Investigation of hydrogen occlusion by molybdenum carbide

Accepted Manuscript Title: Investigation of hydrogen occlusion by molybdenum carbide Author: Ricardo R. Oliveira Jr. Angela S. Rocha Victor Teixeira da Silva Alexandre B. Rocha PII: DOI: Reference:

S0926-860X(13)00573-5 http://dx.doi.org/doi:10.1016/j.apcata.2013.09.031 APCATA 14469

To appear in:

Applied Catalysis A: General

Received date: Revised date: Accepted date:

22-7-2013 6-9-2013 17-9-2013

Please cite this article as: R.R. Oliveira Jr., A.S. Rocha, V.T. Silva, A.B. Rocha, Investigation of hydrogen occlusion by molybdenum carbide, Applied Catalysis A, General (2013), http://dx.doi.org/10.1016/j.apcata.2013.09.031 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Investigation of hydrogen occlusion by molybdenum carbide

Ricardo R. Oliveira Jr.1, Angela S. Rocha2,3, Victor Teixeira da Silva2,*,

Universidade Federal do Rio de Janeiro, Departamento de Físico-Química,

cr

1

ip t

Alexandre B. Rocha1,*

us

Instituto de Química, Cidade Universitária, CT, Bloco A,

2

an

Rio de Janeiro, RJ, CEP 21941-909, Brazil.

Universidade Federal do Rio de Janeiro, NUCAT – Programa de Engenharia

M

Química –COPPE, P.O. Box 68502, 21941-914, RJ, Rio de Janeiro, Brazil.

Universidade do Estado do Rio de Janeiro, Departamento de Físico-Química,

d

3

Ac ce p

te

Instituto de Química, Maracanã, 20.550-013, RJ, Rio de Janeiro, Brazil.

* Corresponding author: [email protected], Phone: +55 21 2562 8344 Fax: +55 21 2562 8300

1

Page 1 of 31

Abstract The incorporation of hydrogen in the bulk of molybdenum carbide is studied by experimental and theoretical methods. The motivation for this study is to explain the

ip t

features of the deactivation of Mo2C catalyst during the hydrogenation of benzene when the reaction is performed at atmospheric pressure. In this condition, deactivation occurs

cr

for different reaction times depending on reaction temperature. By means of thermal-

us

programmed desorption it was observed that there is an evolution of hydrogen for the deactivated catalyst at temperatures higher than 923 K. In order to verify if during the

an

carburization of MoO3 to Mo2C there is hydrogen occlusion in the interior of the carbidic phase, a systematic thermodynamic analysis was performed based on density

M

functional theory (DFT) and the harmonic oscillator-rigid rotor approach to calculate partition functions. Possible diffusion paths of hydrogen atoms in molybdenum carbide

d

were calculated by the nudged-elastic bands method at DFT level. The results of

te

calculations are totally consistent with experimental findings confirming that there is

Ac ce p

indeed hydrogen occlusion during the synthesis of molybdenum carbide.

Keywords: Molybdenum carbide; Benzene hydrogenation; Hydrogen occlusion; DFT.

2

Page 2 of 31

1. Introduction Transition metal carbides, particularly molybdenum carbide, are highly active for a large number of reactions catalyzed by noble metals based materials [1-5]. Results from literature show that the electron density is transferred from Mo to C atoms and

ip t

there is a high density of states at Fermi level as expected for a metal [6].

Molybdenum carbide catalysts have been employed in several reactions in which

cr

hydrogen is used, such as hydrogenation [7-9], hydrogenolysis [10], Fischer-Tropsch

(FT) synthesis [11, 12], hydrocracking [13], and hydrotreating [14-16]. These reactions

us

have also been performed on specific metallic catalysts and the hydrogen adsorption has a key role for the adequate activity of the active phase, therefore a high chemisorption

an

capacity is a critical and desirable characteristic.

The performance of molybdenum carbide in the benzene hydrogenation reaction at

M

atmospheric pressure was previously studied and it has been hypothesized that after the synthesis by temperature-programmed carburization of the molybdenum oxide with a mixture of CH4 and H2, some amount of hydrogen remained on the carbide surface and

d

was responsible for the activity of the material [17]. It was observed that an initial

te

conversion of 100 % was achieved but after 30 minutes of reaction, it started to decrease and at approximately 4 hours after the beginning of reaction, the conversion

Ac ce p

was zero. Based on temperature-programmed desorption experiments and theoretical calculations a mechanism for deactivation was proposed. Apparently, deactivation is caused by the strong adsorption of benzene on the carbide surface, preventing further hydrogen adsorption and thus leading to deactivation. In the beginning of the reaction, the surface is covered by hydrogen and hydrogenation of benzene occurred favorably, apparently through Eley-Rideal mechanism. As reaction goes on, benzene molecules will eventually find a spot on surface free of hydrogen. The benzene strongly adsorbs poisoning the surface, which prevents more hydrogen to be adsorbed. One point not discussed in detail previously is that if only a monolayer of adsorbed hydrogen is available to hydrogenation of benzene, the deactivation would be faster than that observed. Therefore, there must be an additional source of hydrogen. TPD results suggested that hydrogen could diffuse from the bulk of molybdenum carbide towards the surface as soon as the surface hydrogen is consumed. These 3

Page 3 of 31

hydrogen molecules could have been imprisoned in the carbide bulk during the carburization step. The main objective of this study using both theoretical and experimental results is to investigate if it is possible to have hydrogen occlusion during the carburization step. benzene

hydrogenation

and

temperature-programmed

desorption

ip t

Accordingly,

experiments have been carried out and systematic theoretical calculations based on

us

cr

density functional theory have been performed.

2. Experimental

an

2.1 Synthesis and Activity of bulk β-Mo2C

The molybdenum carbide synthesis was described in a previous paper [17]. Briefly, molybdenum oxide (Aldrich) was carburized using a 350 mL min−1 flow of a

M

mixture of 20 % (v/v) CH4/H2 (Linde UP) up to a temperature of 923 K at a 10 K min−1 heating rate and holding the final temperature for 2 h.

d

Immediately after the synthesis, at 923 K, the reactor was cooled down to room

te

temperature (RT) under two different conditions: (i) pure helium flow; (ii) pure hydrogen down to RT and then, the gas was changed to helium.

Ac ce p

After the synthesis and cooling, a thermal programmed experiment was performed

by heating the reactor from RT up to 923 K at a heating rate of 10 K min−1 under flow of pure He (50 mL min-1), and analyzing the gas exiting the reactor in a mass spectrometer (MKS-PPT). The signals were chosen to allow the study of some compounds of interest that were monitored and recorded, i.e., m/z values of 2 (hydrogen), 16 (methane), 18 (water), and 28 (carbon monoxide). After the desorption experiment, the system was cooled down to RT and a new heating and monitoring was performed with the objective of eliminating and analyzing all of the hydrogen trapped in the bulk of the molybdenum carbide. Benzene hydrogenation was carried out at atmospheric pressure, under a 30 mL min-1 of hydrogen saturated with benzene vapor at 296 K (pv = 11.3 Pa) at 323 and 363 K reaction temperature. Experiments were performed immediately after in situ carbide synthesis with 0.375 g of bulk carbide followed by cooling with pure hydrogen (120 mL 4

Page 4 of 31

min-1). Reaction products were analyzed on-line by a Finnigan 9001 gas chromatograph equipped with a flame-ionization detector and a methyl siloxane capillary column (30.0 m x 250 μm x 1.0 μm). Under the employed conditions, cyclohexane was the only

ip t

product formed and detected by gas chromatography.

cr

2.2 Theoretical calculations

In order to verify the possibility of hydrogen adsorption on the Mo2C surface

us

during the cooling step under flow of CH4/H2 gas mixture and its concomitant incorporation in the bulk during the carburization step, several models have been constructed containing different amounts of hydrogen for a given Mo2C unit cell. The

an

unit cells for molybdenum carbide used in this study are of orthorhombic type. The calculated cell constants are a = 4.73 Å, b = 6.06 Å, and c = 5.25 Å [17], which should be compared to experimental values of a = 4.729 Å, b = 6.028 Å, and c = 5.197 Å [18].

M

We will refer to this unit cell as model cell 1 (MC1), to which different amounts of hydrogen were incorporated. Calculation with this unit cell provided an initial

d

exploratory study of incorporation of hydrogen.

te

A second type of unit cell, which we will refer as model cell 2 (MC2), was constructed by propagating the MC1 cell one time in the z direction, i.e. a (1×1×2)

Ac ce p

supercell, and by establishing a vacuum layer of 17 Å, in order to provide a model to surface studies. This supercell was used to study the concomitant adsorption of hydrogen with its imprisoning in its bulk. In both model cells, atomic positions were optimized but cell constants were

kept fixed, since no expansion is experimentally observed when hydrogen is incorporated into molybdenum carbide, i.e., when molybdenum carbide is prepared by carburization of the oxide.

A detailed thermodynamic analysis was done concerning the process of adsorption and occlusion of hydrogen in molybdenum carbide from gas phase, and the diffusion of hydrogen atoms from the bulk to surface was also investigated. Some diffusion paths have been calculated by the Nudged Elastic Band [19] method, which is commonly used to calculate the minimum energy path. In this method several 5

Page 5 of 31

intermediate images can be generated along the path connecting two minima (reactants and products) and each image is optimized defining the minimum energy path. Although not very large, these unit cells provide a clear understanding of the problem we are facing. Larger unit cell could be a better choice but there is a

ip t

computational limitation, especially in what concerns the calculation of phonon frequencies as discussed later.

cr

Calculations were done at spin unpolarized Density Functional Theory (DFT)

level with periodic boundary condition, plane wave basis set and ultrasoft pseudo-

us

potentials [20]. The exchange and correlation functional developed by Perdew, Burke and Ernzerhof [21] was used. Occupation was treated by the cold smearing technique of

an

Marzari et al. [22], with smearing parameter of 0.02 Ry for structural optimization. All calculations have been done with Quantum Espresso suite of programs [23]. For thermodynamic analysis, the Fermi–Dirac distribution has been used to

M

determine the occupation of electronic levels and consequently for the calculation of the electronic partition function as discussed below. For the present case though, it will be

d

shown that results are essentially the same if the cold smearing distribution is used in

te

the calculation of the electronic partition function. The kinetic energy cutoff was 30 Ry. Phonon frequencies have been calculated

Ac ce p

at Density Functional Perturbation Theory (DFPT) [24] level and the obtained values were used to validate the optimized geometries as real minima. The vibrational data were also used to calculate thermodynamic functions. For thermodynamic analysis, partition functions were constructed by

considering the harmonic oscillator-rigid rotor approximation for molecules in the gas phase, and harmonic oscillator for condensed phases. The Gibbs free energy is written as: (Eq.1) which combined with enthalpy and Helmholtz free energy expressions: (Eq.2) (Eq.3) 6

Page 6 of 31

results in: (Eq.4) In the canonical ensemble, the Helmholtz free energy is written as:

cr

Consequently, the Gibbs free energy can be written as:

ip t

(Eq.5)

(Eq.6)

us

Using the ideal gas approximation, the Gibbs free energy becomes:

(Eq.7)

an

For the gas phase, the partition function has electronic, vibrational, translational and rotational contribution. The electronic contribution is equal to the degeneracy of the

In equation (8),

(Eq.8)

te

d

The vibrational contribution is:

M

electronic ground state. In the present case, it is equal to one.

, kB, T, c and h stand for the wave number, the Boltzmann

Ac ce p

constant, the absolute temperature, the speed of light and the Planck constant, respectively.

The rotational contribution for a linear molecule is: (Eq. 9)

In equation (9), B is the rotational constant. The translational contribution is: (Eq. 10)

In equation (10), m is the total mass. The overall partition function is: (Eq.11) 7

Page 7 of 31

In the condensed phase, the partition function has only electronic and vibrational contribution. On the gamma point, the vibrational contribution is:

The electronic contribution for Helmholtz free energy:

ip t

(Eq.12)

cr

(Eq.13)

us

where, fi is the Fermi-Dirac distribution.

For condensed phase, the PV term is negligible and the Gibbs free energy equals

an

the Helmholtz free energy:

(Eq.14)

) stands for the Gibbs function of the system with n hydrogen

te

where G(

(Eq.15)

d

M

The variation of Gibbs function was calculated for the following process:

atoms incorporated to the surface and to the bulk per unit cell, whilst G(Mo2C) and

Ac ce p

G(H2) are the Gibbs function for Mo2C free of hydrogen and of gas phase hydrogen molecules, respectively.

3. Results and Discussion

The hydrogen signal obtained during the temperature-programmed desorption of

molybdenum carbide previously carburized a 673 K for 2 hours and cooled under flow of pure hydrogen followed by purge with helium at room temperature (RT) is shown in Figure 1. After heating up to 923 K under pure He flow, the sample was kept at this temperature for 30 min and re-cooled down to RT. Then, a second desorption procedure was performed, with monitoring of signal m/z = 2, followed by a new cooling and a third desorption. The hydrogen signals obtained for the second and third desorption experiments are also presented in Figure 1. 8

Page 8 of 31

As can be seen in the first desorption profile, some quantity of hydrogen desorbs upon heating presenting four well defined maxima at 340, 389, 429 and 466 K, and the signal returns to the baseline around 700 K. This result is similar to the obtained before [17]. The greatest amount of hydrogen desorbs in temperatures lower than 500 K, but a significant quantity continues to desorb while the temperature is still being increased up

ip t

to the plateau at 923 K, with the hydrogen signal returning to the baseline only after cooling. This indicates that even heating the sample at 923 K during 30 minutes, there

cr

was still hydrogen is evolving from the carbide. This assumption was confirmed by

performing a second desorption, and observing a small hydrogen desorption with two

us

maxima around 480 and 620 K. By carrying out a third desorption procedure, it is possible to verify that only an insignificant quantity of hydrogen evolves from the

an

sample, around 625 K.

The hydrogen desorbed at low temperatures is consistent with weakly chemisorbed species. However, the hydrogen evolution at higher temperature and

M

during successive desorption procedures could be an indication of occlusion of hydrogen in the bulk, which after successive heating under helium is released from the

te

following hypothesis:

d

solid. The genesis of these “different” types of hydrogen could be explained by the

i) Because molybdenum carbide presents noble metal behavior, as soon as it is

Ac ce p

formed during the carburization step under flow of methane/hydrogen, hydrogen spillover can occur. Thus, the activated hydrogen would migrate to the interior of the carbide generating the occluded hydrogen. The same phenomenon could occur during the cooling step either under flow of CH4/H2 mixture or under flow of pure H2. This hypothesis was theoretically proven as will be shown; ii) When the cooling temperature reaches values were the migration of hydrogen

to the bulk is not favored anymore then it chemisorbs at the surface being available for reaction. The temperature-programmed desorption was also performed three consecutive times after cooling the carbide under helium flow after its synthesis in order to verify if the hydrogen chemisorption or migration to the bulk had occurred during this step. The result is presented in Figure 2. In this case, two maxima of hydrogen evolution were observed at 394 and 595 K in the first desorption experiment, both of them 9

Page 9 of 31

approximately with the same intensity. The profiles obtained for second and third successive desorption experiments show only hydrogen evolution at high temperatures (> 500 K) thus indicating that even after cooling the sample with helium, there was hydrogen occluded in the bulk of the carbide.

ip t

By comparing the first two desorption experiments, it is possible to observe, in Figure 3, that the quantity of hydrogen diffused from the bulk to the surface and

released at temperatures lower than 500 K is much higher when the carbide was cooled

cr

under flow of methane/hydrogen mixture after the carburization. This behavior indicates

us

that during the cooling procedure under helium there was the release of the hydrogen occluded. On the other hand, some hydrogen remains occluded in the carbide

an

framework after the carburization.

To verify the influence of this hydrogen occluded into the carbide on the hydrogenation activity of the solid, a catalytic test was performed at two different

M

temperatures. It is noteworthy that for both temperatures employed (323 and 363 K) the initial benzene conversion was 100%, which is the equilibrium conversion for the reaction. Figure 4 presents the activity at 323 and 363 K for β-Mo2C carburized in situ

d

and cooled to room temperature under flow of pure hydrogen. The high initial activities

te

(XBenzene = 100 %) decreased rapidly in both experiments at different temperatures, and conversion was null after 1.7 h for reaction at 323 K, and 3.7 h when the reaction was

Ac ce p

performed at 363 K.

This rapid deactivation was observed previously, and it was proposed that it is

associated to a strong benzene adsorption [17] on the surface of the molybdenum carbide. In order to understand the reason why the higher the reaction temperature the higher the time for complete deactivation, several theoretical calculations were performed with the aim of investigating either if the hydrogen occluded on the bulk of molybdenum carbide was stable or if it could migrate to the surface thus influencing the hydrogenation capacity of molybdenum carbide. The incorporation of hydrogen in the carbide bulk was investigated first, by using Model Cell 1 (MC1). In Table 1, one can find the calculated ΔEele concerning the reaction, (n/2) H2 + Mo2C → Mo2C (nH-bulk). 10

Page 10 of 31

All values are positive, which means that ΔG is positive unless the system is submitted to an enormous pressure. From these results, it is quite clear that the absorption of hydrogen into the bulk of Mo2C is intrinsically unfavorable. Of course, this is a very idealized model since there is no incorporation of hydrogen in the bulk

ip t

without concomitant adsorption on the surface. In order to take this possibility into account, a calculation with model cell 2 (MC2) was performed considering different

cr

amounts of hydrogen on the surface and in the bulk.

MC2 has two times more atoms than MC1 and a vacuum layer in order to expose

us

the (001) surface. Accordingly, from MC2 we can model adsorption on the surface and absorption in the bulk at the same time. The question that immediately arises is what is the amount of hydrogen on surface and in the bulk? Calculations with MC1 showed that

an

incorporation of hydrogen in the bulk is not favorable. So we decide to start the calculation with MC2 with all atoms inside the slab in order to verify whether they spontaneously go to the surface and in what proportion. The initial and final structures

M

are shown in Figure 5 for a cell with 8 hydrogen atoms, which is one half the number of molybdenum atoms. As can be seen, a certain amount of hydrogen spontaneously goes

d

toward the surface. The procedure was repeated for different proportions of hydrogen.

te

The calculated ΔEele concerning the reaction,

Ac ce p

(n/2) H2 + Mo2C → Mo2C(nH),

where Mo2C(nH) stands for structures similar to the final structure in Figure 5, are shown in Table 2.

All values are negative but this cannot guarantee that such processes are

spontaneous at finite temperature. So there was the need to calculate the other terms of G. Translational and rotational contributions to partition function are trivially computed. Vibrational contribution is also trivial for gas phase molecules but quite expensive for slab calculation, since it demands the calculations of phonon frequencies. The calculation for a slab without hydrogen and with different proportions of hydrogen in order to calculate the ΔG were performed. All calculations have been done at gamma point. Results are also shown in Table 2 for 363 K. Negative values indicate that the incorporation of hydrogen to the bulk of Mo2C concomitantly with its adsorption on the surface is a spontaneous process for a cell containing up to 13 11

Page 11 of 31

hydrogen atoms. Figure 6 shows the ΔG values calculated for different temperatures. For each one of them, the point where ΔG becomes positive is different. For cells containing more hydrogen, ΔG becomes positive at a lower temperature. This temperature marks the point where desorption (the reverse reaction) becomes

ip t

spontaneous. These calculations have been done with occupation of electronic levels done by the cold smearing distribution, which does not represent the real physical

situation. That is represented by the Fermi-Dirac distribution as indicated in equation

cr

(13). Nonetheless, the results obtained within the cold smearing occupation are totally

equivalent to that of Fermi-Dirac, as can be seen in Figure 7, where results are

us

compared for two hydrogen compositions. The other cases with hydrogen composition in between that shown in Figure 7 show the same equivalence.

an

From the analysis of ΔG dependence on temperature, the behavior of the benzene hydrogenation with temperature can be rationalized. When the reaction is conducted at

M

323 K the deactivation occurs earlier because a smaller amount of hydrogen is available from the bulk. By increasing the temperature to 363 K more hydrogen is available at the surface the deactivation is postponed by about 2 h (Figure 4) when compared to

d

experiment conducted at 323 K because at higher temperatures desorption is more reaction going.

te

favorable and thus there would be more hydrogen available at the surface to keep the

Ac ce p

The calculations discussed so far do not address the kinetic evolution of hydrogen

from the carbide. So, in order to deal with these aspects, additional calculations based on the NEB method to establish possible diffusion paths for hydrogen were performed. First, the diffusion of a hydrogen atom from subsurface toward the surface was considered and the path is shown in Figure 8. As can be seen there is no barrier involved. This is an expected result since during the structural optimization, shown before in the context of the thermodynamic analysis, the hydrogen atoms spontaneously migrate toward the surface. The importance of it is that, once a surface hydrogen atom is consumed by the hydrogenation reaction, another, lying in the subsurface, takes its place. Finally, the hydrogen diffusion deep inside the bulk was evaluated. MC1 was used in this study and Figure 9 presents an example of diffusion path profile. This time, a barrier of about 15 kcal mol-1 was found. Of course, this is only one of the many 12

Page 12 of 31

possible diffusion paths but aids to illustrate that the process is activated. If in one hand, tunneling could effectively attenuate the barrier increasing effectiveness of diffusion, on the other the solid defects could trap hydrogen making diffusion more difficult. Calculations with larger unit cells could eventually reveal more details of the diffusion

ip t

but such calculation are quite expensive for the time being. The above discussion put some light in the TPD experiments, concerning the evolution of hydrogen at temperatures as high as 923 K. Hydrogen close to surface

cr

evolves without a barrier whilst hydrogen deep inside the bulk can be removed only by

an

us

a certain amount of (thermal) energy.

M

4. Conclusions

In this work, experimental and theoretical evidences were achieved that during the

d

carburization process to form molybdenum carbide from molybdenum oxide, hydrogen

te

is both incorporated to the bulk and adsorbed on the surface of the resulting carbide. The incorporation in the bulk is not thermodynamically favorable but can be

Ac ce p

accomplished by the concomitant adsorption on the surface. This conclusion was attained by a series of DFT calculations done by considering the incorporation of different amounts of hydrogen at several temperatures and two model unit cells. The bulk-imprisoned hydrogen acts as a hydrogen reservoir and, as long the

surface hydrogen is consumed to hydrogenate benzene, it supplies the surface with more hydrogen. The amount of hydrogen that can be released depends on the procedure to cool the carbide after its synthesis (under carburizing mixture or pure He) and on the reaction temperature and explains why the catalyst deactivation occurs at different reaction times when the reaction is conducted at different temperatures. The hydrogen incorporation also explains the evolution of hydrogen at temperatures as high as 923 K in thermal-programmed desorption experiment. Diffusion path profiles were obtained by means of the nudged elastic band method. This 13

Page 13 of 31

calculation shows that diffusion deep inside the bulk is an activated process whilst diffusion from subsurface toward the carbide surface is barrierless.

ip t

Acknowledgements The authors acknowledge Conselho Nacional de Desenvolvimento e Pesquisa

Ac ce p

te

d

M

an

us

cr

(CNPq), FAPERJ and PETROBRAS for financial support.

14

Page 14 of 31

References [1] M. Saito, R. B. Anderson, J. Catal. 63 (1980) 438-446. [2] A. P. E. York, J. B. Claridge, C. Marquez-Alvarez, A. J. Brungs , S. C. Tsang, M. L.

ip t

H. Green Stud. Surf. Sci. Catal. 110 (1997) 711-720. [3] J. A. J. Rodrigues, G. M. Cruz, G. Bugli, M. Boudart, G. Djega-Mariadassou, Catal.

cr

Lett. 45 (1997) 1-3.

[4] J. Wang, M. Castonguay, J. Deng, P. H. McBreen Surf. Sci. 374 (1997) 197-207.

Tsang, M. L. H. Green, J. Catal. 180 (1998) 85-100.

us

[5] J. B. Claridge, A. P. E. York, A. J. Brungs, C. Marquez-Alvarez, J. Sloan, S. C.

an

[6] J. Yu, A. B. Anderson, Surf. Sci. 254 (1991) 320-328.

M

[7] P. M. Patterson, T. K. Das, B. H. Davis, Appl. Catal. A-Gen. 252 (2003) 449–455. [8] A. S. Rocha, V. Teixeira da Silva, A. A. Leitão, M. H. Herbst, A. F. Faro Jr., Catal.

d

Today 98 (2004) 281-288.

137-147.

te

[9] A. S. Rocha, V. Teixeira da Silva, A. F. Faro Jr., Appl. Catal. A-Gen 314 (2006)

Ac ce p

[10] J. S. Lee, S. Locatelli, S. T. Oyama, M. Boudart, J. Catal. 125 (1990) 157–170. [11] D.-V. N. Vo, C. G. Cooper, T.-H. Nguyen; A. A. Adesina, D. B. Bukur, Fuel 93 (2012) 105-116.

[12] G. Vitale, M. L. Frauwallner, C. E. Scott, P. Pereira-Almao, Appl. Catal. A-Gen 408 (2011) 178-186.

[13] K. Chandra Mouli, V. Sundaramurthy, A. K. A. Dalai, J. Mol. Catal. A 304 (2009), 77–84. [14] B. Dhandapani, T. St. Clair, S. T. Oyama, Appl. Catal. A-Gen 168 (1998) 219–228. [15] P. Da Costa, C. Potvin, J.-M. Manoli, B. Genin, G. Djéga-Mariadassou, Fuel 83 (2004) 1717–1726.

15

Page 15 of 31

[16] J.-M. Manoli, P. Da Costa, M. Brun, M. Vrinat, F. Maugé, C. Potvin, J. Catal. 221 (2004) 365–377. [17] A. S. Rocha, A. B. Rocha, V. Teixeira da Silva, Appl. Catal. A-Gen. 379 (2010)

[18] E. Parthe, V. Sadagopan, Acta Crystallogr. 16 (1963) 202–210.

[20] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892–7895.

cr

[19] G. Mills, H. Jonsson, Phys. Rev. Lett. 72 (1994) 1124-1127.

ip t

54–60.

us

[21] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868. [22] N. Marzari, D. Vanderbilt, A. De Vita, M. C. Payne, Phys. Rev. Lett. 82 (1999)

an

3296–3299.

[23] P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D.

M

Ceresoli, G. L. Chiarotti, M. Cococcioni, I. Dabo, A. D. Corso, S. de Gironcoli, S. Fabris, G. Fratesi, R. Gebauer, U. Gerstmann, C. Gougoussis, A. Kokalj, M. Lazzeri, L.

d

Martin-Samos, N. Marzari, F. Mauri, R. Mazzarello, S. Paolini, A. Pasquarello, L. Paulatto, C. Sbraccia, S. Scandolo, G. Sclauzero, A. P. Seitsonen, A. Smogunov, P.

te

Umari, R. M. Wentzcovitch, J. Phys. Condens. Matter. 21 (2009) 395502.

Ac ce p

[24] S. Baroni, S. de Gironcoli, A. Dal Corso, and P. Giannozzi, Rev. Mod. Phys. 73 (2001) 515-562.

16

Page 16 of 31

Caption for Tables

Table 1 - Calculated electronic energies for incorporation of hydrogen atoms in the bulk of molybdenum carbide for model cell 1 (see text), which exhibits no surface. Zero

cr

ip t

point energy correction was not applied.

Table 2 - Calculated electronic energies, and Gibbs free energy at 363 K, within the

us

rigid-rotor-harmonic-oscillator approximation, for incorporation of hydrogen atoms in the bulk of molybdenum carbide and concomitant adsorption on the surface.

Ac ce p

te

d

M

an

Calculations done with model cell 2 (see text).

17

Page 17 of 31

Caption for figures Figure 1 – Successive hydrogen desorption profiles after carburization of molybdenum oxide followed by cooling with hydrogen stream.

ip t

Figure 2 – Successive hydrogen desorption profiles after carburization of molybdenum oxide followed by cooling with helium stream.

cr

Figure 3 – Hydrogen desorption profile after carburization of molybdenum oxide

us

followed by cooling with hydrogen or helium stream.

Figure 4 – Benzene hydrogenation activity at 323 K (z) and 363 K („) with 0.375 g of

an

molybdenum carbide.

Figure 5 – Structural relaxation using MC2. Left: initial structure with all hydrogen atoms regularly distributed inside the bulk. Right: Final structure in which part of

M

hydrogen atoms migrates toward the surface.

Figure 6 – Temperature dependence of ΔG for the incorporation of hydrogen atoms in

d

the bulk of molybdenum carbide and concomitant adsorption on the surface. The point

te

where the curves cross the ΔG=0 axis marks the temperature at which the desorption

Ac ce p

process becomes favorable.

Figure 7 – Comparison of temperature dependence of ΔG with different distribution functions for occupation of electronic levels. Two hydrogen compositions are shown, with 8 and 16 hydrogen atoms per unit cell. The Fermi-Dirac (FD) distribution, which has physical meaning, leads to results equivalent to the cold smearing (CS) distribution of Marzari et al. in the present case. Figure 8 – NEB calculation for diffusion of a hydrogen atom from subsurface toward surface. The hydrogen atom which is been moved is marked in red. Horizontal axis indicates the distance from the surface. Zero value is the surface coordinate. Negative values mean that the atom is below the surface. Positive values mean above. Figure 9 – NEB calculation for diffusion of a hydrogen atom inside the bulk (far from surface) starting in an interstitial position. This is just one of the many possible paths but serves to illustrate that the process is activated. 18

Page 18 of 31

Table 1

ΔE (kcal mol-1)

1

18.2

2

38.7

3

49.4

4

60.5

5

73.3

6

86.6

7

120.2

cr

153.2 188.0

M

9

223.6 285.9 336.5

Ac ce p

te

d

10

12

us

an

8

11

ip t

N(H)

20

Page 19 of 31

ΔE (kcal mol-1)

ΔG (363 K) (kcal mol-1)

8

-94.36

-42.34

9

-103.21

-43.91

10

-103.47

11

-97.35

12

-97.11

13

-89.18

14

-57.70

15

cr us

-37.52

-26.95

an

-21.71 -9.90 28.13

-50.18

38.21

-28.67

72.63

Ac ce p

te

d

16

ip t

N (H)

M

Table 2

21

Page 20 of 31

Ac ce p

te

d

M

an

us

cr

ip t

Highlights Hydrogen is occluded in the lattice during β-Mo2C synthesis The amount of occluded hydrogen depends on the cooling environment During benzene hydrogenation at atmospheric pressure there is deactivation Deactivation time is related to the amount of occluded hydrogen There is hydrogen migration from the bulk to the surface during reaction

22

Page 21 of 31

Ac

ce

pt

ed

M

an

us

cr

i

*Graphical Abstract (for review)

Page 22 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 1

Page 23 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 2

Page 24 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 3

Page 25 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 4

Page 26 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 5

Page 27 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 6

Page 28 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 7

Page 29 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 8

Page 30 of 31

Ac

ce

pt

ed

M

an

us

cr

i

Figure 9

Page 31 of 31