Reaction mechanism of methyl nitrite dissociation during co catalytic coupling to dimethyl oxalate:A density functional theory study☆

Chen Fan ,Man Luo ,Wende Xiao ,*

1 Key Laboratory of Advanced Control and Optimization for Chemical Processes,Ministry of Education,East China University of Science and Technology(ECUST),Shanghai 200237,China

2 School of Chemistry and Chemical Engineering,Shanghai Jiao Tong University(SJTU),Shanghai 200240,China

1.Introduction

Ethylene glycol(EG),a crucial chemical raw material with a global demand of about 25 million tons each year,is primarily used in polyester manufactures or antifreezing fluid production[1-3].At present,the universal industrial approach to produce EG is from ethylene oxidation based on petroleum resource.How ever,due to the shrink of oil resources and the continuous increase of EG demand,an alternative EG synthesis technology called coal to ethylene glycol(CTEG)has attracted substantial attention[1,4-6].CO catalytic coupling with methyl nitrite(MN)to dimethyl oxalate(DMO)is the crucial step in the conversion of inorganic C1 to organic C2 in CTEG,and thus it is considered to be one of the most important applications in C1 chemistry[5].The coupling reaction can be expressed as,

Hydrogenation of DMO produces EG.The most effective catalysts for the coupling reaction are the supported Pd catalysts.How ever,under the coupling reaction conditions,the non-catalytic(thermal)decomposition of MN in the gas phase to produce methanol(ME)and formaldehyde and the catalytic dissociation of MN on Pd catalysts to yield ME and methyl formate(MF)are two important side reactions of the coupling reaction[7,8],which can be expressed as follow s,

How ever,the detailed reaction mechanism for these two side reactions,especially at the atomistic level,remains unclear.In order to suppress the dissociation of MN and thus improve the selectivity of DMO in the coupling reaction,it is therefore of vital importance to address the reaction mechanism.In this contribution,w e report extensive density functional theory(DFT)investigations to explore the reaction mechanism for both non-catalytic dissociation of MN in the gas phase and catalytic dissociation of MN on Pd(111)surface.

Methyl nitrite(CH3ONO,MN)is a very reactive molecule in the gas phase because of its weak O-NO bond(the bond energy is 1.82 eV).The O-NO bond would be further weakened upon adsorption on the active metal surfaces,owing to the surface stabilization of the adsorbed methoxy(CH3O)and NO[9,10].Hence,MN can be used as the adsorbed precursors for the preparation of CH3O intermediates on a number of metal surfaces.Experimentally,He and co-workers studied the thermal dissociation of MN in a static reactor at temperatures in the range 450-520 K[11].The products monitored by Fourier Transform Infrared Spectroscopy and gas-liquid chromatography included ME,formaldehyde,nitric oxide,and carbon monoxide.Their results showed that the initial O-N bond scission is pressure-dependent in the investigated temperature and pressure ranges.The dissociation of MN on metal surfaces has also been investigated by several groups[7-10,12].Peck and coworkers performed auger electron spectroscopy(AES),low-energy electron diffraction(LEED),and temperature-programmed desorption(TPD)studies on MN dissociation on Pt(111)and Pt-Sn alloys[9,10].They found that the activation of MN is very facile to produce adsorbed CH3O and NO,and the adsorbed CH3O could be either decomposed to CO and H or hydrogenated to ME,depending on the initial coverage of MN.At low coverage of MN(θMN＜ 0.15 Monolayer(ML)),the surface reactivity is the highest and the reactions during TPD only produce gas phase CO,H2,and NO.How ever,at higher coverages,as the concentrations of co-adsorbates increase,the surface is deactivated and the gas phase ME is an important product.Nevertheless,the main pathway for CH3O on Pt(111)is complete dehydrogenation to form CO and H2.By contrast,the dissociation of CH3O on the Pt-Sn alloys yields formaldehyde and adsorbed hydrogen that efficiently hydrogenates CH3O to form ME[9,10].Zhuo and Jiang investigated the vapor phase catalytic dissociation of MN on several supported Pd catalysts[7,8].They found that the catalytic activities were ranked in the follow ing order:Pd/γ-Al2O3＞ Pd/AC ＞ Pd/α-Al2O3＞Pd-Ti/α-Al2O3,and the acidic supports such as HY zeolite,γ-Al2O3and SiO2also exhibited moderate activity for the dissociation of MN.As a result,they suggested that the dissociation of MN on acidic support is an attractive approach to synthesize MF[7,8].On the basis of the experimental observations,they proposed a possible successive dehydrogenation mechanism.How ever,the detail reaction mechanism at atomistic level remains unclear.Our group also investigated the thermal dissociation of MN in the gas phase and the catalytic dissociation of MN on a Pd/α-Al2O3catalyst[12].It is found that the thermal dissociation of MN would produce ME,formaldehyde,and NO while the catalytic dissociation of MN mainly produced ME,MF and NO.

To the best of our knowledge,little theoretical work has been devoted to exploring there action mechanism for the dissociation of MNin the gas phase and on transition metal surfaces.By means of DFT calculations,Gomes et al.investigated the adsorption of MN on Au(111)surface,and found that only the cis-conformer is present on the Au surface[13].Although the theoretical work that related to the MN on transition metal surfaces is limited,the chemisorption and reactivity of the dissociative species,i.e.,CH3O on transition metal surfaces have been extensively studied by DFT calculations[14-17].For example,Chen and co-workers systematically investigated the CH3O dissociation pathways on Pd(111),Cu(111),and Pd-Zn alloyed surfaces using periodic DFT calculations.They found that the activation energies for C-O scission of CH3O species on all the three surfaces are 0.93-1.14 eV higher than the corresponding barriers for C-H bond breaking,implying that C-H breaking is clearly favored than C-O cleavage.Moreover,the dehydrogenation of CH3Oto CH2O is a very facile reaction on Pd(111)according to the calculated energy barrier and reaction energy.After that,the same group continues to investigate the dehydrogenation of formaldehyde(CH2O)and formal(CHO)on Pd(111),Cu(111),and Pd-Zn alloyed surfaces[18].Their calculated results convincingly demonstrated that complete dehydrogenation of CH2Oto COon Pd(111)is both thermodynamically and kinetically favorable,but unfavorable on Cu,and Pd-Zn alloyed surfaces[18].Likewise,Jiang et al.investigated ME dehydrogenation to CO and H2on Pd(111)using DFT[16].On the basis of the calculated adsorption energies and activation energies,they suggested that(i)desorption rather than dehydrogenation is preferable for the adsorbed ME due to the weak interaction of ME with the Pd surface.(ii)For the adsorbed CH2O,the possibilities for desorption and dehydrogenation are comparable according to the similar adsorption energy and dehydrogenation activation barrier.(iii)For other species,dehydrogenation is more favorable due to the strong adsorptions.

It is clear from the previous theoretical work that CO and H2are readily formed from the dehydrogenation of the adsorbed CH3O on Pd surface.Meanwhile,MN is very facile to decompose to adsorbed CH3O and NO due to the considerably weak CH3O-NO bond.Consequently,one may easily deduce that CO,H2,and NO are the main products for the dissociation of MN on Pd catalysts.This reasonable suspect is,however,inconsistent with the experimental observations on Pd catalysts w here ME and MF are the main products for the catalytic dissociation of MN on Pd catalysts(see Eq.(3))[7,8,12].Obviously,the reaction mechanism for the dissociation of MN on Pd catalyst is different from that of dissociation of ME or the adsorbed CH3O species.In order to provide a better understanding of the MN dissociation reactions on Pd catalysts as w ell as the non-catalytic dissociation of MN in the gas phase,we report extensive DFT calculations in this contribution to explore the reaction mechanism for MN dissociation on Pd(111)surface and in the gas phase at atomistic level,aiming to shed light on the reaction pathways at different conditions and the origin of the different product distribution between the non-catalytic and catalytic dissociation of MN.

2.Computational Details

In this work,all the DFT calculations were performed with the VASP code[19-22],in which thew ave functionsat each k-point are expanded with a plane w ave basis set with a kinetic cutoff energy up to 400 eV.The interactions between valence electrons and ion cores were treated by Blöchl's all-electron-like projector augmented w ave(PAW)method[23,24].The exchange-correlation functional utilized was the generalized gradient approximation functional proposed by Perdew,Burke,and Ernzerhof,know n as GGA-PBE[25].Brillouin zone sampling was performed using a Monkhorst-Pack grid[26]and electronic occupancies were determined according to a Meth fess el-Paxton scheme with an energy smearing of 0.2 eV[27].The calculated lattice constant for bulk Pd is0.395 nm,which is in reasonable agreement with the experimental value[28](0.389 nm)and previous theoretical results[29,30].

A p(2×2)unit cell was utilized to model the Pd(111)surface,achieving a surface coverage of 0.25 ML for the adsorbates.The(111)surfaces were modeled by the repeated three-layer slabs separated by a vacuum layer aslarge as 1.2 nm along the direction of the surface normal to avoid the periodic interactions.Our previous work showed that increasing the thickness of the slab had negligible effect on the adsorption energies[31].The first Brillouin zone of the p(2×2)unit cell was sampled with Γ-centered 7 × 7 × 1 k-point grids,which were proven to be sufficient for this cell[31].The bottom two layers in the slab models were fixed and the topmost layer and the adsorbates were allowed to relax during geometry optimization.The dipole corrections were found to have negligible effect on the adsorption energies,and therefore were not considered in this work.The zero point energy(ZPE)corrections were not included in this w ork.

The dimer method was used to locate the transition states(TSs)for the elementary reactions.This method has been described in detail elsewhere[32].In all the calculations,a force-based conjugated-gradient method[33]was used to optimize the geometry.Saddle points and minim a were considered to be converged w hen the maximum force in every degree of freedom was less than 0.3 eV·nm-1.In order to obtain accurate forces,the total energy and band structure energy were converged to within 1×10-7eV/atom during the electronic optimization.To verify the adsorption con figurations and TSs,vibrational frequency calculations were carried out.The Hessian matrix was determined by the numerical finite difference method with a step size of 0.002 nm for the displacement of the individual atoms of the adsorbates along each Cartesian coordinate.Then the vibrational frequencies were calculated by diagonalization of the Hessian matrix.

The adsorption energies(Eads)of the adsorbates were calculated by Eq.(4),

where Eadsorbate/surface,Esurface,and Eadsorbateare the DFT total energies of the surface with adsorbed molecules,the bare surface,and the isolated gas-phase molecule,respectively.Eadsorbate/surfaceand Esurfacewere calculated with the same computational setup(k-point sampling,energy cutoff,etc.).Eadsorbatewas calculated by putting the isolated adsorbate in an orthorhombic box with dimensions of 1.5 nm×1.55 nm×1.6 nm and carrying out a spin-polarized Γ-point calculation.With this definition,a more negative value of adsorption energy denotes stronger binding between adsorbate and surface.

3.Results and Discussion

3.1.Non-catalytic dissociation of MN in the gas phase

Experimentally,the non-catalytic(thermal)dissociation of MN in the gas phase produces almost equal amount of ME and formaldehyde[11,12].The w hole process can be simply described by the follow ing elementary steps.

where the·represents the radical species.In order to further clarify the different product distribution between the catalytic and non-catalytic dissociations of MN,the further non-catalytic dissociation of formaldehyde is also taken into account,which can be expressed as,

For these gas-phase elementary reactions,the TSs are difficult to identify.Previous calculations demonstrated that there is no maximum on the potential energy surface and the dissociation reaction presents a“loose”transition state[34].Indeed,we could not locate the saddle points connecting the initial states and the final states on the potential energy surface during the TS search,and perhaps such saddle points do not exist.Instead,the reaction energies for these reactions are calculated,and the results are summarized in Table 1.Note that for the gas phase reactions,the reaction energies are actually the bond energies for the dissociating chemical bonds of the reaction intermediates,and therefore should be a good descriptor in describing the reactivity of the elementary steps that involved in the non-catalytic dissociation reactions.The reaction energies are always positive for the bond breaking reactions and negative for the bond making reactions.

It can be seen from Table 1 that the CH3O-NO bond energy is calculated to be 1.91 eV,which is in reasonable agreement with theexperimental value of 1.82 eV.Then,the produced CH3O radical can be decomposed to CH2O molecule with the reaction energy of 1.11 eV,which is much easier than the initial O-N bond scission.The associative reaction of CH3O and H to form the ME molecule is strongly exothermic with the reaction energy of-4.69 eV due to the strong O-H bond,which indicates that the formation of ME is very favorable thermodynamically.The reaction energy for the further dissociation of CH2O to CHO is calculated to be 3.94 eV,which is very difficult to overcome under the typical thermal dissociation conditions.From these calculated results,the reaction mechanism for the non-catalytic dissociation of MN can be deduced as follow s.First,the MN molecules can be dissociated to the CH3O radical species and NO molecules at appropriate reaction temperatures.Second,because of the intrinsic instability of the produced CH3O radical species and the closed shell nature of the CH2O molecule,the C-H bond energy is rather weak with the bond energy of 1.11 eV,and thus the radical species can be easily decomposed to CH2O molecules.However,the further dissociation of CH2O to CHO is unfavorable due to the strong C-H bond in CH2O molecule with the bond energy as high as 3.94 eV.Therefore,CH2O could be taken as a stable product in the non-catalytic dissociation reaction,despite that the dissociation of CHO to CO is quite facile with the dissociation energy of 1.18 eV.On the other hand,the produced H from the dissociation of CH3O can easily react with the CH3O species to form the ME molecules due to the strong exothermic nature of this reaction.

Table 1 Calculated reaction energies of the elementary stepsthat are involved in the non-catalytic dissociation of MN in the gas phase

3.2.Catalytic dissociation of MN on Pd(111)

Experimentally,it has been demonstrated that MN can be readily decomposed to the adsorbed CH3O and NO on metal surfaces even at low temperatures[9,10].Then,the adsorbed NO would desorb to gas phase as a product without any further reactions.For the adsorbed CH3O,there are three possible reaction pathways for the further conversion:(i)the successive dehydrogenation of CH3O*to CH2O*,CHO*,and CO*.CH2O*and CO*may desorb to the gas phase as products;(ii)the hydrogenation of CH3O*to form CH3OH*,and the later could desorb to the gas phase as the product ME with the atomistic H*originated from the pathway(i);(iii)CH3O*could also react with CO*to form an intermediate COOCH3*.This intermediate can be hydrogenated to CHOOCH3*,which could desorb to the gas phase as the product MF.The full reaction network is illustrated in Fig.1.Here,w e note that it is also possible for CH3O*to break its C-O bond to produce CH3*and atomic O*.The activation barrier for this reaction is,how ever,much higher than that for the dehydrogenation reaction from previous calculations[14].Therefore,the C-O bond scission reactions are not necessarily taken into account.

3.2.1.Adsorption of the reaction intermediates

On the basis of the constructed reaction network show n in Fig.1,the adsorption energies and structures of the reaction intermediates that involved in the reaction network are calculated and discussed in the following subsection.The most stable adsorption structures of the reaction intermediates are presented in Fig.2,and the corresponding adsorption energies are summarized in Table 2.The adsorption of CO and NO on Pd(111)surface has been extensively studied in our previous work[31],and therefore not discussed hereafter.

3.2.1.1.CH3ONO.As shown in Fig.2,MNmoleculebinds to the Pd surface through the Natom at the Atop site.The calculated adsorption energy is-0.51 eV,which indicates that the adsorption of the MN molecule on Pd(111)surface is not very strong.The N-Pd distance is calculated to 0.202 nm.The CH3O-NO bond and CH3ON-O bond lengths are calculated to be 0.143 and 0.121 nm,respectively,which are almost identical to those in the gas phase,in accordance with the moderate adsorption energy.The C-O bond in the CH3O group is almost perpendicular to the Pd(111)surface.We also examined the adsorption structure with the O in the CH3O group bonding to the Pd surface.How ever,such adsorption structure is unstable during the geometry optimization,w here the MN molecule is decomposed to CH3O and NO,consistent with the fact that the CH3O-NO bond is relatively weak.

Fig.1.Schematic representation of reaction network of MN dissociation on Pd(111).

3.2.1.2.CH3O.The adsorption of CH3O on Pd(111)has been extensively investigated by means of DFT calculations[14,16].Our calculated results show that the adsorption of CH3O at the Fcc site is more stable than that at the other adsorption sites,such as Hcp,Bridge,and Atop sites.The adsorption energy of CH3O at the Fcc site on Pd(111)is calculated to be-1.91 eV,in reasonable agreement with the previous calculations[14,16].Ascan be seen from Fig.2,in the CH3Oadsorption con figuration,the O atom is exactly centered over the Fcc site with three equal O-Pd distances of 0.218 nm,while the C-O bond is perpendicular to the Pd(111)surface with a bond length of 0.144 nm.

3.2.1.3.CH2O.Since CH2O is a closed shell molecule,it has been detected asaproduct in the MNdissociation reactions from the experimental observations,despite that the amount is very small[8].There are generally two adsorption modes for CH2O adsorption on transition metal surfaces from the previous calculations,namely,η1-(O),and η1-C-η1-O[14].In the η1-(O)mode,the molecule binds to the surface through the O atom,while in the η1-C-η1-O mode the molecule binds to the surface through both the C and O atoms.In our calculations,it is found that the η1-(O)mode is unstable i.e.,the CH2O molecule placed initially via the η1-(O)mode shifts to the η1-C-η1-O mode during the geometry optimization.As show n in Fig.2,in the η1-C-η1-O mode,the CH2group is at the Atop site of the surface Pd atoms,and the O atom is at the Bridge site.The calculated adsorption energy of CH2O in the η1-C-η1-O mode is-0.59 eV,which is in good agreement with the TPD experimental data of-0.52 eV[35]and the previous theoretical results of-0.45 eV[14,16].Meanwhile,the small adsorption energy indicates that adsorption of CH2O on Pd(111)surface is rather weak,and thus may easily desorb to the gas phase as a product.The calculated C-O bond length of 0.134 nm is stretched upon adsorption as compared to the bond length of 0.121 nm in the gas phase.The C-Pd and O-Pd distances are calculated to be 0.209 and 0.225 nm,respectively,in good agreement with the previous calculations[14,16].

Fig.2.Adsorption structures of the reaction intermediates that are involved in the reaction network for MN dissociation on Pd(111)surface.Large balls denote Pd atoms;small gray balls denote C atoms;small black balls denote N or O atoms;small white balls denote H atoms.

Table 2 Calculated adsorption energies of the reaction intermediates that involved in the reaction network of MN dissociation on Pd(111)

3.2.1.4.CHO.By contrast to the η1-C-η1-O adsorption mode of CH2O,CHO favors the η2-C-η1-O mode,as shown in Fig.2.In the adsorption con figuration,the CH group almost sits at the Bridge site,while the O atom is at the Atop site of the Pd surface.The calculated adsorption energy of-2.43 eV is in reasonable agreement with previous calculations[14,16].The two equal C-Pd distances are calculated to be0.207 nm,and the O-Pd distance is calculated to be 0.217 nm.The calculated C-Obond length of 0.127 nm is stretched as compared to the bond length of 0.119 nm in the gas phase.

3.2.1.5.H.The most stable adsorption site for Hon Pd(111)is the Fcc site,as shown in Fig.2.The adsorption energy of Hat the Fcc site is calculated to be-2.83 eV,which is0.04 eV more negative than that at the Hcp site.In the most stable adsorption con figuration,the H atom is exactly centered over the Fcc site with three equal H-Pd distances of 0.182 nm,which agrees w ell with the previous calculations[14,16].

3.2.2.Reaction pathways

As mentioned before,the dissociation of MN to CH3O*and NO*is very facile due to the considerably weak CH3O-NO bond.Indeed,we obtained a very small energy barrier of 0.04 eV for this dissociation reaction,in accordance with the fact that CH3O*can be readily formed on transition metal surfaces even at very low temperatures[9,10].As show n in Fig.3,the O-N bond length is lengthened from 0.144 nm in the most stable adsorption structure to 0.151 nm in the TS con figuration.The N-Pd and O-Pd distances are calculated to be 0.204 and 0.258 nm,respectively.Only one negative mode is identified in the frequency calculation,which indicates that this TS con figuration is indeed a first-order saddle point on the potential energy surface.

Asillustrated in Fig.1,there are three different reaction path ways for the CH3O*,namely,successive dehydrogenation,hydrogenation and carbonylation of CH3O*.In order to elucidate the dominant reaction pathway for further conversion of CH3O*,the activation energies for the elementary reactions that are involved in the three different reaction pathways are calculated.The TS structures are illustrated in Fig.3,and the potential energy diagram is show n in Fig.4.

3.2.2.1.Successive dehydrogenation of CH3O

3.2.2.1.1.Dissociation of CH3O*.The activation energy for the dissociation of CH3O to CH2O*and H*is calculated to be 0.46 eV,similar with the previous calculation results[14,16].Such a low energy barrier for this reaction indicates that the H abstraction of CH3O*on Pd(111)surface proceeds rapidly.In the TS con figuration(TS2 in Fig.3),the dissociating CH2O is filed over the Bridge site of the Pd surface while the atomic H is at the Atop site.The O-Pd and C-Pd distances are calculated to be0.208 and 0.239 nm,respectively.The detached C-Hdistance is calculated to be 0.151 nm.

3.2.2.1.2.Dissociation of CH2O*.The activation energy for the dissociation of CH2O to CHO*and H*is calculated to be 0.31 eV,with a desorption energy barrier of CH2O*of 0.59 eV,much higher than the dissociation energy barrier,which indicates that CH2O*prefers further dehydrogenation rather than desorption.Actually,in the MN dissociation experiments,only a trace amount of CH2O is found in the product[7,8],which is in good agreement with the present calculation results.As shown in Fig.3(TS3),the dissociating CHO group binds to the Pd surface via the η2-C-η1-O mode,similar to the adsorption structure of CHO on the surface.The detached H atom is at the Atop site of the Pd surface,sharing the same Pd atom with the CHO group,with a detached C-H distance of 0.152 nm.

3.2.2.1.3.Dissociation of CHO*.The dissociation of CHO to CO*and H*proceeds with a small energy barrier of 0.33 eV,implying that this step is facile and fast.As show n in Fig.3(TS4),the dissociating CH and H are both at the Atop sites of the Pd surface with a detached C-H distance of 0.145 nm.It should be mentioned that there is another reaction channel of CHO*reacting with the CH3O*to form CHOOCH3*directly,as shown in Fig.4(TS5),with an energy barrier of 0.65 eV,which is much higher than the further dehydrogenation of CHO*and thus is not taken into account in the reaction network.

3.2.2.2.Hydrogenation of CH3O.The activation energy for the hydrogenation of CH3O*to CH3OH*is calculated to be 0.71 eV,0.25 eV higher than the dehydrogenation of CH3O*.Considering that the coverage of the free site is typically higher than the coverage of H*at the initial stage of MN dissociation,the dehydrogenation of CH3O*is more favored than the hydrogenation reaction.As shown in Fig.3(TS6),the reacting CH3O and H are both at the Atop sites of the Pd surface with a detached O-H distance of 0.160 nm.The produced CH3OH*can easily desorb to the gas phase as the product ME with a small desorption barrier of 0.30 eV.

Fig.3.TS structures for the elementary reactions that are involved in the reaction network for MN dissociation on Pd(111)surface.

Fig.4.Potential energy diagram of MN dissociation on Pd(111).

3.2.2.3.Carbonylation of CH3O.The activation energy for the carbonylation of CH3O*w ith CO*to COOCH3*is calculated to be 1.11 eV,which is0.65 eV and 0.40 eV higher than the dehydrogenation and hydrogenation of CH3O*,respectively,indicating that the cabonylation reaction is the most unfavorable reaction at the initial stage of MN dissociation.As show n in Fig.3(TS7),the reacting CH3O is at the Bridge site while the CO is at the Atop site of the Pd surface with a detached C-O distance of 0.193 nm.The produced COOCH3*can be further hydrogenated to form CHOOCH3*with an energy barrier of 0.81 eV(TS8 in Fig.3).Finally,the CHOOCH3*could easily desorb to the gas phase as the product MF with a very small desorption barrier of 0.12 eV.

3.2.3.Surface coverage effect

It is clear from Fig.4 that the successive dehydrogenation of CH3O*is the most favorable reaction pathway at the initial stage of MN dissociation due to the relatively low est activation energies along this reaction pathway.The hydrogenation and carbonylation reactions are hindered by much higher activation energies.How ever,one can find that the complete dehydrogenation species,i.e.,CO*is difficult to desorb to the gas phase as a product due to the considerably high desorption energy barrier(more than 2 eV).As a result,the CO*would accumulate on the surface to achieve a high coverage under this circumstance.For the samereason,the coverage of NO*that originated from the direct dissociation of MN would also be high as the desorption energy barrier is as high as 2.29 eV.Therefore,the effect of the co-adsorbed species,particularly CO*and NO*,on the activities of the different reaction pathways should be taken into account.To this end,w e further investigated the effect of co-adsorbed CO*and NO*on three key elementary reactions of the three different reaction pathways,as listed in Table 3.These three key elementary reactions are singled out to represent the activities for the three different reaction pathways because the activation energies for these reactions are the highest along each reaction pathway.

Table 3 Calculated activation energies for the key elementary reactions at different surface coverage on Pd(111)

As can be seen from Table 3,with the co-adsorbed CO and NO molecules on the Pd(111)surface,the activation energies for the dehydrogenation of CH3O*are increased from 0.46 eV to 1.28 and 1.23 eV,respectively,which means that the dehydrogenation reactions are greatly suppressed at high coverage.By contrary,the activation energies for the hydrogenation and the carbonylation reactions are greatly reduced,indicating that these reactions become faster at high coverage.Therefore,the wholere action mechanism for MN dissociation on Pd catalysts can be deduced as follow s.At the initial stage,the successive dehydrogenation of CH3O*is dominant at low surface coverage,and the surface is quickly covered by the dissociated species to achieve a high surface coverage.Then,the hydrogenation and carbonylation reactions are greatly enhanced at high coverage to yield ME and MF,which can easily desorb to the gas phase as the main products.

In order to elucidate the physical origin of effects of the co-adsorbed CO and NO molecules on the reactivity of the reactions,the variation in the activation energies(ΔEa)due to the co-adsorbed species is decomposed into two parts[36]:

For a typical surface reaction AB⇌A+B with the transition state[A-B]#,the variation in the dissociation activation energy(ΔEa(disso))and association activation energy(ΔEa(asso))can be written as,respectively,

w here ΔEX(X=AB,A,B or[A-B]#)is the variation in the adsorption energy of species X due to the presence of the co-adsorbed molecules.All these individual terms in Eqs.(11)and(12)for the three key elementary steps can be calculated exactly using DFT.The results are summarized in Table 4.

It is clear from Table 4 that for the dehydrogenation reaction,the destabilization to the TS is much more significant than that to the IS and thus the activation energy is increased(ΔEa＞0).By contrary,for the hydrogenation and carbonylation reactions,the destabilization to the TS is much smaller than that to the IS.As a result,the activation energies for the associating reactions are all reduced(ΔEa＜ 0).The principle behind such adifference between the dissociation and association reactionsliessimply with the different number of the reactants:there are tw o reactants in the association reactions and only one in the dissociation reaction.

Table 4 Destabilization(eV)of the IS and the TS of the key elementary reactions due to the presence of the co-adsorbed CO and NO on Pd(111)

3.3.Implication for MN dissociation

The calculated results convincingly show that the dissociation of CH3Oto CH2O is considered as the rate-determining step of the catalytic dissociation reaction on Pd(111),which is quite different from that of the non-catalytic dissociation reaction where the dissociation of MN to CH3O and NO is the rate-determining step.The apparent activation energy for non-catalytic dissociation of MN(1.91 eV)is much higher than that for catalytic dissociation(1.28 eV),which is consistent with the general chemical consensus that an effective catalyst can greatly reduce the reaction barrier for a reaction.It should be noted that for the gas-phase reactions,the transition states,if existing,must be higher in energy than the initial states and final states,and therefore the actual activation energies are generally higher than the calculated reaction energies shown in Table 1.

On the basis of the detailed reaction mechanism for both the nocatalytic and catalytic dissociation of MN at atomistic level,one may find ways to prevent the dissociation reactions in the CO coupling reactions to improve the selectivity of the objective product DMO.To prevent the non-catalytic dissociation of MN,one has to carefully control the reaction temperature of the coupling reaction to minimize the dissociation of MN in the gas phase.On the other hand,a promising way to suppress the catalytic dissociation of MN on Pd surface is to increase the CO/MN ratio in the feedstock to achieve a high CO coverage on the surface so that the dissociation of CH3O*could be hindered.Actually,in the CO coupling experiments,the catalytic dissociation of MN on Pd catalysts is greatly suppressed when the CO/MN ratio exceeds 1.5,with only a trace amount of MF in the products[12].

3.4.Comparison to available experimental data

It is interesting to compare our DFT calculated results from MN dissociation with the available experimental data.As mentioned above,the products from MN dissociation depend on the initial surface coverage of MN from the TPD experiments.At low MN coverage(θMN＜ 0.15 ML),the surface reactivity is highest and the reactions during TPD only produce gas phase CO,H2,and NO.How ever,at higher coverages,as the concentrations of coadsorbates increases,the surface is deactivated and the gas phase ME is an important product[9,10].The experimental observations on Pt(111)are quite similar to the present calculated results on Pd(111),i.e.,the successive dehydrogenation of CH3O*to produce the CO*is preferred at low coverage while the hydrogenation reaction to produce ME is more favorable at high coverage.Experimentally,Zhuo and Jiang investigated the vapor phase catalytic dissociation of MN on a series of supported Pd catalysts[8].The main products from the experimental observations are ME and MF,which is in accordance with our calculated results at high coverage.The optimum reaction temperature to obtain the highest yield of MF is about 433 K,in line with the activation energy for the dissociation of CH3O*at high coverage(1.23 eV).Our group also investigated the thermal dissociation of MN in the gas phase and the catalytic dissociation of MN on a Pd/α-Al2O3catalyst[12].It is found that the thermal dissociation of MN would produce ME,formaldehyde,and NO while the catalytic dissociation of MN mainly produced ME and MF.This different product distribution of thermal and catalytic dissociation of MN can be w ell rationalized by our DFT calculated results.Apparently,the key to determine whether the MF or formaldehyde would be formed in the MN dissociation reactions is that whether the further dissociation of formaldehyde could occur.From our DFT calculations,we know that the further dissociation of formaldehyde is very facile on Pd(111)due to the relatively low activation energy as compared to the desorption energy barrier of CH2O*.Hence,only a trace amount of formaldehyde can be detected in the catalytic dissociation of MN,which is consistent with the experiments[8].On the other hand,the calculated results convincingly demonstrate that the further dissociation of formaldehyde in the gas phase is unfavorable in the gas phase without the help of the catalyst due to the considerably high dissociation energy(～4 eV).

4.Conclusions

This work represents the first theoretical attempt to explore the reaction mechanism for non-catalytic dissociation of MN in the gas phase and the catalytic dissociation of MN on Pd(111)surface.The main conclusions are summarized as follow s.

The non-catalytic dissociation of MN initiates from the breaking of the weak CH3O-NO bond to the CH3O radical species and NO molecules.Then,the dissociation of CH3O radical species to form formaldehyde is quite facile due to the weak C-H bond while the further dissociation of formaldehyde in the gas phase is difficult because of the strong C-H bond of the CH2O molecule.The formation of the ME in the gas phase is strongly exothermic and energetically favorable.Therefore,ME and formaldehyde are the main products for non-catalytic dissociation of MN,which is in accordance with the experimental observations.

On the other hand,for the catalytic dissociation of MN on Pd(111),the calculated results show that the dissociation of MN to the adsorbed CH3O and NO is facile with a small energy barrier of 0.03 eV.The calculated activation energies along different reaction pathways indicate that at the initial stage,the successive dehydrogenation of CH3O*is dominant at low surface coverage,and the surface is quickly covered by the dissociated species to achieve a high surface coverage.Then,the hydrogenation and carbonylation reactions are greatly enhanced at high coverage to yield ME and MF,which can easily desorb to the gas phase as the main products.

On the basis of the detailed reaction mechanism,we suggest that controlling the reaction temperature and increasing the CO/MN ratio in the feedstock can effectively suppress both the non-catalytic and catalytic dissociation of MN during the coupling reactions.In addition,our calculated results can provide a rational explanation of the experimental observations for the current DMO manufacturing process.

[1]L.Zhao,Y.Zhao,S.Wang,H.Yue,B.Wang,J.Lv,X.Ma,Hydrogenation of dimethyl oxalate using extruded Cu/SiO2catalysts:Mechanical strength and catalytic performance,Ind.Eng.Chem.Res.51(43)(2012)13935-13943.

[2]Z.Meng,J.Sun,J.Wang,J.Zhang,Z.Fu,W.Cheng,X.Zhang,An efficient and stable ionic liquid system for synthesis of ethylene glycol via hydrolysis of ethylene carbonate,Chin.J.Chem.Eng.18(6)(2010)962-966.

[3]H.Zhou,S.Zhang,F.Gao,X.Bai,Z.Sha,Solubility of ammonia in ethylene glycol between 303 K and 323 K under low pressure from 0.030 to 0.101 MPa,Chin.J.Chem.Eng.22(2)(2014)181-186.

[4]X.Ma,H.Chi,H.Yue,Y.Zhao,Y.Xu,J.Lv,S.Wang,J.Gong,Hydrogenation of dimethyl oxalate to ethylene glycol over mesoporous Cu-MCM-41 catalysts,AIChE J.59(7)(2013)2530-2539.

[5]S.Y.Peng,Z.N.Xu,Q.S.Chen,Y.M.Chen,J.Sun,Z.Q.Wang,M.S.Wang,G.C.Guo,An ultra-low Pd loading nanocatalyst with high activity and stability for CO oxidative coupling to dimethyl oxalate,Chem.Commun.49(51)(2013)5718-5720.

[6]Z.N.Xu,J.Sun,C.S.Lin,X.M.Jiang,Q.S.Chen,S.Y.Peng,M.S.Wang,G.C.Guo,High performance and long-lived Pd nanocatalyst directed by shape effect for CO oxidative coupling to dimethyl oxalate,ACS Catal.3(2)(2013)118-122.

[7]G.L.Zhuo,X.Z.Jiang,An attractive synthetic approach to methyl formate from methanol via methyl nitrite,Catal.Lett.80(3-4)(2002)171-174.

[8]G.L.Zhuo,X.Z.Jiang,Catalytic decomposition of methyl nitrite over supported palladium catalysts in vapor phase,React.Kinet.Catal.Lett.77(2)(2002)219-226.

[9]J.W.Peck,D.E.Beck,D.I.Mahon,Methyl nitrite adsorption as a novel route to the surface methoxy intermediate,J.Phys.Chem.B 102(18)(1998)3321-3323.

[10]J.W.Peck,D.I.Mahon,D.E.Beck,B.Bansenaur,B.E.Koel,TPD,HREELS and UPS study of the adsorption and reaction of methyl nitrite(CH3ONO)on Pt(111),Surf.Sci.410(2-3)(1998)214-227.

[11]Y.He,W.A.Sand ers,M.C.Lin,Thermal decomposition of methyl nitrite:Kinetic modeling of detailed product measurements by gas-liquid chromatography and Fourier-transform infrared spectroscopy,J.Phys.Chem.92(19)(1988)5474-5481.

[12]Y.Ji,Study of Catalyst,Mechanism,and Intrinsic Kinetics of CO Catalytic Coupling to Dimethyl Oxalate(DMO)(Ph.D thesis)East China University of Science and Technology(ECUST),Shanghai,2010.

[13]J.R.B.Gomes,F.Illas,The adsorption of methyl nitrite on the Au(111)surface,Catal.Lett.71(1-2)(2001)31-35.

[14]Z.X.Chen,K.M.Neyman,K.H.Lim,N.Rösch,CH3O decomposition on Pd Zn(111),Pd(111),and Cu(111).A theoretical study,Langmuir 20(19)(2004)8068-8077.

[15]Z.X.Chen,K.H.Lim,K.M.Neyman,N.Rösch,Effect of steps on the decomposition of CH3O at Pd Zn alloy surfaces,J.Phys.Chem.B 109(10)(2005)4568-4574.

[16]R.Jiang,W.Guo,M.Li,D.Fu,H.Shan,Density functional investigation of methanol dehydrogenation on Pd(111),J.Phys.Chem.C 113(10)(2009)4188-4197.

[17]R.Jiang,W.Guo,M.Li,X.Lu,J.Yuan,H.Shan,Dehydrogenation of methanol on Pd(100):comparison with the results of Pd(111),PCCP 12(28)(2010)7794-7803.

[18]K.H.Lim,Z.X.Chen,K.M.Neyman,N.Rösch,Comparative theoretical study of formaldehyde decomposition on Pd Zn,Cu,and Pd surfaces,J.Phys.Chem.B 110(30)(2006)14890-14897.

[19]G.Kresse,J.Furthmüller,Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-w ave basis set,Comput.Mater.Sci.6(1)(1996)15-50.

[20]G.Kresse,J.Furthmüller,Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,Phys.Rev.B 54(16)(1996)11169-11186.

[21]G.Kresse,J.Hafner,Ab initio molecular dynamics for liquid metals,Phys.Rev.B 47(1)(1993)558-561.

[22]G.Kresse,J.Hafner,Ab initio molecular-dynamics simulation of the liquid-metalamorphous-semiconductor transition in germanium,Phys.Rev.B 49(20)(1994)14251-14269.

[23]P.E.Blöchl,Projector augmented-w ave method,Phys.Rev.B 50(24)(1994)17953-17978.

[24]G.Kresse,D.Joubert,From ultras oft pseudopotentials to the projector augmented wave method,Phys.Rev.B 59(3)(1999)1758-1775.

[25]J.P.Perdew,K.Burke,M.Ernzerhof,Generalized gradient approximation made simple,Phys.Rev.Lett.77(18)(1996)3865-3868.

[26]H.J.Monkhorst,J.D.Pack,Special points for Brillouin-zone integrations,Phys.Rev.B 13(12)(1976)5188-5192.

[27]M.Methfessel,A.T.Paxton,High-precision sampling for Brillouin-zone integration in metals,Phys.Rev.B 40(6)(1989)3616-3621.

[28]D.R.Lide,CRC Handbook of Chemistry and Physics,79th ed.CRC Press,Boca Raton,Florida,1998.

[29]D.Loffreda,D.Simon,P.Sautet,Dependence of stretching frequency on surface coverage and adsorbate-adsorbate interactions:a density-functional theory approach of CO on Pd(111),Surf.Sci.425(1)(1999)68-80.

[30]Z.H.Zeng,J.L.F.Da Silva,W.X.Li,Density functional theory and ab initio molecular dynamics study of NO adsorption on Pd(111)and Pt(111)surfaces,Phys.Rev.B 81(8)(2010).

[31]C.Fan,W.D.Xiao,Origin of site preference of CO and NO adsorption on Pd(111)at different coverages:a density functional theory study,Comput.Theor.Chem.1004(2013)22-30.

[32]G.Henkelman,H.Jonsson,A dimer method for finding saddle points on high dimensional potential surfaces using only first derivatives,J.Chem.Phys.111(15)(1999)7010-7022.

[33]D.Sheppard,R.Terrell,G.Henkelman,Optimization methods for finding minimum energy paths,J.Chem.Phys.128(13)(2008)134106.

[34]A.Fernández-Ramos,E.Martínez-Núñez,M.A.Ríos,J.Rodríguez-Otero,S.A.Vázquez,C.M.Estévez,Direct dynamics study of the dissociation and elimination channels in the thermal decomposition of methyl nitrite,J.Am.Chem.Soc.120(30)(1998)7594-7601.

[35]J.L.Davis,M.A.Barteau,Polymerization and decarbonylation reactions of aldehydes on the Pd(111)surface,J.Am.Chem.Soc.111(5)(1989)1782-1792.

[36]Z.P.Liu,S.J.Jenkins,D.A.King,Car exhaust catalysis from first principles:Selective NO reduction under excess O2conditions on Ir,J.Am.Chem.Soc.126(34)(2004)10746-10756.