Perspective pubs.acs.org/JPCL
Controlling Catalytic Selectivity via Adsorbate Orientation on the Surface: From Furfural Deoxygenation to Reactions of Epoxides Simon H. Pang and J. Will Medlin* Department of Chemical and Biological Engineering, University of Colorado Boulder, Boulder, Colorado 80309, United States ABSTRACT: Specificity to desired reaction products is the key challenge in designing solid catalysts for reactions involving addition or removal of oxygen to/from organic reactants. This challenge is especially acute for reactions involving multifunctional compounds such as biomass-derived aromatic molecules (e.g., furfural) and functional epoxides (e.g., 1-epoxy-3-butene). Recent surface-level studies have shown that there is a relationship between adsorbate surface orientation and reaction selectivity in the hydrogenation pathways of aromatic oxygenates and the ring-opening or ring-closing pathways of epoxides. Control of the orientation of reaction intermediates on catalytic surfaces by modifying the surface or near-surface environment has been shown to be a promising method of affecting catalytic selectivity for reactions of multifunctional molecules. In this Perspective, we review recent model studies aimed at understanding the surface chemistry for these reactions and studies that utilize this insight to rationally design supported catalysts. compounds termed α,β-unsaturated aldehydes containing aldehyde and olefin functionality in close proximity. As shown conceptually in Scheme 1a, it is intuitive that promoting adsorption selectivity (or molecular orientation relative to the surface) would have important implications for selectivity. In fact, detailed studies of such reactions show that although adsorbate orientation does play a key role in controlling surface reaction chemistry, the relationship between surface orientation and selectivity is often more complex than the ideal case shown in Scheme 1a.5−7 Another area of research where controlled surface orientation is vital is in the area of asymmetric heterogeneous catalysis, through the use of chiral surface modifiers.8,9 These areas have been reviewed extensively and so will not be discussed in detail here; the reader is directed to the aforementioned references. This Perspective gives an overview of recent studies that elucidate mechanisms governing reaction selectivity, and efforts to utilize the knowledge gained from surface-level studies to rationally design selective catalysts. We focus on three types of reactions: (1) furfural hydrogenation and deoxygenation, (2) selective hydrogenation of 1-epoxy-3-butene, and (3) selective epoxidation of 1,3-butadiene and ethylene using molecular oxygen. The compounds and reaction pathways that will be discussed throughout this Perspective are summarized in Scheme 1. Furf ural Hydrogenation and Deoxygenation. The hydrogenation and hydrodeoxygenation (HDO) reactions of furfural and its derivatives have recently been of interest due to the prevalence of these compounds during the dehydration of C5 and C6 sugars and their potential for use as building blocks in
R
eaction selectivity control is a key focus for heterogeneous catalysis research in the search for “greener” processes that produce little byproduct and avoid waste of reactant. For multifunctional oxygenatescompounds that contain two or more diverse functional groups, including alcohols, aldehydes, olefins, and aromatic moietiesthis is a particular challenge due to the varying degree of interaction each functional group has with the catalyst surface. In particular, upgrading of biomass-derived molecules via selective hydrogenation presents a challenge for the catalysis community due to the high degree of functionality present in these molecules.1 In addition, selective oxidations using molecular oxygen are also difficult due to competition with overoxidation including combustion. In this Perspective, we discuss recent work aimed at controlling selectivity in two key classes of reactions for platform chemicals: (1) selective hydrogenation and deoxygenation of bioderived furanic compounds, and (2) reactions involving formation and further functionalization of epoxides. Reactions of furfural have received intense focus in attempts to develop biorenewable fuels and chemicals,1,2 whereas olefin epoxidation has been one of the most important processes in the chemical industry for decades.3,4 Reaction selectivity can be influenced by many factors, including adsorbate−surface, adsorbate−adsorbate, and solvent−adsorbate interactions. One conceptually appealing method for controlling reaction selectivity is to focus on one factor, adsorbate−surface interactions, by controlling the orientation of adsorbed reactive intermediates. For example, furfural presents two functional groups, an aldehyde and an aromatic furyl group. One would anticipate that the rates of competing surface reactions may be related to competition between these two groups for catalyst adsorption sites; that is, the orientation of the adsorbed reactant may be critical in controlling reactivity. Most widely studied has been a class of © 2015 American Chemical Society
Received: February 16, 2015 Accepted: March 26, 2015 Published: March 26, 2015 1348
DOI: 10.1021/acs.jpclett.5b00347 J. Phys. Chem. Lett. 2015, 6, 1348−1356
The Journal of Physical Chemistry Letters
Perspective
Resasco and co-workers found that the weak interaction between the furan ring and Cu(111) causes furfural to adsorb through its carbonyl oxygen in an upright or tilted conformation with the ring away from the surface. The tendency for furfural to adsorb in this configuration was attributed to the repulsive interaction between the 3d band of Cu and the antibonding orbital of the aromatic system.22 Because the preferred binding geometry is very different between Pd and Cu surfaces, it is expected that the reactivity of furfural will also be very different. Experimentally, the surface chemistry of furfural and furfuryl alcohol has been studied on Pd(111).23 TPD investigations reveal that at coverages well below saturation, both furfural and furfuryl alcohol undergo decarbonylation, producing furan, carbon monoxide, and hydrogen. As depicted in Scheme 1b, furfural was found to be an intermediate product during decarbonylation of furfuryl alcohol, that is, furfuryl alcohol decarbonylation was preceded by dehydrogenation. At larger surface coverages, furfuryl alcohol additionally undergoes deoxygenation, producing methylfuran, as seen in Figure 1. Experiments performed with preadsorbed deuterium revealed that the major deoxygenation pathway resulted in production of methylfuran with no deuterium incorporation, implicating a bimolecular pathway where there is direct hydrogen exchange between adsorbed molecules. A higher-temperature methylfuran desorption peak resulted in inclusion of deuterium atoms, implicating a surface reaction. Deoxygenation has also been observed for large exposures of furfuryl alcohol to an oxygenprecovered Pd(111) surface; methylfuran was observed to desorb concurrently with furfural, suggesting a disproportionation reaction resulting in evolution of both molecules.24 The decarbonylation, hydrogenation, and deoxygenation pathways have been studied in detail using DFT calculations by the group of Vlachos.20,25 Agreement is found between experimental and computational results suggesting that both furfural and furfuryl alcohol will undergo facile decarbonylation on Pd to produce furan and carbon monoxide. It is useful to compare the results for furyl oxygenates to other aromatic oxygenates, such as benzyl alcohol and benzaldehyde. Similar coverage-dependent phenomena have been observed for benzyl alcohol adsorbed on Pd(111); at low coverage, decarbonylation is the primary pathway, resulting in production of benzene, whereas at larger exposures, deoxygenation occurs, producing toluene.26 Complementary high resolution electron energy loss spectroscopy (HREELS) investigations revealed that at low exposures, the strong
One conceptually appealing method for controlling reaction selectivity is to focus on one factor, adsorbate−surface interactions, by controlling the orientation of adsorbed reactive intermediates. biorefining.10,11 Surface-level studies have been performed to understand the adsorption geometry of furan on catalytic surfaces and to determine how this geometry affects subsequent reactivity. On Pd(111), furan adsorbs through its aromatic πelectrons, with the plane of the C atoms essentially lying flat and the oxygen atom bent away from the surface, as seen by investigations using scanning tunneling microscopy,12,13 nearedge X-ray absorption fine structure spectroscopy, photoelectron diffraction,14 and angle-resolved ultraviolet photoelectron spectroscopy.15 This structure has also been confirmed by density functional theory (DFT) calculations which additionally show that furan adsorbs with its ring centered around a 3-fold hollow site and an adsorption energy of approximately 100 kJ/mol.16,17 TPD studies show that this strongly adsorbed species decomposes near 300 K to CO and a surface C3H3 species.18,19 The adsorption geometry of furfural (which contains a pendant aldehyde function at the 2-position of the furan ring) has also been studied on Pd(111) by DFT calculations.20 At low coverage, the strong interaction between the furan ring and the surface causes the molecule to adsorb in a flat configuration with an adsorption energy of around 175 kJ/mol. In this geometry, both the furan ring and the aldehyde are in close proximity to the surface. Due to the multifunctionality of the furfural molecule, additional, less energetically favorable binding modes can be identified: Vlachos and co-workers calculated that furfural can bind in a di-σ configuration through its aldehyde, and also in an η1 configuration through an oxygen lone pair, with adsorption energies ranging from 140 to 80 kJ/ mol. A natural question is whether the various orientations can yield different reaction products under hydrogenation conditions, as discussed below. In contrast to Pd(111) surfaces, aromatic furan rings interact very weakly with Cu surfaces. TPD studies show that furan is weakly bound on Cu(111), desorbing intact near 165 K with no evidence for decomposition.21 DFT studies performed by
Scheme 1. Hypothetical Surface Orientations and Reactions of Multifunctional Molecules Considered in This Perspectivea
a (a) Hypothetical surface orientations of multifunctional molecule leading to different reaction products. Reactions of multifunctional molecules considered in this Perspective: (b) Furfural hydrogenation pathways over Pd catalysts. (c) 1,3-butadiene partial oxidation and 1-epoxy-3-butene hydrogenation pathways.
1349
DOI: 10.1021/acs.jpclett.5b00347 J. Phys. Chem. Lett. 2015, 6, 1348−1356
The Journal of Physical Chemistry Letters
Perspective
benzyl alcohol resulted in large yields of dehydrogenation and decarbonylation, producing benzene. On a surface covered with adamantanethiolate (AT, fractional surface coverage ∼0.17),29 the selectivity to HDO increased markedly. Similar experiments performed on Pd(111) covered with a SAM of octadecanethiolate (C18, fractional coverage ∼0.33)30 completely eliminated the decarbonylation pathway, while toluene formation was still observed. These results suggested that thiolates could be used to modify supported Pd catalysts for promotion of selectivity in benzyl alcohol HDO. Experiments performed under hydrogenation conditions confirmed that benzyl alcohol did not undergo decarbonylation on C18-modified Pd/Al2O3, resulting in high selectivity to toluene. However, the overall rate of reaction was decreased significantly by the dense C18 coating. In contrast, AT-modified Pd/Al2O3 was found to offer improved selectivity and overall reaction rate, with the latter effect being due to a reduced tendency toward deactivation through coking compared to the uncoated catalyst. The decreased tendency toward coking was attributed to suppression of aromatic adsorption in flat-lying adsorption modes on the crowded surface. A similar effect was observed for furfural hydrogenation over SAM-modified Pd/Al2O3 catalysts.31,32 On uncoated Pd catalysts, furan was produced with over 95% selectivity, whereas furfuryl alcohol and methylfuran were minor products. In this case, the rate of decarbonylation was about 2 orders of magnitude larger than that of aldehyde hydrogenation and hydrodeoxygenation. Using CO as a probe molecule, diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) studies were performed to identify accessible reaction sites. Unsurprisingly for the uncoated Pd catalyst, CO primarily adsorbed on 3-fold terrace sites, indicating that large terraces were accessible during furfural hydrogenation. The Pd surface was then modified with a C18-SAM. The selectivity to furan dropped to under 30%, consistent with an over two order-of-magnitude decrease in the rate of decarbonylation. CO DRIFTS studies showed a decrease in the magnitude of 3-fold binding, indicating a high coverage of thiolate on the Pd terraces. However, the apparent activation energy for decarbonylation was nearly the same on uncoated and C18-coated Pd catalysts, suggesting that the mechanism of decarbonylation was unchanged after modification with the C18-SAM. It was proposed that the effect of the C18 monolayer was to decrease the average size of contiguous terrace active sites, thereby limiting the amount of furfural that could exist in a flat-lying conformation; however, once furfural did adopt this surface orientation, it proceeded through the same decarbonylation mechanism regardless of catalyst surface. Surprisingly, the rate of aldehyde hydrogenation changed by less than an order of magnitude, and the rate of HDO was essentially unchanged, suggesting that these reaction pathways were not as hindered by the presence of the thiolate monolayer. Additionally, CO binding on top sites, edges and defects was found to be largely unaffected by the presence of the thiolates in DRIFTS studies. Thus, it was proposed that corner, edge, and defect sites were important for aldehyde hydrogenation and dominant for hydrodeoxygenation, likely associated with a state where the ring did not have as much interaction with the surface.32 The proposed reaction pathways on uncoated and C18-coated catalysts are summarized in Scheme 2, demonstrating the site-blocking terrace restriction and proposed surface orientation of adsorbed furfural.
Figure 1. TPD spectra for carbon monoxide (m/z = 28) and methylfuran (m/z = 82) as a function of increasing exposure of furfuryl alcohol on Pd(111). Whereas CO is the dominant volatile product at low precoverage of furfuryl alcohol, methylfuran becomes a major product at high precoverage. Reprinted from ref 23.
interaction between the phenyl ring of benzyl alcohol and the Pd(111) surface causes the molecule to lie flat, resulting in an adsorption orientation that is conducive to decarbonylation. However, at high exposures, crowding between neighboring adsorbates causes the phenyl ring to bend away from the surface, leading to an upright structure primarily bound through the oxygen atom of the alcohol. It was proposed that deoxygenation occurs from this structure. These spectra and the proposed structures are shown in Figure 2. Similar exposure-dependent structures and products were observed for benzyl alcohol on oxygen-precovered Pd(111) as well.27 These surface science studies show the importance of surface orientation in determining the reaction selectivity of multifunctional molecules; in this case, the orientation was influenced by the total coverage (i.e., crowding) of coadsorbed molecules. Controlling adsorbate coverage is not a typical method for design of technical catalysts; normally, surface coverages are dictated in large part by reaction conditions and are not directly manipulated. However, one method for manipulating surface crowding is to prepare surfaces with adsorbed spectator species of adjustable density. For example, alkanethiolates are known to form self-assembled monolayers (SAMs) on late transition metal surfaces, where the surface coverage of the thiolates can be controlled by the steric bulk of the ligands. These effects have recently been studied for thiolate-modified Pd(111).28 TPD experiments performed on unmodified Pd(111) with
Figure 2. HREEL spectra as a function of increasing exposure of benzyl alcohol on Pd(111), reprinted from ref 26. Features at 740 and 990 cm−1 are associated with out-of-plane C−H stretches on a flat lying structure, whereas the feature at 3050 cm−1 is associated with inplane C−H stretches on an upright structure. 1350
DOI: 10.1021/acs.jpclett.5b00347 J. Phys. Chem. Lett. 2015, 6, 1348−1356
The Journal of Physical Chemistry Letters
Perspective
Scheme 2. Surface Orientations and Corresponding Reaction Pathways for Furfural and Furfuryl Alcohol on Uncoated and Alkanethiolate-Coated Pd Catalystsa
a
Reprinted from ref 32.
Adamantanethiolate (AT) and 1,2-benzenedithiolate (BDT) were used to probe the effect of thiolate coverage on surface reactivity. AT forms a sparsely packed monolayer (0.17 monolayer coverage on a (111) surface)29 that was found to minimally affect reaction selectivity compared to the uncoated catalyst. CO DRIFTS studies indicated that this catalyst had large patches of contiguous active sites available for furfural to bind flat on the surface and decarbonylate, leading to similar selectivity for this catalyst as the uncoated catalyst.31 BDT forms a more densely packed monolayer than C18 (0.50 monolayer coverage of sulfur),33 and correspondingly had even higher selectivity to furfuryl alcohol and methylfuran. Interestingly, the activation energies for all processes were statistically identical between the C18-coated and BDT-coated catalysts, suggesting that the effect of high coverage of surface sulfur is to effectively isolate Pd active sites but that additional sulfur did not alter the rate-limiting mechanism.32 The observations about the influence of surface orientation on reaction selectivity from surface science and catalysis studies for aromatic alcohol reactions have been corroborated by DFT and microkinetic modeling studies.34 As surface coverage of furfural or hydrogen increases, the adsorption energy of flatlying furfural decreases and a tilted or upright geometry becomes more favorable, indicating a shift in surface orientation. In addition, the activation barrier for decarbonylation was calculated to increase as the surface became very crowded. This led to a decrease in the microkinetic modelpredicted rate of reaction for decarbonylation. In contrast, the rate of reaction was expected to increase slightly at high hydrogen coverage; correspondingly, the selectivity to furfuryl alcohol was calculated to exhibit a maximum at high surface coverage. It is expected that this high coverage limit is relevant to catalysis studies where the surface is covered with thiolate and adsorbed hydrogen. Bimetallic catalysts have also been used to control the selectivity of furfural hydrogenation through surface orientation effects. NiFe catalysts are over 65% selective to methylfuran, whereas pure Ni catalysts have very low selectivity and pure Fe catalysts barely exhibit any activity.35 DFT studies revealed that the presence of oxophilic Fe enhances the interaction between the carbonyl and the surface, lengthening the C−O bond, which facilitates deoxygenation. On the other hand, PdCu catalysts exhibit high selectivity to furfuryl alcohol compared to Pd, at the expense of overall catalytic activity.36 In this case, the presence of surface Cu causes the aromatic ring to be repelled from the surface while allowing the carbonyl to be hydrogenated.22
CuNi catalysts also can take advantage the repulsion between the aromatic ring and Cu to increase selectivity to furfuryl alcohol and methylfuran.37 However, under reaction conditions, Ni tends to surface segregate, leading to a loss of selectivity as the amount of Ni at the surface is increased. Unique synergistic effects were observed upon addition of a C18 monolayer to a Cu-rich bimetallic catalyst (Cu4Ni): high selectivity was retained and, surprisingly, the activity of aldehyde hydrogenation increased. Ambient-pressure XPS studies revealed that the presence of the C18 monolayer kinetically stabilized Cu at the surface such that Ni was unable to surface segregate during the reaction, as seen in Figure 3. However, retention of small amounts of surface Ni hypothetically allowed for hydrogen spillover from surface Ni to surface Cu sites, likely providing the origin for increased hydrogenation activity while retaining high selectivity. As demonstrated in these studies, the surface orientation of furfural is crucial to understanding its reactivity and reaction selectivity. For highly active metals such as Pd or Ni, flat-lying structures allow for ring-surface interaction that promotes decarbonylation; aldehyde hydrogenation and hydrodeoxygenation selectivity is correlated with upright structures where the ring does not contact the surface. This kind of structure can be achieved through the use of surface modifiers such as alkanethiolates or noble metals. Reactions of other multifunctional molecules can also benefit from control of surface orientation. Reaction of Epoxides. Multifunctional epoxides also present a challenge for selective hydrogenations over heterogeneous catalysts due to the high reactivity of the strained epoxide ring. The surface chemistry of 1-epoxy-3-butene (EpB) has been studied on Pt(111) and Pd(111) with the goal of improving hydrogenation selectivity to 1-epoxybutane over ring-opened products such as crotonaldehyde. On Pt(111),38 adsorption occurs through the olefin; electron energy loss spectroscopy studies conducted after adsorption at
