- Email: [email protected]
Density functional theory study of ethanol synthesis from dimethyl ether and syngas over cobalt catalyst
Molecular Catalysis 432 (2017) 115–124
Contents lists available at ScienceDirect
Molecular Catalysis journal homepage: www.elsevier.com/locate/mcat
Editor’s choice paper
Density functional theory study of ethanol synthesis from dimethyl ether and syngas over cobalt catalyst Xinbao Li a , Shurong Wang b,∗ , Yingying Zhu a , Geng Chen a , Guohua Yang a a b
Faculty of Maritime and Transportation, Ningbo University, Ningbo, 315211, PR China State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, PR China
a r t i c l e
i n f o
Article history: Received 2 July 2016 Received in revised form 26 January 2017 Accepted 9 February 2017 Keywords: Density functional theory Ethanol Dimethyl ether Syngas Carbonylation
a b s t r a c t The reaction mechanism of ethanol synthesis from dimethyl ether (DME) and syngas was studied via density functional theory calculations. Various possible pathways for ethanol formation, and byproduct formations of methanol, acetic acid, methane, carbon dioxide, and water over an active cobalt stepped surface were calculated. The most favorable pathway for ethanol synthesis starts with the dissociation of dimethyl ether to CH2 , followed by carbon monoxide insertion to form CH2 CO, and then undergoes successive hydrogenations to give ethanol. CH3 CHO hydrogenation to CH3 CHOH becomes the rate–determining step with a reaction barrier of 1.48 eV. DME decomposition and CH2 carbonylation occur easily with low barriers of 0.61 eV and 0.48 eV, respectively. Carbon monoxide insertion into CH2 is more facile than into CH3 and CH. Hydrogenation at the carbon atom occurs prior to the oxygen atom in the order of ␣–carbon > carbonyl carbon > oxygen. The calculations demonstrate that ethanol synthesis via DME carbonylation and hydrogenation is thermodynamically favored and is kinetically faster than that via syngas direct synthesis. Methane and acetic acid are two dominant competing byproducts. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Biomass–derived ethanol (CH3 CH2 OH) has been recognized as an alternative fuel to traditional fossil fuels since that its raw material, biomass, is renewable and abundant on the earth. The increasing utilization of ethanol as motor fuel will reduce CO2 emission. Because of the easy and low cost production of syngas via biomass gasification, direct CH3 CH2 OH synthesis from syngas (CO + H2 ) has been studied widely [1–4]. Four categories of heterogeneous catalysts have been tested and optimized: modified Fischer–Tropsch synthesis catalysts (cobalt–based or iron–based), modified methanol synthesis catalysts (copper–based), modified molybdenum–based catalysts, and rhodium–based catalysts [4]. However, drawbacks of severe reaction conditions, low one pass CO conversion, and low CH3 CH2 OH selectivity has inhibited its industrial application. Recently, an improved method for CH3 CH2 OH synthesis from dimethyl ether (DME) and syngas has been proposed [5–7]. The synthesis involves the simultaneous or stepwise reactions of DME vapor–phase carbonylation and hydrogenation. It is considered a promising technology because of the high reaction activity and the
∗ Corresponding author. E-mail address: [email protected] (S. Wang). http://dx.doi.org/10.1016/j.mcat.2017.02.014 2468-8231/© 2017 Elsevier B.V. All rights reserved.
efficient production of DME and syngas feedstock from renewable biomass. Iglesia and coworkers found that HMOR and HFER catalyzed DME carbonylation to methyl acetate had stable rates and more than 99% selectivity at 423 463 K, and the rate of DME carbonylation was much higher than that of CH3 OH carbonylation [8,9]. Tsubaki and coworkers reported that DME was converted completely with a CH3 CH2 OH selectivity of 48.4% in the one step CH3 CH2 OH synthesis from DME, CO, and H2 over affordable HMOR and Cu/ZnO dual catalysts [5–7,10,11]. Lu et al. used different noble metals (Pt, Pd, and Ru) to impregnate HMOR and to promote the catalytic activity of CH3 CH2 OH synthesis from DME and syngas, and they found that the combination of 0.4 wt% Pt/HMOR and Cu/ZnO catalyst presented excellent catalytic performance [12]. We have studied DME carbonylation over metal modified HMOR catalysts (Cu, Ni, Co, Zn and Ag), and found that the catalytic performances of Co, Cu, and Ni modified HMOR were better than the performance of HMOR [13]. The activity of Co was comparable with that of Cu in DME conversion (close to 100%) and methyl acetate selectivity (higher than 99%), which shows the effectiveness of Co for DME carbonylation. Bartek et al. reported that CH3 CH2 OH/acetaldehyde (CH3 CHO) selectivities were promoted significantly when Co based catalysts were used in the reductive carbonylation of DME [14]. Co is highly active for ester and aldehyde hydrogenation to produce alcohols [15–17]. Co catalysts are
116
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
considered as promising candidates for CH3 CH2 OH synthesis from DME and syngas via synergic carbonylation and hydrogenation. In this work, we study the mechanism of CH3 CH2 OH synthesis from DME and syngas over a Co catalyst via density functional theory (DFT) calculations. A stepped surface was used in the calculation because it is recognized as the most active site of transition metals for heterogeneous catalytic reactions [18,19]. Adsorptions of key intermediates were firstly calculated to identify their preferred surface adsorption geometries. Then the elementary reactions for DME decomposition, subsequent CO insertion into CHx (x = 1–3) and successive hydrogenation to CH3 CH2 OH, and the competing reactions of direct CO hydrogenation to CH3 OH and CH3 CH2 OH were calculated. Acetic acid (HOAc), CO2 , water (H2 O), and methane (CH4 ) byproduct formations were also considered. Finally, the most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 were identified. To the best of our knowledge, this is the first theoretical study that presents thermodynamic and kinetic reaction barriers of CH3 CH2 OH synthesis from DME and syngas.
2. Computational Methods All spin–polarized periodic DFT calculations were carried out using the Vienna Ab initio Simulation Package (VASP) [20–22], and the Perdew–Burke Ernzerhof (PBE) generalized gradient correction (GGA) was used to treat exchange and correlation effects. The core electrons of all atoms were described by the projector–augmented wave pseudopotential, and the energy cutoff was 400 eV. Monkhorst Pack k–point mesh of 6 × 6 × 6 and 3 × 3 × 1 for bulk and surface calculations were used to sample the Brillouin zone. The Fermi level was smeared using the Methfessel Paxton method with a width of 0.1 eV. The electronic self consistent field converged to 1 × 10−4 eV and forces on all atoms converged to less than 0.05 eV/Å. A fcc Co close-packed unit cell with an original lattice constant of 3.544 Å was selected and optimized by a conjugate-gradient algorithm. The calculated equilibrium bulk lattice constant for Co is 3.524 Å, which agrees well with theoretical (3.521 Å) and experimental values (3.545 Å) [23,24]. Co steps were modeled by using a four layer thick slab of 3 × 3 Co(111) surface supercell, from which two closed packed rows were removed to form a step. The atoms of the bottom two layers were fixed in the bulk positions, whereas the others were allowed to relax to their lowest energy configurations. A vacuum space of 15 Å was applied to avoid the interaction between the slab and its images. The side and top views of the model Co stepped surface are shown in Fig. 1. The transition states (TS) were determined by the climbing nudged elastic band method (CINEB) [25,26]. This method gives a precise estimate of the saddle point with a small modification of the regular efficient NEB method. It has been widely used for TS searching and the minimum energy path finding [18,23,27–30]. In our calculation, an elastic band with 8 structural images along the path of the initial state (IS) and the final state (FS) of each elementary reaction were constructed by linear interpolation. A spring interaction between adjacent images was added to ensure continuity of the path, and optimized using quick-min algorithm. After being converged, the image with the highest energy (TS) was not affected by the spring forces at all. The TS structures were further confirmed by vibrational frequency analysis and determining that only a single imaginary frequency presented along the proper reaction coordinate. All vibrational frequencies were calculated by allowing the entirety of the adsorbed species on the surface to fully relax while all the Co atoms were kept fixed. The reaction rate constants were calculated using the harmonic transition-state theory. The adsorption energy Eads is defined as Eads = Eads/slab − Eslab − Egas , where Eads/slab is the total energy of the slab with
adsorbates, Eslab is the slab energy, and Egas is the energy of adsorbates in the gas phase. The reaction energy (H) and activation energy (Ea ) are defined as H = EFS − EIS , and Ea = ETS − EIS , respectively. EIS , ETS , and EFS are the total energies of the IS, TS, and FS, respectively.
3. Results 3.1. Adsorptions of Intermediates Adsorption geometries of some key reaction intermediates on the Co stepped surface are presented in Fig. 2. The corresponding adsorption geometrical parameters, configurations, and adsorption energies are listed in Table 1. The nomenclature i –j was used to designate i atoms of the adsorbate bound to j atoms on the metal surface. Saturated species, dimethyl ether (CH3 OCH3 ), CH3 OH, and CH3 CH2 OH, have small adsorption energies that range from −0.42 to −0.62 eV, and which demonstrate weak binding to the surface. The adsorptions are induced by the lone pair electrons of O atoms in those adsorbates, forming 1 –1 (O) configurations with an O–s Co distance of 2.09–2.14 Å (Figs. 2(a–c)). For CH3 CHO and HCHO (Fig. 2(d,e)), the adsorption energy increases to −1.42 and −1.59 eV, respectively, and give stronger binding to the surface than that of CH3 OCH3, CH3 OH, and CH3 CH2 OH. This can be ascribed to a relatively strong d– interactions between the d orbital of s Co atoms and the bond of C O in CH CHO and HCHO. Both 3 2 –2 (C,O) configurations were generated with an O atom located at the bridge site and a carbonyl C atom located at the top site of the s Co atoms. The O–s Co distances are shortened by ∼0.15 Å compared with those of CH3 OCH3, CH3 OH, and CH3 CH2 OH species. Unsaturated species bind strongly to the surface through stepped Co atoms with adsorption energies that range from −2.02 to −4.00 eV, which are larger than the saturated species. The CH3 OCH species is the most strongly adsorbed intermediates on the surface with binding energy of −4.00 eV. The corresponding configuration is 1 –2 (C). With the elimination of hydrogen atoms and a decreasing symmetry for CH3 OCH3 decomposition intermediates (Figs. 2(a, f–h)), the adsorption sites change from oxygen to carbon atoms and the corresponding adsorption configurations follow 1 –1 (O) → 2 –2 (C,O) → 1 –2 (C). The reason may be assigned to the favorable formation of a tetrahedral bonding geometry for carbon atoms on the surface [31]. The adsorption of COH on the surface is strong. The C atom located at the bridge site is linked to two s Co atoms on the stepped surface, giving a 1 –2 (C) configuration (Fig. 2(n)). The corresponding adsorption energy is −3.84 eV with dC– s Co of 1.81 and 1.82 Å. The alkoxide species, CH3 O and CH3 CH2 O are also two strong adsorption intermediates, which have the binding energies of −3.55 and −3.52 eV, respectively. The resulting binding configurations are 1 –2 (O) (Figs. 2(i) and (o)). The binding atoms change from carbon to oxygen compared with COH. The length of O s Co bonds for CH3 O and CH3 CH2 O are 1.92 and 1.93 Å, respectively, which are 0.11 Å longer than that for dC– s Co in COH. Both CH3 CHOH and CH2 OH have the adsorption configurations of 2 –2 (C,O) through their unsaturated C atoms and hydroxyl O atoms on the surface, while CHOH is linked to the surface through its C atom only to give a 1 –2 (C) configuration. The calculated configurations are in good agreement with the previous DFT results [29,32]. It can be ascribed to that with the increase of the valence saturation of C atom (the maximum is 4) in adsorbates, the adsorption induced by C atom becomes weaker and weaker, while that induced by O atom becomes dominant. The O and unsaturated C atoms are competitive adsorption on the surface, with C being pre-
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
117
Fig. 1. Side and top views of the model Co stepped surface used in this study. Dotted red line in the top view indicates the unit cell. s Co and f Co represent the stepped and flat atoms, respectively.
Fig. 2. Side and top view of preferred adsorption intermediates. Blue, brown, red, and white spheres represent Co, C, O, and H, respectively.
ferred. Here, the saturation of active C in CH3 CHOH and CH2 OH are 3, and that for CHOH is 2. Similarly, for CH3 OCH2 and CH3 OCH, the saturations of active C (in CH2 and CH groups) are 3 and 2, respectively. Therefore, CH3 OCH2 binds to the surface through its C and O atoms together to form 2 –2 (C,O) configuration, while CH3 OCH binds to the surface through C atom only to form 1 –2 (C) configuration. It also
happens for CH3 OC. On the other hand, as discussed above, the species with completely saturated C, like CH3 OCH3 , CH3 CH2 OH, and CH3 OH, are bonding to the surface through the atomic O only. Given that, it can be concluded that with atomic C saturation, in general, the adsorption configuration changes from C binding to C O binding, and finally to O binding.
118
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Table 1 Adsorption energies, configurations, and geometrical parameters of reaction intermediates on a Co stepped surface. Species
Eads (eV)
Configurations
CH3 OCH3 CH3 OH CH3 CH2 OH CH3 CHO HCHO CH3 OCH2 CH3 OCH CH3 OC CH3 O CH2 OH CHOH CHO COH CH3 CH2 O CH3 CHOH
−0.42 −0.62 −0.58 −1.42 −1.59 −2.47 −4.00 −2.33 −3.55 −2.15 −3.27 −2.64 −3.84 −3.52 −2.02
1 –1 (O) 1 –1 (O) 1 –1 (O) 2 –2 (C,O) 2 –2 (C,O) 2 –2 (C,O) 1 –2 (C) 1 –2 (C) 1 –2 (O) 2 –2 (C,O) 1 –2 (C) 2 –2 (C,O) 1 –2 (C) 1 –2 (O) 2 –2 (C,O)
Geometrical parameters (Å) dC– s Co
3.2. Elementary Reactions In this section, we investigated each elementary reaction involved in DME decomposition; subsequent carbonylation and hydrogenation for CH3 OH and CH3 CH2 OH synthesis; and HOAc, CH4 , CO2 , and H2 O byproduct formations. Potential energy surfaces for the most likely reaction pathways are discussed and depicted. The obtained geometries of all TS and their imaginary frequencies are presented in the Supporting Information.
3.2.1. DME Decomposition A schematic of DME decomposition on a Co stepped surface is presented in Fig. 3. Activation energies, reaction energies, and geometries of transition states are also provided. DME (CH3 OCH3 ) decomposition starts from dehydrogenation to give CH3 OCH2 with a low activation energy of 0.48 eV. The reaction is exothermic by −0.25 eV. The cleaving O H bond in TS1 is elongated from 1.11 Å to 1.38 Å. CH3 OCH2 has two compatible dissociation reactions: O CH2 scission to give CH3 O and CH2 with a barrier of 0.61 eV, and dehydrogenation from the CH2 group to give CH3 OCH with a barrier of 0.58 eV. Both low reaction barriers demonstrate the readily formation of CH2 and CH3 OCH intermediates. The energy barrier difference for those two reactions is as low as 0.03 eV. The corresponding reaction heat for O CH2 scission is strongly exothermic by −0.80 eV, and that for dehydrogenation is slightly endothermic by 0.12 eV, showing a large difference. As a result, CH3 OCH2 dissociation via O CH2 scission to give CH2 is thermodynamically and kinetically favored. The C O bond is elongated significantly by 0.51 Å to 2.12 Å in TS2. For the dehydrogenation, the O H bond is elongated by 0.52 Å to 1.63 Å in TS3. As a CH3 OCH2 dehydrogenation product, CH3 OCH might undergo decomposition via O CH cleavage to give CH3 O and CH (Ea = 0.51 eV) and/or dehydrogenation at CH to give CH3 OC and H (Ea = 0.31 eV). Both reactions are exothermic, with −1.23 eV versus −0.22 eV, respectively. The dehydrogenation is slightly kinetically preferred, whereas the O CH cleavage is strongly thermodynamically favored. The following decomposition of CH3 OC via CH3 O C scission has a reaction barrier of 0.63 eV. Based on the above results of kinetic reaction barriers and thermodynamic reaction energies for possible elementary reactions of DME decomposition, it is clear that DME can be converted easily to give CH2 and CH via C O scissions after dehydrogenation on the Co stepped surface. CH2 formation is even preferred due to its low reaction barrier and strongly exothermic in the early step of DME decomposition.
1.98 1.96 1.98 1.94, 1.93 1.81, 1.83 1.98 1.95, 1.93 1.87, 2.10 1.81, 1.82 1.98
dO– s Co 2.14 2.09 2.10 1.97, 1.95 1.97, 1.94 1.98
1.92, 1.92 2.03 1.97 1.93, 1.93 2.01
3.2.2. Ethanol and Methanol Synthesis The reaction network of CH3 CH2 OH and CH3 OH synthesis via carbonylation and hydrogenation from DME and CO is shown in Fig. 4. Adsorbed H2 is easily dissociated on the Co surface with an extremely low reaction barrier of 0.03 eV, supplying adequate H intermediates for the hydrogenation reactions. H2 dissociation is also thermodynamically favored with a mild exothermicity of −0.55 eV. Intermediates of COH and CHO are two products of CO hydrogenation at its O and C atoms, respectively. Both reactions are strongly endothermic (1.54 and 0.91 eV, respectively). COH formation is energetically unlikely because of its extremely high barrier of 2.13 eV. The energy barrier for CHO formation is 0.93 eV, which is 1.20 eV lower than that for COH formation, and indicates that the first hydrogenation of CO at the C atom to give CHO is preferred rather than the hydrogenation at the O atom to give COH. This agrees well with published results of CO hydrogenation to CHO and COH over Cu and Rh surfaces [29,33,34]. The reported activation energy of COH formation (1.89–2.07 eV) was found to be much higher than that of CHO formation (0.88–1.26 eV). Similarly, for the further hydrogenation of CHO, the addition of H to C atom to form HCHO is more favorable than the addition to O atom to form CHOH, since the barrier for the former is 0.72 eV lower than the later (0.38 eV versus 1.10 eV). HCHO formation is almost thermoneutral. In addition, it is interesting to find that the reaction barrier for CHO deoxygenation to CH is relatively low by 0.63 eV, which is a small 0.25 eV higher than HCHO formation, suggesting the easy deoxygenation of CHO. This reaction is mildly exothermic by −0.38 eV. CHO hydrogenation to HCHO and deoxygenation to CH are two compatible and preferable reactions for CHO conversion. The intermediate of HCHO may undergo three reactions: two hydrogenations at the C and O atoms to form CH3 O and CH2 OH, respectively, and one deoxygenation to form CH2 . The corresponding activation energies are 0.25, 1.04, and 0.85 eV, respectively, in which CH3 O formation is energetically preferred. CH3 O and CH2 formations are exothermic, whereas the formation of CH2 OH is endothermic. The subsequent hydrogenation of CH3 O results in CH3 OH formation. However, the reaction has a high activation energy of 1.67 eV and a strong endothermicity of 1.03 eV, which implies the unlikely formation of CH3 OH and the relative stability of CH3 O species on the surface. This is consistent with a strong binding of CH3 O to the surface. The barrier for CH3 O deoxygenation to CH3 is high to 1.46 eV, which indicates that CH3 production from CO is difficult. The formation of CH3 OH via sequent hydrogenation starts from CHOH at an unsaturated C atom has a relatively low reaction barrier (0.40, 0.80 eV) and is nearly thermoneutral. Consequently, for direct CO hydrogenation to CH3 OH, it is clear that the
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
119
Fig. 3. Schematic of DME decomposition and snapshots of transition states (eV).
hydrogenation at carbon atom is energetically preferred over the hydrogenation at oxygen atom. Studt et al. also found that hydrogenation of the oxygen end involved intermediates of COH, HCOH, and H2 COH, is associated with much higher barriers, and that all intermediates are higher in energy than their carbon hydrogenated counterparts [34]. According to the above discussion of direct CO hydrogenation, intermediates of CH, CH2 , and CH3 are produced by direct deoxygenation of CHO, CH2 O, and CH3 O with reaction barriers of 0.63, 0.85, and 1.46 eV, respectively. All these three deoxygenation reactions are exothermic. CH2 and CH are also predominant products of DME decomposition via C O scissions with relatively low reaction barriers (∼0.60 eV). To produce CH3 CH2 OH, subsequent CO insertions into these CHx (x = 1 − 3) are essential and vital. CO insertions of CH, CH2 , and CH3 to form CHCO, CH2 CO, and CH3 CO have activation energies of 0.64, 0.48, and 1.11 eV with an increasing endothermicity of 0.03, 0.29, and 0.54 eV, respectively. The formation of CH2 CO has the lowest reaction barrier, which indicates that the insertion of CO into CH2 is most likely to occur. In addition to this most preferred reaction, CHCO formation is an energetically close reaction because of a slight barrier difference of 0.16 eV. Compared with CH2 CO and CHCO, CH3 CO formation has a much higher reaction barrier, which shows that it is more difficult to produce directly by CO insertion of CH3 . However, CH3 CO is favorable to be generated by the hydrogenation at the ␣–carbon of CHCO and CH2 CO. The reaction barriers are relatively low at 0.49 and 0.29 eV for CHCO hydrogenation to CH2 CO and CH2 CO hydrogenation to CH3 CO, respectively. These reactions are exothermic by −0.42 and −0.23 eV, respectively, which implies that both reactions are kinetically and thermodynamically preferred. With respect to CO insertion, competing reactions for CHx conversion are their direct hydrogenations. The highest reaction barrier is 0.85 eV, which is associated with hydrogenation of CH3 to form CH4 . This reaction has a mild endothermicity of 0.33 eV. The length of forming C H bond is 1.57 Å at TS, consistent with the
reported DFT result (1.54 Å) [27]. The lowest barrier in CHx hydrogenation is CH2 hydrogenation to CH3 (Ea = 0.28 eV) which is mildly exothermic by −0.37 eV. The barrier for CH hydrogenation to CH2 is 0.30 eV with a mild exothermicity of −0.56 eV. Wang et al. carried out DFT calculations of ethanol decomposition on Co(0001) surface, which can be treated as a reverse reaction of ethanol formation, and found that the barriers for CH2 , CH3 , and CH4 formations were 0.44, 0.46, and 0.99 eV, which are similar to our results [27]. The relatively low reaction barriers obtained in these CHx hydrogenation reactions imply the readily formation of CH4 . This result agrees with experimental observations of alcohol synthesis from syngas, for which CH4 is recognized as an easily generated byproduct [35–37]. In addition to hydrogenation at the ␣–carbon, another favored hydrogenation site for CHCO, CH2 CO, and CH3 CO is the carbonyl carbon. The reactions for CH2 CO hydrogenation to CH2 CHO and CH3 CO hydrogenation to CH3 CHO have relatively low barriers of 0.48 and 0.31 eV, respectively. Both reactions are exothermic (−0.49 and −0.24 eV, respectively). Therefore, hydrogenation at the carbonyl carbon is kinetically and thermodynamically preferable for CH2 CO and CH3 CO. However, the hydrogenation of CHCO to CHCHO is not favored because of its relatively high reaction barrier of 1.02 eV compared with the hydrogenations of CHCO and CH2 CO. Although the formation of CHCHO is difficult, its transformation via hydrogenation is easy. CHCHO undergoes a two–step hydrogenation to CH3 CHO through CH2 CHO with low activation energies of 0.30 eV (CHCHO + H → CH2 CHO) and 0.41 eV (CH2 CHO + H → CH3 CHO). Sites for H addition are both ␣–carbon. Consequently, CH3 CHO intermediates are likely generated through CO insertions into CH and CH2 followed by hydrogenations. Further hydrogenation of CH3 CHO will produce two intermediates of CH3 CH2 O and CH3 CHOH with the reaction barriers of 0.30 and 1.48 eV, respectively. The formation of CH3 CH2 O is exothermic by −0.36 eV, whereas CH3 CHOH formation is endothermic by 0.73 eV. Thus, hydrogenation at the carbonyl carbon atom of CH3 CHO to CH3 CH2 O is more likely than hydrogenation at the oxygen atom to CH3 CHOH.
120
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Fig. 4. Reaction network of CH3 CH2 OH and CH3 OH synthesis via carbonylation and hydrogenation (eV).
However, CH3 CH2 O hydrogenation at the oxygen atom to CH3 CH2 OH is both kinetically and thermodynamically unfavored due to its extremely high reaction barrier of 1.73 eV and its strongly endothermicity of 1.05 eV. The reason may be ascribed to the extremely strong adsorption of CH3 CH2 O species (–3.52 eV) on the surface, which will suppress its reaction with vicinal H atoms. This is consistent with the calculation carried out by Wang et al., in which the reaction of CH3 CH2 O + H → CH3 CH2 OH also has a high barrier of 1.41 eV and is endothermic by 0.62 eV [27]. The length of changing O H bond at TS in our calculation on Co stepped surface is 1.40 Å, the same as that on Co(0001) surface [27]. In contrast with CH3 CH2 O hydrogenation, hydrogenation of CH3 CHOH at the ␣–carbon to form CH3 CH2 OH is energetically facilitated with a relatively low activation energy of 0.66 eV and an exothermicity of −0.17 eV. As a result, the highest reaction barriers for CH3 CH2 OH formation from CH3 CHO mediated CH3 CHOH and CH3 CH2 O are 1.48 eV (CH3 CHO + H → CH3 CHOH) versus 1.73 eV (CH3 CH2 O + H → CH3 CH2 OH), which suggests that CH3 CHOH mediated path for CH3 CH2 OH formation is energetically preferred over
that of CH3 CH2 O mediated path. Liu et al. suggested that ethanol synthesis from syngas on Rh(111) surface goes through intermediates of CH3 CHOH [38]. Zuo et al. also reported that the route for CH3 CH2 OH formation from syngas on the Cu/ZnO catalyst involved the vital steps of CH3 CHO → CH3 CHOH → CH3 CH2 OH [3]. Another possible route for CH3 CH2 OH formation with CH3 CO hydrogenation through CH3 COH and CH3 CHOH was also considered. The highest activation energy in this route is 1.42 eV (CH3 CO + H → CH3 COH), which is 1.11 eV higher than that of CH3 CO + H → CH3 CHO. The reaction is endothermic by 0.46 eV. Subsequent hydrogenation of CH3 COH to form CH3 CHOH has low activation energy of 0.47 eV, associated with a slight endothermicity of 0.23 eV.
3.2.3. Acetic Acid Synthesis Reaction networks for the conversion of O species and the formation of dominant byproducts, HOAc, are presented in Fig. 5. According to the above analysis, O intermediates can be produced easily from CO by direct deoxygenation of CHO and HCHO
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
121
Fig. 5. Reaction network of O species conversion and HOAc formation (eV).
with low activation energies of 0.63 eV (CHO → O + CH) and 0.85 eV (HCHO → O + CH2 ), respectively. By reacting with H, O will be converted to OH with a high activation energy of 1.04 eV and a mild exothermicity of −0.42 eV. Meanwhile, a parallel reaction of O with CO to give CO2 (Ea = 1.05 eV) will occur because of its close reaction barrier to that of OH formation. This reaction is endothermic by 0.27 eV. OH intermediates undergo further hydrogenation to H2 O and CO insertion to COOH with high activation energies of 1.68 eV and 1.87 eV, respectively. Both reactions are strongly endothermic by more than 1 eV. Therefore, OH conversions with H and CO are not preferred. O and OH intermediates are anticipated to carry out oxidation reactions to form carboxylic acid HOAc. The oxidation reactions here are supposed to start from CHCO, CH2 CO, and CH3 CO, which are identified as active intermediates in the reaction network of CH3 CH2 OH synthesis. All three intermediates may undergo direct addition of OH or two–step additions of O and H. Formations of CHCOOH, CH2 COOH, and CH3 COOH via direct OH addition of CHCO, CH2 CO, and CH3 CO have activation energies of 1.10, 0.41, and 0.99 eV, respectively. Among these three reactions, CH2 COOH formation (CH2 CO + OH → CH2 COOH) is kinetically preferred because of its relatively low reaction barriers. The reactions of CHCOOH and CH3 COOH formations are mildly endothermic by 0.54 and 0.50 eV, respectively, whereas that for CH2 COOH formation is exothermic by −0.49 eV. As a consequence, CH2 COOH formation is also thermodynamically facilitated. Activation energies for O additions into CHCO and CH3 CO at carbonyl carbon atoms to generate CHCOO and CH3 COO are 0.98 and 0.73 eV, respectively. Whereas the addition of O into CH2 CO to form CH2 COO has a low activation energy of 0.40 eV, which is 0.58 and 0.33 eV lower than the other two reactions. This reaction is mildly exothermic by −0.42 eV. Consequently, oxidations of CH2 CO to CH2 COO and/or CH2 COOH are more likely than oxidations of CHCO and CH3 CO. Reaction barriers for the further hydrogenations of CHCOO and CH2 COO to CHCOOH and CH2 COOH are 0.75 and 0.61 eV, respectively, implying an energetic preference of CH2 COO + H → CH2 COOH. The hydrogenation of CH3 COO to
CH3 COOH has an extremely high reaction barrier of 1.62 eV, which is strongly endothermic by 1.05 eV and makes the transformation of CH3 COO to CH3 COOH difficult. The barrier for CH3 COOH formation via CH2 COOH hydrogenation is 0.34 eV lower than that of via CH3 COO hydrogenation. Given that, we conclude that CH3 COOH formation is likely starting from CH2 CO oxidation rather than that from CHCO and CH3 CO oxidations. Because of the high activation barrier of OH formation (O + H → OH), CH2 CO oxidation with O is more favored than that with OH. 4. Discussion 4.1. Preferred Pathways for Ethanol Synthesis Four preferred reaction pathways for CH3 CH2 OH synthesis via DME carbonylation and hydrogenation on a Co stepped surface are shown in Fig. 6. As discussed in Section 3.2.1, for P1 and P2 pathways during DME decomposition, the steps with the highest reaction barriers are CH3 OCH2 → CH3 O + CH2 (Ea = 0.61 eV). After that, carbonylation occurs via CO insertion to generate an active intermediate of ketene CH2 CO. The carbonylation reaction occurs easily because of its low energy barrier of 0.48 eV. The following successive hydrogenations of CH2 CO to produce final CH3 CH2 OH in both P1 and P2 proceed through intermediates of CH3 CHO and CH3 CHOH. The highest reaction barriers during the hydrogenation are both 1.48 eV and are associated with CH3 CHOH formation (CH3 CHO + H → CH3 CHOH), which is also the highest barrier in entire P1 and P2 pathways. The difference between P1 and P2 is the starting hydrogenation site of CH2 CO. In P1, CH2 CO hydrogenation is performed at the ␣–carbon to form CH3 CO (CH2 CO + H → CH3 CO, Ea = 0.29 eV). In P2, it is performed at the carbonyl carbon to form CH2 CHO (CH2 CO + H → CH2 CHO, Ea = 0.48 eV). The barrier of the later reaction is 0.19 eV higher than that of the former reaction. CH3 CHO formations via hydrogenations of CH3 CO (P1) and CH2 CHO (P2) have energy barriers of 0.31 and 0.41 eV, respectively. As a result, it is clear that the hydrogenation of CH2 CO → CH3 CO → CH3 CHO
122
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Fig. 6. Four preferred reaction pathways for CH3 CH2 OH synthesis via DME carbonylation and hydrogenation on a Co stepped surface.
in P1 has lower reaction barriers than the hydrogenation of CH2 CO → CH2 CHO → CH3 CHO in P2. Therefore, P1 is more kinetically favored than P2. In addition to CH2 in P1 and P2, CH is another dominant product of DME decomposition in P3 and P4. It is produced through CH3 OCH3 → CH3 OCH2 → CH3 OCH → CH. The highest barrier is 0.58 eV (CH3 OCH2 → CH3 OCH + H). This is close to the highest barrier for CH2 formation in P1 and P2 (CH3 OCH2 → CH3 O + CH2 , Ea = 0.61 eV). Subsequent carbonylation occurs via CO insertion into CH with a barrier of 0.64 eV (CH + CO → CHCO), which is 0.16 eV higher than the barrier of CH2 + CO → CH2 CO in P1 and P2. Similar to CH2 CO, CHCO prefers to be hydrogenated at the ␣–carbon to CH2 CO in P3 and P4 rather than to be hydrogenated at the carbonyl carbon. The reason is that the barrier of ␣–carbon hydrogenation (CHCO + H → CH2 CO, Ea = 0.49 eV) is much lower than that of carbonyl carbon hydrogenation (CHCO + H → CHCHO, Ea = 1.02 eV). The hydrogenations of CH2 CO to produce final CH3 CH2 OH in P3 and P4 are the same as those to P1 and P2, respectively. The highest reaction barrier in the hydrogenation is also 1.48 eV as associated with CH3 CHOH formation, which is also the highest barrier over the entire pathways. Therefore the rate–determining steps (RDSs) for the four P1–P4 pathways are all CH3 CHO + H → CH3 CHOH. Additionally, similar to P1 and P2, P3 is preferable over P4. Although the RDSs are the same and barriers for DME decomposition are energetically close, P1 has an even lower reaction barrier than P3 for the vital CO insertion reaction (0.48 eV versus 0.64 eV), which implies that P1 is further preferred over P3. Therefore, we conclude that in these four competing pathways, P1 is the most kinetically favorable pathway for CH3 CH2 OH synthesis which starts from DME decomposition, followed by carbonylation and hydrogenation on the Co stepped surface. Nevertheless, the other three pathways P2–P4 are also feasible and are likely to occur since the difference in reaction barriers between P1 and P2–P4 is less than 0.20 eV. 4.2. Preferred Pathways for Byproduct Formation As shown in Fig. 4, the kinetically preferred pathway for CH2 formation from CO is CO → CHO → CH → CH2 . The highest barrier is 0.93 eV, which is associated with the first CO hydrogenation step and is 0.32 eV higher than the highest barrier of CH2 formation in P1 (CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH). Consequently, CH2 formation via CO hydrogenation is more difficult than its formation via DME decomposition. In the CO route, the ensuing favored carbonylation and hydrogenations for CH2 to produce CH3 CH2 OH are the same as that of P1. The RDS is also CH3 CHO + H → CH3 CHOH over the entire CH3 CH2 OH synthesis pathway. Therefore, the most likely pathway for CH3 CH2 OH synthesis via direct CO hydrogenation, P(CO), is
CO → CHO → CH → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH. To compare the formation routes of CH3 CH2 OH from DME and from syngas, the forward reaction rate constants of the steps involved in the pathways of P1 and P(CO) were listed in Table 2. It is shown that the slowest reaction both in P1 and P(CO) is CH3 CHO + H → CH3 CHOH. Its rate constants are low to 8.54 × 10−3 , 2.49 × 10−1 , 4.05 × 100 , and 4.22 × 101 s−1 at 473, 523, 573, and 623 K, respectively. Therefore, this step is confirmed as the rate-determining step in CH3 CH2 OH synthesis. Additionally, as discussed above, the difference for P1 and P(CO) is the route for CH2 formation. For P1, CH2 formation is determined by CH3 OCH2 → CH3 O + CH2 . For P(CO), CH2 formation is determined by CO + H → CHO. The corresponding rate constants for those two reactions at 573 K are 3.02 × 108 and 1.51 × 104 s−1 , respectively, which shows that CH2 formation in P1 is 4 orders of magnitude faster than that in P(CO). Therefore, CH3 CH2 OH synthesis via P1 is significantly faster than that via P(CO). The most likely pathway for CH3 OH synthesis from CO is CO → CHO → HCHO → CH2 OH → CH3 OH with the highest barrier of 1.04 eV (HCHO + H → CH2 OH). Additionally, CH4 will be produced as a main byproduct according to the low energy barrier pathway of CH3 OCH3 → CH3 OCH2 → CH2 → CH3 → CH4 . The highest reaction barrier in this path is 0.85 eV associated with CH3 hydrogenation to CH4 . Besides that, the likely route for CH4 formation from CO is CO → CHO → CH → CH2 → CH3 → CH4 . The highest reaction barrier in this route is 0.93 eV with CO hydrogenation to CHO. This barrier is 0.32 and 0.08 eV higher than the barriers of CH3 OCH2 → CH3 O + CH2 and CH3 + H → CH4 , respectively. Therefore, CH4 formation from CO is less preferred than that from DME. HOAc formation is favored via CH2 CO oxidation with O species followed by hydrogenation, as discussed above. By combining an optimal route for CH2 CO formation, the most likely HOAc formation pathway is considered to be CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH2 COO → CH2 COOH → CH3 COOH. The RDS is CH2 COOH hydrogenation to CH3 COOH with a reaction barrier of 1.28 eV. This is 0.20 eV lower than the RDS barrier of P1, and indicates kinetic competition of HOAc formation with respect to CH3 CH2 OH synthesis. Fig. 7 presents a summary of the potential energy surfaces of the most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 synthesis from CO and DME. The overall CH3 CH2 OH synthesis starting from DME decomposition is exothermic by −0.67 eV, but that starting from CO hydrogenation is endothermic by 0.35 eV. Therefore, ethanol formation from DME is thermodynamically preferred and benefitted. CH3 OH synthesis is thermodynamically unfavorable because of the increasing reaction energies, whereas HOAc and CH4 synthesis commencing from DME is favored thermodynamically with overall exothermic reactions.
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
123
Table 2 Forward reaction rate constants (kf ) of the steps involved in the most likely pathways of CH3 CH2 OH synthesis from DME and CO. Temperature ranges from 473 to 623 K. Reaction
kf (s−1 ) T = 473 K
CH3 OCH3 → CH3 OCH2 + H CH3 OCH2 → CH3 O + CH2 CO + H → CHO CHO → CH + O CH + H → CH2 CH2 + CO → CH2 CO CH2 CO + H → CH3 CO CH3 CO + H → CH3 CHO CH3 CHO + H → CH3 CHOH RDS CH3 CHOH + H → CH3 CH2 OH RDS
T = 523 K
1.36 × 10 2.70 × 107 1.94 × 102 8.42 × 105 1.12 × 1010 3.34 × 107 5.99 × 109 1.81 × 109 8.54 × 10−3 1.50 × 106 9
T = 573 K
3.17 × 10 1.01 × 108 2.11 × 103 3.52 × 106 2.30 × 1010 1.06 × 108 1.22 × 1010 4.14 × 109 2.49 × 10−1 6.77 × 106 9
T = 623 K
6.41 × 10 3.02 × 108 1.51 × 104 1.15 × 107 4.16 × 1010 2.75 × 108 2.19 × 1010 8.20 × 109 4.05 × 100 2.34 × 107 9
1.16 × 1010 7.61 × 108 7.87 × 104 3.12 × 107 6.87 × 1010 6.12 × 108 3.59 × 1010 1.45 × 1010 4.22 × 101 6.65 × 107
Rate-determining step.
CH4 and HOAc are identified as two predominant competing byproducts. Acknowledgements The authors acknowledge financial support from the National Natural Science Foundation of China (Nos. 51406090 and 51406093), Ningbo Natural Science Foundation of China (No. 2014A610119), and K.C. Wong Magna Fund in Ningbo University. We are also grateful for the Special Program for Applied Research on Super Computation of the NSFC–Guangdong Joint Fund (the second phase). Appendix A. Supplementary data
Fig. 7. Potential energy surfaces of most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 synthesis from DME and CO.
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.mcat.2017.02. 014. References
5. Conclusions Reaction mechanisms of efficient CH3 CH2 OH synthesis from DME and syngas were studied systematically via DFT calculations. Possible elementary reactions that involve DME decomposition, CO insertion, subsequent hydrogenation and oxidation, and competing reactions of direct CO hydrogenation on an active Co stepped surface were calculated. Preferred adsorption geometries, parameters, and adsorption energies of key reaction intermediates were obtained. The adsorption energies of saturated species CH3 OCH3 , CH3 OH, and CH3 CH2 OH are lower than those of CH3 CHO and HCHO, and are much lower than the unsaturated species. With the increase of the valence saturation of C atom, the adsorption configuration changes from C binding to C O binding, and finally to O binding. The most favorable pathway for CH3 CH2 OH synthesis from DME and syngas is CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH. CH3 CHO hydrogenation to CH3 CHOH becomes the rate-determining step with an activation energy of 1.48 eV. The barriers for other steps in this pathway are less than 0.7 eV. CO insertion into CH2 to form ketene CH2 CO is more facile than the insertions into CH and CH3 . The corresponding barriers are 0.48, 0.64, and 1.11 eV, respectively. Hydrogenation at the carbon atom occurs prior to that at the oxygen atom with an order of ␣–carbon > carbonyl carbon > oxygen. The calculations demonstrate that CH3 CH2 OH synthesis via DME carbonylation and hydrogenation is thermodynamically favored and is kinetically faster than that via syngas direct synthesis.
[1] W.G. Liu, S. Wang, S.D. Wang, Effect of impregnation sequence of Ce promoter on the microstructure and performance of Ce-promoted Rh-Fe/SiO2 for the ethanol synthesis, Appl. Catal. A 510 (2016) 227–232. [2] S. G. Prieto, M.L. Beijer, M. Smith, Y. He, Z. Au, D.A. Wang, K.P. Bruce, J.J. Jong, P.E. Jongh Spivey, Design and synthesis of copper–cobalt catalysts for the selective conversion of synthesis gas to ethanol and higher alcohols, Angew. Chem. Int. Ed. 53 (2014) 6397–6401. [3] Z.J. Zuo, L. Wang, L.M. Yu, P.D. Han, W. Huang, Experimental and theoretical studies of ethanol synthesis from syngas over cuznal catalysts without other promoters, J. Phys. Chem. C 118 (2014) 12890–12898. [4] J.J. Spivey, A. Egbebi, Heterogeneous catalytic synthesis of ethanol from biomass-derived syngas, Chem. Soc. Rev. 36 (2007) 1514–1528. [5] X.G. San, Y. Zhang, W.J. Shen, N. Tsubaki, New synthesis method of ethanol from dimethyl ether with a synergic effect between the zeolite catalyst and metallic catalyst, Energy Fuels 23 (2009) 2843–2844. [6] Y. Zhang, X.G. San, N. Tsubaki, Y.S. Tan, J.F. Chen, Novel ethanol synthesis method via C1 chemicals without any agriculture feedstocks, Ind. Eng. Chem. Res. 49 (2010) 5458–5488. [7] X.G. Li, X.G. San, Y. Zhang, T. Ichii, M. Meng, Y.S. Tan, N. Tsubaki, Direct synthesis of ethanol from dimethyl ether and syngas over combined h-mordenite and cu/zno catalysts, ChemSusChem 3 (2010) 1192–1199. [8] P. Cheung, A. Bhan, G.J. Sunley, E. Iglesia, Selective carbonylation of dimethyl ether to methyl acetate catalyzed by acidic zeolites, Angew. Chem. Int. Ed. 45 (2006) 1617–1620. [9] P. Cheung, A. Bhan, G.J. Sunley, D.J. Law, E. Iglesia, Site requirements and elementary steps in dimethyl ether carbonylation catalyzed by acidic zeolites, J. Catal. 245 (2007) 110–123. [10] G.H. Yang, X.G. San, N. Jiang, Y. Tanaka, X.G. Li, Q. Jin, K. Tao, F.Z. Meng, N. Tsubaki, A new method of ethanol synthesis from dimethyl ether and syngas in a sequential dual bed reactor with the modified zeolite and Cu/ZnO catalysts, Catal. Today 164 (2011) 425–428. [11] D. Wang, G.H. Yang, Q.X. Ma, Y. Yoneyama, Y.S. Tan, Y.Z. Han, N. Tsubaki, Facile solid-state synthesis of Cu–Zn–O catalysts for novel ethanol synthesis from dimethyl ether (DME) and syngas (CO + H2), Fuel 109 (2013) 54–60. [12] P. Lu, G.H. Yang, Y. Tanaka, N. Tsubaki, Ethanol direct synthesis from dimethyl ether and syngas on the combination of noble metal impregnated zeolite with Cu/ZnO catalyst, Catal. Today 232 (2014) 22–26.
124
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
[13] S.R. Wang, W.W. Guo, L.J. Zhu, H.X. Wang, K.Z. Qiu, K.F. Cen, Methyl acetate synthesis from dimethyl ether carbonylation over mordenite modified by cation exchange, J. Phys. Chem. C 119 (2015) 524–533. [14] R. Bartex, M.N. Habin, W.R. Pertzer, Cobalt-catalyzed reductive carbonylation of dimethyl ether, J. Mol. Catal. 33 (1985) 245–258. [15] M. Audemar, C. Ciotonea, K.D.O. Vigier, S. Royer, A. Ungureanu, B. Dragoi, E. Dumitriu, F. Jérôme, Selective hydrogenation of furfural to furfuryl alcohol in the presence of a recyclable cobalt/sba-15 catalyst, ChemSusChem 8 (2015) 1885–1891. [16] D. Srimani, A. Mukherjee, A.F.G. Goldberg, G. Leitus, Y.D. Posner, L.J.W. Shimon, Y.B. David, D. Milstein, Cobalt-catalyzed hydrogenation of esters to alcohols: unexpected reactivity trend indicates ester enolate intermediacy, Angew. Chem. Int. Ed. 54 (2015) 1–5. [17] Y. Oliveira, Selective hydrogenation of methyl oleate into unsaturated alcohols in the presence of cobalt–tin supported over zinc oxide catalysts, J. Catal. 204 (2001) 230–237. [18] X.B. Li, S.R. Wang, Y.Y. Zhu, G.H. Yang, P.J. Zheng, DFT study of bio-oil decomposition mechanism on a Co stepped surface: acetic acid as a model compound, Int. J. Hydrogen Energy 40 (2015) 330–339. [19] K. Honkala, A. Hellman, I.N. Remediakis, A. Logadottir, A. Carlsson, S. Dahl, C.H. Christensen, J.K. Nørskov, Science 307 (2005) 555–558. [20] G. Kresse, J. Furthmuller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a planewave basis set, J. Comp. Mater. Sci. 6 (1996) 15–50. [21] G. Kresse, J. Furthmuller, Efficient iterative schemes for abinitio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169–11186. [22] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals, Phys. Rev. B 47 (1993) 558–561. [23] W.J. Luo, A. Asthagiri, Density functional theory study of methanol steam reforming on co(0001) and co(111) surface, J. Phys. Chem. C 118 (2014) 15274–15285. ˜ O’shea, I.P.R. de Moreira, A. Roldán, F. Illas, ELectronic and [24] V.A. de la Pena ˜ ’, and á - phase from density magnetic structure of bulk cobalt: the n, functional theory calculations, J. Chem. Phys. 133 (2010) 024701. [25] G. Henkelman, B.P. Uberuaga, H. Jonsson, A climbing image nudged elastic band method for finding saddle points and minimum energy paths, J. Chem. Phys. 113 (2000) 9901–9904.
[26] G. Henkelman, H. Jonsson, Improved tangent estimate in the nudged elastic band method for finding minimum energy paths and saddle points, J. Chem. Phys. 113 (2000) 9978–9985. [27] M.R. Li, J. Chen, G.C. Wang, Reaction mechanism of ethanol on model cobalt catalysts: DFT calculations, J. Phys. Chem. C. 120 (2016) 14198–14208. [28] M.R. Li, G.C. Wang, Ethanol reaction on Co3 O4 (110) and Co(0001): a microkinetic model analysis, J. Phys. Chem. C 120 (2016) 28110–28124. [29] J.C. Wang, Z.X. Liu, R.G. Zhang, B.J. Wang, Ethanol Synthesis from Syngas on the Stepped Rh(211) Surface: Effect of Surface Structure and Composition, J. Phys. Chem. C 118 (2014) 22691–22701. [30] X.B. Li, S.R. Wang, Y.Y. Zhu, C. Lv, G.H. Yang, Density functional theory and microkinetic studies of bio-oil decomposition on a cobalt surface: formic acid as a model compound, Energy Fuels (2017), http://dx.doi.org/10.1021/acs. energyfuels.6b03005. [31] L. Xu, Y. Xu, Activation of methyl acetate on Pd(111) surface, Surf. Sci. 604 (2010) 887–892. [32] R.G. Zhang, G.R. Wang, B.J. Wang, Insights into the mechanism of ethanol formation from syngas on Cu and an expanded prediction of improved Cu-based catalyst, J. Catal. 305 (2013) 238–255. [33] H.Y. Zheng, R.G. Zhang, Z. Li, B.J. Wang, Insight into the mechanism and possibility of ethanol formation from Syngas on Cu(100) surface, J. Mol. Catal. A: Chem. 404 (2015) 115–130. [34] F. Studt, F.A. Pedersen, Q.X. Wu, A.D. Jensen, B. Temel, J.D. Grunwaldt, J.K. Nørskov, CO hydrogenation to methanol on Cu–Ni catalysts: theory and experiment, J. Catal. 293 (2012) 51–60. [35] Y.Z. Yang, X.Z. Qi, X.X. Wang, D. Lv, F. Yu, L.S. Zhong, H. Wang, Y.H. Sun, Deactivation study of CuCo catalyst for higher alcohol synthesis via syngas, Catal. Today 270 (2016) 101–107. [36] V.R. Surisetty, A.K. Dalai, J. Kozinski, Intrinsic reaction kinetics of higher alcohol synthesis from synthesis gas over a sulfided alkali-promoted Co-Rh-Mo trimetallic catalyst supported on multiwalled carbon nanotubes (MWCNTs), Energy Fuels 24 (2010) 4130–4137. [37] T. Toyoda, T. Minami, E.W. Qian, Mixed alcohol synthesis over sulfided molybdenum-based catalysts, Energy Fuels 27 (2013) 3769–3777. [38] Y.M. Choi, P. Liu, Mechanism of Ethanol Synthesis from Syngas on Rh(111), J. Am. Chem. Soc. 131 (2009) 13054–13061.
Contents lists available at ScienceDirect
Molecular Catalysis journal homepage: www.elsevier.com/locate/mcat
Editor’s choice paper
Density functional theory study of ethanol synthesis from dimethyl ether and syngas over cobalt catalyst Xinbao Li a , Shurong Wang b,∗ , Yingying Zhu a , Geng Chen a , Guohua Yang a a b
Faculty of Maritime and Transportation, Ningbo University, Ningbo, 315211, PR China State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou 310027, PR China
a r t i c l e
i n f o
Article history: Received 2 July 2016 Received in revised form 26 January 2017 Accepted 9 February 2017 Keywords: Density functional theory Ethanol Dimethyl ether Syngas Carbonylation
a b s t r a c t The reaction mechanism of ethanol synthesis from dimethyl ether (DME) and syngas was studied via density functional theory calculations. Various possible pathways for ethanol formation, and byproduct formations of methanol, acetic acid, methane, carbon dioxide, and water over an active cobalt stepped surface were calculated. The most favorable pathway for ethanol synthesis starts with the dissociation of dimethyl ether to CH2 , followed by carbon monoxide insertion to form CH2 CO, and then undergoes successive hydrogenations to give ethanol. CH3 CHO hydrogenation to CH3 CHOH becomes the rate–determining step with a reaction barrier of 1.48 eV. DME decomposition and CH2 carbonylation occur easily with low barriers of 0.61 eV and 0.48 eV, respectively. Carbon monoxide insertion into CH2 is more facile than into CH3 and CH. Hydrogenation at the carbon atom occurs prior to the oxygen atom in the order of ␣–carbon > carbonyl carbon > oxygen. The calculations demonstrate that ethanol synthesis via DME carbonylation and hydrogenation is thermodynamically favored and is kinetically faster than that via syngas direct synthesis. Methane and acetic acid are two dominant competing byproducts. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Biomass–derived ethanol (CH3 CH2 OH) has been recognized as an alternative fuel to traditional fossil fuels since that its raw material, biomass, is renewable and abundant on the earth. The increasing utilization of ethanol as motor fuel will reduce CO2 emission. Because of the easy and low cost production of syngas via biomass gasification, direct CH3 CH2 OH synthesis from syngas (CO + H2 ) has been studied widely [1–4]. Four categories of heterogeneous catalysts have been tested and optimized: modified Fischer–Tropsch synthesis catalysts (cobalt–based or iron–based), modified methanol synthesis catalysts (copper–based), modified molybdenum–based catalysts, and rhodium–based catalysts [4]. However, drawbacks of severe reaction conditions, low one pass CO conversion, and low CH3 CH2 OH selectivity has inhibited its industrial application. Recently, an improved method for CH3 CH2 OH synthesis from dimethyl ether (DME) and syngas has been proposed [5–7]. The synthesis involves the simultaneous or stepwise reactions of DME vapor–phase carbonylation and hydrogenation. It is considered a promising technology because of the high reaction activity and the
∗ Corresponding author. E-mail address: [email protected] (S. Wang). http://dx.doi.org/10.1016/j.mcat.2017.02.014 2468-8231/© 2017 Elsevier B.V. All rights reserved.
efficient production of DME and syngas feedstock from renewable biomass. Iglesia and coworkers found that HMOR and HFER catalyzed DME carbonylation to methyl acetate had stable rates and more than 99% selectivity at 423 463 K, and the rate of DME carbonylation was much higher than that of CH3 OH carbonylation [8,9]. Tsubaki and coworkers reported that DME was converted completely with a CH3 CH2 OH selectivity of 48.4% in the one step CH3 CH2 OH synthesis from DME, CO, and H2 over affordable HMOR and Cu/ZnO dual catalysts [5–7,10,11]. Lu et al. used different noble metals (Pt, Pd, and Ru) to impregnate HMOR and to promote the catalytic activity of CH3 CH2 OH synthesis from DME and syngas, and they found that the combination of 0.4 wt% Pt/HMOR and Cu/ZnO catalyst presented excellent catalytic performance [12]. We have studied DME carbonylation over metal modified HMOR catalysts (Cu, Ni, Co, Zn and Ag), and found that the catalytic performances of Co, Cu, and Ni modified HMOR were better than the performance of HMOR [13]. The activity of Co was comparable with that of Cu in DME conversion (close to 100%) and methyl acetate selectivity (higher than 99%), which shows the effectiveness of Co for DME carbonylation. Bartek et al. reported that CH3 CH2 OH/acetaldehyde (CH3 CHO) selectivities were promoted significantly when Co based catalysts were used in the reductive carbonylation of DME [14]. Co is highly active for ester and aldehyde hydrogenation to produce alcohols [15–17]. Co catalysts are
116
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
considered as promising candidates for CH3 CH2 OH synthesis from DME and syngas via synergic carbonylation and hydrogenation. In this work, we study the mechanism of CH3 CH2 OH synthesis from DME and syngas over a Co catalyst via density functional theory (DFT) calculations. A stepped surface was used in the calculation because it is recognized as the most active site of transition metals for heterogeneous catalytic reactions [18,19]. Adsorptions of key intermediates were firstly calculated to identify their preferred surface adsorption geometries. Then the elementary reactions for DME decomposition, subsequent CO insertion into CHx (x = 1–3) and successive hydrogenation to CH3 CH2 OH, and the competing reactions of direct CO hydrogenation to CH3 OH and CH3 CH2 OH were calculated. Acetic acid (HOAc), CO2 , water (H2 O), and methane (CH4 ) byproduct formations were also considered. Finally, the most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 were identified. To the best of our knowledge, this is the first theoretical study that presents thermodynamic and kinetic reaction barriers of CH3 CH2 OH synthesis from DME and syngas.
2. Computational Methods All spin–polarized periodic DFT calculations were carried out using the Vienna Ab initio Simulation Package (VASP) [20–22], and the Perdew–Burke Ernzerhof (PBE) generalized gradient correction (GGA) was used to treat exchange and correlation effects. The core electrons of all atoms were described by the projector–augmented wave pseudopotential, and the energy cutoff was 400 eV. Monkhorst Pack k–point mesh of 6 × 6 × 6 and 3 × 3 × 1 for bulk and surface calculations were used to sample the Brillouin zone. The Fermi level was smeared using the Methfessel Paxton method with a width of 0.1 eV. The electronic self consistent field converged to 1 × 10−4 eV and forces on all atoms converged to less than 0.05 eV/Å. A fcc Co close-packed unit cell with an original lattice constant of 3.544 Å was selected and optimized by a conjugate-gradient algorithm. The calculated equilibrium bulk lattice constant for Co is 3.524 Å, which agrees well with theoretical (3.521 Å) and experimental values (3.545 Å) [23,24]. Co steps were modeled by using a four layer thick slab of 3 × 3 Co(111) surface supercell, from which two closed packed rows were removed to form a step. The atoms of the bottom two layers were fixed in the bulk positions, whereas the others were allowed to relax to their lowest energy configurations. A vacuum space of 15 Å was applied to avoid the interaction between the slab and its images. The side and top views of the model Co stepped surface are shown in Fig. 1. The transition states (TS) were determined by the climbing nudged elastic band method (CINEB) [25,26]. This method gives a precise estimate of the saddle point with a small modification of the regular efficient NEB method. It has been widely used for TS searching and the minimum energy path finding [18,23,27–30]. In our calculation, an elastic band with 8 structural images along the path of the initial state (IS) and the final state (FS) of each elementary reaction were constructed by linear interpolation. A spring interaction between adjacent images was added to ensure continuity of the path, and optimized using quick-min algorithm. After being converged, the image with the highest energy (TS) was not affected by the spring forces at all. The TS structures were further confirmed by vibrational frequency analysis and determining that only a single imaginary frequency presented along the proper reaction coordinate. All vibrational frequencies were calculated by allowing the entirety of the adsorbed species on the surface to fully relax while all the Co atoms were kept fixed. The reaction rate constants were calculated using the harmonic transition-state theory. The adsorption energy Eads is defined as Eads = Eads/slab − Eslab − Egas , where Eads/slab is the total energy of the slab with
adsorbates, Eslab is the slab energy, and Egas is the energy of adsorbates in the gas phase. The reaction energy (H) and activation energy (Ea ) are defined as H = EFS − EIS , and Ea = ETS − EIS , respectively. EIS , ETS , and EFS are the total energies of the IS, TS, and FS, respectively.
3. Results 3.1. Adsorptions of Intermediates Adsorption geometries of some key reaction intermediates on the Co stepped surface are presented in Fig. 2. The corresponding adsorption geometrical parameters, configurations, and adsorption energies are listed in Table 1. The nomenclature i –j was used to designate i atoms of the adsorbate bound to j atoms on the metal surface. Saturated species, dimethyl ether (CH3 OCH3 ), CH3 OH, and CH3 CH2 OH, have small adsorption energies that range from −0.42 to −0.62 eV, and which demonstrate weak binding to the surface. The adsorptions are induced by the lone pair electrons of O atoms in those adsorbates, forming 1 –1 (O) configurations with an O–s Co distance of 2.09–2.14 Å (Figs. 2(a–c)). For CH3 CHO and HCHO (Fig. 2(d,e)), the adsorption energy increases to −1.42 and −1.59 eV, respectively, and give stronger binding to the surface than that of CH3 OCH3, CH3 OH, and CH3 CH2 OH. This can be ascribed to a relatively strong d– interactions between the d orbital of s Co atoms and the bond of C O in CH CHO and HCHO. Both 3 2 –2 (C,O) configurations were generated with an O atom located at the bridge site and a carbonyl C atom located at the top site of the s Co atoms. The O–s Co distances are shortened by ∼0.15 Å compared with those of CH3 OCH3, CH3 OH, and CH3 CH2 OH species. Unsaturated species bind strongly to the surface through stepped Co atoms with adsorption energies that range from −2.02 to −4.00 eV, which are larger than the saturated species. The CH3 OCH species is the most strongly adsorbed intermediates on the surface with binding energy of −4.00 eV. The corresponding configuration is 1 –2 (C). With the elimination of hydrogen atoms and a decreasing symmetry for CH3 OCH3 decomposition intermediates (Figs. 2(a, f–h)), the adsorption sites change from oxygen to carbon atoms and the corresponding adsorption configurations follow 1 –1 (O) → 2 –2 (C,O) → 1 –2 (C). The reason may be assigned to the favorable formation of a tetrahedral bonding geometry for carbon atoms on the surface [31]. The adsorption of COH on the surface is strong. The C atom located at the bridge site is linked to two s Co atoms on the stepped surface, giving a 1 –2 (C) configuration (Fig. 2(n)). The corresponding adsorption energy is −3.84 eV with dC– s Co of 1.81 and 1.82 Å. The alkoxide species, CH3 O and CH3 CH2 O are also two strong adsorption intermediates, which have the binding energies of −3.55 and −3.52 eV, respectively. The resulting binding configurations are 1 –2 (O) (Figs. 2(i) and (o)). The binding atoms change from carbon to oxygen compared with COH. The length of O s Co bonds for CH3 O and CH3 CH2 O are 1.92 and 1.93 Å, respectively, which are 0.11 Å longer than that for dC– s Co in COH. Both CH3 CHOH and CH2 OH have the adsorption configurations of 2 –2 (C,O) through their unsaturated C atoms and hydroxyl O atoms on the surface, while CHOH is linked to the surface through its C atom only to give a 1 –2 (C) configuration. The calculated configurations are in good agreement with the previous DFT results [29,32]. It can be ascribed to that with the increase of the valence saturation of C atom (the maximum is 4) in adsorbates, the adsorption induced by C atom becomes weaker and weaker, while that induced by O atom becomes dominant. The O and unsaturated C atoms are competitive adsorption on the surface, with C being pre-
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
117
Fig. 1. Side and top views of the model Co stepped surface used in this study. Dotted red line in the top view indicates the unit cell. s Co and f Co represent the stepped and flat atoms, respectively.
Fig. 2. Side and top view of preferred adsorption intermediates. Blue, brown, red, and white spheres represent Co, C, O, and H, respectively.
ferred. Here, the saturation of active C in CH3 CHOH and CH2 OH are 3, and that for CHOH is 2. Similarly, for CH3 OCH2 and CH3 OCH, the saturations of active C (in CH2 and CH groups) are 3 and 2, respectively. Therefore, CH3 OCH2 binds to the surface through its C and O atoms together to form 2 –2 (C,O) configuration, while CH3 OCH binds to the surface through C atom only to form 1 –2 (C) configuration. It also
happens for CH3 OC. On the other hand, as discussed above, the species with completely saturated C, like CH3 OCH3 , CH3 CH2 OH, and CH3 OH, are bonding to the surface through the atomic O only. Given that, it can be concluded that with atomic C saturation, in general, the adsorption configuration changes from C binding to C O binding, and finally to O binding.
118
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Table 1 Adsorption energies, configurations, and geometrical parameters of reaction intermediates on a Co stepped surface. Species
Eads (eV)
Configurations
CH3 OCH3 CH3 OH CH3 CH2 OH CH3 CHO HCHO CH3 OCH2 CH3 OCH CH3 OC CH3 O CH2 OH CHOH CHO COH CH3 CH2 O CH3 CHOH
−0.42 −0.62 −0.58 −1.42 −1.59 −2.47 −4.00 −2.33 −3.55 −2.15 −3.27 −2.64 −3.84 −3.52 −2.02
1 –1 (O) 1 –1 (O) 1 –1 (O) 2 –2 (C,O) 2 –2 (C,O) 2 –2 (C,O) 1 –2 (C) 1 –2 (C) 1 –2 (O) 2 –2 (C,O) 1 –2 (C) 2 –2 (C,O) 1 –2 (C) 1 –2 (O) 2 –2 (C,O)
Geometrical parameters (Å) dC– s Co
3.2. Elementary Reactions In this section, we investigated each elementary reaction involved in DME decomposition; subsequent carbonylation and hydrogenation for CH3 OH and CH3 CH2 OH synthesis; and HOAc, CH4 , CO2 , and H2 O byproduct formations. Potential energy surfaces for the most likely reaction pathways are discussed and depicted. The obtained geometries of all TS and their imaginary frequencies are presented in the Supporting Information.
3.2.1. DME Decomposition A schematic of DME decomposition on a Co stepped surface is presented in Fig. 3. Activation energies, reaction energies, and geometries of transition states are also provided. DME (CH3 OCH3 ) decomposition starts from dehydrogenation to give CH3 OCH2 with a low activation energy of 0.48 eV. The reaction is exothermic by −0.25 eV. The cleaving O H bond in TS1 is elongated from 1.11 Å to 1.38 Å. CH3 OCH2 has two compatible dissociation reactions: O CH2 scission to give CH3 O and CH2 with a barrier of 0.61 eV, and dehydrogenation from the CH2 group to give CH3 OCH with a barrier of 0.58 eV. Both low reaction barriers demonstrate the readily formation of CH2 and CH3 OCH intermediates. The energy barrier difference for those two reactions is as low as 0.03 eV. The corresponding reaction heat for O CH2 scission is strongly exothermic by −0.80 eV, and that for dehydrogenation is slightly endothermic by 0.12 eV, showing a large difference. As a result, CH3 OCH2 dissociation via O CH2 scission to give CH2 is thermodynamically and kinetically favored. The C O bond is elongated significantly by 0.51 Å to 2.12 Å in TS2. For the dehydrogenation, the O H bond is elongated by 0.52 Å to 1.63 Å in TS3. As a CH3 OCH2 dehydrogenation product, CH3 OCH might undergo decomposition via O CH cleavage to give CH3 O and CH (Ea = 0.51 eV) and/or dehydrogenation at CH to give CH3 OC and H (Ea = 0.31 eV). Both reactions are exothermic, with −1.23 eV versus −0.22 eV, respectively. The dehydrogenation is slightly kinetically preferred, whereas the O CH cleavage is strongly thermodynamically favored. The following decomposition of CH3 OC via CH3 O C scission has a reaction barrier of 0.63 eV. Based on the above results of kinetic reaction barriers and thermodynamic reaction energies for possible elementary reactions of DME decomposition, it is clear that DME can be converted easily to give CH2 and CH via C O scissions after dehydrogenation on the Co stepped surface. CH2 formation is even preferred due to its low reaction barrier and strongly exothermic in the early step of DME decomposition.
1.98 1.96 1.98 1.94, 1.93 1.81, 1.83 1.98 1.95, 1.93 1.87, 2.10 1.81, 1.82 1.98
dO– s Co 2.14 2.09 2.10 1.97, 1.95 1.97, 1.94 1.98
1.92, 1.92 2.03 1.97 1.93, 1.93 2.01
3.2.2. Ethanol and Methanol Synthesis The reaction network of CH3 CH2 OH and CH3 OH synthesis via carbonylation and hydrogenation from DME and CO is shown in Fig. 4. Adsorbed H2 is easily dissociated on the Co surface with an extremely low reaction barrier of 0.03 eV, supplying adequate H intermediates for the hydrogenation reactions. H2 dissociation is also thermodynamically favored with a mild exothermicity of −0.55 eV. Intermediates of COH and CHO are two products of CO hydrogenation at its O and C atoms, respectively. Both reactions are strongly endothermic (1.54 and 0.91 eV, respectively). COH formation is energetically unlikely because of its extremely high barrier of 2.13 eV. The energy barrier for CHO formation is 0.93 eV, which is 1.20 eV lower than that for COH formation, and indicates that the first hydrogenation of CO at the C atom to give CHO is preferred rather than the hydrogenation at the O atom to give COH. This agrees well with published results of CO hydrogenation to CHO and COH over Cu and Rh surfaces [29,33,34]. The reported activation energy of COH formation (1.89–2.07 eV) was found to be much higher than that of CHO formation (0.88–1.26 eV). Similarly, for the further hydrogenation of CHO, the addition of H to C atom to form HCHO is more favorable than the addition to O atom to form CHOH, since the barrier for the former is 0.72 eV lower than the later (0.38 eV versus 1.10 eV). HCHO formation is almost thermoneutral. In addition, it is interesting to find that the reaction barrier for CHO deoxygenation to CH is relatively low by 0.63 eV, which is a small 0.25 eV higher than HCHO formation, suggesting the easy deoxygenation of CHO. This reaction is mildly exothermic by −0.38 eV. CHO hydrogenation to HCHO and deoxygenation to CH are two compatible and preferable reactions for CHO conversion. The intermediate of HCHO may undergo three reactions: two hydrogenations at the C and O atoms to form CH3 O and CH2 OH, respectively, and one deoxygenation to form CH2 . The corresponding activation energies are 0.25, 1.04, and 0.85 eV, respectively, in which CH3 O formation is energetically preferred. CH3 O and CH2 formations are exothermic, whereas the formation of CH2 OH is endothermic. The subsequent hydrogenation of CH3 O results in CH3 OH formation. However, the reaction has a high activation energy of 1.67 eV and a strong endothermicity of 1.03 eV, which implies the unlikely formation of CH3 OH and the relative stability of CH3 O species on the surface. This is consistent with a strong binding of CH3 O to the surface. The barrier for CH3 O deoxygenation to CH3 is high to 1.46 eV, which indicates that CH3 production from CO is difficult. The formation of CH3 OH via sequent hydrogenation starts from CHOH at an unsaturated C atom has a relatively low reaction barrier (0.40, 0.80 eV) and is nearly thermoneutral. Consequently, for direct CO hydrogenation to CH3 OH, it is clear that the
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
119
Fig. 3. Schematic of DME decomposition and snapshots of transition states (eV).
hydrogenation at carbon atom is energetically preferred over the hydrogenation at oxygen atom. Studt et al. also found that hydrogenation of the oxygen end involved intermediates of COH, HCOH, and H2 COH, is associated with much higher barriers, and that all intermediates are higher in energy than their carbon hydrogenated counterparts [34]. According to the above discussion of direct CO hydrogenation, intermediates of CH, CH2 , and CH3 are produced by direct deoxygenation of CHO, CH2 O, and CH3 O with reaction barriers of 0.63, 0.85, and 1.46 eV, respectively. All these three deoxygenation reactions are exothermic. CH2 and CH are also predominant products of DME decomposition via C O scissions with relatively low reaction barriers (∼0.60 eV). To produce CH3 CH2 OH, subsequent CO insertions into these CHx (x = 1 − 3) are essential and vital. CO insertions of CH, CH2 , and CH3 to form CHCO, CH2 CO, and CH3 CO have activation energies of 0.64, 0.48, and 1.11 eV with an increasing endothermicity of 0.03, 0.29, and 0.54 eV, respectively. The formation of CH2 CO has the lowest reaction barrier, which indicates that the insertion of CO into CH2 is most likely to occur. In addition to this most preferred reaction, CHCO formation is an energetically close reaction because of a slight barrier difference of 0.16 eV. Compared with CH2 CO and CHCO, CH3 CO formation has a much higher reaction barrier, which shows that it is more difficult to produce directly by CO insertion of CH3 . However, CH3 CO is favorable to be generated by the hydrogenation at the ␣–carbon of CHCO and CH2 CO. The reaction barriers are relatively low at 0.49 and 0.29 eV for CHCO hydrogenation to CH2 CO and CH2 CO hydrogenation to CH3 CO, respectively. These reactions are exothermic by −0.42 and −0.23 eV, respectively, which implies that both reactions are kinetically and thermodynamically preferred. With respect to CO insertion, competing reactions for CHx conversion are their direct hydrogenations. The highest reaction barrier is 0.85 eV, which is associated with hydrogenation of CH3 to form CH4 . This reaction has a mild endothermicity of 0.33 eV. The length of forming C H bond is 1.57 Å at TS, consistent with the
reported DFT result (1.54 Å) [27]. The lowest barrier in CHx hydrogenation is CH2 hydrogenation to CH3 (Ea = 0.28 eV) which is mildly exothermic by −0.37 eV. The barrier for CH hydrogenation to CH2 is 0.30 eV with a mild exothermicity of −0.56 eV. Wang et al. carried out DFT calculations of ethanol decomposition on Co(0001) surface, which can be treated as a reverse reaction of ethanol formation, and found that the barriers for CH2 , CH3 , and CH4 formations were 0.44, 0.46, and 0.99 eV, which are similar to our results [27]. The relatively low reaction barriers obtained in these CHx hydrogenation reactions imply the readily formation of CH4 . This result agrees with experimental observations of alcohol synthesis from syngas, for which CH4 is recognized as an easily generated byproduct [35–37]. In addition to hydrogenation at the ␣–carbon, another favored hydrogenation site for CHCO, CH2 CO, and CH3 CO is the carbonyl carbon. The reactions for CH2 CO hydrogenation to CH2 CHO and CH3 CO hydrogenation to CH3 CHO have relatively low barriers of 0.48 and 0.31 eV, respectively. Both reactions are exothermic (−0.49 and −0.24 eV, respectively). Therefore, hydrogenation at the carbonyl carbon is kinetically and thermodynamically preferable for CH2 CO and CH3 CO. However, the hydrogenation of CHCO to CHCHO is not favored because of its relatively high reaction barrier of 1.02 eV compared with the hydrogenations of CHCO and CH2 CO. Although the formation of CHCHO is difficult, its transformation via hydrogenation is easy. CHCHO undergoes a two–step hydrogenation to CH3 CHO through CH2 CHO with low activation energies of 0.30 eV (CHCHO + H → CH2 CHO) and 0.41 eV (CH2 CHO + H → CH3 CHO). Sites for H addition are both ␣–carbon. Consequently, CH3 CHO intermediates are likely generated through CO insertions into CH and CH2 followed by hydrogenations. Further hydrogenation of CH3 CHO will produce two intermediates of CH3 CH2 O and CH3 CHOH with the reaction barriers of 0.30 and 1.48 eV, respectively. The formation of CH3 CH2 O is exothermic by −0.36 eV, whereas CH3 CHOH formation is endothermic by 0.73 eV. Thus, hydrogenation at the carbonyl carbon atom of CH3 CHO to CH3 CH2 O is more likely than hydrogenation at the oxygen atom to CH3 CHOH.
120
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Fig. 4. Reaction network of CH3 CH2 OH and CH3 OH synthesis via carbonylation and hydrogenation (eV).
However, CH3 CH2 O hydrogenation at the oxygen atom to CH3 CH2 OH is both kinetically and thermodynamically unfavored due to its extremely high reaction barrier of 1.73 eV and its strongly endothermicity of 1.05 eV. The reason may be ascribed to the extremely strong adsorption of CH3 CH2 O species (–3.52 eV) on the surface, which will suppress its reaction with vicinal H atoms. This is consistent with the calculation carried out by Wang et al., in which the reaction of CH3 CH2 O + H → CH3 CH2 OH also has a high barrier of 1.41 eV and is endothermic by 0.62 eV [27]. The length of changing O H bond at TS in our calculation on Co stepped surface is 1.40 Å, the same as that on Co(0001) surface [27]. In contrast with CH3 CH2 O hydrogenation, hydrogenation of CH3 CHOH at the ␣–carbon to form CH3 CH2 OH is energetically facilitated with a relatively low activation energy of 0.66 eV and an exothermicity of −0.17 eV. As a result, the highest reaction barriers for CH3 CH2 OH formation from CH3 CHO mediated CH3 CHOH and CH3 CH2 O are 1.48 eV (CH3 CHO + H → CH3 CHOH) versus 1.73 eV (CH3 CH2 O + H → CH3 CH2 OH), which suggests that CH3 CHOH mediated path for CH3 CH2 OH formation is energetically preferred over
that of CH3 CH2 O mediated path. Liu et al. suggested that ethanol synthesis from syngas on Rh(111) surface goes through intermediates of CH3 CHOH [38]. Zuo et al. also reported that the route for CH3 CH2 OH formation from syngas on the Cu/ZnO catalyst involved the vital steps of CH3 CHO → CH3 CHOH → CH3 CH2 OH [3]. Another possible route for CH3 CH2 OH formation with CH3 CO hydrogenation through CH3 COH and CH3 CHOH was also considered. The highest activation energy in this route is 1.42 eV (CH3 CO + H → CH3 COH), which is 1.11 eV higher than that of CH3 CO + H → CH3 CHO. The reaction is endothermic by 0.46 eV. Subsequent hydrogenation of CH3 COH to form CH3 CHOH has low activation energy of 0.47 eV, associated with a slight endothermicity of 0.23 eV.
3.2.3. Acetic Acid Synthesis Reaction networks for the conversion of O species and the formation of dominant byproducts, HOAc, are presented in Fig. 5. According to the above analysis, O intermediates can be produced easily from CO by direct deoxygenation of CHO and HCHO
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
121
Fig. 5. Reaction network of O species conversion and HOAc formation (eV).
with low activation energies of 0.63 eV (CHO → O + CH) and 0.85 eV (HCHO → O + CH2 ), respectively. By reacting with H, O will be converted to OH with a high activation energy of 1.04 eV and a mild exothermicity of −0.42 eV. Meanwhile, a parallel reaction of O with CO to give CO2 (Ea = 1.05 eV) will occur because of its close reaction barrier to that of OH formation. This reaction is endothermic by 0.27 eV. OH intermediates undergo further hydrogenation to H2 O and CO insertion to COOH with high activation energies of 1.68 eV and 1.87 eV, respectively. Both reactions are strongly endothermic by more than 1 eV. Therefore, OH conversions with H and CO are not preferred. O and OH intermediates are anticipated to carry out oxidation reactions to form carboxylic acid HOAc. The oxidation reactions here are supposed to start from CHCO, CH2 CO, and CH3 CO, which are identified as active intermediates in the reaction network of CH3 CH2 OH synthesis. All three intermediates may undergo direct addition of OH or two–step additions of O and H. Formations of CHCOOH, CH2 COOH, and CH3 COOH via direct OH addition of CHCO, CH2 CO, and CH3 CO have activation energies of 1.10, 0.41, and 0.99 eV, respectively. Among these three reactions, CH2 COOH formation (CH2 CO + OH → CH2 COOH) is kinetically preferred because of its relatively low reaction barriers. The reactions of CHCOOH and CH3 COOH formations are mildly endothermic by 0.54 and 0.50 eV, respectively, whereas that for CH2 COOH formation is exothermic by −0.49 eV. As a consequence, CH2 COOH formation is also thermodynamically facilitated. Activation energies for O additions into CHCO and CH3 CO at carbonyl carbon atoms to generate CHCOO and CH3 COO are 0.98 and 0.73 eV, respectively. Whereas the addition of O into CH2 CO to form CH2 COO has a low activation energy of 0.40 eV, which is 0.58 and 0.33 eV lower than the other two reactions. This reaction is mildly exothermic by −0.42 eV. Consequently, oxidations of CH2 CO to CH2 COO and/or CH2 COOH are more likely than oxidations of CHCO and CH3 CO. Reaction barriers for the further hydrogenations of CHCOO and CH2 COO to CHCOOH and CH2 COOH are 0.75 and 0.61 eV, respectively, implying an energetic preference of CH2 COO + H → CH2 COOH. The hydrogenation of CH3 COO to
CH3 COOH has an extremely high reaction barrier of 1.62 eV, which is strongly endothermic by 1.05 eV and makes the transformation of CH3 COO to CH3 COOH difficult. The barrier for CH3 COOH formation via CH2 COOH hydrogenation is 0.34 eV lower than that of via CH3 COO hydrogenation. Given that, we conclude that CH3 COOH formation is likely starting from CH2 CO oxidation rather than that from CHCO and CH3 CO oxidations. Because of the high activation barrier of OH formation (O + H → OH), CH2 CO oxidation with O is more favored than that with OH. 4. Discussion 4.1. Preferred Pathways for Ethanol Synthesis Four preferred reaction pathways for CH3 CH2 OH synthesis via DME carbonylation and hydrogenation on a Co stepped surface are shown in Fig. 6. As discussed in Section 3.2.1, for P1 and P2 pathways during DME decomposition, the steps with the highest reaction barriers are CH3 OCH2 → CH3 O + CH2 (Ea = 0.61 eV). After that, carbonylation occurs via CO insertion to generate an active intermediate of ketene CH2 CO. The carbonylation reaction occurs easily because of its low energy barrier of 0.48 eV. The following successive hydrogenations of CH2 CO to produce final CH3 CH2 OH in both P1 and P2 proceed through intermediates of CH3 CHO and CH3 CHOH. The highest reaction barriers during the hydrogenation are both 1.48 eV and are associated with CH3 CHOH formation (CH3 CHO + H → CH3 CHOH), which is also the highest barrier in entire P1 and P2 pathways. The difference between P1 and P2 is the starting hydrogenation site of CH2 CO. In P1, CH2 CO hydrogenation is performed at the ␣–carbon to form CH3 CO (CH2 CO + H → CH3 CO, Ea = 0.29 eV). In P2, it is performed at the carbonyl carbon to form CH2 CHO (CH2 CO + H → CH2 CHO, Ea = 0.48 eV). The barrier of the later reaction is 0.19 eV higher than that of the former reaction. CH3 CHO formations via hydrogenations of CH3 CO (P1) and CH2 CHO (P2) have energy barriers of 0.31 and 0.41 eV, respectively. As a result, it is clear that the hydrogenation of CH2 CO → CH3 CO → CH3 CHO
122
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
Fig. 6. Four preferred reaction pathways for CH3 CH2 OH synthesis via DME carbonylation and hydrogenation on a Co stepped surface.
in P1 has lower reaction barriers than the hydrogenation of CH2 CO → CH2 CHO → CH3 CHO in P2. Therefore, P1 is more kinetically favored than P2. In addition to CH2 in P1 and P2, CH is another dominant product of DME decomposition in P3 and P4. It is produced through CH3 OCH3 → CH3 OCH2 → CH3 OCH → CH. The highest barrier is 0.58 eV (CH3 OCH2 → CH3 OCH + H). This is close to the highest barrier for CH2 formation in P1 and P2 (CH3 OCH2 → CH3 O + CH2 , Ea = 0.61 eV). Subsequent carbonylation occurs via CO insertion into CH with a barrier of 0.64 eV (CH + CO → CHCO), which is 0.16 eV higher than the barrier of CH2 + CO → CH2 CO in P1 and P2. Similar to CH2 CO, CHCO prefers to be hydrogenated at the ␣–carbon to CH2 CO in P3 and P4 rather than to be hydrogenated at the carbonyl carbon. The reason is that the barrier of ␣–carbon hydrogenation (CHCO + H → CH2 CO, Ea = 0.49 eV) is much lower than that of carbonyl carbon hydrogenation (CHCO + H → CHCHO, Ea = 1.02 eV). The hydrogenations of CH2 CO to produce final CH3 CH2 OH in P3 and P4 are the same as those to P1 and P2, respectively. The highest reaction barrier in the hydrogenation is also 1.48 eV as associated with CH3 CHOH formation, which is also the highest barrier over the entire pathways. Therefore the rate–determining steps (RDSs) for the four P1–P4 pathways are all CH3 CHO + H → CH3 CHOH. Additionally, similar to P1 and P2, P3 is preferable over P4. Although the RDSs are the same and barriers for DME decomposition are energetically close, P1 has an even lower reaction barrier than P3 for the vital CO insertion reaction (0.48 eV versus 0.64 eV), which implies that P1 is further preferred over P3. Therefore, we conclude that in these four competing pathways, P1 is the most kinetically favorable pathway for CH3 CH2 OH synthesis which starts from DME decomposition, followed by carbonylation and hydrogenation on the Co stepped surface. Nevertheless, the other three pathways P2–P4 are also feasible and are likely to occur since the difference in reaction barriers between P1 and P2–P4 is less than 0.20 eV. 4.2. Preferred Pathways for Byproduct Formation As shown in Fig. 4, the kinetically preferred pathway for CH2 formation from CO is CO → CHO → CH → CH2 . The highest barrier is 0.93 eV, which is associated with the first CO hydrogenation step and is 0.32 eV higher than the highest barrier of CH2 formation in P1 (CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH). Consequently, CH2 formation via CO hydrogenation is more difficult than its formation via DME decomposition. In the CO route, the ensuing favored carbonylation and hydrogenations for CH2 to produce CH3 CH2 OH are the same as that of P1. The RDS is also CH3 CHO + H → CH3 CHOH over the entire CH3 CH2 OH synthesis pathway. Therefore, the most likely pathway for CH3 CH2 OH synthesis via direct CO hydrogenation, P(CO), is
CO → CHO → CH → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH. To compare the formation routes of CH3 CH2 OH from DME and from syngas, the forward reaction rate constants of the steps involved in the pathways of P1 and P(CO) were listed in Table 2. It is shown that the slowest reaction both in P1 and P(CO) is CH3 CHO + H → CH3 CHOH. Its rate constants are low to 8.54 × 10−3 , 2.49 × 10−1 , 4.05 × 100 , and 4.22 × 101 s−1 at 473, 523, 573, and 623 K, respectively. Therefore, this step is confirmed as the rate-determining step in CH3 CH2 OH synthesis. Additionally, as discussed above, the difference for P1 and P(CO) is the route for CH2 formation. For P1, CH2 formation is determined by CH3 OCH2 → CH3 O + CH2 . For P(CO), CH2 formation is determined by CO + H → CHO. The corresponding rate constants for those two reactions at 573 K are 3.02 × 108 and 1.51 × 104 s−1 , respectively, which shows that CH2 formation in P1 is 4 orders of magnitude faster than that in P(CO). Therefore, CH3 CH2 OH synthesis via P1 is significantly faster than that via P(CO). The most likely pathway for CH3 OH synthesis from CO is CO → CHO → HCHO → CH2 OH → CH3 OH with the highest barrier of 1.04 eV (HCHO + H → CH2 OH). Additionally, CH4 will be produced as a main byproduct according to the low energy barrier pathway of CH3 OCH3 → CH3 OCH2 → CH2 → CH3 → CH4 . The highest reaction barrier in this path is 0.85 eV associated with CH3 hydrogenation to CH4 . Besides that, the likely route for CH4 formation from CO is CO → CHO → CH → CH2 → CH3 → CH4 . The highest reaction barrier in this route is 0.93 eV with CO hydrogenation to CHO. This barrier is 0.32 and 0.08 eV higher than the barriers of CH3 OCH2 → CH3 O + CH2 and CH3 + H → CH4 , respectively. Therefore, CH4 formation from CO is less preferred than that from DME. HOAc formation is favored via CH2 CO oxidation with O species followed by hydrogenation, as discussed above. By combining an optimal route for CH2 CO formation, the most likely HOAc formation pathway is considered to be CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH2 COO → CH2 COOH → CH3 COOH. The RDS is CH2 COOH hydrogenation to CH3 COOH with a reaction barrier of 1.28 eV. This is 0.20 eV lower than the RDS barrier of P1, and indicates kinetic competition of HOAc formation with respect to CH3 CH2 OH synthesis. Fig. 7 presents a summary of the potential energy surfaces of the most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 synthesis from CO and DME. The overall CH3 CH2 OH synthesis starting from DME decomposition is exothermic by −0.67 eV, but that starting from CO hydrogenation is endothermic by 0.35 eV. Therefore, ethanol formation from DME is thermodynamically preferred and benefitted. CH3 OH synthesis is thermodynamically unfavorable because of the increasing reaction energies, whereas HOAc and CH4 synthesis commencing from DME is favored thermodynamically with overall exothermic reactions.
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
123
Table 2 Forward reaction rate constants (kf ) of the steps involved in the most likely pathways of CH3 CH2 OH synthesis from DME and CO. Temperature ranges from 473 to 623 K. Reaction
kf (s−1 ) T = 473 K
CH3 OCH3 → CH3 OCH2 + H CH3 OCH2 → CH3 O + CH2 CO + H → CHO CHO → CH + O CH + H → CH2 CH2 + CO → CH2 CO CH2 CO + H → CH3 CO CH3 CO + H → CH3 CHO CH3 CHO + H → CH3 CHOH RDS CH3 CHOH + H → CH3 CH2 OH RDS
T = 523 K
1.36 × 10 2.70 × 107 1.94 × 102 8.42 × 105 1.12 × 1010 3.34 × 107 5.99 × 109 1.81 × 109 8.54 × 10−3 1.50 × 106 9
T = 573 K
3.17 × 10 1.01 × 108 2.11 × 103 3.52 × 106 2.30 × 1010 1.06 × 108 1.22 × 1010 4.14 × 109 2.49 × 10−1 6.77 × 106 9
T = 623 K
6.41 × 10 3.02 × 108 1.51 × 104 1.15 × 107 4.16 × 1010 2.75 × 108 2.19 × 1010 8.20 × 109 4.05 × 100 2.34 × 107 9
1.16 × 1010 7.61 × 108 7.87 × 104 3.12 × 107 6.87 × 1010 6.12 × 108 3.59 × 1010 1.45 × 1010 4.22 × 101 6.65 × 107
Rate-determining step.
CH4 and HOAc are identified as two predominant competing byproducts. Acknowledgements The authors acknowledge financial support from the National Natural Science Foundation of China (Nos. 51406090 and 51406093), Ningbo Natural Science Foundation of China (No. 2014A610119), and K.C. Wong Magna Fund in Ningbo University. We are also grateful for the Special Program for Applied Research on Super Computation of the NSFC–Guangdong Joint Fund (the second phase). Appendix A. Supplementary data
Fig. 7. Potential energy surfaces of most likely reaction pathways for CH3 CH2 OH, CH3 OH, HOAc, and CH4 synthesis from DME and CO.
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.mcat.2017.02. 014. References
5. Conclusions Reaction mechanisms of efficient CH3 CH2 OH synthesis from DME and syngas were studied systematically via DFT calculations. Possible elementary reactions that involve DME decomposition, CO insertion, subsequent hydrogenation and oxidation, and competing reactions of direct CO hydrogenation on an active Co stepped surface were calculated. Preferred adsorption geometries, parameters, and adsorption energies of key reaction intermediates were obtained. The adsorption energies of saturated species CH3 OCH3 , CH3 OH, and CH3 CH2 OH are lower than those of CH3 CHO and HCHO, and are much lower than the unsaturated species. With the increase of the valence saturation of C atom, the adsorption configuration changes from C binding to C O binding, and finally to O binding. The most favorable pathway for CH3 CH2 OH synthesis from DME and syngas is CH3 OCH3 → CH3 OCH2 → CH2 → CH2 CO → CH3 CO → CH3 CHO → CH3 CHOH → CH3 CH2 OH. CH3 CHO hydrogenation to CH3 CHOH becomes the rate-determining step with an activation energy of 1.48 eV. The barriers for other steps in this pathway are less than 0.7 eV. CO insertion into CH2 to form ketene CH2 CO is more facile than the insertions into CH and CH3 . The corresponding barriers are 0.48, 0.64, and 1.11 eV, respectively. Hydrogenation at the carbon atom occurs prior to that at the oxygen atom with an order of ␣–carbon > carbonyl carbon > oxygen. The calculations demonstrate that CH3 CH2 OH synthesis via DME carbonylation and hydrogenation is thermodynamically favored and is kinetically faster than that via syngas direct synthesis.
[1] W.G. Liu, S. Wang, S.D. Wang, Effect of impregnation sequence of Ce promoter on the microstructure and performance of Ce-promoted Rh-Fe/SiO2 for the ethanol synthesis, Appl. Catal. A 510 (2016) 227–232. [2] S. G. Prieto, M.L. Beijer, M. Smith, Y. He, Z. Au, D.A. Wang, K.P. Bruce, J.J. Jong, P.E. Jongh Spivey, Design and synthesis of copper–cobalt catalysts for the selective conversion of synthesis gas to ethanol and higher alcohols, Angew. Chem. Int. Ed. 53 (2014) 6397–6401. [3] Z.J. Zuo, L. Wang, L.M. Yu, P.D. Han, W. Huang, Experimental and theoretical studies of ethanol synthesis from syngas over cuznal catalysts without other promoters, J. Phys. Chem. C 118 (2014) 12890–12898. [4] J.J. Spivey, A. Egbebi, Heterogeneous catalytic synthesis of ethanol from biomass-derived syngas, Chem. Soc. Rev. 36 (2007) 1514–1528. [5] X.G. San, Y. Zhang, W.J. Shen, N. Tsubaki, New synthesis method of ethanol from dimethyl ether with a synergic effect between the zeolite catalyst and metallic catalyst, Energy Fuels 23 (2009) 2843–2844. [6] Y. Zhang, X.G. San, N. Tsubaki, Y.S. Tan, J.F. Chen, Novel ethanol synthesis method via C1 chemicals without any agriculture feedstocks, Ind. Eng. Chem. Res. 49 (2010) 5458–5488. [7] X.G. Li, X.G. San, Y. Zhang, T. Ichii, M. Meng, Y.S. Tan, N. Tsubaki, Direct synthesis of ethanol from dimethyl ether and syngas over combined h-mordenite and cu/zno catalysts, ChemSusChem 3 (2010) 1192–1199. [8] P. Cheung, A. Bhan, G.J. Sunley, E. Iglesia, Selective carbonylation of dimethyl ether to methyl acetate catalyzed by acidic zeolites, Angew. Chem. Int. Ed. 45 (2006) 1617–1620. [9] P. Cheung, A. Bhan, G.J. Sunley, D.J. Law, E. Iglesia, Site requirements and elementary steps in dimethyl ether carbonylation catalyzed by acidic zeolites, J. Catal. 245 (2007) 110–123. [10] G.H. Yang, X.G. San, N. Jiang, Y. Tanaka, X.G. Li, Q. Jin, K. Tao, F.Z. Meng, N. Tsubaki, A new method of ethanol synthesis from dimethyl ether and syngas in a sequential dual bed reactor with the modified zeolite and Cu/ZnO catalysts, Catal. Today 164 (2011) 425–428. [11] D. Wang, G.H. Yang, Q.X. Ma, Y. Yoneyama, Y.S. Tan, Y.Z. Han, N. Tsubaki, Facile solid-state synthesis of Cu–Zn–O catalysts for novel ethanol synthesis from dimethyl ether (DME) and syngas (CO + H2), Fuel 109 (2013) 54–60. [12] P. Lu, G.H. Yang, Y. Tanaka, N. Tsubaki, Ethanol direct synthesis from dimethyl ether and syngas on the combination of noble metal impregnated zeolite with Cu/ZnO catalyst, Catal. Today 232 (2014) 22–26.
124
X. Li et al. / Molecular Catalysis 432 (2017) 115–124
[13] S.R. Wang, W.W. Guo, L.J. Zhu, H.X. Wang, K.Z. Qiu, K.F. Cen, Methyl acetate synthesis from dimethyl ether carbonylation over mordenite modified by cation exchange, J. Phys. Chem. C 119 (2015) 524–533. [14] R. Bartex, M.N. Habin, W.R. Pertzer, Cobalt-catalyzed reductive carbonylation of dimethyl ether, J. Mol. Catal. 33 (1985) 245–258. [15] M. Audemar, C. Ciotonea, K.D.O. Vigier, S. Royer, A. Ungureanu, B. Dragoi, E. Dumitriu, F. Jérôme, Selective hydrogenation of furfural to furfuryl alcohol in the presence of a recyclable cobalt/sba-15 catalyst, ChemSusChem 8 (2015) 1885–1891. [16] D. Srimani, A. Mukherjee, A.F.G. Goldberg, G. Leitus, Y.D. Posner, L.J.W. Shimon, Y.B. David, D. Milstein, Cobalt-catalyzed hydrogenation of esters to alcohols: unexpected reactivity trend indicates ester enolate intermediacy, Angew. Chem. Int. Ed. 54 (2015) 1–5. [17] Y. Oliveira, Selective hydrogenation of methyl oleate into unsaturated alcohols in the presence of cobalt–tin supported over zinc oxide catalysts, J. Catal. 204 (2001) 230–237. [18] X.B. Li, S.R. Wang, Y.Y. Zhu, G.H. Yang, P.J. Zheng, DFT study of bio-oil decomposition mechanism on a Co stepped surface: acetic acid as a model compound, Int. J. Hydrogen Energy 40 (2015) 330–339. [19] K. Honkala, A. Hellman, I.N. Remediakis, A. Logadottir, A. Carlsson, S. Dahl, C.H. Christensen, J.K. Nørskov, Science 307 (2005) 555–558. [20] G. Kresse, J. Furthmuller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a planewave basis set, J. Comp. Mater. Sci. 6 (1996) 15–50. [21] G. Kresse, J. Furthmuller, Efficient iterative schemes for abinitio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54 (1996) 11169–11186. [22] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals, Phys. Rev. B 47 (1993) 558–561. [23] W.J. Luo, A. Asthagiri, Density functional theory study of methanol steam reforming on co(0001) and co(111) surface, J. Phys. Chem. C 118 (2014) 15274–15285. ˜ O’shea, I.P.R. de Moreira, A. Roldán, F. Illas, ELectronic and [24] V.A. de la Pena ˜ ’, and á - phase from density magnetic structure of bulk cobalt: the n, functional theory calculations, J. Chem. Phys. 133 (2010) 024701. [25] G. Henkelman, B.P. Uberuaga, H. Jonsson, A climbing image nudged elastic band method for finding saddle points and minimum energy paths, J. Chem. Phys. 113 (2000) 9901–9904.
[26] G. Henkelman, H. Jonsson, Improved tangent estimate in the nudged elastic band method for finding minimum energy paths and saddle points, J. Chem. Phys. 113 (2000) 9978–9985. [27] M.R. Li, J. Chen, G.C. Wang, Reaction mechanism of ethanol on model cobalt catalysts: DFT calculations, J. Phys. Chem. C. 120 (2016) 14198–14208. [28] M.R. Li, G.C. Wang, Ethanol reaction on Co3 O4 (110) and Co(0001): a microkinetic model analysis, J. Phys. Chem. C 120 (2016) 28110–28124. [29] J.C. Wang, Z.X. Liu, R.G. Zhang, B.J. Wang, Ethanol Synthesis from Syngas on the Stepped Rh(211) Surface: Effect of Surface Structure and Composition, J. Phys. Chem. C 118 (2014) 22691–22701. [30] X.B. Li, S.R. Wang, Y.Y. Zhu, C. Lv, G.H. Yang, Density functional theory and microkinetic studies of bio-oil decomposition on a cobalt surface: formic acid as a model compound, Energy Fuels (2017), http://dx.doi.org/10.1021/acs. energyfuels.6b03005. [31] L. Xu, Y. Xu, Activation of methyl acetate on Pd(111) surface, Surf. Sci. 604 (2010) 887–892. [32] R.G. Zhang, G.R. Wang, B.J. Wang, Insights into the mechanism of ethanol formation from syngas on Cu and an expanded prediction of improved Cu-based catalyst, J. Catal. 305 (2013) 238–255. [33] H.Y. Zheng, R.G. Zhang, Z. Li, B.J. Wang, Insight into the mechanism and possibility of ethanol formation from Syngas on Cu(100) surface, J. Mol. Catal. A: Chem. 404 (2015) 115–130. [34] F. Studt, F.A. Pedersen, Q.X. Wu, A.D. Jensen, B. Temel, J.D. Grunwaldt, J.K. Nørskov, CO hydrogenation to methanol on Cu–Ni catalysts: theory and experiment, J. Catal. 293 (2012) 51–60. [35] Y.Z. Yang, X.Z. Qi, X.X. Wang, D. Lv, F. Yu, L.S. Zhong, H. Wang, Y.H. Sun, Deactivation study of CuCo catalyst for higher alcohol synthesis via syngas, Catal. Today 270 (2016) 101–107. [36] V.R. Surisetty, A.K. Dalai, J. Kozinski, Intrinsic reaction kinetics of higher alcohol synthesis from synthesis gas over a sulfided alkali-promoted Co-Rh-Mo trimetallic catalyst supported on multiwalled carbon nanotubes (MWCNTs), Energy Fuels 24 (2010) 4130–4137. [37] T. Toyoda, T. Minami, E.W. Qian, Mixed alcohol synthesis over sulfided molybdenum-based catalysts, Energy Fuels 27 (2013) 3769–3777. [38] Y.M. Choi, P. Liu, Mechanism of Ethanol Synthesis from Syngas on Rh(111), J. Am. Chem. Soc. 131 (2009) 13054–13061.