Selective Lactic Acid Hydrogenation: DFT, Microkinetic Modeling and Experiments

Suyash Singh, James A Dumesic and Manos Mavrikakis* Department of Chemical and Biological Engineering, University of WisconsinMadison, Madison, WI 53706 *manos@engr.wisc.edu

Significance The overarching goal of this project is to address the two key issues in biomass conversion reactions: (a) Selective removal of oxygen by dehydration, relevant to the formation of diesel fuels and (b) Selective hydrogenation of oxygen containing functional groups, relevant to the formation of commodity chemicals from biomass derived feed stocks. Figure 1. Thermochemical potential energy surface for LA hydrogenation on the close packed facets of 10 transition metals. Various Is represent stable intermediates along the reaction coordinate. Figure 2. Reactivity volcano plot, as obtained via a descriptor-based catalyst screening approach employing DFT-based microkinetic modeling. Red region corresponds to more active catalysts. Operating conditions at which this volcano is developed: Temperature = 400C; Pressure = 1 bar; Inlet feed composition = 1% LA, 5% H2O, 24% H2 and 74% He. The reactor is modeled as a CSTR and all points on this volcano plot correspond to 10% conversion. TOF stands for Turnover Frequency References 1. Simonov MN, Simakova IL, & Parmon VN., Reaction Kinetics and Catalysis Letters 2009; 97: 157-162. 2. Cortright RD, Sanchez-Castillo M, & Dumesic JA., Applied Catalysis B-Environmental 2002; 39: 353-359. Acknowledgement This material is based upon work supported as part of the Institute for Atom-efficient Chemical Transformations (IACT), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences.

Introduction Lactic Acid (LA) is the smallest organic compound that contains a methyl, a hydroxyl and a carboxylic acid functional group. For that, LA can be considered as an ideal model molecule for highly oxygenated biomass derived feed stocks. To date, only a few attempts at describing the selective hydrogenation [1] of carboxylic acids have been published, and the mechanism for this important class of reactions is not well established. We present a systematic study for selective LA hydrogenation to propylene glycol (PG) and propanoic acid (PA), the two major products, that combines first principles calculations, experiments and microkinetic modeling to gain insights into the reaction mechanism. Materials and Methods Periodic, self-consistent density functional theory calculations (DFT) were performed with the PW91-GGA exchange correlation functional using the Vienna ab initio simulation package (VASP). The catalysts were modeled by the (111) facets for the facecentered-cubic (fcc) metals (Cu, Pt, Pd, Ir, Ni, Ag, Au, Rh) and (0001) facets for the hexagonal-closed-packed (hcp) metals (Co, Ru). All the surfaces were represented by fourlayer slabs with a p(3x3) surface unit cell, periodically repeated in super cell geometry with a minimum of six equivalent layers of vacuum (> 14 ) in between any two successive metal slabs. The atoms of the top two layers of the slabs were allowed to relax. Spin polarized calculations were performed for adsorption on Ni(111) and Co(0001). The electron wave function was expanded using plane waves with an energy cutoff of 400 eV, and the core electron interactions were described using the projector augmented wave method (PAW). The surface Brillouin zone was sampled using a 4 x 4 x 1 Monkhorst-Pack k-point mesh. Descriptor based catalyst screening was performed using a mean-field microkinetic model with parameters all calculated from DFT. Results and Discussion DFT calculations were employed to probe the thermochemistry of LA hydrogenation on several close-packed transition metal surfaces. The results from these calculations suggest Cu(111) to be an efficient catalyst for this reaction (Figure 1). Detailed activation energy barrier calculations were then performed for all the elementary steps on Cu(111) and Pt(111), and the most favorable pathways for the formation of PG and PA were identified. Microkinetic Modeling was then employed to compare the calculated reaction rates and selectivities with the experimentally measured [2] values. A careful comparison between the DFT derived parameters (binding energies and activation energy barriers), and those obtained upon fitting the microkinetic model provides valuable insights into the nature of active site on Cu catalysts. In a subsequent step, we employed our detailed microkinetic model to develop trends across metals (Figure 2), to identify improved (active and selective) alloy catalysts for this reaction, and to determine the optimal operating conditions for these promising alloy catalysts.

Figure 1

Figure 2