Article pubs.acs.org/JPCA
1,2-Ethanediol and 1,3-Propanediol Conversions over (MO3)3 (M = Mo, W) Nanoclusters: A Computational Study Zongtang Fang, Patrick Zetterholm, and David A. Dixon* Department of Chemistry, The University of Alabama, Shelby Hall, Box 870336, Tuscaloosa, Alabama 35487, United States S Supporting Information *
ABSTRACT: The dehydration and dehydrogenation reactions of one and two 1,2-ethanediol and 1,3-propanediol molecules on (MO3)3 (M = Mo, W) nanoclusters have been studied computationally using density functional and coupled cluster (CCSD(T)) theory. The reactions are initiated by the formation of a Lewis acid−base complex with an additional hydrogen bond. Dehydration is the dominant reaction proceeding via a metal bisdiolate. Acetaldehyde, the major product for 1,2-ethanediol, is produced by α-hydrogen transfer from one CH2 group to the other. For 1,3-propanediol, the C−C bond breaking pathways to produce C2H4 and HCH O simultaneously and proton transfer to generate propylene oxide have comparable barrier energies. The barrier to produce propanal from the propylene oxide complex is less than that for epoxide release from the cluster. On the Mo3O9 cluster, a redox reaction channel for 1,2-ethanediol to break the C−C bond to form two formaldehyde molecules and then to produce C2H4 is slightly more favorable than the formation of acetaldehyde. For WVI, the energy barrier for the reduction pathway is larger due to the lower reducibility of W3O9. Similar reduction on MoVI for 1,3-propanediol to form propene is not a favorable pathway compared with the other pathways as additional C−H bond breaking is required in addition to breaking a C−C bond. The dehydrogenation and dehydration activation energies for the selected glycols are larger than the reactions of ethanol and 1propanol on the same clusters. The CCSD(T) method is required because density functional theory with the M06 and B3LYP functionals does not predict quantitative energies on the potential energy surface. The M06 functional performs better than does the B3LYP functional.
■
INTRODUCTION Biomass is an important alternative energy source for the replacement of nonrenewable fossil fuels to meet the current increasing energy demands as well as to reduce anthropogenic CO2 emissions as the biomass is, in principle, renewable. Critical steps in transforming biomass into fuels are to reduce the oxygen content and to generate C−C bonds from biomassderived intermediates.1−3 Biomass also can be used as feedstock for the synthesis of many different types of chemicals.3−5 Among the most important biomass precursors are polyols containing multiple hydroxyl groups, which have been widely investigated to produce chemical feedstocks using metals or metal oxides as catalysts.6−9 The catalytic conversion reactions of alcohols, the simplest polyols, have been studied thoroughly not only on the bulk scale10−12 but also in the molecular level.13−17 An extension of this research is the conversion of diols such as 1,2-ethanediol and 1,3-propanediol. Dehydration and dehydrogenation of 1,2-ethanediol have been studied on both titanium oxide18,19 and cerium oxide surfaces.20,21 On the TiO2 (110) surface, the coverage of ethylene glycol plays an important role for the selectivity of the dehydration and dehydrogenation channels. Acetaldehyde and ethylene are the major carbon-containing products, and small amounts of molecular hydrogen have also been observed.18,19 © 2016 American Chemical Society
The activation barrier decreases as a function of the number of defect sites.19 The reaction on the CeO2 (111) surface proceeds via a stable (−OCH2CH2O−) intermediate, leading to C−C bond breaking, whereas a different surface enolate (−CH2− CHO−) intermediate can be formed on a reduced CeO2−x surface and can further decompose to C2 products.20,21 Dohnálek and coworkers studied deuterated 1,2-propylene glycol conversion on rutile TiO2 (110) surfaces with temperature-programmed desorption (TPD).22 The products D2O and D2 were first observed at low temperatures before the desorption of C-containing products on the reduced surface, which is consistent with the reaction of 1,2-ethanediol on TiO2 (110) surfaces.18 The reactivity on a hydroxylated or an oxidized surface is comparable to a normal crystal surface as the oxygen vacancy sites (active sites) regenerate at a temperature lower than the C-containing product formation temperature. Deoxydehydration of larger vicinal diols to produce alkenes was also studied on trioxorhenium catalysts.23,24 Reducing agents such as alcohols or sulfite are required to activate the catalyst. The conversion of 1,3-propylene glycol differs from the Received: January 6, 2016 Revised: February 19, 2016 Published: February 22, 2016 1897
DOI: 10.1021/acs.jpca.6b00158 J. Phys. Chem. A 2016, 120, 1897−1907
Article
The Journal of Physical Chemistry A
Table 1. Calculated Molecular Adsorption and Dissociative Adsorption Enthalpies (ΔHad,0K, kcal/mol) (One and Two Glycols) and Reaction Barriers of the Dissociative Chemisorption of the Glycols from the Reactant Complexes (ΔH‡0K, kcal/mol) at the CCSD(T)/aD//B3LYP/VDZ Level at 0 K ΔH0K‡ proton transfer
ΔHad,0K dissociative adsorption cluster
glycol
ΔHad,0K molecular adsorption
1st H transfer
2nd H transfer
1st H transfer
2nd H+ transfer
W3O9
1st 1,2-ethanediol addition 2nd 1,2-ethanediol addition 1st 1,3-propanediol addition 2nd 1,3-propanediol addition 1st 1,2-ethanediol addition 2nd 1,2-ethanediol addition 1st 1,3-propanediol addition 2nd 1,3-propanediol addition
−37.2 −27.4 −41.6 −34.0 −28.8 −20.9 −32.9 −30.6
−47.7 −20.1 −48.6 −24.4 −28.9 −5.3 −29.8 −13.4
−45.5 −16.9 −48.5 −26.4 −30.4 −2.9 −31.9 −15.3
16.3 33.8 22.3 37.7 20.1 36.4 26.6 40.7
20.6 10.6 14.2 14.2 15.2 9.9 10.0 14.6
Mo3O9
+
+
+
selected for the density functional theory (DFT) geometry optimizations. The structures of global minima and transition states were characterized by vibrational frequencies, which are used to obtain the zero-point energy corrections (ZPEs) as well as the thermal corrections at 298 K derived from normal statistical mechanical expressions.32 The synchronous transitguided quasi-Newton (STQN) method33,34 was used for the transition-state optimizations, and the transition states are characterized by a single imaginary frequency. The geometry optimizations were performed using the DFT-optimized DZVP2 basis sets35 for C, H, O and the cc-pVDZ-pp basis sets coupled to relativistic effective core potentials (RECPs)36−38 used for Mo and W. These basis sets are denoted as VDZ. The B3LYP/VDZ geometries were then used in single-point calculations with the B3LYP functional with the aug-cc-pVDZ basis sets for H, C, and O39 and the augmented aug-cc-pVDZ-pp basis sets for Mo and W; this combination of basis sets is denoted as aD. The B3LYP/VDZ geometries were also in single-point calculations at the DFT level with the M06 exchange-correlation functional40 and at the correlated molecular orbital theory coupled cluster CCSD(T)41−44 level with the aD basis sets. Selected geometry optimizations with M06 functional showed energy differences to those without optimization of B3LYP. For the second glycol addition, slightly more exothermic dissociative adsorption energies are predicted by DFT in some cases. The DFT errors as compared with CCSD(T)/aD//B3LYP/VDZ are 3 to 10 kcal/mol, and the M06 functional generally performs better than the B3LYP. For the barriers for proton transfer, DFT with the B3LYP and M06 functionals give errors
