Vapor-phase hydrodeoxygenation of guaiacol over carbon-supported Pd, Re and PdRe catalysts

Applied Catalysis A, General 563 (2018) 105–117

Contents lists available at ScienceDirect

Applied Catalysis A, General journal homepage: www.elsevier.com/locate/apcata

Vapor-phase hydrodeoxygenation of guaiacol over carbon-supported Pd, Re and PdRe catalysts Simon T. Thompson, H. Henry Lamb

T

⁎

Department of Chemical and Biomolecular Engineering, North Carolina State University, Campus Box 7905, Raleigh, NC, 27695, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Anisole EXAFS XANES TPR HAADF-STEM EDX

Vapor-phase hydrodeoxygenation (HDO) of guaiacol was investigated over a commercial Pd/C (A) catalyst (Evonik) and Pd/C (B), Re/C and PdRe/C catalysts prepared by incipient wetness impregnation of Norit SX-1 G activated carbon. The Pd/C catalysts had equivalent dispersions after reduction at 300 °C; however, Pd/C (B) had very low dispersion after reduction at 400 °C. CO chemisorption, Re LIII edge extended x-ray absorption fine structure (EXAFS) spectroscopy, and high-angle annular dark field (HAADF)-scanning transmission electron microscopy (STEM) of the Re/C catalyst after reduction at 400 °C evidenced the formation of supported Re clusters. EXAFS spectroscopy of the PdRe/C catalyst after in situ reduction at 300 °C indicated the presence of Pd nanoparticles and Re clusters; a 2.70 Å Pd-Re contribution was required to adequately fit the Re LIII EXAFS spectrum. HAADF-STEM with energy-dispersive x-ray (EDX) analysis of the PdRe/C catalyst after reduction at 400 °C revealed Re clusters and Pd nanoparticles, some in intimate contact. In guaiacol HDO at 300 °C and 1 atm, Pd/C (A) was selective to phenol and cyclohexan-one/-ol and did not produce significant yields of benzene and cyclohexane, despite its high activity. Turnover frequencies for phenol (and cyclohexan-one/-ol) formation over the Pd/C catalysts were equivalent. Phenol, benzene and anisole were major products over Re/C after in situ reduction at 400 °C. The highest yield (52%) of fully deoxygenated products was obtained over PdRe/C after in situ reduction at 400 °C. We infer that the bimetallic catalyst combines synergistically the demethoxylation and hydrogenation functions of Pd/C with the capability of Re/C to deoxygenate phenol [Ghampson, et al., Catal. Sci. Technol. 2016].

1. Introduction Modern transportation relies heavily on liquid fuels derived from petroleum. Concerns over the limited supply of petroleum and global climate change have motivated research on biorenewable transportation fuels [1]. Woody biomass is a viable source of renewable liquid fuels for the near- to mid-term and is, in principle, carbon-neutral [2,3]. Fast pyrolysis oils (bio-oils) derived from lignocellulose are not suitable for direct use as transportation fuels because of their low energy densities, pH values, and shelf lives associated with oxygenated organic constituents (e.g., acids, ketones and aldehydes) [1,2,4,5]. Hydrodeoxygenation (HDO) of bio-oils has been demonstrated using petroleum hydrotreating catalysts (e.g., sulfided CoMo and NiMo/Al2O3); however, these catalysts deactivate via sulfur leaching, coking, and acid attack on the Al2O3 support [6–9]. Carbon-supported noble metal catalysts are promising alternatives that provide higher yields of liquid fuels with lower residual oxygen contents when compared to noble metals on metal oxide supports [10,11]. Guaiacol is a model compound for the phenolic fast pyrolysis ⁎

Corresponding author. E-mail address: [email protected] (H.H. Lamb).

https://doi.org/10.1016/j.apcata.2018.06.031 Received 20 March 2018; Received in revised form 22 June 2018; Accepted 25 June 2018 Available online 25 June 2018 0926-860X/ © 2018 Elsevier B.V. All rights reserved.

products that comprise a significant fraction of bio-oils [12,13]. Guaiacol HDO has been investigated extensively over conventional hydroprocessing catalysts [6,7] and supported noble metals, including Pt, Pd, and Ru/C [12,14–16], Pt/Al2O3 [17,18] and Pt/MgO [19]. Noble metals on non-acidic supports typically show a strong propensity for aromatic ring hydrogenation and demethoxylation of guaiacol to phenol; however, these monofunctional catalysts lack the capacity to fully deoxygenate guaiacol to benzene and cyclohexane [16]. For example, Gao, et al. screened carbon-supported Ru, Rh, Pd, and Pt monometallic catalysts for vapor-phase HDO of guaiacol at 300 °C and 1 atm [14]. Of these catalysts, Pt/C had the highest activity and slowest deactivation rate; the main product was phenol. Similarly, the main products of guaiacol HDO over Pd/C were phenol and cyclohexan-one/ol [14]. Bifunctional catalysts typically are more effective at complete HDO of guaiacol to benzene and cyclohexane [20]. Metal-zeolite catalysts, such as Ni/H-ZSM-5 [21] and Pt/HY [22,23], have been demonstrated to achieve phenol and guaiacol HDO via synergistic metal (hydrogenation) and acid-catalyzed (dehydration) pathways. Alternatively, a more oxophilic metal may be added to supported noble

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

metals to enhance direct deoxygenation (DDO) activity [20]. For example, PdFe/C catalysts have been shown to provide high yields of fully deoxygenated products (benzene and cyclohexane) when operated at 350–400 °C [16]. Recently, supported rhenium catalysts (Re/C and Re/SiO2) were shown to be active for HDO of guaiacol and related phenolics in the liquid phase at 300 °C and 5 MPa [24,25]. Catalytic activity for CeO bond scission resulting in high yields fully deoxygenated products was ascribed to partially reduced rhenium oxide species (ReOx). Addition of Cu to Re/SiO2 was found to facilitate Re reduction and boost guaiacol conversion and selectivity to benzene and cyclohexane [26]. Conversely, rhenium has been demonstrated to enhance selectivity and catalytic activity for selective hydrogenation [27–33] and CeO bond hydrogenolysis [34–38] when paired with a platinum-group metal (PGM). Tomishige and coworkers have studied several supported PGMRe catalysts for selective hydrogenation of mono- and dicarboxylic acids [27–29] and hydrogenolysis of ethers and polyols [38]. Their results suggest that the catalytically active sites comprise low-valent ReOx clusters in intimate contact with PGM nanoparticles; however, a role for zero-valent Re clusters cannot be excluded [39]. Koso, et al. described supported Re nanoparticles covered by ReOx species and lowvalent ReOx clusters attached to Rh metal particles in Re/SiO2 and RhRe/SiO2 catalysts, respectively, after reduction at ∼330 °C [40]. Hakim, et al. concluded that the active sites for selective hydrogenolysis of 2-(hydroxymethyl)tetrahydropyran (HMTHP) to 1,6-hexanediol comprise small noble metal particles (Rh, Pt) adjacent to highly reduced moieties of a more oxophilic metal (Mo, Re) on a Vulcan carbon support [37]. Focusing on RhRe/C catalysts, Chia, et al. reported that Rh-rich nanoparticles with a partial shell of metallic Re are active for selective hydrogenolysis of HMTHP to 1,6-hexanediol and that Re penetration into the core after reduction at 450 °C correlated with a decrease in activity for HMTHP hydrogenolysis [36]. Alloy formation is less favorable for PdRe than RhRe because of the larger miscibility gap in the bulk phase diagram [41]. Takeda, et al. proposed that Ren+ species on the surfaces of metallic Re clusters and Pd nanoparticles were the active sites in RePd/SiO2 (Re/Pd = 8) catalysts for hydrogenation of stearic [29] and succinic acids [28]; the percentage of Re° clusters and their interaction with the Pd-rich nanoparticles depended on the precise reduction conditions. Shao, et al., however, demonstrated using high-resolution TEM that a PdRe/C (Re/Pd = 0.6) catalyst for succinic acid hydrogenation contained Pd and Pd-Re alloy nanoparticles [31]. In this work, a commercial Pd/C (A) catalyst (Evonik) and Pd/C (B), Re/C and PdRe/C catalysts prepared by incipient wetness impregnation of Norit SX-1 G activated carbon were investigated for vapor-phase HDO of guaiacol at 300 °C and 1 atm. The catalysts were characterized by temperature-programmed reduction (TPR), CO chemisorption, x-ray absorption spectroscopy (XAS), and scanning transmission electron microscopy (STEM) with energy-dispersive x-ray (EDX) analysis. Guaiacol HDO product distributions were evaluated for each catalyst in a continuous flow reactor at integral conversion, and turnover frequencies (TOFs) for primary products were determined at differential conversion.

Table 1 Catalyst metal loadings and CO chemisorption results. Catalyst

Metal loadings (wt.%)

Pd

Reduction temperaturea (°C)

Strongly bound CO (μmol/ gcat)b

CO/ metalc

Metal dispersiond (%)

39.3 39.3 6.7 222 217 224

0.087 0.10 0.017 0.54 0.28 0.29

17 20 3.5 54 – –

in air 0.064 0.010 0.30 0.43 0.17 0.19

13 2.1 30 43 – –

Re

Catalysts after in situ reduction Pd/C (A) 4.80 – 300 Pd/C (B) 4.11 – 300 400 Re/C – 7.64 400 PdRe/C 3.96 7.50 300 400

Catalysts previously reduced ex situ at 400 °C and passivated Pd/C (B) 4.11 – 300 24.6 400 4.0 Re/C – 7.64 300 123 400 178 PdRe/C 3.96 7.50 300 134 400 147

a In situ reduction at indicated temperature for 1 h followed by evacuation for 4 h. b Measured by difference isotherm method at 35 °C. c Mols strongly bound CO/mol metal (total). d Stoichiometry factors: 0.5 CO/Pd and 1 CO/Re.

and the resultant powder was stored in a desiccator until use. Metal loadings (Table 1) were determined by inductively coupled plasmaoptical emission spectrometry at Eastman Chemical Company, Kingsport, TN. 2.2. TPR Measurements were performed using a Micromeritics 2920 AutoChem II. Catalyst powder (∼100 mg) was loaded into a quartz Utube and purged with He (UHP, National Welders) at ∼40 °C. Subsequently, the sample was heated at 10 °C/min in flowing 5% H2/Ar (Certified mixture, Machine and Welding) to a final temperature of 400, 500 or 800 °C. H2 uptake during TPR was monitored using a AgO-calibrated thermal conductivity detector (TCD). 2.3. CO and H2 chemisorption Volumetric CO chemisorption measurements were made using a Micromeritics 2020c ASAP instrument. Catalyst samples were evacuated for 1 h at 100 °C, reduced in flowing H2 (Research grade, National Welders) at either 300 °C or 400 °C for 1 h, and evacuated at the reduction temperature for 4 h. After a negative leak test, an adsorption isotherm using CO (Research grade, National Welders) was measured at 35 °C. Subsequently, the sample was evacuated to remove weakly bound CO, and then the analysis repeated. A difference isotherm corresponding to strongly bound CO was used in dispersion calculations. Metal dispersions were calculated using surface atom stoichiometry factors of 0.5 CO/Pd [42] and 1 CO/Re [43]. Volumetric H2 chemisorption measurements were performed at 35, 70 and 100 °C on selected samples using an analogous procedure.

2. Experimental 2.1. Catalysts

2.4. X-ray absorption spectroscopy (XAS) A commercial 5% Pd/C catalyst (Evonik E117), denoted Pd/C (A), was received as a reduced 50% water-wet powder and dried at 110 °C in air prior to use. Carbon-supported Pd, Re, and PdRe catalysts were prepared by incipient wetness impregnation of Norit SX-1 G activated carbon. The catalyst precursors: Pd(NO3)2·H2O (99.9% Pd basis, Strem) and a 76.5 wt% solution of HReO4 (99.99% Re, Acros Organics) were dissolved in 18 MΩ-cm deionized water. After impregnation, the paste was dried at 110 °C in air, the solid was crushed with mortar and pestle,

Pd K (24,350 eV) and Re LIII (10,535 eV) x-ray absorption spectra were measured in transmission mode at the National Synchrotron Light Source, Brookhaven National Laboratory using beam lines X-10C and X11 A. The Si(311) x-ray monochromators were calibrated using Pd foil and Re powder standards. XAS measurements were made on lab-prepared catalysts that had been reduced ex situ at 400 °C, cooled to 25 °C, and exposed slowly to air. Powdered samples were pressed into 106

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

analyze for organics in the reactor effluent stream prior to the condenser. The split ratio was 10, and the GC oven temperature program included a 5 min hold at 30 °C followed by a 30 °C/min ramp to 120 °C, a 45 °C/min ramp from 120 °C to 275 °C and a 5-min final hold. Product peaks were identified and calibrated using known standards in MeOH. Nine C6-C7 products were identified and calibrated. Cyclohexanol and cyclohexanone co-eluted as a single peak with a retention time of 8.871 min. This peak was quantified using the FID response factor for cyclohexanol. Two significant unidentified products, UNK 1 and UNK 2, with retention times of 7.787 and 9.718 min, respectively, were detected. Potential product species, including 1,2-cyclohexanediol, toluene, cresols, xylenes and veratrole, were ruled out using authentic standards. Non-condensable species (CH4, CO and CO2) in the effluent stream were analyzed using a SRI 8610C GC equipped with a Restek ShinCarbon ST100 (2 m x 1.0 mm ID) column and a TCD. A gas mixture (5% H2, 5% CO, 5% CO2, 1% CH4, balance He, National Welders) was used to calibrate the GC-TCD. Reactant conversion in integral reactor experiments was calculated from the reactant injection rate and the measured effluent concentration. Closure of the carbon balance was 93.7 ± 5.9% in these experiments. In differential reactor experiments, conversion was calculated by summing the C yields of measured products.

stainless steel holders and loaded into an in situ XAS cell equipped with Be windows [44]. Initial energy scans were recorded in flowing He (UHP, Airgas) at 25 °C. Subsequently, catalysts were reduced in 5% H2/ He (Airgas, certified) at 300 or 400 °C for 1 h. After cooling to 25 °C, energy scans were recorded under 5% H2/He. Subsequently, the cell was purged with He and heated to 100 °C and held for 30 min to decompose any β-PdHx phase. The cell was cooled to 25 °C, and additional energy scans were recorded. The XAS data were processed using Athena (part of the IFEFFIT package [45]) and fit using Artemis [46]. FEFF references were generated using ATOMS [47] and relevant crystallographic data for NH4ReO4 [48], Pd and Re [49]. Amplitude-reduction factors, S02 for Pd and Re, 0.76 and 0.78, respectively, were determined from fitting the first-shell EXAFS of Pd foil and Re powder. Heterometallic backscattering paths (Pd-Re and Re-Pd) were modeled in ATOMS by replacing half of the first nearest neighbors within an fcc structure with heteroatoms at the same distance (2.75 Å). EXAFS spectra were fit in R space using kn (n = 1, 2 and 3) weighting. The Pd K and Re LIII EXAFS spectra of PdRe/C were fit simultaneously using constraints to minimize the number of free parameters. The heterometallic paths were constrained to have the same coordination number (N), distance (R) and Debye-Waller factor (σ2). A single energy shift parameter, ΔE0, was used for all paths of each absorber atom. Fourier transform (Δk) and fitting ranges (ΔR) for each sample are provided in Supplementary Information (Table S1).

3. Results 3.1. Catalyst characterization

2.5. STEM with EDX TPR profiles of the lab-prepared catalysts are shown in Fig. 1a. Multiple H2 consumption features for Pd/C (B) appear between 50 and 250 °C. The sharp negative peak at 64 °C is assigned to β-PdHx decomposition [51] and indicates that metallic Pd particles were present before the temperature ramp began. Because the catalyst was not calcined prior to reduction, most of the observed H2 consumption is associated with reduction of nitrogen oxides (NOx), as verified by on-line quadrupole mass spectrometry (QMS). Reduction of Re/C begins at ∼200 °C, and the rate exhibits a sharp maximum at 285 °C. H2 consumption continues above 400 °C with a weak maximum at ∼480 °C. Cumulative H2 consumption during TPR to 400 °C (9.7 H/Re) exceeds the quantity required to reduce Re7+ to Re0. Excess H2 consumption that occurs primarily above 300 °C is associated with methanation of the carbon support, as evidenced by on-line QMS (Supplementary Information, Fig. S1). PdRe/C exhibits a sharp TPR peak at ∼115 °C with a low-temperature shoulder. The weak maximum at ∼265 °C and broad high-temperature feature are similar to those found in the TPR profile of Re/C. In comparison to Re/C, most of the supported ReOx species are reduced at significantly lower temperatures in the bimetallic catalyst. To parallel catalyst treatment procedures for XAS, TPR to 400 °C was performed on catalysts previously reduced ex situ at 400 °C, cooled to 25 °C, and exposed slowly to air. Low-temperature H2 consumption features are absent in the TPR profile of Pd/C (B) (Fig. 1b) indicating that any PdOx species formed by air exposure were reduced prior to initiating the temperature ramp. The sharp H2 evolution peak at 64 °C is assigned to β-PdHx decomposition [51]. The TPR profile of Re/C is closely similar to that of the untreated catalyst consistent with complete re-oxidization of Re upon air exposure. In contrast, the TPR profile of PdRe/C does not resemble that of the untreated catalyst. The PdHx decomposition feature at ∼62 °C indicates that Pd nanoparticles were present prior to TPR ramp initiation. Three ReOx reduction features appear: one at < 100 °C, another at ∼175 °C, and a shoulder at ∼260 °C. The latter is assigned to reduction of isolated perrhenate species because it overlaps with the main TPR peak for Re/C. These H2 consumption features can be assigned to ReOx reduction because residual NOx species would have been removed during ex situ reduction at 400 °C. CO chemisorption results for previously untreated catalysts and catalysts reduced ex situ at 400 °C and passivated in air are provided in

An aberration-corrected FEI Titan 80–300 electron microscope operated by the NC State Analytical Instrumentation Facility (AIF) was used in this work. The microscope is equipped with a high-angle annular dark field (HAADF) detector and a SuperX EDS system comprising four Bruker Si drift detectors. Catalysts were reduced in flowing H2 for 1 h at 400 °C, cooled to 25 °C, and removed from the reactor tube in a N2-purged glove box. TEM samples were prepared by dipping 300 mesh carbon-coated copper grids (Ted Pella) into dry catalyst powder in the glove box and transported to the microscope under N2; however, brief exposure to ambient air was unavoidable during sample insertion. 2.6. Catalyst testing Guaiacol and anisole HDO experiments were performed in a fixedbed micro-reactor with dual on-line gas chromatographs (GCs) [50]. The reactor comprised a 0.5-in 316 stainless steel (SS) tube positioned in a vertical tube furnace. A quartz wool plug supported the catalyst bed (typically, 0.5 g), and catalyst temperature was monitored using a Type K thermocouple in direct contact with the bed. Ultra-high purity H2 and He (National Welders) were metered to the reactor using mass flow controllers (Brooks). Liquid guaiacol (TCI) and anisole (TCI) were fed via syringe pump (Teledyne Isco 260D) and evaporated into the feed stream via a 6-ft. heated coil of 0.25-in. SS tubing. Catalysts were reduced in situ in flowing H2 (100 sccm) for 1 h at 300 or 400 °C. Reaction temperature was 300 °C, and the H2:reactant molar ratio was ∼60:1. Unless otherwise noted, the weight hourly space velocity (WHSV) was ∼1 h−1. Tests with an empty tube showed negligible thermal decomposition of guaiacol and anisole under these conditions. In order to get lower contact times (1/WHSV) for differential conversion experiments at constant reactant concentration, less catalyst was used (typically 0.15 g), and the bed was diluted with borosilicate glass beads (ChemGlass) to minimize channeling. The reactor effluent passed serially through a high-temperature GC sampling valve (Valco) held at 300 °C, a water-cooled tube-in-tube condenser, and a low-temperature GC sampling valve. All lines upstream of the condenser were wrapped with heating tape and held at 200 °C. A Shimadzu 2010 GC equipped with an Alltech EconoCap-1 column (30 m x 0.53 mm ID) and flame ionization detector (FID) was used to 107

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

untreated Re/C and PdRe/C catalysts are given in Table 2. H2 uptake for Re/C at 35 °C is much lower than CO uptake indicating that the supported Re particles do not readily dissociatively chemisorb H2 [52,53]. The low H/CO chemisorption ratio at 35 °C (0.12) is consistent with previous results for Re/Al2O3 catalysts [32]. Higher adsorption temperature resulted in an increase in H2 uptake for Re/C. H2 uptake by PdRe/C at 35 °C is even lower than for Re/C suggesting that CO chemisorption occurs primarily on Re and that Re may suppress H2 dissociative chemisorption on Pd [32,53]. H2 uptake for PdRe/C was approximately the same at 70 °C but decreased significantly at 100 °C. Unfortunately, the difference isotherm technique does not yield accurate H2 chemisorption measurements on poorly dispersed supported Pd catalysts, such as Pd/C (B) after in situ reduction at 400 °C [54]. CO uptakes and metal dispersions measured after ex situ reduction at 400 °C and passivation in air at 25 °C are 20–35% lower than after direct in situ reduction at the same temperature; losses in dispersion are greater for Pd-containing catalysts. Increasing the reduction temperature of Re/C from 300 to 400 °C results in a significant increase in CO uptake and metal dispersion suggesting Re reduction was incomplete at 300 °C. Re LIII XANES spectra of the Re/C and PdRe/C catalysts after ex situ reduction at 400 °C and passivation in air are shown in Fig. 2. The 2p-5d white line intensities and edge shifts (∼4 eV relative to Re powder) are virtually identical and indicative of reoxidation of Re to the 7+ state, as confirmed by EXAFS spectroscopy of the Re/C catalyst (vide infra). XANES spectra of the Re/C and PdRe/C catalysts after in situ reduction at 400 and 300 °C, respectively, for 1 h are also shown. The edge positions are equivalent; however, the white line is slightly more intense for PdRe/C than Re/C. The edge shifts (∼1 eV relative to Re powder) and white line intensities suggest a low positive average oxidation state. We infer that the 2p-5d white line for the Re metal standard was attenuated by thickness effects common to transmission XANES measurements [55], however, we can roughly estimate the average Re oxidation state from the edge shift. Using the linear calibration provided in Fig. S2 (Supplementary Information) results in an average Re oxidation state of ∼2+ for the Re/C and PdRe/C catalysts after in situ reduction. Fourier transform (FT) EXAFS spectra of the Re/C and PdRe/C catalysts after ex situ reduction at 400 °C and passivation in air are shown in Fig. 3. Qualitatively, these spectra support the following inferences: (1) lack of significant oxidation of Pd in PdRe/C after air exposure; (2) complete reoxidation of Re in Re/C after air exposure; and (3) incipient reduction of ReOx in PdRe/C after H2 exposure at 25 °C. The Pd K FT EXAFS spectra of Pd/C (B) (Fig. 3a, i) and PdRe/C (Fig. 3a, ii) after in situ reduction at 400 and 300 °C, respectively, contain similar first- and higher shell Pd-Pd peaks indicative of supported nanoparticles. The spectra of PdRe/C before (Fig. 3a, iii) and after in situ

Fig. 1. TPR profiles of Pd/C (B), Re/C and PdRe/C catalysts (a) after initial drying in air and (b) after ex situ reduction at 400 °C and subsequent air exposure at 25 °C. Table 2 H2 chemisorption results for catalysts reduced in situ at 400 °C. Analysis Temperature 35 °C

70 °C

Catalyst

H2 uptakea (μmol/g)

H/metal

Re/C PdRe/C

13.7 8.4

0.067 0.022

a b

b

100 °C

H2 uptakea (μmol/g)

H/metal

26.9 8.0

0.132 0.021

b

H2 uptakea (μmol/g)

H/metalb

19.8 4.6

0.097 0.012

Strongly bound hydrogen (difference isotherm method). Mols strongly bound H/mol metal (total).

Table 1. The previously untreated Pd/C catalysts have equivalent uptakes of strongly bound CO and closely similar dispersions when reduced in situ at 300 °C; however, in situ reduction of Pd/C (B) at 400 °C results in very low dispersion. Uptakes of strongly bound CO are equivalent for Re/C and PdRe/C after in situ reduction at 400 °C. The large CO uptake by the Re/C catalyst is consistent with the presence of supported Re0 particles. The PdRe/C catalyst has much greater CO uptake than Pd/C (B). Moreover, the CO uptakes for PdRe/C after reduction at 300 and 400 °C are equivalent suggesting complete reduction of Re at each temperature. H2 chemisorption results for previously

Fig. 2. Re LIII XANES spectra of Re/C and PdRe/C catalysts previously reduced ex situ at 400 °C and passivated in air; spectra measured before and after in situ reduction in XAS cell at 400 °C and 300 °C, respectively. 108

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

evidencing incipient reduction. In addition to a short Re-O contribution at ∼1.4 Å (without phase-shift correction), there are overlapping contributions from backscatterers at longer distances, i.e., a longer Re-O distance and Re-metal contribution(s). Because of the complexity of this spectrum, fitting was not attempted. The FT EXAFS spectra of Re/C (Fig. 3b, i) and PdRe/C (Fig. 3b, ii) after in situ reduction at 400 and 300 °C, respectively, also contain multiple low-intensity peaks in the first-shell region, and the quantitative fitting results (Table 3) are described below. R factors are less than 1% indicating excellent fits to the experimental data. An excellent fit of the Re LIII EXAFS of the Re/C catalyst after in situ reduction at 400 °C and purging with He was obtained using first-shell Re-Re and ReO contributions, as illustrated in Fig. 4a. In addition to the overall fit, contributions from the individual backscattering paths are shown in Fig. 4b (real part of FT). The Re-Re contribution is dominant, and the Re-Re distance (Table 3) is equivalent to the average value in the bulk hcp metal. The Re-Re coordination number is indicative of small supported metal particles, and the Debye-Waller factor for the ReRe path is in excellent agreement with that reported by Yang, et al. [58]. The Debye-Waller factor for the Re-O contribution was fixed at 0.005 Å2 to reduce the number of free parameters; the observed range when this parameter was allowed to vary was 0.0038 to 0.0067 Å2. Although the presence of residual perrhenate species cannot be excluded, adding a short Re = O path at ∼1.73 Å did not significantly improve the fit. Very similar fitting results (Table 3) were obtained for the Re/C catalyst immediately after in situ reduction at 400 °C in 5% H2. The slight differences in the fitting parameters are not considered significant. Fig. 5 illustrates simultaneous fits of the Pd K and Re LIII EXAFS spectra of PdRe/C after in situ reduction at 300 °C and purging with He at 100 °C. The first-shell Pd K EXAFS can be fit satisfactorily using a single Pd-Pd path; however, multiple contributions are apparent in the Re LIII EXAFS. In addition to Re-Re and Re-O contributions (similar to those found for Re/C), a Re-Pd contribution improves the fit markedly. From the real part of the FT (Fig. 5c), it is clear that the Re-Pd path destructively interferes with the Re-Re path [33]. Because of the dominant Pd-Pd contribution, adding a Pd-Re contribution only improves the Pd K EXAFS fit modestly. Nevertheless, in order to obtain internally consistent results at both edges and given the necessity of a Re-Pd path at the Re LIII edge, a Pd-Re path was included (Fig. 5b). The Pd-Re and Re-Pd coordination numbers were constrained to be equal (because Pd and Re are present in a 1:1 atomic ratio), and only a single heterometallic distance and Debye-Waller factor were used. The DebyeWaller factors for the Re-Re and Re-Pd contributions were fixed at 0.007 Å2 (equivalent to that found for Re/C) and 0.004 Å2 (equivalent to that found in Ref [33].), respectively, to further reduce the number of free parameters. The EXAFS fitting results (Table 3) are indicative of a mixture of small Re particles and larger Pd-rich nanoparticles. The RePd and Re-Re bond distances are shorter than the average interatomic distances in bulk Re and Pd. Similar fitting results were obtained for the PdRe/C sample immediately after in situ reduction at 300 °C in 5% H2. The longer Pd-Pd distance and higher Debye-Waller factor under H2 indicate β-PdHx formation [57], as confirmed by TPHD (Fig. S3, Supplementary Information). For comparison, fitting results for an EXAFS model excluding the heterometallic contributions are provided in Supplementary Information (Table S2). Clearly, as indicated by comparison of R factors, the Pd-Re and Re-Pd contributions significantly improve the fits at each edge. HAADF-STEM images of the Re/C catalyst (Fig. 6) after reduction at 400 °C in H2 reveal very small Re particles dispersed on the activated carbon support. In the HAADF-STEM image shown in Fig. 6a, the largest particles are < 5 nm in size, and most are much smaller (1–2 nm). A higher-magnification image (Fig. 6b) and corresponding EDX map (Fig. 6c) of the highlighted region reveal Re particles ranging from 0.5 to 4 nm in size. The smallest features in this image may be small clusters or single-atom (mononuclear) Re species [55]. The larger supported Re

Fig. 3. a) Pd K edge FT EXAFS spectra (k2-weighting): (i) Pd/C (B) reduced in situ at 400 °C and purged with He; (ii) PdRe/C reduced in situ at 300 °C and purged with He; (iii) PdRe/C reduced ex situ at 400 °C and passivated in air. (b) Re LIII edge FT EXAFS spectra (k2-weighting): (i) Re/C reduced in situ at 400 °C and purged with He; (ii) PdRe/C reduced in situ at 300 °C and purged with He; (iii) PdRe/C reduced ex situ at 400 °C, passivated, and purged with H2 at 25 °C; (iv) Re/C reduced ex situ at 400 °C and passivated in air.

reduction (Fig. 3a, ii) are closely similar; there is only a small decrease in low R backscattering concomitant with reduction. EXAFS fitting results for Pd/C (B) after in situ reduction at 400 °C are given in Table 3. Because the first-shell Pd-Pd coordination number (N) exceeds 10, a detailed analysis of the higher shells would be required to accurately infer particle size and morphology [56]. The longer Pd-Pd distance and higher Debye-Waller factor for Pd/C (B) under H2 indicate β-PdHx formation [57], as confirmed by temperature-programmed hydride decomposition (TPHD) (Fig. S3, Supplemental Information). After the β-PdHx phase is decomposed by purging in He at 100 °C, the first-shell Pd-Pd distance becomes equivalent to that in the bulk, and there is a concomitant decrease in disorder (σ2) [57]. The Re LIII FT EXAFS spectrum of Re/C after ex situ reduction at 400 °C and passivation in air (Fig. 3b, iv) contains only a single low R contribution. Fitting this simple spectrum yielded a Re-O bond distance, coordination number, and Debye-Waller factor (Table 3) that are consistent with tetrahedral perrhenate ion [ReO4]− [48]. The Re LIII EXAFS spectrum (Fig. 3b, iii) of the reduced and passivated PdRe/C catalyst after purging briefly with 5% H2 at 25 °C contains multiple low-R peaks 109

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

Table 3 EXAFS spectroscopy fitting results. In situ treatment

Patha

N

400 °C reduction, in 5% H2 400 °C reduction, after He purge

Pd-Pd Pd-Pd

11.3 ± 0.6 11.1 ± 0.4

none (in air) 400 °C reduction, in 5% H2

Re-O Re-Re Re-O Re-Re Re-O

3.8 7.9 1.3 7.4 1.4

± ± ± ± ±

0.3 0.7 0.2 0.8 0.3

Re/C 1.73 2.74 2.01 2.74 2.01

± ± ± ± ±

Pd-Pd Pd-Re Re-Re Re-Pd Re-O Pd-Pd Pd-Re Re-Re Re-Pd Re-O

9.0 0.9 6.4 0.9 1.8 8.4 0.9 6.4 0.9 1.9

± ± ± ± ± ± ± ± ± ±

1.0 0.2 0.4 0.2 0.2 1.6 0.2 0.6 0.2 0.3

PdRe/C 2.83 2.69 2.72 2.69 2.00 2.74 2.70 2.72 2.70 2.01

± ± ± ± ± ± ± ± ± ±

400 °C reduction, after He purge

300 °C reduction, 5% H2

300 °C reduction, after He purge

a b

σ2 (10−3 Å2)

ΔE0 (eV)

R factorb

9.1 ± 0.5 5.8 ± 0.2

−0.3 ± 0.4 0.0 ± 0.2

0.0036 0.0009

0.01 0.01 0.01 0.01 0.02

0.6 ± 0.6 7.2 ± 0.5 (5.0) 6.4 ± 0.6 (5.0)

10.5 ± 1.2 7.6 ± 1.1 7.6 ± 1.1 7.5 ± 1.3 7.5 ± 1.3

0.0055 0.0048

0.01 0.02 0.01 0.02 0.01 0.01 0.01 0.01 0.02 0.01

9.0 ± 1.0 (4.0) (7.0) (4.0) (5.0) 5.6 ± 1.1 (4.0) (7.0) (4.0) (5.0)

−2.3 ± 0.8 −2.3 ± 0.8 5.4 ± 1.6 5.4 ± 1.6 5.4 ± 1.6 −1.7 ± 1.4 −1.7 ± 1.4 6.2 ± 1.8 6.2 ± 1.8 6.2 ± 1.8

0.0090

R (Å) Pd/C (B) 2.82 ± 0.01 2.74 ± 0.01

0.0089

0.0072

Absorber-backscatterer pairs. Fit sum of squares of differences divided by data sum of squares.

particles do not appear to be equiaxed consistent with sheet or raft-like structures [58]. HAADF-STEM images of the PdRe/C catalyst after reduction at 400 °C in H2 are shown in Fig. 7. The larger (10–30 nm) particles seen in Fig. 7a comprise primarily Pd, as evidenced by EDX (Fig. 7b). In contrast, most Re appears to be dispersed over the carbon support in very small particles or clusters (Fig. 7c). In some cases, the Re clusters appear to be in direct contact with much larger Pd nanoparticles. A higher magnification HAADF-STEM image (Fig. 7d) shows an ∼35-nm Pd nanoparticle with very small (< 2 nm) Re entities covering the nanoparticle and the carbon support (Fig. 7f). Yellow pixels in the composite EDX map indicate superposition of Pd (red) and Re (green) pixels. 3.2. Guaiacol HDO Catalysts were tested for guaiacol HDO at 300 °C and 1 atm in a continuous flow reactor at ∼1 h−1 WHSV. Fig. 8 shows guaiacol conversion with time on stream (TOS) for the lab-prepared catalysts. Pd/C (B) after in situ reduction at 300 °C had modest activity that decayed slowly with TOS. Pd/C (B) exhibited lower activity after in situ reduction at 400 °C consistent with loss of Pd surface area. In contrast, guaiacol conversion for Pd/C (A) was ∼100% for the duration of the experiment (∼6 h TOS). Re/C showed a sharp initial decline in activity followed by slow deactivation to a stable conversion (∼45%) after 4–5 h TOS. PdRe/C exhibited less deactivation than Re/C, and its guaiacol conversion stabilized at ∼80% after 3 h TOS. Product distributions were evaluated for each catalyst after in situ reduction at 300 and 400 °C, except Pd/C (A) which was only tested after in situ reduction at 300 °C. The carbon yields of C1 (total), C6 and C7 products are shown for each catalyst and reduction condition in Fig. 9. Transalkylation products (e.g., methylguaiacols, cresols, 3-methyl catechol) were not detected over any of the catalysts. Total yields of fully deoxygenated products (benzene and cyclohexane) are indicated above each bar. Several C1 products, including methanol, CH4, CO, and CO2, were detected suggesting different guaiacol HDO pathways (vide infra). Pd/C (A) was highly active and selective to phenol and cyclohexan-one/-ol, and CO and methanol were the main C1 products. Notably, guaiacol HDO over Pd/C (A) did not produce significant yields of benzene and cyclohexane despite very high conversion. The major product over Pd/C (B) was 2-methoxy-cyclohexanol (2-MCH) irrespective of catalyst reduction temperature. Other significant

Fig. 4. Re LIII FT EXAFS spectra (k2-weighting) of Re/C after in situ reduction at 400 °C and purging with He at 100 °C: (a) magnitude and (b) real part.

110

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

reported in guaiacol HDO over Pd/C catalysts under similar operating conditions [14,15]. UNK 2 is likely 2-methoxy-cyclohexanone [15] based on its retention time and high selectivity over Pd/C (B) at low conversion (vide infra). Phenol selectivity of Pd/C (B) was higher after in situ reduction at 400 °C. Phenol was the major product of guaiacol HDO over Re/C; however, anisole, benzene, catechol, and 2-MCH were also produced in significant yields. Selectivity to 2-MCH was low, and yields of other ring-saturation products (e.g., cyclohexan-one/-ol, methoxycyclohexane, and cyclohexane) were negligible. Methane was the most abundant C1 product. In situ reduction of Re/C at 400 °C markedly increased guaiacol conversion and benzene yield. PdRe/C after in situ reduction at 300 °C produced a broad slate of products similar to that of Pd/C (B), except much greater yields of benzene and cyclohexane. Increasing catalyst reduction temperature resulted in higher activity; the combined yield of benzene and cyclohexane over PdRe/C after in situ reduction at 400 °C was higher than over any other catalyst. 3.3. Anisole HDO Anisole HDO was investigated under equivalent conditions (300 °C, 1 atm, ∼1 h−1 WHSV) to probe catalyst selectivity toward phenylOCH3 and OeCH3 bond hydrogenolysis. Trends in catalytic activity (Fig. 10) are similar to those observed for guaiacol, except the lower activity of Pd/C (A). Pd/C (A) and Pd/C (B) after in situ reduction at 300 °C had similar activities for anisole HDO. In addition to phenol, cyclohexan-one/-ol and methoxycyclohexane, the Pd/C catalysts produced significant yields of fully deoxygenated products. MeOH was the most abundant C1 product over the Pd/C catalysts. Re/C and PdRe/C were highly selective to C6 hydrocarbons and CH4. PdRe/C had higher hydrocarbon yield but produced more cyclohexane than Re/C. 3.4. Kinetics and reaction pathways Additional guaiacol HDO experiments were conducted at higher WHSVs (4-28 h−1) to achieve differential conversions (< 10%) allowing primary product identification and reaction rate measurements. Several conditions were tested in order to construct first-order Delplots (Fig. S4, Supplementary Information); primary products were identified by non-zero intercepts when selectivity was extrapolated to zero conversion [59]. Primary products are defined as product species desorbed from the catalyst surface following a single guaiacol adsorption event, i.e., without desorption and re-adsorption of intermediates [59]. Phenol, anisole, 2-MCH and catechol are primary products over each catalyst. UNK 2 is also a primary product over the Pd-containing catalysts, consistent with its assignment to 2-methoxycyclohexanone. Benzene is also a primary product over the Re-containing catalysts. Product formation rates calculated from product concentrations measured at differential conversion are given in Table 4. The average relative error (standard deviation divided by mean) of these measurements is ∼15%. The formation rates of C1 products are 10–25% lower than the corresponding formation rates of C6 products; the source of this systematic error is unknown. All products except 2-MCH have higher formation rates over Pd/C (A) than Pd/C (B). Over Pd/C (A), there is good agreement between the phenol and MeOH formation rates catalyst indicating primary hydrogenolysis of the phenyl−OCH3 bond, i.e., demethoxylation (DMO). Also, the catechol and CH4 formation rates are nearly equal evidencing primary hydrogenolysis of the O−CH3 bond, i.e., demethylation (DME). The high formation rate of UNK 2 over Pd/C (A) is consistent with its assignment as an aromatic ring hydrogenation (HYD) product. The HYD product 2-MCH has the highest rate of formation over Pd/C (B). Good agreement between the phenol and methanol formation rates is also observed over Pd/C (B); however, there is a significant deficit in CH4 production expected based on catechol formation. Over Re/C, phenol has the highest formation rate, consistent with DMO; however, the relatively high CH4 formation

Fig. 5. EXAFS spectra (k2-weighting) of PdRe/C after in situ reduction at 300 °C and purging with He at 100 °C: (a) FT magnitudes of Pd K (upper) and Re LIII (lower) EXAFS; (b) real part of Pd K EXAFS; and (c) real part of Re LIII EXAFS, including FEFF paths used in fits.

products over Pd/C (B) included phenol, cyclohexan-one/-ol, catechol, anisole, methoxycyclohexane, UNK 1, and UNK 2. UNK 1 is tentatively assigned to cyclopentanone; this unanticipated product was recently 111

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

Fig. 6. HAADF-STEM images of Re/C catalyst following reduction at 400 °C and transfer to the microscope under N2: (a) wide field of view, (b) higher magnification view of region indicated by box in part (a), and (c) EDX map of region in part (b).

similar dispersions after in situ reduction at 300 °C. Reduction of Pd/C (B) at 400 °C decreases its dispersion to less than 5%, consistent with sintering of the carbon-supported Pd nanoparticles. TPR of previously reduced/passivated Pd/C (B) demonstrated removal of superficial Pd oxide and concomitant β-PdHx formation under H2 at 40 °C. EXAFS spectroscopy and TPHD of Pd/C (B) after in situ reduction at 400 °C and cooling in H2 confirmed the formation of β-PdHx nanoparticles [51,57]. The higher guaiacol HDO activity of Pd/C (A) after in situ reduction at 300 °C may be related to the type of activated carbon and/or the presence of promoters (e.g., alkali metals). TPR of the Re/C catalyst indicates complete reduction of Re7+ species to metallic Re at ∼300 °C. Arnoldy, et al. assigned a similar TPR peak at 265–270 °C for a 5.2 wt.% Re/C catalyst prepared from NH4ReO4 to reduction of perrhenate crystallites to metallic Re [60]. A broad high-temperature peak was inferred to arise from reduction of isolated Re7+ ions interacting strongly with the carbon support; however, our QMS results indicate that Re-catalyzed methanation of the carbon support is primarily responsible for this feature [31,61]. Simonetti, et al. reported a TPR peak at 367 °C for HReO4 supported on Vulcan carbon black with quantitative H2 consumption sufficient for complete reduction of Re7+ to Re0 [62]. Similarly, Shao, et al. reported a TPR peak at ∼370 °C for a 4 wt.% Re/C catalyst prepared from NH4ReO4 and previously reduced ex situ at 300 °C [31]. Despite the range of reported TPR peak temperatures, there is consensus that reduction of carbon-supported Re7+ species to Re0 occurs at temperatures ≤400 °C.

rate indicates that guaiacol also reacts via DME. Over PdRe/C, the formation rates of phenol and anisole are substantially greater than the sum of the corresponding rates over Pd/C (B) [or Pd/C (A)] and Re/C, and there is good agreement between the phenol formation rate and the sum of the MeOH and CO formation rates. The anisole formation rate indicative of DDO of guaiacol is notably higher over PdRe/C. CH4 formation outpaces catechol formation consistent with rapid HDO of catechol surface intermediates. Significant formation rates of benzene and cyclohexane are observed over PdRe/C even at differential conversion. TOFs (Table 4) were calculated from individual product formation rates and the CO chemisorption uptake for each catalyst after in situ reduction. Each irreversibly adsorbed CO molecule was assumed to occupy one catalytic site. The TOFs for phenol and cyclohexanone/-ol over the two Pd/C catalysts are in excellent agreement; however, TOFs for aromatic ring HYD products (2-MCH and UNK 2) are much higher over Pd/C (B). TOFs for anisole and catechol formation are also higher over Pd/C (B). The TOF for phenol formation over Re/C is approximately an order of magnitude lower than over the Pd/C catalysts. PdRe/C exhibits significantly higher TOFs for primary products, including benzene, than Re/C.

4. Discussion Two Pd/C catalysts were investigated in this work: a commercial sample (A) and a lab-prepared catalyst (B). These catalysts have very 112

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

Fig. 7. HAADF-STEM images of PdRe/C catalyst following reduction at 400 °C and transfer to the microscope under N2 (a, d) HAADF, (b, e) Pd EDX and (c, f) Pd and Re composite EDX.

Fig. 8. Catalyst performance with TOS at 300 °C, 1 atm and ∼1 h−1 WHSV; catalyst reduction temperature 400 °C, except as indicated in legend. Pd/C refers to Pd/C (B).

Fig. 9. Carbon-based yields for products of guaiacol HDO over Pd/C, Re/C and PdRe/C catalysts at 300 °C and 1 atm H2. Catalysts are labeled by composition and in situ reduction temperature (°C). Total yields of fully deoxygenated products (benzene and cyclohexane) are indicated above each bar.

Chemisorption measurements and HAADF-STEM images of the Re/ C catalyst after initial reduction at 400 °C are indicative of very small supported Re particles. The large irreversible CO uptake (220 μmol/g) after initial reduction almost certainly corresponds to chemisorption on metallic Re [33,62,63]. In contrast, the supported Re clusters do not chemisorb a comparable amount of H2 at 35 °C, in agreement with the widely reported observation that supported Re particles do not dissociatively chemisorb H2 [32,52,53,62]. The low H/Re ratio at 35 °C

and increase in H2 uptake with temperature is typical for Re nanoparticles supported on carbon [62] and γ-Al2O3 [32,53]. For comparison, Simonetti, et al. reported that a 5 wt.% Re/C catalyst after reduction at 450 °C had a CO uptake of 30 μmol/g and negligible H2 uptake [62]. Consistent with high dispersion of the Re/C catalyst, HAADF-STEM images show sub-5-nm particles and very small (< 1 nm) entities that may be Re adatoms or clusters [55]. The larger Re particles appear to have a sheet or raft-like morphology [58]. Exposure of Re/C 113

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

LIII EXAFS of a 14 wt.% Re/SiO2 catalyst that had been reduced in situ at 500 °C [28]. Similarly, Koso, et al. found a 2.07-Å Re-O contribution (N = 2.0) for small silica-supported Re particles (NRe-Re = 6.6) after reduction of a 3.6 wt.% Re/SiO2 catalyst at ∼380 °C in 5% H2/Ar [40]. If we estimate the average Re oxidation state from the Re-O coordination number [40], the result (∼1.4+) is in good agreement with the Re LIII edge shift. Because of the relatively high Re dispersion, it is possible that the Re-O contribution originates from surface species (e.g., hydroxyl groups [35]) and/or from the metal-support interface [67]. Activated carbons, such as Norit SX-1 G, possess oxygen-containing surface functional groups [65,68]. The presence of hydroxyl groups or other adsorbed oxygen species could also explain the relatively insignificant H2 chemisorption uptake by the Re/C catalyst. TPR and EXAFS spectroscopy of PdRe/C demonstrate that Pd facilitates low-temperature reduction of supported perrhenate species via hydrogen spillover and/or migration of ReOx species to Pd particles. Similar observations have been reported for PdRe/Al2O3 [32,51] and analogously for PtRe/C [34,62] catalysts. The Re LIII EXAFS spectrum of the reduced and passivated PdRe/C catalyst following brief exposure to flowing 5% H2 at 25 °C indicates reduction of perrhenate species—consistent with intimacy of Pd° and ReOx species. EXAFS spectroscopy of the PdRe/C catalyst after in situ reduction in 5% H2 at 300 °C evidences that the catalyst contains small Re clusters and larger Pd nanoparticles. The total first-shell coordination number for metallic Re (∼7.3) is equivalent to that of Re/C after in situ reduction at 400 °C. Similar to Re/C, there is a Re-O contribution at a distance consistent with a single bond. For PdRe/C, the Re-O coordination number is marginally higher in agreement with its greater white line intensity. The total first-shell coordination number for Pd (∼9.9) indicates that the particles are smaller than those characterizing Pd/C (B) after in situ reduction at 400 °C. The EXAFS fitting results also include a statistically significant Pd-Re contribution at ∼2.70 Å. The heterometallic distance is significantly shorter than the homometallic distances in the bulk metals. Similar EXAFS spectroscopy results for PdRe catalysts include: a 2.67-Å Pd-Re distance in a 1:1 PdRe/Al2O3 catalyst [69]; a 2.68-Å PdRe distance in a PdRe/CeO2 catalyst [70]; and a 2.72 Å Pd-Re distance in a PdRe/Al2O3 catalyst prepared from [Pd(NH3)4][ReO4]2 [33]. The Cowley short-range order parameter ( α) was calculated at the Re and Pd edges from the partial coordination numbers in Table 3 [71]. Both values are ∼0.8 indicating a strong tendency for metal segregation in the bimetallic catalyst. Moreover, β-PdHx formation is sensitive to alloying [51]. EXAFS spectroscopy and TPHD (Fig. S3, Supplementary Information) evidence hydride formation with very similar H/Pd ratios for Pd/C (B) and PdRe/C arguing against bulk alloy formation. Consistent with these results, HAADF-STEM images and EDX maps of the catalyst after reduction at 400 °C show very small Re particles (clusters) covering the carbon support nearby and in contact with larger Pd nanoparticles.

Fig. 10. Carbon-based yields for products of anisole HDO over Pd/C, Re/C and PdRe/C catalysts at 300 °C and 1 atm H2. Catalysts are labeled by composition and in situ reduction temperature (°C). Total yields of fully deoxygenated products (benzene and cyclohexane) are indicated above each bar.

previously reduced at 400 °C to ambient air results in oxidation of the supported Re particles forming perrhenate species. Subsequent TPR and in situ EXAFS spectroscopy demonstrate reduction of these Re7+ species to Re° after heating to 400 °C in 5% H2. CO chemisorption indicates that this redox cycle results in some loss of Re dispersion. The first-shell Re-Re distance and corresponding higher shell peaks found in the EXAFS spectrum of the Re/C catalyst after in situ reduction at 400 °C are fully consistent with small metallic Re particles; however, the Re LIII XANES exhibits greater white line intensity and a higher edge energy than Re powder. Although the Re powder white line may be attenuated by thickness effects [55,65], the edge shift indicates that the supported Re clusters have a low positive oxidation state (∼2+). These observations are typical for small Re particles supported on MOR/alumina [55], ZSM-5 [64], SiO2 [40], and carbon [34,36,65]. Re clusters containing up to 13 atoms are too small to exhibit bulk electronic properties and may become electron deficient when supported on metal oxides [55]. In addition to Re backscatterers, O backscatterers were detected at a distance consistent with a Re-O single bond. The Re-O distance coincides with the average first-shell distance in ReO2 (2.01 Å) [66]. If we assume that the Re-O contribution arises from a minority ReO2 species, ∼23% (1.4/6) of the Re atoms would be in this phase based on octahedral coordination of Re4+ (e.g., ReO6 octahedra). A similar Re-O contribution has been reported for supported Re nanoparticles produced by “wet reduction” of rhenium oxide [58]. Takeda, et al. reported a small (N = 0.6) Re-O contribution at 2.06 Å to the Re Table 4 Primary product formation rates and turnover frequencies (TOFs) from guaiacol. Catalyst

Pd/C (A)a

Pd/C (B)b

Product phenol cyclohexanone/-ol anisole catechol 2-MCH UNK 2 benzene MeOH CO CH4 a b

Re/Cb

PdRe/Cb

Pd/C (A)

Pd/C (B)

0.38 0.06 0.38 0.45 0.63 0.91 0.00 0.41 0.00 0.26

PdRe/C

5.1 0.0 0.8 0.8 0.7 0.1 0.4 1.3 0.8 3.4

19.7 3.3 6.8 2.1 2.2 1.2 3.2 15.6 0.0 8.2

TOF (h−1)

Rate (mmol/g/h) 2.40 0.40 0.46 0.72 0.39 2.17 0.00 2.15 0.34 0.75

Re/C

1.14 0.00 0.18 0.19 0.16 0.02 0.09 0.28 0.17 0.75

4.42 0.73 1.52 0.48 0.49 0.28 0.72 3.50 0.00 1.84

Reduced in situ at 300 °C. Reduced in situ at 400 °C. 114

61.1 10.3 11.7 18.3 10.0 55.3 0.0 54.8 8.6 19.2

57.0 9.1 57.3 67.6 94.4 136 0.2 61.9 0.0 38.9

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

Fig. 11. Proposed reaction pathways for guaiacol over Pd/C, Re/C and PdRe/C catalysts at 300 °C and 1 atm [15,17,25]. H2 addition has been omitted for simplicity. Red arrows denote predominantly Re-catalyzed pathways. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Leiva, et al. reported that Re/SiO2 catalysts were active for guaiacol HDO in the liquid phase at 300 °C under 5 MPa H2 [24]. Under equivalent conditions, Ghampson, et al. reported that Re supported on carbon nanofibers (Re/CNF) was active for guaiacol HDO to benzene and cyclohexane [25]. Over these supported Re catalysts, phenol was the initial product of guaiacol HDO, and it subsequently was deoxygenated to benzene and cyclohexane. Phenol HDO was demonstrated independently over Re/CNF at 300 °C under 3 MPa H2, and DDO (rather than hydrogenation-dehydration) was the dominant reaction pathway [25]. Partially reduced ReOx moieties were proposed as the catalytically active sites for guaiacol and phenol DDO. The Re/CNF and Re/ SiO2 catalysts were dried and calcined in air at 300 °C prior to testing, and post-reaction x-ray photoelectron spectroscopy performed ex situ evidenced a mixture of Re7+, Re4+ and Re δ+ (0 < δ < 4) species [24,25]. The PdRe/C catalyst after in situ reduction at 400 °C produces the highest yield of fully deoxygenated products from guaiacol. Moreover, it can achieve complete HDO of guaiacol to benzene even at differential conversions. The PdRe/C and Re/C catalysts have equivalent CO site densities, but the latter is much more active. PdFe/C catalysts have been reported to exhibit similar guaiacol HDO performance albeit at higher temperatures (350–450 °C) [16]. Phenol and anisole are key intermediates in complete HDO of guaiacol at integral conversions (see Fig. 11). The formation rates of phenol and anisole over PdRe/C are substantially greater than the sum of the corresponding rates over Re/C and Pd/C (B) [or the more active Pd/C (A)]. In contrast, the apparent rate of catechol formation is lower over PdRe/C than Pd/C (B) [or Pd/C (A)]. Very rapid deoxygenation of catechol surface intermediates may explain this observation, and rapid HDO of vicinal diols (e.g., 1,4-anhydroerythritol) over PdRe/CeO2 catalysts has been reported [70]. Anisole is converted in high yield to benzene and cyclohexane over

Proposed reaction pathways for guaiacol over Pd/C, Re/C and PdRe/C are illustrated in Fig. 11 [15,17,25]. Red arrows denote predominantly Re-catalyzed pathways. Our results for Pd/C (A) confirm previous reports that (1) guaiacol HDO over Pd/C is selective to phenol (and cyclohexan-one/-ol) and (2) hydrocarbons are not produced in significant yields even at high conversions [14–16]. DMO and aromatic ring HYD are dominant over Pd/C (A) and Pd/C (B), respectively. DME and DDO are more prevalent over Pd/C (B). UNK1 is a significant product at high guaiacol conversion over both Pd/C catalysts, and it is assigned to cyclopentanone based on recent reports [14,15]. The observed CO production may result from cyclopentanone formation via decarbonylation [14,19] and/or decomposition of methanol or surface methoxy species [72]. Anisole conversion over the Pd/C catalysts yields multiple products, including substantial amounts of benzene and cyclohexane. DMO, DME, and HYD of anisole result in benzene, phenol, and methoxycyclohexane, respectively. Because phenol is not converted to benzene over Pd/C [15], selectivity to fully deoxygenated products from anisole is limited to 50–60%. Guaiacol HDO over Re/C is selective to phenol; however, benzene, anisole, catechol, and 2-MCH are produced in significant yields. Secondary HYD products (i.e., cyclohexan-one/-ol and cyclohexane) are strongly suppressed, and neither cresols nor xylenes are observed as reaction products. Moreover, HDO activity and benzene yield increase significantly after in situ reduction of Re/C at 400 °C, and complete deoxygenation of guaiacol to benzene is observed over this catalyst even at very low conversions. Because the CH4 formation rate over Re/ C is significantly greater than the catechol formation rate, rapid HDO of catechol surface intermediates is inferred. The Re/C catalyst is active for vapor-phase HDO of anisole and highly selective to benzene. Because CH4 is the main C1 product, we infer that over Re/C, anisole undergoes DME to phenol followed by deoxygenation to benzene. 115

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

PdRe/C. Because Pd/C catalysts do not convert guaiacol to benzene (or cyclohexane) in significant yields even at high conversion [16], we infer that Re plays an active role, specifically, by enabling DDO of phenol [25] and catechol [70]. The higher yields of fully deoxygenated products observed following in situ reduction of the PdRe/C catalyst at 400 °C suggest that the active sites comprise Pd nanoparticles and lowvalent Re species, consistent with the characterization data.

[13] D.E. Resasco, S.P. Crossley, Catal. Today 257 (2015) 185–199. [14] D. Gao, C. Schweitzer, H.T. Hwang, A. Varma, Ind. Eng. Chem. Res. 53 (2014) 18658–18667. [15] H. Shafaghat, P.S. Rezaei, W. Daud, RSC Adv. 5 (2015) 33990–33998. [16] J.M. Sun, A.M. Karim, H. Zhang, L. Kovarik, X.H.S. Li, A.J. Hensley, J.S. McEwen, Y. Wang, J. Catal. 306 (2013) 47–57. [17] R.C. Runnebaum, T. Nimmanwudipong, D.E. Block, B.C. Gates, Catal. Sci. Tech. 2 (2012) 113–118. [18] T. Nimmanwudipong, R.C. Runnebaum, D.E. Block, B.C. Gates, Energy Fuels 25 (2011) 3417–3427. [19] T. Nimmanwudipong, C. Aydin, J. Lu, R.C. Runnebaum, K.C. Brodwater, N.D. Browing, D.E. Block, B.C. Gates, Catal. Lett. 142 (2012) 1190–1196. [20] A.M. Robinson, J.E. Hensley, J.W. Medlin, ACS Catal. 6 (2016) 5026–5043. [21] W. Song, Y. Liu, E. Baráth, C. Zhao, J.A. Lercher, Green Chem. 17 (2015) 1204–1218. [22] D.-Y. Hong, S.J. Miller, P.K. Agrawal, C.W. Jones, Chem. Commun. 46 (2010) 1038–1040. [23] H. Lee, H. Kim, M.J. Yu, C.H. Ko, J.-K. Jeon, J. Jae, S.H. Park, S.-C. Jung, Y.-K. Park, Sci. Rep. 6 (2016), http://dx.doi.org/10.1038/srep28765. [24] K. Leiva, N. Martinez, C. Sepulveda, R. Garcia, C.A. Jimenez, D. Laurenti, M. Vrinat, C. Geantet, J.L.G. Fierro, I.T. Ghampson, N. Escalona, Appl. Catal. A 490 (2015) 71–79. [25] I.T. Ghampson, C. Sepulveda, R. Garcia, J.L.G. Fierro, N. Escalona, Catal. Sci. Tech. 6 (2016) 4356–4369. [26] N. Martinez, R. Garcia, J.L.G. Fierro, C. Wheeler, R.N. Austin, J.R. Gallagher, J.T. Miller, T.R. Krause, N. Escalona, C. Sepulveda, Fuel 186 (2016) 112–121. [27] Y. Takeda, Y. Nakagawa, K. Tomishige, Catal. Sci. Tech. 2 (2012) 2221–2223. [28] Y. Takeda, M. Tamura, Y. Nakagawa, K. Okumura, K. Tomishige, Catal. Sci. Tech. 6 (2016) 5668–5683. [29] Y. Takeda, M. Tamura, Y. Nakagawa, K. Okumura, K. Tomishige, ACS Catal 5 (2015) 7034–7047. [30] B.K. Ly, B. Tapin, M. Aouine, P. Delichere, F. Epron, C. Pinel, C. Especel, M. Besson, ChemCatChem 7 (2015) 2161–2178. [31] Z. Shao, C. Li, X. Di, Z. Xiao, C. Liang, Ind. Eng. Chem. Res. 53 (2014) 9638–9645. [32] S.T. Thompson, H.H. Lamb, ACS Catal 6 (2016) 7438–7447. [33] S.T. Thompson, H.H. Lamb, J. Catal. 350 (2017) 111–121. [34] O.M. Daniel, A. DeLaRiva, E.L. Kunkes, A.K. Datye, J.A. Dumesic, R.J. Davis, ChemCatChem 2 (2010) 1107–1114. [35] M. Chia, Y.J. Pagan-Torres, D. Hibbitts, Q.H. Tan, H.N. Pham, A.K. Datye, M. Neurock, R.J. Davis, J.A. Dumesic, J. Am. Chem. Soc. 133 (2011) 12675–12689. [36] M. Chia, B.J. O’Neill, R. Alamillo, P.J. Dietrich, F.H. Ribeiro, J.T. Miller, J.A. Dumesic, J. Catal. 308 (2013) 226–236. [37] S.H. Hakim, C. Sener, A.C. Alba-Rubio, T.M. Gostanian, B.J. O’Neill, F.H. Ribeiro, J.T. Miller, J.A. Dumesic, J. Catal. 328 (2015) 75–90. [38] S. Koso, H. Watanabe, K. Okumura, Y. Nakagawa, K. Tomishige, Appl. Catal. B 111–112 (2012) 27–37. [39] K. Tomishige, Y. Nakagawa, M. Tamura, Green Chem. 19 (2017) 2876–2924. [40] S. Koso, H. Watanabe, K. Okumura, Y. Nakagawa, K. Tomishige, J. Phys. Chem. C 116 (2012) 3079–3090. [41] B. Predel, O.E. Madelung (Ed.), Landolt-Börnstein - Group IV Physical Chemistry, Springer Materials, 1999. [42] G. Fagherazzi, P. Canton, P. Riello, N. Pernicone, F. Pinna, A. Battagliarin, Langmuir 16 (2000) 4539–4546. [43] R. Ducros, J. Fusy, J. Jupille, P. Pareja, S. Tatarenko, Appl. Surf. Sci. 29 (1987) 179–193. [44] F.W.H. Kampers, T.M.J. Maas, J. Vangrondelle, P. Brinkgreve, D.C. Koningsberger, Rev. Sci. Instrum. 60 (1989) 2635–2638. [45] M. Newville, J. Synchrotron Rad. 8 (2001) 322–324. [46] B. Ravel, M. Newville, J. Synchrotron Rad. 12 (2005) 537–541. [47] B. Ravel, J. Synchrotron Rad. 8 (2001) 314–316. [48] G.J. Kruger, E.C. Reynhardt, Acta Crystallograph. B 34 (1978) 259–261. [49] C. Kittel, Introduction to Solid State Physics, 7th ed., Wiley, New York, 1996. [50] K. Sun, T.C. Schulz, S.T. Thompson, H.H. Lamb, Catal. Today 269 (2016) 93–102. [51] S.B. Ziemecki, G.A. Jones, J.B. Michel, J. Catal. 99 (1986) 207–217. [52] G.W. Chadzynski, H. Kubicka, Thermochim. Acta 158 (1990) 353–367. [53] M.P. Latusek, R.M. Heimerl, B.P. Spigarelli, J.H. Holles, Appl. Catal. A 358 (2009) 79–87. [54] J.E. Benson, H.S. Hwang, M. Boudart, J. Catal 30 (1973) 146–153. [55] S.R. Bare, S.D. Kelly, F.D. Vila, E. Boldingh, E. Karapetrova, J. Kas, G.E. Mickelson, F.S. Modica, N. Yang, J.J. Rehr, J. Phys. Chem. C 115 (2011) 5740–5755. [56] A.I. Frenkel, C.W. Hills, R.G. Nuzzo, J. Phys. Chem. B 105 (2005) 12689. [57] J.A. McCaulley, J. Phys. Chem. 97 (1993) 10372–10379. [58] N. Yang, G.E. Mickelson, N. Greenlay, S.D. Kelly, F.D. Vila, J. Kas, J.J. Rehr, S.R. Bare, AIP Conf. Proc. 882 (2007) 591–593. [59] N.A. Bhore, M.T. Klein, K.B. Bischoff, Ind. Eng. Chem. Res. 29 (1990) 313–316. [60] P. Arnoldy, E.M. van Oersa, O.S.L. Bruinsma, V.H.J. de Beer, J.A. Moulijn, J. Catal. 93 (1985) 231–245. [61] X. Di, A. Shao, C. Li, W. Li, C. Liang, Catal. Sci. Technol. 5 (2015) 2441–2448. [62] D.A. Simonetti, E.L. Kunkes, J.A. Dumesic, J. Catal. 247 (2007) 298–306. [63] W. Daniell, T. Weingand, H. Knozinger, J. Mol. Catal. A 204 (2003) 519–526. [64] H.S. Lacheen, P.J. Cordeiro, E. Iglesia, Chem. Eur. J. 13 (2007) 3048–3057. [65] P.U. Karanjkar, S.P. Burt, X. Chen, K.J. Barnett, M.R. Ball, M.D. Kumbhalkar, X. Wang, J.B. Miller, I. Hermans, J.A. Dumesic, G.W. Huber, Catal. Sci. Technol. 6 (2016) 7841–7851. [66] R.W.G. Wyckoff, Crystal Structures, 2nd ed., Interscience Publishers, New York, NY, 1963.

5. Conclusions TPR, CO chemisorption and EXAFS spectroscopy of the Re/C catalyst indicate reduction of carbon-supported perrhenate species to metallic Re at 400 °C or less. HAADF-STEM reveals very small (< 1 nm) Re entities (clusters) and larger nanoparticles covering the activated carbon support. TPR and EXAFS spectroscopy evidence that Pd facilitates low-temperature reduction of Re in the PdRe/C catalyst, and a 2.70 Å Pd-Re backscattering path was required to adequately fit the Re LIII EXAFS spectrum of PdRe/C. HAADF-STEM with EDX reveals intimate contact between Re clusters and larger Pd-rich nanoparticles. βPdHx formation in the bimetallic catalyst argues against PdRe alloying. The Re/C and PdRe/C catalysts exhibit activity for complete HDO of guaiacol to benzene and cyclohexane. In contrast, Pd/C (A) is highly selective to phenol (and cyclohexanone/-ol) and does not produce significant benzene (or cyclohexane) even at high conversion. The Re/C catalyst produces phenol, benzene and anisole, but it deactivates rapidly and lacks the HYD activity to convert benzene to cyclohexane. The PdRe/C catalyst combines the capacity of Re/C to fully deoxygenate guaiacol with the DMO and HYD functions of Pd/C. Formation rates of phenol and anisole are substantially greater over PdRe/C than the sum of the corresponding rates over Pd/C (B) and Re/C demonstrating bimetallic synergy. Phenol, anisole, 2-MCH and catechol are primary products over all the catalysts. Benzene is also a primary product over the Re-containing catalysts. Acknowledgements This work was supported by the U.S. Department of Energy under an award to the Integrated Biomass Research Initiative at North Carolina State University (DE-FG36-08GO88053). The NC State AIF is supported by the State of North Carolina and the National Science Foundation (ECCS-1542015). We appreciate the assistance of Dr. Xiahan Sang of AIF. We acknowledge ExxonMobil Corporation and the Naval Research Laboratory for beam time at the National Synchrotron Light Source, Brookhaven National Laboratory. ICP-OES analysis was provided by Eastman Chemical Company. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.apcata.2018.06.031. References [1] Y.C. Lin, G.W. Huber, Energy Env. Sci. 2 (2009) 68–80. [2] A.V. Bridgwater, Appl. Catal. A 116 (1994) 5–47. [3] U.S. Department of Energy, Biotechnologies Office, Multiyear Research, Development & Deployment Plan, 2016. [4] E. Laurent, B. Delmon, Appl. Catal. A 109 (1994) 97–115. [5] A.V. Bridgwater, G.V.C. Peacocke, Renew. Sustainable Energy Rev. 4 (2000) 1–73. [6] E. Laurent, B. Delmon, Appl. Catal. A 109 (1994) 77–96. [7] E. Furimsky, Appl. Catal. A 199 (2000) 147–190. [8] D.C. Elliott, Energy Fuels 21 (2007) 1792–1815. [9] D.C. Elliott, T.R. Hart, G.G. Neuenschwander, L.J. Rotness, A.H. Zacher, Env. Prog. Sus. Energy 28 (2009) 441–449. [10] J. Wildschut, F.H. Mahfud, R.H. Venderbosch, H.J. Heeres, Ind. Eng. Chem. Res. 48 (2009) 10324–10334. [11] J. Wildschut, M. Iqbal, F.H. Mahfud, I. Melian-Cabrera, R.H. Venderbosch, H.J. Heeres, Energy Env. Sci. 3 (2010) 962–970. [12] D.C. Elliott, T.R. Hart, Energy Fuels 23 (2009) 631–637.

116

Applied Catalysis A, General 563 (2018) 105–117

S.T. Thompson, H.H. Lamb

3213. [71] A.I. Frenkel, Q. Wang, S.I. Sanchez, M.W. Small, R.G. Nuzzo, J. Chem. Phys 138 (2013) 064202. [72] W.J. Shen, Y. Matsumura, J. Mol. Catal. A 153 (2000) 165–168.

[67] D.C. Koningsberger, B.C. Gates, Catal. Lett. 14 (1992) 271–277. [68] J.L. Figueiredo, M.F.R. Pereira, M.M.A. Freitas, J.J.M. Órfão, Carbon 37 (1999) 1379–1389. [69] G. Meitzner, G.H. Via, F.W. Lytle, J.H. Sinfelt, J. Chem. Phys. 87 (1987) 6354–6363. [70] N. Ota, M. Tamura, Y. Nakagawa, K. Okumura, K. Tomishige, ACS Catal. 6 (2016)

117