough substrate surface

    cannot change the hydrogen evolution mode, but areconductive to the hydrogen production rate, especially on the

    stainless substrate, where the hydrogen evolution reaction

    mainly occurs on the bottom of the liquid metal. Overall, the

    present metal enhanced liquid phase hydrogen generation

    method could be possibly used for straightforward real-time

    and on-demand hydrogen production and utilization, espe-

    cially in situations where water or salt water is readily avail-

    able, thus only bulk Al and a small amount of GaeIn alloy are

    needed.

    Acknowledgment 

    This work is partially supported by the Dean's Research

    Funding of the ChineseAcademy of Sciences. Great thanks are

    given to Mr. Si-cong Tan, Mr. Yu-jie Ding and Ms. Lu Tian in

    the lab for their help in the experiments.

    r e f e r e n c e s

    [1]  Nichele V, Signoretto M, Menegazzo F, Rossetti I, Cruciani G.

    Hydrogen production by ethanol steam reforming: effect of 

    the synthesis parameters on the activity of Ni/TiO2 catalysts.Int J Hydrogen Energy 2015;39:4252e8.

    [2]  Jin H, Chen YN, Ge ZW, Liu SK, Ren CS, Guo LJ. Hydrogen

    production by Zhundong coal gasification in supercritical

    water. Int J Hydrogen Energy 2015;40(46):16096e103.

    [3]   Iulianelli A, Ribeirinha P, Mendes A, Basile A. Methanol

    steam reforming for hydrogen generation via conventional

    and membrane reactors: a review. Renew Sustain Energy Rev

    2014;29:355e68.

    [4]   Wang MY, Wang Z, Gong XZ, Guo ZC. The intensification

    technologies to water electrolysis for hydrogen production e

    a review. Renew Sustain Energy Rev 2014;29:573e88.

    [5]  Wang F, Wen M, Feng K, Liang WJ, Li XB, Chen B, et al.

    Amphiphilic polymeric micelles as microreactors: improving 

    the photocatalytic hydrogen production of the [FeFe]-

    hydrogenase mimic in water. Chem Commun (Camb)2016;52(3):457e60.

    [6]   Luo JS, Im JH, Mayer MT, Schreier M, Nazeeruddin MK,

    Park NG, et al. Water photolysis at 12.3% efficiency via

    perovskite photovoltaics and earth-abundant catalysts.

    Science 2014;345:1593e6.

    [7]   Uan JY, Cho CY, Liu KT. Generation of hydrogen from

    magnesium alloy scraps catalyzed by platinum-coated

    titanium net in NaCl aqueous solution. Int J Hydrogen Energy

    2007;32:2337e43.

    [8]   Grosjean MH, Zidoune M, Roue L, Huot JY. Hydrogen

    production via hydrolysis reaction from ball-milled Mg-

    based materials. Int J Hydrogen Energy 2006;31:109e19.

    [9]  Gai WZ, Liu WH, Deng ZS, Zhou JG. Reaction of Al powder

    with water for hydrogen generation under ambient

    condition. Int J Hydrogen Energy 2012;37:13132e

    40.

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 96198

    http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref9http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref8http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref7http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref6http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref5http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref4http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref3http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref2http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1http://refhub.elsevier.com/S0360-3199(15)31451-8/sref1

  • 8/18/2019 1-s2.0-S0360319915314518-main

    7/7

    [10]  Wang HZ, Leung DYC, Leung MKH, Ni M. A review on

    hydrogen production using aluminum and aluminum alloys.

    Renew Sustain Energy Rev 2009;13:845e53.

    [11]   Kravchenko OV, Semenenko KN, Bulychev BM, Kalmykov KB.

    Activation of aluminum metal and its reaction with water. J

    Alloys Compd 2005;397:58e62.

    [12]   Ziebarth JT, Woodall JM, Kramer RA, Choi G. Liquid phase-

    enabled reaction of AleGa and AleGaeIneSn alloys with

    water. Int J Hydrogen Energy 2011;36:5271e

    9.[13]  Wang W, Zhao XM, Chen DM, Yang K. Insight into the

    reactivity of AleGaeIneSn alloy with water. Int J Hydrogen

    Energy 2012;37:2187e94.

    [14]   Rosenband V, Gany A. Application of activated aluminum

    powder for generation of hydrogen from water. Int J

    Hydrogen Energy 2010;35:10898e904.

    [15]   Elitzur S, Rosenband V, Gany A. Study of hydrogen

    production and storage based on aluminumewater reaction.

    Int J Hydrogen Energy 2014;39:6328e34.

    [16]   Alinejad B, Mahmoodi K. A novel method for generating 

    hydrogen by hydrolysis of highly activated aluminum

    nanoparticles in pure water. Int J Hydrogen Energy

    2009;34:7934e8.

    [17]   Fan MQ, Sun LX, Xu F. Study of the controllable reactivity of 

    aluminum alloys and their promising application forhydrogen generation. Energy Convers Manag 2010;51:594e9.

    [18]   Rajagopalan, Bhatia MA, Tschopp MA, Srolovitz DJ,

    Solanki KN. Atomic-scale analysis of liquid-gallium

    embrittlement of aluminum grain boundaries. Acta Mater

    2014;73:312e25.

    [19]   Ilyukhina AV, Kravchenko OV, Bulychev BM, Shkolnikov EI.

    Mechanochemical activation of aluminum with gallams for

    hydrogen evolution from water. Int J Hydrogen Energy

    2010;35:1905e10.

    [20]   Flamini DO, Saidman SB, Bessone JB. Aluminium activation

    produced by gallium. Corros Sci 2006;48:1413e25.

    [21] Yuan B, Tan SC, Liu J. Dynamic hydrogen generation

    phenomenon of aluminum fed liquid phase Ga-In alloy

    inside NaOH electrolyte. Int J Hydrogen Energy2016;41:1453e9. http://dx.doi.org/10.1016/

     j.ijhydene.2015.10.044.

    [22]  Sheng L, He ZZ, Yao YY, Liu J. Transient state machine

    enabled from the colliding and coalescence of a swarm of 

    autonomously running liquid metal motors. Small

    2015;11:5253e61.

    [23]   Yuan B, Tan SC, Zhou YX, Liu J. Self-powered macroscopic

    Brownian motion of spontaneously running liquid metal

    motors. Sci Bull 2015;60:1203e10.

    [24]  Regan MJ, Kawamoto EH, Lee S, Pershan PS, Maskil N,

    Deutsch M, et al. Surface layering in liquid gallium: an X-Ray

    reflectivity study. Phys Rev Lett 1995;75:2498e501.

    [25]  Zhang J, Yao YY, Sheng L, Liu J. Self-fueled biomimetic liquid

    metal mollusk. Adv Mater 2015;27:2648e55.

    [26]  Wang J, Chen LF, Wu N, Kong ZZ, Zeng MQ, Zhang T, et al.Uniform graphene on liquid metal by chemical vapour

    deposition at reduced temperature. Carbon 2016;96:799e804.

    [27]  Tang SY, Sivan V, Petersen P, Zhang W, Morrison PD,

    Kalantar-zadeh K, et al. Liquid metal actuator for inducing 

    chaotic advection. Adv Funct Mater 2014;24:5851e8.

    i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 6 1 9 3 e6 1 9 9   6199

    http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://dx.doi.org/10.1016/j.ijhydene.2015.10.044http://dx.doi.org/10.1016/j.ijhydene.2015.10.044http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://dx.doi.org/10.1016/j.ijhydene.2016.03.020http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref27http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref26http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref25http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref24http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref23http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://refhub.elsevier.com/S0360-3199(15)31451-8/sref22http://dx.doi.org/10.1016/j.ijhydene.2015.10.044http://dx.doi.org/10.1016/j.ijhydene.2015.10.044http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://refhub.elsevier.com/S0360-3199(15)31451-8/sref20http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref19http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref18http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref17http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref16http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref15http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref14http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref13http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref12http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref11http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10http://refhub.elsevier.com/S0360-3199(15)31451-8/sref10