Bifunctional Catalysts for Upgrading of Biomass-Derived Oxygenates

Subscriber access provided by UNIVERSITY OF KENTUCKY

Review

Bifunctional catalysts for upgrading of biomass-derived oxygenates: A Review Allison M Robinson, Jesse E. Hensley, and J. Will Medlin ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.6b00923 • Publication Date (Web): 21 Jun 2016 Downloaded from http://pubs.acs.org on June 25, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

Bifunctional catalysts for upgrading of biomass-derived oxygenates: A Review Allison M. Robinson a, Jesse E. Hensley b, J. Will Medlin a,* a

Department of Chemical and Biological Engineering, University of Colorado – Boulder, UCB 596, Boulder, Colorado 80309, United States * Email [email protected] b National Bioenergy Center, National Renewable Energy Laboratory, 15013 Denver West Pkwy, Golden, Colorado 80401, United States TOC Graphic Intermediates Derived from Fast Pyrolysis Phenolics R'

Acids O

RO

R' R

OH

Deoxygenated chemicals and fuels R

R

H2

O R

Bifunctional Catalysis

R

OH

OH

R'

O R"

Ketones and Aldehydes

Material 1

Material 2

R

R'

Abstract Deoxygenation is an important reaction in the conversion of biomass-derived oxygenates to fuels and chemicals. A key route for biomass refining involves the production of pyrolysis oil through rapid heating of the raw biomass feedstock. Pyrolysis oil as produced is highly oxygenated, so the feasibility of this approach depends in large part on the ability to selectively deoxygenate pyrolysis oil components to create a stream of high-value finished products. Identification of catalytic materials that are active and selective for deoxygenation of pyrolysis oil components has therefore represented a major research area. One catalyst is rarely capable of performing the different types of elementary reaction steps required to deoxygenate biomass-derived compounds. For this reason, considerable attention has been placed on bifunctional catalysts, where two different active materials are used to provide catalytic sites for diverse reaction steps. Here, we review recent trends in the development of catalysts, with a focus on catalysts for which a bifunctional effect has been proposed. We summarize recent studies of hydrodeoxygenation (HDO) of pyrolysis oil and model compounds for a range of materials, including supported metal and bimetallic catalysts as well as transition metal oxides, sulfides, carbides, nitrides, and phosphides. Particular emphasis is placed on how catalyst structure can be related to performance via molecular-level mechanisms. These studies demonstrate the importance of catalyst bifunctionality, with each class of materials requiring hydrogenation and C-O scission sites to perform HDO at reasonable rates.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Key Words Bifunctional, catalyst, hydrodeoxygenation, bimetallic, metal-metal oxide, tethered bifunctional catalysts 1. Introduction Increasing environmental, economic, and political motivation to develop renewable liquid fuels and chemicals has generated significant interest in using biomass as a sustainable carbon source. Upgrading of lignocellulosic biomass provides a unique opportunity to utilize a renewable resource to create these fuels and chemicals from waste materials and plants that do not compete with food production. Through fast pyrolysis, the biomass is rapidly heated in the absence of oxygen to produce a condensable mixture of oxygenates along with char and light gases.1-3 Pyrolysis oil composition varies depending on the feedstock and process conditions, but contains high concentrations of phenolic compounds from lignin decomposition and furanic compounds derived from cellulose, in addition to organic acids and other low molecular weight oxygenates.4 This complex mixture of oxygenates generally has poor properties as a fuel.1,5,6 The undesirable fuel characteristics stem from a high concentration of highly oxygenated compounds, making upgrading through deoxygenation critical for improving heating value, stability, and physical properties such as acidity and viscosity. Conversion of the compounds making up pyrolysis oil requires new types of catalysts that can selectively remove oxygen from the oil.9-11 One common strategy for deoxygenating whole pyrolysis oil is hydrodeoxygenation (HDO), which involves adding hydrogen to accomplish C-O bond scission via removal of water. This is typically done sequentially, with a hydrogenation-dehydration-hydrogenation route in which dehydration often occurs over an acid site. Alternatively, direct deoxygenation (DDO) does not require intermediate hydrogenation steps; instead, the C-O bond is directly cleaved from the reactant through hydrogenolysis. For clarification, it should be noted that while these definitions of HDO and DDO are used throughout this manuscript, some previous reports have used different terminology for the same reactions. Other reactions can also take place under these conditions, including decarboxylation/decarbonylation, cracking, and hydrogenation (Figure 1). 7,8 It should be noted that while HDO reactions often produce saturated compounds and DDO may lead directly to aromatic compounds, aromatic and saturated products can also be interconverted subsequent to being formed. For example, increasing the temperature shifts the thermodynamic equilibrium towards aromatic products such as toluene production over methylcyclohexane.

ACS Paragon Plus Environment

Page 2 of 44

Page 3 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

Decarbonylation/Decarboxylation O R1

R1H + CO H

O R1

R1H + CO2 OH

Hydrodeoxygenation R1

R1 OR2

H2

+ R2OH

Direct Deoxygenation R1

R1 OR2

H2

+ R2OH

Hydrogenation R1

R1 OR2

OR2

H2

Cracking R1

R1 OR2

R1

OR2

R2

R1

+ R2

Figure 1: Example reactions discussed in this review A wide variety of materials have been studied for pyrolysis oil upgrading, ranging from carbides and nitrides to supported transition metals, but despite their diversity the most promising materials are bifunctional: they contain disparate active sites that catalyze complementary reaction steps. In many cases upgrading reactions demand a combination of the different reaction types described in Figure 1.; it is natural that each step would require its own catalyst. In this contribution, we review advances in catalysts currently used for HDO with a focus on bifunctional catalysts. There have been a wide variety of studies attempting to better understand how HDO proceeds over different catalytic materials using various probe molecules or whole oils. Comparing these approaches may help elucidate which types of materials are most promising for the end goal of making pyrolysis oil a viable alternative to petroleum in fuel and chemical

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

production. While pyrolysis oil is known to contain many different compounds and types of functional groups9, the majority of the studies referenced in this report use phenolic compounds as a probe molecule. This will aid in gaining an understanding of not only how the current state of the art deoxygenation catalysts preform, but also how they relate to one another. 2. Bifunctional Catalysis The general concept behind bifunctional heterogeneous catalysis is that there are two distinct types of active sites that function in tandem to perform a surfacecatalyzed reaction. Often, these two types of sites are expected to catalyze different elementary steps within an overall reaction. In principle, however, the two sites could participate in the same step, for example through interaction of different parts of an adsorbed reactant molecule with the catalyst as in a concerted reaction. In this section, we highlight two major types of materials used as bifunctional catalysts before considering the application of similar types of materials in biomass refining reactions. Proposed mechanisms for two major classes of catalysts discussed in this review are shown schematically below in Figure 2 a-b. To point out the connections between these examples for simple reactants of well-studied reactions and the oxygenate upgrading reactions of primary interest for this review, an example of bifunctional catalysts for deoxygenation of phenolic oxygenates is shown in Figure 2c.

ACS Paragon Plus Environment

Page 4 of 44

Page 5 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

a. Bimetallic Catalysts CO adsorption and O2 dissociation on Pd sites CO

O2 O O

CO

CO O

Pd Au Au Pd Pd Au Pd

O

CO

Pd Au Au Pd Pd Au Pd

O spillover to Au sites

CO oxidation by Au sites

CO2

CO2

Pd Au Au Pd Pd Au Pd b. Metal-Metal Oxide Catalysts CO2 adsorption to O on ZrO2 and CH4 decomposition on Pt

CH4

CO2

H2

CO

C OCO H H

Pt

COH

Pt

ZrO2

H

CO formation over Pt and formate production over ZrO2

ZrO2 CO H

Pt

Formate decomposition over ZrO2 to CO and OH

H

ZrO2 c. Deoxygenation over Bimetallic Catalysts H2

H2 dissociation on Pt sites Ring adsorption on Pt and oxygen adsorption on Mo sites

H H

OH H H

H H

Pt Pt Mo Pt Pt

O

H H

Pt Pt Mo Pt Pt Tautomerization on Mo site

H H

H2O

Deoxygenation and ring hydrogenation

Pt Pt Mo Pt Pt

Figure 2: Schematic representation of proposed reaction mechanisms by a) bimetallic catalysts10, b) metal-metal oxide catalysts11 , and c) bimetallic catalysts specifically for deoxygenation12. 2.1. Examples of bifunctional catalysis 2.1.1 Bimetallic catalysts One of the most common methods for controlling the catalytic performance of supported metal catalysts is via control over the composition of bimetallic

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

catalysts. In a large number of cases, adding one metal to another is not intended to introduce a new catalytic function, but instead modifies the activity of the original metal. A second metal may influence the properties of the first via ligand effects, which is often interpreted in terms of a shifting of the d-band density of states of the first metal. A much-studied example of this phenomenon is in the performance of Pt “skin” catalysts that have been modified by 3d elements such as Ni, Co, Fe, and Ti in the subsurface for improved performance in the electrocatalytic oxygen reduction reaction.13-15 In many of the systems dominated by this effect, the modifying metal is assumed to not be present in the surface layer, and thus has little opportunity to directly create a bifunctional effect.16 Another mechanism by which addition of a second metal can modify a catalyst surface is via the so-called “ensemble” effect.17 For example, addition of (relatively unreactive) Au to Pd olefin hydrogenation catalysts has yielded an improved activity that has been attributed to a decrease in the average number of contiguous Pd atoms; contiguous surface Pd clusters have been associated with accumulation of unreactive spectator species.18 In some cases, however, the presence of two metals at the surface under reaction conditions has been associated with a bifunctional effect. Bifunctional effects have been invoked, for example, in some studies involving PdAu catalysts. Xu and coworkers found that biphasic PdAu nanoparticles—where segregation of the two metals into relatively Pd-rich and Au-rich—were highly active for lowtemperature CO oxidation.10 The authors found that Pd-rich sites were needed to carry out the dissociative adsorption of oxygen, whereas Au-rich sites were active for CO oxidation, as shown schematically in Figure 2a. The dispersion of the two metals in an active catalyst has been seen experimentally through STM images of PdAu alloy electrodes. In this example, PdAu thin films were electrodeposited on Au(111), resulting in Pd monomers, dimers, and trimers on the surface (Figure 3).19 Scanning tunneling microscopy was used to obtain TM images of different PdAu compositions, which were found to contain highly dispersed Pd atoms. This dispersion is correlated with activity in Figure 3; the coverage of hydrogen and CO on the surface was measured along with the coverage of monomers, dimers, and trimers of Pd.19 Because CO adsorbs on Pd sites, the coverage of CO increases with Pd loading. Furthermore, IR results showed that the CO adsorbed on a top site in these disperse samples, while large Pd clusters (or a pure Pd surface) had a high concentration of bridge/threefold bound CO.19 This high dispersion creates many of the neighboring Pd-Au sites responsible for the bifunctional reaction shown schematically in Figure 2a.

ACS Paragon Plus Environment

Page 6 of 44

Page 7 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

Figure 3: The surface coverage of monomers, dimers, and trimers of Pd for two surfaces along with experimental results for the coverage of H and CO on these two electrodes. Reprinted from 19 with permission from Science. Similar effects have been proposed in olefin hydrogenation, where Pd sites are necessary for dissociative hydrogen adsorption while Au sites are active for hydrogenation.20 Bimetallic catalysts have also been employed such that different types of sites are effective for reactions under different conditions. For example, PtAu nanoparticles have been applied as electrocatalysts for lithium-air batteries. The Au component was found to be more active for the oxygen reduction reaction (which occurs during battery discharging), while Pt sites were associated with the reverse oxygen evolution reaction (which occurs during charging).21 The utility of combining two active sites in one material has also been shown for use in unitized regenerative fuel cells (URFCs). These electrochemical cells operate in one mode to electrolyze water, producing hydrogen and oxygen, then operate in fuel cell mode to recombine them and produce electricity.22,23 The challenge with this concept is designing an electrocatalyst that is active for all of the reactions involved; catalysts active for the oxygen reduction reaction (ORR) perform poorly for the oxygen evolution reaction (OER).22,23 Finding a catalyst that is both stable and active for OER, which involves oxidizing water to molecular oxygen in acidic media, has represented a major research focus in electrocatalysis. This is due in part to the limited studies on the relationship between catalyst active site structure and OER activity.24 While Pt is nearly ideal for ORR, it has shown poor OER activity. The lower rates have been attributed to reduced surface area and formation of a poorly conductive PtOx layer that resists current flow under the conditions of OER. This may be overcome by using a bifunctional catalyst consisting of the conventional Pt particles used for ORR supported on an oxide such as IrO2 for the oxygen evolution reaction.22,23 2.1.2. Metal-metal oxide catalysts A frequently used type of bifunctional catalyst combines metal sites that are active for hydrogenation/dehydrogenation and acid sites that are active for protonation/deprotonation. For example, catalytic metals may be supported on a zeolite. In the isomerization of alkanes, the saturated hydrocarbons are

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

dehydrogenated to alkenes on the metal sites; the alkenes can subsequently migrate to acid sites, which protonate the alkenes to form carbenium ions.25 Structural rearrangement of the carbenium ion can lead to formation of branched structures, which upon deprotonation yield branched alkenes. Finally, the branched alkenes are hydrogenated on metal sites. While these reactions can be catalyzed by physical mixtures of metal and zeolite catalysts, achieving intimate contact between the different types of reactive sites significantly improves the kinetics. Producing zeolite-supported metals with high dispersion is important for improving the intimacy of contact and achieving rapid isomerization kinetics.26 The intimacy of metal-metal oxide contact has also been found to be important in other reactions, with considerable focus devoted to Au nanoparticles dispersed on reducible oxide supports.27 For example, Au supported on ceria28,29 or titania30 has been found to be active for the water-gas shift reaction at relatively low temperatures, ranging from approximately 200-350°C. The metal oxide support has been reported to be active for the required water dissociation step, whereas CO oxidation has been associated with sites at the Au-metal oxide interface. A related mechanism has been demonstrated on Pt/ZrO2 catalysts for dry reforming of methane (Figure 2(b)). It was shown that the Pt metal serves to activate methane, dehydrogenating it to form CHx, while sites at the Pt-ZrO2 interface are responsible for CO2 dissociation. Achieving good catalyst performance was found to require optimization of the interfacial contact area.11 Furthermore, bifunctional metal – metal oxide catalysts in which the oxide serves as a base have also been reported. For example, isobutanol may be produced using syngas (CO and H2) on bifunctional catalysts consisting of Cu supported on basic metal oxides such as MgO.31 The syngas initially forms alcohols that may be coupled through condensation reactions. In the proposed mechanism, the alcohol dissociatively adsorbs on the basic oxide surface, creating hydrogen and alkoxide species.31 Hydrogen is removed by migration to Cu sites, where they recombine and desorb as H2, while the alkoxide species may be further dehydrogenated and eventually take part in condensation reactions to form longer chain compounds.31 Chain growth rates are low due to the steric hindrance that occurs as branches are added to the initially formed linear alcohols, resulting in production of isobutanol at high selectivity.31 2.2. Importance of bifunctionality for biomass reforming A common theme across the classes of materials studied for biomass deoxygenation is their bifunctionality. As discussed later in this review, the importance of having sites capable of activating hydrogen in close proximity to sites that are effective for C-O scission has been proposed many times throughout the literature for materials ranging from supported transition metals to traditional hydrotreating catalysts such as sulfides.32-44 Often these two reaction steps cannot be effectively catalyzed by one type of active site. As a result, the most commonly proposed mechanisms for deoxygenation of these compounds involve one active site (such as a transition metal) that readily activates hydrogen and another type of site nearby, such as an acid site on an oxide support, for C-O bond activation. While

ACS Paragon Plus Environment

Page 8 of 44

Page 9 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

this can be studied using whole pyrolysis oil, the complex mixture of oxygenates present makes it very difficult to elucidate reaction pathways to understand mechanisms. For this reason it can be useful to instead consider more detailed studies using model compounds. Phenolic compounds are useful probe molecules for pyrolysis upgrading reactions; they make up a large fraction of the oil and the CO bonds in these molecules are very difficult to break.33,45-47 Other processes more representative of alternative conversion approaches, such as aqueous phase processing of polyols, are also described below. 3. Bifunctional catalysts for deoxygenation 3.1 Supported Metals Many supported metal catalysts have been investigated for deoxygenation reactions. While it is difficult to directly compare the variety of different catalysts tested under varying reaction conditions, it may be informative to look at general trends in performance that has been observed in the literature. Figure 4 shows the selectivity to deoxygenated products when feeding aromatic oxygenates including cresol and guaiacol.

Figure 4: Selectivity to deoxygenated products from feeding vapor phase aromatic oxygenates (cresol or guaiacol) over Pt (diamonds), Ru (crosses), Fe (squares), Pd (triangles), and Ni (circles) catalysts. Color indicates the reference the data point was taken from. Across a range of conversions, Pt catalysts appear to have the highest deoxygenation selectivity of these monometallic samples. The data used in this figure can also be seen in Table 1. However, the performance of these monometallic catalysts can be dramatically changed through incorporation of a second metal, as will be discussed below.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 44

Table 1: Data used in Figure 4 to compare selectivity to deoxygenation of m-cresol or guaiacol over various supported metals. Catalyst Pt/Al2O3 Pt/Al2O3 Pt/Al2O3 Pt/SiO2 Pt/ Al2O3 Pt/ TiO2 Pt/ ZrO2 Ni/SiO2 Ni/C Ni/SiO2 Pd/C Pd/SiO2 Pd/C Fe/SiO2 Fe/SiO2

Temperature (K) 533 533 533 533 533 573 573 523 623 573 573 523 623 573 623

P (atm) 1 1 1 1 1 1 1 1 39.5 1 1 1 39.5 1 1

PH2 (atm) 0.5 0.5 0.5 0.5 0.5 1 1 1 37.6 1 0.4 1 37.6 1 0.9

Ru/SiO2 Ru/C

573 623

1 39.5

1 37.6

wt% metal 0.5 1 1.7 1.6 1.7 1 1 5 5 5 5 1 1 5 not reported 9.4 1

reactant m-cresol m-cresol m-cresol m-cresol m-cresol m-cresol m-cresol m-cresol guaiacol m-cresol m-cresol m-cresol guaiacol m-cresol guaiacol

conversion (%) 9.3 18 38.3 54.9 30 17 12 55.8 30.7 16.2 10 54.9 15.5 8.8 50

% deoxygenation 79.2 82.6 80.5 82.1 76.1 97 88.5 7.2 17.5 14.2 78 24.1 23.3 60.2 37

Ref. 33 33 33 33 32 66 66 68 60 34 61 68 60 34 76

m-cresol guaiacol

4.5 34.2

38.5 39.6

67 60

3.1.1. Supported Noble Metals One of the most widely used types of catalysts for upgrading of biomassderived oxygenates are supported metals. To understand reactivity trends for these systems, it is often useful to study reactivity on model surfaces such as single crystals. Reactions of simple organic oxygenates that can serve as probe molecules for components in deconstructed biomass have been extensively studied using surface spectroscopies and model surfaces. There have been past reviews focusing on the decomposition of simple oxygenates including alcohols, aldehydes, ketones, and carboxylic acids over transition metal surfaces.48-51 For example, formic acid decomposition has been studied as a probe for organic acids. A past review of surface science studies over single crystals focused formic acid adsorption and decomposition processes.48 Formic acid tends to decompose through a formate intermediate. The formation of a carboxylate intermediate has been observed for larger carboxylic acids as well, over a wide array of transition metals.51 Formate decomposition can then occur either through dehydrogenation to produce CO2 and H2, or C-O scission to form CO and H2O.48 The pathway that a given transition metal will favor depends on how strongly it binds oxygen: metals like Pt that form relatively weak metal-oxygen bonds favor dehydrogenation52 while metals further left on the periodic table such as Fe favor C-O scission.53 On Pd(111) both pathways were observed, although selectivity was greater towards dehydrogenation.54 On Ni(111) the binding configuration was proposed to be different, creating formic anhydride as an intermediate, but the dominate decomposition pathway was C-O scission.55 A major conclusion from these studies is that C-O bond scission is often

ACS Paragon Plus Environment

Page 11 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

not favored on metals like Pt and Pd that do not strongly bind oxygen, even when the C-O bond is relatively weak. Similar observations have been made for reactions of other oxygenates such as alcohols and aldehydes. On noble metals such as Pt and Pd, decarbonylation via C-C scission is often a dominant pathway, and C-O scission is generally not observed.49 An exception is the reaction of unsaturated alcohols and aldehydes such as prenal, furfural, and benzyl alcohol, where the C-O bond is much weaker than in aliphatic or phenolic reactants. However, noble metals are attractive for their ability to activate hydrogen and to conduct hydrogenation reactions. This suggests that deoxygenation for reforming pyrolysis oil may be best accomplished using a bifunctional catalyst that combines hydrogenation sites with sites that form strong metal-oxygen bonds. This hypothesis is also supported by studies of technical catalysts composed of combinations of metal and acid sites. Late transition metal catalysts have several properties that make them potentially attractive for pyrolysis oil upgrading; they are typically very active, are effective at hydrogenation, and have well characterized structures. Additionally, they do not suffer from some of the issues of traditional hydrotreating catalysts discussed below, such as the need for co-feeding toxic H2S to maintain activity as for sulfides.56 When an active support material is used to impart bifunctionality these catalysts can be used for HDO of relatively recalcitrant compounds such as phenolics. Before discussing the bifunctional materials, it is useful to first discuss the individual components. Noble metals including Ru, Rh, Pd, and Pt are a class of promising upgrading catalysts; they activate hydrogen, are not deactivated by water, and are already widely used in commercial catalytic processes.57 Under typical upgrading conditions the differences in the chemistry catalyzed by the various noble metals are subtle8, and these materials have been highlighted in a number of reviews discussing pyrolysis oil upgrading.1,7,8,39,58 As an illustration of their HDO activity as monofunctional catalysts, Pd/C, Pt/C, and Ru/C were studied for vapor phase guaiacol (2-methoxyphenol) HDO.59 While the hydrogenation activity was high, the paper concluded that carbon-supported precious metals are poor HDO catalysts because they favor ring saturation. At 250°C the C-O bond between the ring and methoxy group was broken, but C-O scission of the strong aromatic-hydroxyl bond did not readily occur, producing phenol as the major product.59 Additionally, ring hydrogenation to form cyclohexanone and cyclohexanol occurred, and a small amount of benzene was formed as well.59 Even at harsher reaction conditions (350°C and 39.5 atm) noble metal catalysts show poor selectivity to complete deoxygenation.60 Supports with acid/base activity have been widely used to achieve better deoxygenation yields from phenolic reactants. The aromatic-O bonds of phenolic compounds are weakened following ring hydrogenation, so it has been proposed that the deoxygenation mechanism over noble metals involves hydrogenation followed by dehydration rather than direct C-O scission.32-34 In this model, one site is proposed to activate C-O bonds. This component is often proposed to be a

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 44

Brønsted acid site on the oxide support materials. In addition, sites are needed to supply atomic hydrogen to the reaction; this role is played by the noble metal.39 Proposed mechanisms from model compound studies on noble metal catalysts consistently invoke hydrogenation followed by dehydration, after which further hydrogenation steps can occur.8,32,33 This mechanism is shown in Figure 5. M = transition metal or bimetallic particle R

R

R

Hydrogenation on M sites OH H

H

OH H

H

M

OH

M

Support

Support

Dehydration over acid sites on bimetallic particle

Dehydration over acidic support

R

R

R

R

H 2O H

H M

OH OH

Acidic support (ex. Al2O3 )

H

OH H2O OH

Support = hydrogenation site = oxophilic site

Figure 5: Schematic representation of metal-acid catalysis proposed for aromatic deoxygenation on noble metal based catalysts. The deoxygenated product in Figure 5 is shown converting between the aliphatic and aromatic compound to suggest that depending on the catalyst and the reaction conditions either could be possible. Aromatic oxygenates may be fully saturated over the metal particles, particularly if the reaction conditions favor this (e.g., at lower temperatures).32,33 Bimetallic particles may be used to eliminate the need for an acidic support by providing oxophilic sites that catalyze dehydration. For example, Fe can be used to promote Pd deoxygenation activity either by depositing Pd on an Fe2O3 support or by creating PdFe particles on a carbon support.59,61 In other systems, an acidic support such as alumina may still be needed for dehydration activity. For example, PtCo and PtNi bimetallic particles were prepared to increase hydrogenation activity and therefore increase the rate of deoxygenation through sequential hydrogenation followed by dehydration on the support.32

ACS Paragon Plus Environment

Page 13 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

One report studying the vapor phase HDO of m-cresol (3-methylphenol) over Pt/Al2O3 found that a bifunctional mechanism was required, involving hydrogenation activity to form alcohol intermediates as well as acidity for dehydration.32 It was proposed that ring hydrogenation first occurs to saturate the phenyl ring and a partially or fully hydrogenated alcohol intermediate is formed. On Pt/Al2O3 the partially hydrogenated alcohols may then deoxygenate on acidic sites on the support to form toluene.32 This hypothesis was further supported by reacting the fully hydrogenated alcohol intermediate (3-methylcyclohexanol) on a bare Al2O3 support, where it was deoxygenated to form methylcyclohexene isomers.32 The rate was much higher when the saturated intermediate 3-methylcyclohexanol was fed compared to the rate when m-cresol or an alkene was fed. This result suggested that at 533 K ring hydrogenation over Pt sites to form the alcohol intermediates was rate limiting. The proposed rate-limiting step of ring hydrogenation was used to explain the higher selectivity of Pt/Al2O3 towards toluene than methylcyclohexane, because fewer steps are required to dehydrate the unsaturated alcohol intermediates than to produce saturated alkanes. Note that this mechanism requires intimate contact between the hydrogenation and deoxygenation sites to prevent over-hydrogenation of the relatively unstable intermediate. Also, despite bare alumina easily catalyzing dehydration of the saturated alcohol no reaction was observed when feeding mcresol; at least partial hydrogenation was required to allow cleavage of the C-O bond.33 To improve the HDO activity, Ni and Co were added to form bimetallic catalysts. These 3d metals have previously been shown to increase hydrogenation activity of unsaturated compounds.62 By increasing the hydrogenation activity through these bimetallic catalysts the HDO activity increased and the product distribution shifted to produce more methylcyclohexane relative to the amount of toluene formed. Total deoxygenation also increased over the bimetallic catalysts, proposed to be a result of generating aluminate species through interaction between the 3d metal and the support, which are active for alcohol dehydration.32

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Pt/SiO2 was also studied to compare the alumina-supported material to a less acidic catalyst.33 While both contain acid sites, the Lewis and Brønsted acidity on Al2O3 63,64 is stronger than that of SiO2.65 The less acidic Pt/SiO2 was found to be ineffective at dehydrating 3-methylcyclohexanol to the alkane product. A kinetic study of Pt/SiO2 also concluded that this dehydration pathway is limited.66 The dominant deoxygenation pathway was instead proposed to be dehydration of unsaturated alcohol intermediates to form toluene. While SiO2 alone was not active for dehydration of 3-methylcyclohexanol, in the presence of Pt its weakly acidic hydroxyl groups can dehydrate partially hydrogenated alcohol intermediates present on the Pt particles as shown in Figure 5.33 Both the Pt and SiO2 sites are necessary for this reaction. When the support was doped with potassium to suppress the acidity the product distribution was substantially altered. Rather than primarily toluene, m-cresol formed 3-methylcyclohexanol with 84% selectivity.33 This is in agreement with DFT calculations over Pt(111), which show that m-cresol HDO is less energetically favorable than ring hydrogenation.67 However, increasing the reaction temperature may increase the favorability of the HDO pathway.68 Another study of support effects on noble metals also concluded that acidic supports facilitate phenolic HDO. Pt, Rh, Pd, and Ru supported on Al2O3, SiO2-Al2O3, and nitric-acid-treated carbon black (NAC) were studied for the liquid phase batch upgrading of guaiacol.35 The support was found to have a greater influence on the product distribution than the noble metal. Using ammonia TPD, the acidity of the catalysts was found to decrease in the order SiO2- Al2O3 > Al2O3 > NAC.35 The same trend was found in production of cyclohexane, the fully deoxygenated and hydrogenated product. The least acidic catalysts, those supported on NAC, had the highest yields of 2-methoxycyclohexanol, a result of ring hydrogenation without any C-O scission and an indication of low HDO activity.35 Therefore, HDO was again proposed to need a bifunctional catalyst where acid sites are required to produce a deoxygenated product rather than stop at a ring saturated oxygenate. In addition to HDO through hydrogenation reactions followed by dehydration, direct deoxygenation (DDO) has been observed over supported metal catalysts. As shown in Figure 1, DDO does not require intermediate hydrogenation steps; instead, the C-O bond is directly cleaved from the aromatic reactant. This may be desirable for reducing the hydrogen demand of the deoxygenation process. One example of DDO is using noble metals supported on reducible oxides such as TiO2. For instance, Ru/TiO2 has been shown to catalyze DDO of phenol to benzene through a bifunctional mechanism at the interface between the metal and oxide.69,70 Using both experimental and computational techniques it was demonstrated that at 573 K and 3.8 MPa water that is present on the TiO2 may either accept or donate protons across the metal-metal oxide interface. This allows the phenolic OH group to be directly removed through donation of a proton, shown in Figure 6. Although this is typically difficult to achieve due to the strong aromatic-oxygen bond, the bifunctional sites present at the interface containing both metal and acid functionalities were calculated to have a lower energy barrier than hydrogenation. On large Ru particles, where the interfacial area is much smaller, ring hydrogenation

ACS Paragon Plus Environment

Page 14 of 44

Page 15 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

was found to be most favorable as expected from previous studies of noble metals. Isotopic labeling experiments in combination with DFT calculations indicated that the Ru particles bind the phenol and activate hydrogen to catalyze the hydride attack (which regenerates the water co-catalyst on the support) while Brønsted acidic water on TiO2 catalyzes C-O scission. H

H H H a.

Ru

Ru

O Ti Ti

H

H O

Ti Ti H

b.

Ru

O

H O H

H O

Ti Ti

H

H Ru

O Ti Ti

c.

Figure 6: Proposed mechanism of DDO at the bifunctional Ru/TiO2 interface. (a.) H2 is activated on Ru particles and heterolytically dissociates across the interface to protonate a bridging hydroxyl group on TiO2, generating a Brønsted acid site. (b.) This then assists in phenol deoxygenation. (c.) Side view of DFT calculation of waterassisted phenol DDO on Ru/TiO2, including initial state (left panel), transition state (center panel), and final state (right panel). Color code: hydrogen, white; oxygen, red; carbon, dark gray; ruthenium, teal; titanium, light gray. Reproduced from 69 with permission from ACS Catalysis. This illustrates the importance of better understanding the metal-support interface, especially for reducible metal oxides like TiO2 that may act as either an acid or a base. Similar results have been observed for Pt/TiO2. The combination of a metal and a reducible oxide support were shown to enhance m-cresol deoxygenation by introducing additional energetically favorable reaction pathways.71 While Pt/C preferred ring hydrogenation, Pt/TiO2 catalyzed tautomerization and DDO. This synergistic effect was particularly important at elevated temperatures (623 K).71 There has also been some interest in using zeolites as the solid acid component of these bifunctional systems. While their low activity towards phenolic reforming and high rates of coke generation make zeolites potentially less attractive than the other upgrading catalysts highlighted here, these issues can be improved by adding metals to the zeolite. Using guaiacol as a model compound, only transalkylation was found to occur over HY zeolite, resulting in no deoxygenated products.72 Similar results were observed for Pt/Al2O3 in the absence of hydrogen.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

However, phenol could be hydrogenated and deoxygenated over several zeolites that had been doped with Pt.73 Pt/HY has also been used for HDO of phenolic compounds, with deoxygenation as well as coupling reactions forming larger molecules.74 In addition to these purely heterogeneous examples, supported metal catalysts have also been combined with liquid acids to achieve high deoxygenation selectivity. For example, an aqueous phase phenolic mixture of pyrolysis oil model compounds was upgraded over carbon-supported noble metal catalysts and a mineral acid.38 In neutral solution the noble metal catalysts favored hydrogenation of phenol to cyclohexanol. However, in the presence of phosphoric acid the alcohol was easily dehydrated. The proposed pathway for this process is initial metalcatalyzed hydrogenation of the ring followed by acid-catalyzed dehydration to form cyclohexene and subsequent metal-catalyzed hydrogenation to ultimately produce cyclohexane.38 This combination of a carbon-supported noble metal with a mineral acid was found to be effective for selective deoxygenation of a variety of phenolic compounds in the aqueous phase, performing multistep reactions involving hydrogenation, hydrolysis, and dehydration steps. For instance, in neutral solution all catalysts produced no deoxygenated products at 100% conversion; however, upon addition of phosphoric acid to provide bifunctionality to the catalytic mixture in the form of acid sites, the deoxygenated alkane product was formed with about 90% selectivity over all metals.38 3.1.2 Supported Base Metals While supported noble metal catalysts are promising for biomass HDO, their high cost has sparked interest in studying less expensive base metals such as Ni, Co, and Fe. As with the noble metal catalysts, it is useful to first investigate the monofunctional behavior of these metals. Fe/SiO2 was used for the gas-phase upgrading of guaiacol as a model compound to study iron’s ability to break hydroxyl and methoxyl bonds on an aromatic ring.75 The catalyst was found to be selective for production of aromatic hydrocarbons over a range of temperatures.75 Kinetic studies of Fe/SiO2 for guaiacol upgrading have also been performed.76 Results from Fe catalysts were compared to the same reaction over a Co catalyst. Cobalt was found to be more active than Fe, but was shown to catalyze decomposition of the desirable products.76 A simplified pathway is shown in Figure 7. Transalkylation reactions were proposed to occur over the acid sites, explaining the formation of toluene from guaiacol.

ACS Paragon Plus Environment

Page 16 of 44

Page 17 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

CO, coke, polyaromatic hydrocarbons Methanol, CH4 + H2

+ H2

Desired products

O OH

OH

+ H2 + H2

Alkanes, CH 4

Only over cobalt catalysts

Figure 7: Simplified pathway of guaiacol HDO on iron and cobalt catalysts. Both iron and cobalt catalysts follow black arrows, only cobalt catalyzes additional reactions shown in blue. Adapted from 76. A study comparing carbon-supported Fe and Cu to noble metals obtained similar results for guaiacol HDO.59 Guaiacol HDO was found to proceed through a phenol intermediate over all catalysts, but unlike the noble metals these base metals did not lead to ring hydrogenation and C-C scission products. For example, Fe/C was able to convert some of the phenol to aromatic deoxygenated products such as benzene and toluene.59 However, the activity of the base metal catalysts was lower than the noble metals. On Pd, Pt, and Ru the intermediate phenol was primarily hydrogenated to form cyclohexanone and cyclohexanol as expected in the absence of an acid site for deoxygenation. In addition, ring opening to form light gases also occurred. Meanwhile, Fe and Cu were not active for ring saturation and ring opening reactions, predominately forming benzene.59 This proposed network can be seen in Figure 8.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Pd, Pt, Ru

Cu, Fe, PdFe OH

O

OH OH Catechol

Guaiacol

OH

Phenol

O

Cyclohexanone OH

Cyclohexanol Benzene C1 and C2 gasses

Figure 8: Proposed reaction network for guaiacol HDO on precious metals (Pd, Pt, Ru, pathways shown with black arrows) and base metals (Cu, Fe, pathways shown with red arrows). Adapted from 59. These base metals show even more promising HDO activity when used as part of a bifunctional metal-acid catalytic system. For example, RANEY® Ni and Nafion on SiO2 were used for the aqueous phase HDO of a series of phenolic compounds.36 The Ni was found to act as a hydrogenation catalyst and the Nafion/SiO2 provided acid sites for hydrolysis and dehydration, resulting in high selectivity to hydrocarbon production.36 The proposed reaction pathway for phenolics over this catalyst system is depicted in Figure 9.

ACS Paragon Plus Environment

Page 18 of 44

Page 19 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

R = H, OCH3 R1 = alkyl OH R

OH

O OCH3

+ H2

R

R1

OCH3

+ H2

R

OCH3

R1

R1

+ H+ OH R

OH

+ H2

R

O

- H 2O /

H+

R

OH + CH3OH

R1

R1

R1

+ H2 OH

- H2O / H+

+ H2

+ RH R1

R1

R1

Figure 9: Proposed reaction pathways of phenolic compounds over RANEY ® Ni and Nafion/SiO2. Adapted from 36. With Nafion/SiO2 still present in the aqueous mixture to provide acid sites, Ni/SiO2 has also shown activity for deoxygenation of phenol with the properly selected reaction temperature.77 Temperatures below 523 K did not catalyze hydrogenolysis to form benzene; this was attributed to thermodynamic limitations. Instead, cyclohexanone and cyclohexanol were produced. Higher Ni loading was shown to increase activity as well as selectivity to benzene production, while lower loadings of Ni produced more cyclohexanol and cyclohexanone.77 This was proposed to be due to a greater population of hydrogen on the surface that can be used for hydrogenolysis.77 It was further proposed through H2 TPD results that on the larger particles the majority of the hydrogen available is at metal-support interfacial sites, which are most active for hydrogenolysis. Hydrogen TPD showed a sharp peak at 663 K on large particles, which was proposed to come from hydrogen at the metalsupport interface. The TPD trace for the smallest Ni loading had a small shoulder at this temperature along with a larger peak and additional high temperature shoulder at 720 and 760 K.77 The peak at 720 K was attributed to hydrogen that is associated with the surface Ni sites, which has also been proposed to be responsible for hydrogenation selectivity, while the peak at 760 K was attributed to hydrogen that had spilled over onto the support. Because the interfacial area was much higher for the catalysts with high Ni loading, it was suggested that increases in the Ni loading increased the availability of weakly bound hydrogen accessible for hydrogenolysis

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

to benzene. A different bifunctional approach to deoxygenation using Ni is to utilize MgO as a basic support. Liquid phase guaiacol upgrading was studied over Ni/MgO at 433 K and 29.6 atm H2.78 At 98% conversion there was 100% selectivity to demethoxylation to produce cyclohexanol. This high selectivity was attributed to the acid-base interaction between the phenolic OH group of guaiacol and the MgO support, with Ni particles providing hydrogenation sites.78 Guaiacol demethoxylation has also been observed over Ru/MgO79 and Ru-MnOx/C, where in the latter case the MnOx provides basicity.80 Selective demethoxylation and hydrogenation may be desirable as a renewable method of producing cyclohexanol as a feedstock for polymer production or as an intermediate for fuel production. Although the previous literature convincingly demonstrates that a combination of acid and metal sites leads to more complete deoxygenation, it is important to point out that metals themselves have some activity toward HDO reactions. For reactants with relatively weak C-O bonds such an allyl alcohol, furfuryl alcohol, and benzyl alcohol, deoxygenation can even be observed on single crystals in ultrahigh vacuum.81-83 Deoxygenation of these compounds on supported metals has often not been interpreted in terms of the support having a role. Studies of such reactants on supported Pd catalysts have shown that selective poisoning of different types of Pd sites can be used to favor selectivity for deoxygenation.84,85 However, as the studies investigating support effects show, the interaction between the metal particle and the support can significantly impact the surface chemistry. Going forward a major challenge will be to develop better models for the structure of metal-support interface and to learn how to control interfacial active sites. This is discussed in greater detail in the final section. 3.2. Bimetallics The previous examples of transition metal catalyzed HDO of phenolic compounds show that these catalysts may be promising materials for pyrolysis oil upgrading if acid sites are present. This mechanism (as shown in Figure 5) may also take place over bimetallic catalysts. Oxide supports such as alumina are often used as the source of acid sites but they also can strongly bind oxygenates like phenols, leading to intermediates that block sites and can be precursors to coke formation.33,86,87 Phenols and polyphenols have been shown to be coke precursors over a wide range of catalysts, including supported metals, oxides, and zeolites.88 This has prompted studies of the effects of bimetallic catalysts containing a hydrogenating metal and an oxophilic metal (an element that forms strong metaloxygen bonds) to provide intimate contact between diverse sites for pyrolysis oil upgrading.89 Many of these have been noble metal-based alloys. Figure 10 shows one proposed mechanism for the interaction of these bimetallic catalysts with an organic oxygenate. The metals used as a modifier site may be present in their metallic state or as an oxide; regardless, their role is to introduce an additional

ACS Paragon Plus Environment

Page 20 of 44

Page 21 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

active site to produce a bifunctional nanoparticle. Table 2 shows metal-oxygen bond strengths for some commonly used oxophilic modifiers. Table 2: Metal-oxygen bond strengths for oxophilic metals used in bimetallic catalysts. Values represent the bond dissociation energies measured at 298 K. Both modifier sites used as the oxophilic site and hydrogenating metals are included for comparison. Metal-Oxygen Bond Bond Strength (kJ * mol-1) Reference Oxophilic metals Co-O 397 90 Cr-O 461 90 Fe-O 407 90 W-O 653 91 Mo-O 607 91 a Re-O 644 92 Hydrogenating metals Ni-O 366 90 Pd-O 234 91 Pt-O 347 91 Rh-O 377 91 a Specifically for the Re-O bond in ReO3 = hydrogenating metal (ex. Pt, Pd, Ru, Ni) R' = oxophilic metal (ex. Re, Mo, W, Fe) H

RO

H

Figure 10: Hypothesized mechanism for interaction of oxygenates with bimetallic surfaces containing a metal capable of dissociating hydrogen and a metal that forms strong M-O bonds. The hydrogenating sites provide H while the oxophilic metal more strongly binds the reactant to the surface through its oxygen functionality, allowing easier C-O scission and potentially creating new reaction pathways unavailable on a monometallic catalyst. The majority of bimetallic catalysts investigated have involved 3d metals as a modifier to noble metals. These systems are known to create subsurface alloys with the surface noble metal atoms significantly electronically modified.93,94 Subsurface transition metal alloys have been shown to increase HDO activity as well as selectivity towards saturated hydrocarbons.32,95-97 DFT results show that the Pt terminated alloys bind adsorbates less strongly than pure Pt, resulting in a more optimized adsorption strength which increases

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

hydrogenation.32 An investigation of the hydrogenation of the unsaturated compounds benzene and 1,3-butadiene over Pt and PtNi catalysts also observed that hydrogenation rates increased with the Ni loading.97 DFT calculations have suggested this could be due to shifting the d-band center of surface Pt atoms by subsurface Ni, resulting in weaker binding of the unsaturated compounds.98 The hydrogenation activity was shown to have a volcano curve-type relationship with the binding energy of the reactant.98 Therefore tuning of this binding strength allows maximization of the hydrogenation activity. In addition to influencing hydrogenation rates, it was proposed that the bimetallic catalysts contained additional sites for dehydration. Metals such as Co and Ni can create additional acid sites through interaction with the alumina support to increase the rates of dehydration reactions.32 This is proposed to be due to formation of an aluminate species, which previous work has shown increases the acidity on alumina.99 Because acid-catalyzed dehydration steps are critical for C-O scission in phenolic compounds, this increases HDO rates. Many previous studies with bimetallic catalysts have been performed using probe molecules such as glycerol. While the precise role of the modifier is unknown, many studies have shown increased activity and selectivity for C-O bond scission reactions for a variety of probe molecules and metal combinations.37,100-104 PtRe/C catalysts were shown to be more active than monometallic Pt for converting glycerol to syngas.37 Here, the role of the oxophilic metal may not be directly related to its affinity for oxygen. Rather, CO-TPD showed that CO binding energies were reduced on PtRe compared to either monometallic sample, suggesting that reduced site blocking could be at least partially responsible for the increased activity.37 Another glycerol reforming study in which PtRe/C was used for hydrogenolysis to form propanediols reported both increased rate and selectivity towards deoxygenation of the secondary alcohol.37 Similar results were obtained for RhRe/SiO2 catalysts.100 In trying to understand the nature of these modifier sites, two main hypotheses have been proposed. First, it was suggested that hydroxylated Re-OH groups may be responsible for C-O activation by acting as Brønsted acid sites, with the strong metal-oxygen bond weakening the hydroxyl’s oxygen-hydrogen bond, thus allowing the hydrogen to be more easily donated to a neighboring oxygenate.37 NH3 TPD has shown increased acidity on PtRe catalysts in the presence of water relative to what is observed over the monometallic Pt sample.105 The second hypothesis is that these oxophilic sites are strongly anchoring the reactant to the catalyst surface through its oxygen functionality, allowing surrounding sites to donate hydrogen for deoxygenation.100 This strong interaction of the oxophilic site and the oxygenated reactant can result in altered reaction pathways. For example, m-cresol was shown to preferentially follow the ring hydrogenation route to deoxygenation over pure group VIII metals such as Ni, Pd, and Pt (shown by the saturated compounds in Figure 5).34,71 However, when an oxophilic modifier was added to the catalyst a tautomerization pathway became accessible, creating a keto intermediate which may then be dehydrated to form toluene. By making this additional deoxygenation pathway energetically favorable the deoxygenation

ACS Paragon Plus Environment

Page 22 of 44

Page 23 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

activity is increased. This has been shown for m-cresol HDO over PtMo/Al2O3, as illustrated in Figure 2c.12 Although more work is needed to fully understand the role of oxophilic modifiers, model compound studies show their promise for biomass deoxygenation. In addition to improving deoxygenation selectivity, some studies using bimetallic catalysts also reported DDO mechanisms. This is advantageous if the goal is to remove oxygen with minimal hydrogen consumption. Increased selectivity to C-O scission products has been observed over noble metal catalysts modified with oxophilic metals such as Fe, Re, Mo, and W; the modifier sites could potentially bind oxygenates more strongly to the surface or create acid sites when they are hydroxylated by water present under typical reaction conditions.37,58,101,102,104,106,107 In one study, PdFe alloys were used for the vapor-phase upgrading of guaiacol.59 PdFe/C showed better selectivity towards production of deoxygenated aromatics than either of the monometallic catalysts, and no ring saturation or ring opening products were observed. This study led to the hypothesis that phenolics prefer to adsorb on the Pd-modified Fe sites. Compared to the monometallic Fe surface, these sites hypothetically have altered electronic properties and are in a more reduced state due to increased hydrogen activation by Pd. This results in a catalyst that can activate phenolic C-O bonds without first needing to saturate the ring. This mechanism was explored through DFT calculations by substituting Pd atoms into a Fe(110) surface. Phenol was adsorbed with varying distances from the Pd atoms. The results indicated that adsorption and C-O cleavage are both more favorable on the Fe sites.59 TPR results on the PdFe/C catalysts showed that Pd assists in the reduction of FeOx.59 This evidence, along with EXAFS and STEM/EDS results, was used to propose that Fe is the active site for HDO and the role of Pd is facilitating reduction of FeOx as well as modifying the Fe sites through electronic effects due to alloy formation.59 STEM images of the morphologies of these materials can be seen in Figure 11. Bright regions with a diameter of 1-2 nm are Pd clusters on the Fe surface; the darker spots underneath these are Fe atoms.59 Both reduced and spent catalysts were imaged; the comparison between these was used to suggest that there is negligible change in morphology during reaction conditions, and therefore the Pd-Fe structure is robust.59

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 11: Representative aberration corrected high angle annular dark field scanning transmission electron microscopy (AC-HAADF-STEM) images of reduced and (left) and spent (right) Pd/Fe2O3 studied for m-cresol HDO. Brighter spots show Pd atoms and darker spots show Fe atoms. Reproduced from 59 with permission from ACS Catalysis. Similar results were obtained from a Pd/Fe2O3 catalyst for HDO of m-cresol: Pd was proposed to activate hydrogen and allow reduction of the Fe2O3 as well as prevent oxidation by water formed during the HDO reaction.61 Addition of Pd was also found to increase the activity compared to monometallic Fe while maintaining a similar selectivity towards the production of aromatic deoxygenated products.61 Cyclic oxygenates have also been studied for selective hydrogenolysis on bimetallic catalysts. A common conclusion among many of these reports is that the combination of a noble metal and an oxophilic metal such as Re, Mo, or W can dramatically increase selectivity to cleaving more sterically hindered C-O bonds. For example, Rh catalysts were modified with Re to improve selectivity to 1,6hexanediol production from tetrahydropyran-2-methanol (THPM).108,109 While Rh/C had only 15% selectivity to this desired product, Rh-ReOx/C had 96% selectivity as well as increasing the reaction rate from 2 to 180 mmol - h-1 - g catalyst-1. This improved activity was proposed to be due to strong Re-OH interactions that allowed an alkoxide species to form, strongly tethering the THPM to the surface. Hydrogenolysis of the C-O bond adjacent to the alkoxide group is favored and therefore results in high selectivity to 1,6-hexanediol.108 A similar mechanism was proposed using Rh-ReOx/SiO2 for tetrahydrofurfuryl alcohol110 and glycerol100,111 hydrogenolysis. Another study using Rh-ReOx/C for hydrogenolysis of cyclic ethers proposed that this increase in selectivity to C-O bond scission at the more sterically hindered carbon may be due to the ability of OH groups on ReOx sites to stabilize the intermediate carbenium ions.112 To illustrate this mechanism, Figure 12 shows the proposed reactant and intermediate adsorption states for tetrahydrofurfuryl alcohol hydrogenolysis on Rh(111) modified with an Re-OH site.

ACS Paragon Plus Environment

Page 24 of 44

Page 25 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

Figure 12: DFT-calculated reactant and transition state adsorption structures for tetrahydrofurfuryl alcohol hydrogenolysis. The mechanism is proposed to be a concerted protonation, hydride transfer, and ring opening over Re-OH modified Rh(111). Reproduced with permission from 112. Increased hydrogenolysis selectivity has been observed for acyclic oxygenates as well.112 Ir-ReOx/SiO2 showed higher selectivity to 1,4-butanediol production from erythritol hydrogenolysis than unmodified Ir.113 This increase in selectivity from 0% to 33% was again proposed to be due to formation of an alkoxide species on ReOx sites followed by hydrogenolysis by a hydride species on a neighboring Ir site. Similar results were found for glycerol hydrogenolysis on Ir-ReOx/SiO2, with increased C-O scission attributed to formation of an alkoxide.114-116 Characterization work on Ir-ReOx surfaces has supported the hypothesis that interaction between the two sites is important in the enhanced hydrogenolysis; a combination of XAS, TPR, XPS, and CO chemisorption suggested that there is significant interaction between the Ir sites and the low-valent ReOx.117 A similar investigation for Rh-ReOx also found that the two sites were well mixed and interacting.118 Although bimetallics containing noble metals have been the most extensively studied, another promising area is in the use of bimetallics formed from base metals. NiFe/SiO2 catalysts have been studied for m-cresol HDO.34 Over monometallic Ni, methylcyclohexanone was found to be the major product, while on the bimetallic catalyst toluene was dominant. A physical mixture of Ni/ SiO2 and Fe/ SiO2 did not show this change in reaction pathway, indicating that interactions between the two metals are important. It was proposed that alloying with oxophilic metals such as Fe weakened the interaction between the aromatic ring and the surface, favoring strong binding with the carbonyl group and leading to production of unsaturated compounds.34 Studies like these have recently generated a great deal of interest in better understanding bimetallic catalysts containing less expensive base metals for deoxygenation applications. A summary of these studies and the improvement of deoxygenation over the monometallic catalyst can be seen in Table 3.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 44

Table 3: Summary of bimetallic catalysts studied for deoxygenation. Reactants include m-cresol, glycerol, guaiacol, tetrahydropyran-2- methanol (THPM), and Tetrahydrofurfuryl alcohol (THFA) Catalyst, wt% metals PtCo/Al2O3

Ref.

97.1

% Increase over monometallic 21

63

94.2

18.1

32

Glycerol

20

26a

26

37

39.5

Glycerol

10

22a

17

100

523

4.6

m-cresol

95

96

76

12

5.1 % Rh 1.8% ReOx Pd-Fe/C

Vapor (5 atm pressure) 623

0.4

Guaiacol

90

29

26

59

2 % Pd 10% Fe Pd-Fe/C

Vapor (1 atm pressure) 723

0.4

Guaiacol

95

85

80

59

2 % Pd 10% Fe

Vapor, (1 atm pressure) 373

39.5

THPM

36

97

58

108

39.5

THFA

57

94b

76

110

1.3% Pt 4.3% Co PtNi/Al2O3 1.5% Pt 4.9% Ni Pt-Re/C 5.7 % Pt 4.6% Re RhReOx/SiO2

Temperature (K), Phase

PH2 (atm)

Reactant

Conversion (%)

% Deoxygenation

533

0.5

m-cresol

57

0.5

m-cresol

39.5

32

Vapor (1 atm pressure) 533 Vapor (1 atm pressure) 443 Aqueous 393 Aqueous

4 % Rh, 0.5 molar ratio Rh/ReOx PtMo/Al2O3

Rh-ReOx/C 4 % Rh, 0.5 molar ratio Rh/ReOx

Aqueous

RhReOx/SiO2

393 Aqueous

4 % Rh,

ACS Paragon Plus Environment

Page 27 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

0.5 molar ratio Rh/ReOx a Selectivity b Selectivity

to 1,3-propanediol to 1,5-pentanediol

For both noble metal and base metal containing bimetallic catalysts one of the main challenges is understanding and controlling the active sites at the interface of the two metals. Continuing to work towards a better, atomic-scale picture of how the two metals interact under reaction conditions will be a key challenge to overcome going forward. 3.3. Oxides The discussion above mainly considers metal oxides that serve as a carrier for supported metal nanoparticles, and which require the presence of the metal to effectively catalyze deoxygenation. However, some metal oxide surfaces contain diverse types of active sites so that they themselves may be considered bifunctional without the addition of other metals. In particular, reducible metal oxides have shown deoxygenation activity. For example, benzoic acid deoxygenation was studied over ZnO and ZrO2.119 Hydrogen was used to create oxygen vacancies in the oxide surface which could then be reoxidized by the benzoic acid reactant to form benzaldehyde.119 However, these vacancies were susceptible to poisoning by compounds like CO2 and water. In addition to model compound studies, oxides have been studied for catalytic upgrading of whole pyrolysis vapors. A ZrO2 and TiO2 mixed oxide catalyst increased the detected percent of hydrocarbons to 13.1%, compared to 0.1% in the raw vapor.120 One of the most promising uses for oxide catalysts in biomass upgrading is for ketonization. The carboxylic acids present in pyrolysis oil are especially important to remove due to their corrosivity and reactivity. One approach to eliminating these functional groups while also increasing carbon chain length is ketonization, or ketonic decarboxylation, which converts two carboxylic acid molecules into a ketone with water and CO2 formed as byproducts.121,122 R1COOH + R2COOH
R1COR2 + CO2 +H2O

(1)

This is also beneficial because ketones can undergo aldol condensation reactions to form larger molecules in the gasoline to diesel range. The majority of studies for ketonization have been done using oxide catalysts. Reducible metal oxides such as CeO2, ZrO2, and TiO2 have been found to be more active and selective than purely acidic or basic oxides.121 It has been proposed that reducible oxides allow the carboxylic acids to dissociatively adsorb to form carboxylates and hydroxyls on the surface. These oxides also have available cation sites that promote the adsorbed acids to interact and form the required carboxylate intermediates.121 These two distinct types of active sites were proposed to create a bifunctional surface capable

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 44

of improving activity for ketonization. An example of this mechanism is shown in Figure 13. R + H2 O M O

O

O M

O M O

O M O

O

O M

O M

O M O

O

OH

M+

M+

O

O

O

R' O M

O

O

+

OH M+

O

O M O

O

R R'

O HO OH M O

O M+

O

O

M+

M

O

O +

OH O

O

M+

O

R

M

M O

O

M+

O

O

R'

O

O

M+

M+

O

O

O M

O

OH +

M O

O

- H 2O R

R R'

O O

O

M O

M O

+

O

M+ O

M

O +

O

O

M+ O

R'

O O

- CO2

M

O M+

M O

O

O

O

M+ O

M

+

O

M+

M

O

O

Figure 13: Ketonization mechanism proposed over reducible metal oxides. M = metal site. Adapted from 121. Reduced M+ sites act as Lewis acids to bind the carboxylic acid and oxygen anions provide Lewis basicity to deprotonate the adsorbed acid. A combination of surface science and reactor studies has led to proposing that the active site for these reactions contains coordinatively unsaturated metal cations that act as Lewis acid sites for binding of intermediates and oxygen anions, acting as Lewis bases for deprotonation of the acid. The availability of these sites as well as the redox properties of the material can be influenced by adding dopants such as transition metals and metal oxides.121,122 This can improve catalyst activity because both types of active sites are important for ketonization to occur; metal oxides that are not reducible and therefore lack the metal cations for binding the oxygen functional group do not show the same reactivity. These bifunctional surfaces are effective because they provide a surface containing Lewis acid sites for binding the intermediate in close proximity to Lewis basic sites for deprotonation. More information on ketonization catalysts and mechanisms can be found in past

ACS Paragon Plus Environment

Page 29 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

reviews.51,123 3.4. Sulfides The most commonly used industrial hydrotreating catalysts are sulfided molybdenum or tungsten promoted by cobalt or nickel on an alumina support; these materials are widely used in petroleum refining for hydrodesulurization (HDS) and hydrodenitrogation (HDN).8 Due to this large-scale production, these catalysts can already be made economically and are well characterized and understood in the context of HDS and HDN. The wide variety of past studies has shown that edge area of these materials is important for HDS activity, leading to the conclusion that edges are the active sites for this reaction.124-127 However, the nature of the active sites for HDO is still under debate. It is typically proposed that the materials are bifunctional, with one site for activation of heteroatom-C bonds and another for hydrogenation.39,40 Oxygen can adsorb to coordinatively unsaturated sites (sulfur anion vacancies) at the catalyst edges.39,40 An example of this is shown schematically in Figure 14. M = Mo, W H H S S

Oxygenate adsorption in S vacancy

H H S S

S

M M M M M

M M M M M C-O scission on S vacancy

+H2 H S

H HR S S S

OR'H S

R R' S O

S

M M M M M

OR'

M M M M M

Hydrogenation by surface SH groups

Figure 14: Proposed mechanism of deoxygenation over transition metal sulfides. Modifiers including Ni and Co donate electrons to the metal, weakening the metal-S bond and allowing more vacancies to be created.39,128 Hydrogen is thought to come from S-H groups on the surface (formed either through activation of molecular hydrogen or from decomposition of molecules such as water or alcohols).8 This hydrogen can be used for HDO by breaking the C-O bond or for decarbonylation/decarboxylation by using acid sites to remove end groups from aldehydes or carboxylic acids respectively128 (see Figure 1). The selectivity to a given pathway depends on reaction conditions as well as the transition metal being used. For the HDO pathway, unsaturated carbon bonds as well as carbonyl groups are often first hydrogenated to form alcohols, which are then dehydrated. An example of this can be seen for phenolics and ketones in Figure 15.

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

R1

R1 OH

O R1

H2

H2

OH

OH

H2 R2

R1

R1

R2

+ H2O

H2 R1

R2 + H2O

Figure 15: Example mechanism proposed for the HDO pathway over sulfides Recent work has focused on materials that more clearly fit the definition of bifunctional, i.e. that incorporate additional metals to create sites that assist in HDO of compounds relevant to pyrolysis oil reforming.129-133 Previous investigations have shown that Co and Ni can promote the hydrodeoxygenation pathway over hydrogenation or decarboxylation/decarbonylation for Mo2S catalysts.130,131,134 Experimental and DFT work has suggested that incorporation of Ni-Mo sites lowers the activation barrier for breaking aldehyde C=O bonds and alcohol C-OH bonds by altering the preferred adsorption orientation, enhancing HDO over decarbonylation/decarboxylation or hydrogenation. Instead of binding with just the oxygen group on top of Mo site, it was proposed that the bimetallic surface allows bidentate binding where the carbon atom of the functional group (for example, in a carbonyl) is in close proximity to an Ni-H site. This lowers the energy of the transition state for HDO. However, if too much Ni is incorporated, Ni3S2 is formed, which increases selectivity to decarbonylation over HDO.131,134 These studies have also expanded the composition of transition metal sulfides being investigated, incorporating multiple metals and modifiers, as well as studying support effects. For example, on a W-Mo-S catalyst promoted by Ni, tuning the W/Mo ratio was found to change the morphology of the catalyst, influencing the number of coordinatively unsaturated sites that have been previously proposed as the active site for activating C-O bonds.129 This resulted in increased activity as well as selectivity to deoxygenation. 3.5. Carbides, Nitrides and Phosphides Transition metal carbides, nitrides, and phosphides have also been explored for HDO, showing promising hydrotreating abilities because they combine the stability of ceramics with the electronic properties of a metal.135,136 These materials are well-characterized HDS and HDN catalysts due to their use in petroleum refining, but less attention has been given to their applications for HDO until recently. A current hypothesis for the mechanism of deoxygenation over transition metal phosphides relies on the presence of both metallic and acidic groups.41-43 For example, it has been proposed that for Ni2P, P-OH groups could contribute Brønsted acid sites while Ni sites add metallic behavior.41-43 Nickel phosphide has received

ACS Paragon Plus Environment

Page 30 of 44

Page 31 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

the most attention as an HDO catalyst due to its activity and selectivity.136 For example, anisole HDO over supported Ni2P has been investigated; the proposed reaction scheme involves demethylation to form phenol followed by hydrogenolysis and hydrogenation.42 The active sites on this catalyst are proposed to be Ni, modified by its P neighbors.8 Ni sites with 4 neighboring P atoms, or Ni(1) sites, have a tetrahedral geometry and are active for decarbonylation; Ni(2) sites have 5 P neighbors in a square pyramidal geometry and catalyze HDO.8,136,137 Bulk Ni2P contains equal amounts of Ni(1) and Ni(2), but the surface is enriched with Ni(2) and this enrichment increases as the particle size decreases. This has been confirmed experimentally using a range of particle sizes and measuring both coordination numbers as well as Ni/P ratios.136 Increasing the number of Ni(2) sites correlated with an increase in hydrotreating activity and a decrease in decarbonylation.136 A series of phosphides including Ni2P, Co2P, Fe2P, WP, and MoP supported on SiO2 were studied for guaiacol HDO and compared to a Pd/Al2O3 catalyst at the same conditions.138 A proposed reaction scheme for guaiacol HDO on these catalysts can be seen in Figure 16.

OH

cyclohexanol OH

benzene

OH OH

OCH3 OH

catechol

phenol CH3

guaiacol

CH3 OH

O

methoxy-benzene

o-cresol

CH3 toluene

Figure 16: Proposed guaiacol HDO reaction pathways on transition metal phosphides. Proposed intermediates shown in parentheses. Adapted from 138

ACS Paragon Plus Environment

ACS Catalysis

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Pd catalyst had a much higher activity than the phosphides under the experimental conditions, but formed primarily catechol, an undesired product that also is a precursor to coke formation.138 Ni2P was the most active of the phosphides tested and avoided this pathway, favoring production of phenol, benzene, and methoxybenzene.138 The decrease in catechol formation may be attributed to the POH groups on the phosphide that are proposed to add acidity to the surface.41-43 The suppression of catechol production on phosphides is promising, however additional studies are needed to improve selectivity to fully deoxygenated products; the selectivity towards phenol and methoxybenzene seen in this study are not ideal for a final upgrading catalyst. Carbides have been compared to platinum group metals, showing similar catalytic properties; reactions that would not occur on W or Mo alone are observed over the carbide surface, since binding of the metal to carbon atoms shifts the energy of the transition metal d electrons to lower energies.137,139-141 The ability of tungsten and molybdenum carbides to remove heteroatoms suggests they could be promising catalysts for biomass deoxygenation. For example, through DFT calculations as well as surface science and gas-phase reactor studies using propanol and propanal as reactants, WC was shown to activate C-O bonds but not C-C; the dominant product formed was propene.139 The catalysts have again been suggested to be bifunctional in nature. The presence of oxygen on the WC surface was proposed to introduce acidic sites, on which dehydration reactions are faster than when oxidized W or WC is not present.139 A similar effect of surface oxygen has been proposed for other carbide surfaces as well. For the HDO of model compounds over NiMo carbide, it was proposed that metallic sites as well as metal-OH sites are generated under reaction conditions, leading to enhanced dehydration activity.44 Mo2C has also been studied for HDO and has shown promising selectivity similar to that of WC.142-144 In addition to these C3 oxygenates, studies have also been performed using larger, aromatic oxygenates. For instance, a study of group 4, 5, and 6 metal carbides and nitrides investigated the HDO of benzofuran at 643 K and 3.1 MPa.145 The carbides tested (Mo2C, WC, NbC, and VC) showed poor selectivity to HDO (about 9, 10, 5, and 3 % respectively). As a comparison, sulfided Ni-Mo/Al2O3, which is a common industrial hydrotreating catalyst, showed over 60% selectivity to HDO. Nitrides have been studied extensively, particularly for hydrogenation/dehydrogenation reactions. As discussed for carbides, the electronic effects induced by alloying can create a material that has some catalytic properties similar to noble metals.146 They have been explored as hydrotreating catalysts, primarily for HDS and HDN, but there has been increasing interest in using these materials for HDO.146,147 Nitrides are also thought to be bifunctional, with active sites being both coordinatively unsaturated metal atoms as well as 4-fold nitrogen vacancies.148,149 Through studying guaiacol HDO, it was proposed that demethoxylation and dehydroxylation occur over the unsaturated Mo sites and

ACS Paragon Plus Environment

Page 32 of 44

Page 33 of 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Catalysis

demethylation is promoted by nitrogen vacancies.148 When Mo2N was studied for guaiacol HDO similar results were obtained for that of the phosphides reported above, avoiding the pathway to produce catechol but producing mainly phenol rather than achieving complete deoxygenation.150 Mo2N also showed low selectivity to HDO in the study by Ramanathant and Oyama using benzofuran.145 All nitrides tested had low selectivity (