Alumina Supported Au–Ni: Surface Synergism in the Gas Phase

Article pubs.acs.org/JPCC

Alumina Supported Au−Ni: Surface Synergism in the Gas Phase Hydrogenation of Nitro-Compounds Fernando Cárdenas-Lizana,†,§ Santiago Gómez-Quero,† Gary Jacobs,‡ Yaying Ji,‡ Burtron H. Davis,‡ Lioubov Kiwi-Minsker,§ and Mark A. Keane*,† †

Chemical Engineering, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, Scotland Center for Applied Energy Research, 2540 Research Park Drive, Lexington, Kentucky 40511, United States § Group of Catalytic Reaction Engineering, Ecole Polytechnique Fédérale de Lausanne, Lausanne CH-1015, Switzerland ‡

ABSTRACT: The catalytic gas phase hydrogenation of 1,3dinitrobenzene over Au/Al2O3 delivered exclusive reduction of a single −NO2 substituent to generate 1,3-nitroaniline (Ea = 131 kJ mol−1), reaction over Ni/Al2O3 resulted in full hydrogenation to 1,3-phenylenediamine (Ea = 38 kJ mol−1), whereas both products were isolated over Au−Ni/Al2O3. In the hydrogenation of 1,3,5-trinitrobenzene, Au/Al2O3 promoted preferential partial hydrogenation (3,5-dinitroaniline), Ni/Al2O3 generated 1,3,5-triaminobenzene as the predominant product, while Au−Ni/Al2O3 exhibited an intermediate catalytic response. The catalysts have been characterized in terms of temperature programmed reduction (TPR), H2 chemisorption, powder XRD, high-resolution TEM, XPS, and XANES/EXAFS measurements. Post-TPR, there was evidence of a metal particle redispersion resulting from the introduction of Au to Ni/Al2O3 where HRTEM-EDX mapping has established close proximity of Au and Ni on Au−Ni/Al2O3 with electron transfer from Ni to Au (from XPS analysis). Hydrogen chemisorption on Au−Ni/Al2O3 was 3 times lower than that recorded for Ni/Al2O3, suggesting Au−Ni interaction that inhibits H2 uptake. Simulation of the XANES/EXAFS response provided a better fit when incorporating Au−Ni interaction. The results demonstrate control of selectivity in poly nitroarene hydrogenation through the use of mono- (Au and Ni) and bi- (Au−Ni) metallic catalysts, where catalytic response is governed by surface composition.

■

theless, there is evidence in the literature16,17 of unique hydrogenation selectivity over supported Au. The application of bimetallic systems (containing Au) has been considered to some extent where modifications to the catalytic response were achieved with the incorporation of the second metal. Serna et al.,18 studying the gas phase hydrogenation of 3-nitrostyrene, recorded an enhanced −NO2 group reduction selectivity over Pt−Au/TiO2 when compared with Pt/TiO2. Menegazzo and co-workers19 have reported a greater resistance to sulfur poisoning in the liquid phase hydrogenation of benzaldehyde due to the incorporation of Au with Pd/C. Supported Au−Ni bimetallics have exhibited improved stability in steam reforming as a result of decreased coke deposition.20,21 Moreover, an increased activity in the HDS of thiophene and hydrogenation of phenylacetylene over SiO222 and Al2O323 supported Au−Ni has been reported. In this paper, we study the catalytic performance of alumina supported Au−Ni in the hydrogenation of di- and trinitrobenzene. Aromatic amino-compounds are used in the manufacture of a diversity of fine chemicals.24 Standard production routes involve batch liquid phase operation using

INTRODUCTION Up to 40% of all chemical processes in the pharmaceutical and fine chemical industries involve hydrogenation steps.1 In the catalytic hydrogenation of multifunctional hydrocarbons, notably aromatics, selectivity is a crucial process challenge.2 Several factors have been proposed to impact selectivity, including catalyst structure (metal/support interactions and metal dispersion),3,4 use of modifiers (inhibitors or promoters),5,6 and reaction variables such as temperature, pressure, and (where applicable) solvent.7,8 A move from mono- to bimetallic catalyst systems has been demonstrated to have far ranging selectivity consequences.9,10 Supported bimetallic catalyst preparation can involve either the stepwise or simultaneous introduction of both metals to the support.11 The mobility of the two metals on the support during the subsequent temperature programmed activation can determine the ultimate surface composition.12 Catalyst characterization to establish the surface distribution and interaction between the two metals is essential in order to account for the observed catalytic performance. This calls for refined characterization techniques, notably HRTEM, XPS, and EXAFS.13 To date, the application of Au in hydrogenation reactions is still limited, largely due to the low activity when compared with conventional metal catalysts (Pd, Ni, and/or Pt), a result of the lower capacity of Au to adsorb/activate hydrogen.14,15 Never© 2012 American Chemical Society

Received: March 16, 2012 Revised: May 7, 2012 Published: May 7, 2012 11166

dx.doi.org/10.1021/jp3025528 | J. Phys. Chem. C 2012, 116, 11166−11180

The Journal of Physical Chemistry C

Article

10−3 to 9 h. The H2 content was far in excess of the stoichiometric requirements, and the flow rate was monitored using a Humonics (model 520) digital flowmeter; GHSV = 2 × 104 h−1. In a series of blank tests, passage of 1,3-dinitrobenzene (1,3-DNB) or 1,3,5-trinitrobenzene (1,3,5-TNB) in a stream of H2 through the empty reactor or over the support alone, i.e., in the absence of Au and/or Ni, did not result in any detectable conversion. The reactor effluent was frozen in a liquid N2 trap for subsequent analysis, which was made using a Perkin-Elmer Auto System XL gas chromatograph equipped with a programmed split/splitless injector and a flame ionization detector, employing a DB-1 50 m × 0.20 mm i.d., 0.33 μm film thickness capillary column (J&W Scientific).33 Data acquisition and manipulation were performed using the TotalChrom Workstation (version 6.3.2 (Windows)) chromatography data system. The reactant/product molar fractions (xi) were obtained using detailed calibration plots (not shown). The 1,3-DNB and 1,3,5-TNB reactants (Aldrich, ≥ 98%) and solvent (1-butanol, Riedel-de Häen) were used as supplied, without further purification. Fractional conversion, taking 1,3DNB (X1,3‑DNB) as a reactant, was obtained from

Fe based catalysts in acidic media where low selectivities and the generation of toxic waste represent serious sustainability issues.25 There is currently a pressing need for a cleaner alternative. In previous publications,26−30 we have demonstrated high selectivity (but low activity) in the hydrogenation of substituted nitroarenes over supported Au. Moreover, in a recent study,31 we reported differences in product distribution over supported Au and Au−Ni. In this paper, we extend that work and examine the nature of the surface Au−Ni interaction, comparing the catalytic action of Au/Al2O3 and Ni/Al2O3 with Au−Ni/Al2O3, which we correlate with critical catalyst characteristics probed by H2 chemisorption, XRD, HRTEMEDX, XPS, and XANES/EXAFS measurements.

■

METHODS Catalyst Preparation and Activation. The Al2O3 support (Puralox) was obtained from Condea Vista Co. and used as received. Ni/Al2O3 (10 mol %) was prepared by impregnation where a 2-butanolic Ni(NO3)2 solution (1.8 g cm−3) was added dropwise at 353 K to the support with constant agitation (600 rpm) and oven-dried at 393 K for 12 h. The Au−Ni/Al2O3 precursor was prepared by first reducing Ni/Al2O3 in a stream of H2 at 2 K min−1 to 723 ± 1 K, which was maintained for 2.5 h. The gas flow was switched to He, cooled to room temperature, and passivated in 1% v/v O2/He. Passivation served to provide a protective oxide overlayer that prevented bulk oxidation upon exposure to the atmosphere. The passivated sample was contacted with a HAuCl4 solution (Aldrich, 0.0025 g cm−3, pH 2) to deliver a 1/10 Au/Ni mol ratio. The slurry was heated (ca. 2 K min−1) to 353 K and maintained under agitation (600 rpm) with a constant He purge. The solid residue was dried in a flow of He at 383 K for 5 h and stored under He in the dark at 277 K. A 1 mol % Au/ Al2O3 sample was also prepared by impregnation with HAuCl4, post-treatment as above. Prior to use in catalysis, the catalyst precursors (sieved into a batch of 75 μm average particle diameter) were activated in 60 cm3 min−1 H2 at 2 K min−1 to 723 K (Ni/Al2O3) or 603 K (Au/Al2O3 and Au−Ni/Al2O3). The metal contents were measured by inductively coupled plasma-optical emission spectrometry (ICP-OES, Vista-PRO, Varian Inc.) from the diluted extract of aqua regia. A physical mixture of Au/Al2O3 + Ni/Al2O3 (1/10 Au/Ni mol ratio) was also examined. Catalyst Testing. Reactions were carried out under atmospheric pressure, in situ immediately after activation, in a fixed bed vertical continuous flow glass reactor (l = 600 mm, i.d. = 15 mm) over the temperature range 443 K ≤ T ≤ 573 K. The catalytic reactor and operating conditions to ensure negligible heat/mass transport limitations have been fully described elsewhere,32 but some features, pertinent to this study, are given below. A preheating zone (layer of borosilicate glass beads) ensured that the nitroarene reactant was vaporized and reached the reaction temperature before contacting the catalyst. Isothermal conditions (±1 K) were maintained by thoroughly mixing the catalyst with ground glass (75 μm) before insertion into the reactor. The temperature was continuously monitored by a thermocouple inserted in a thermowell within the catalyst bed. The reactants were delivered as butanolic solutions in a cocurrent flow of H2, via a glass/Teflon airtight syringe and a Teflon line, using a microprocessor controlled infusion pump (Model 100 kd Scientific) at a fixed calibrated flow rate; the molar metal (n) to inlet molar −NO2 feed rate ratio (F) spanned the range 5 ×

X1,3 ‐ DNB =

[1, 3‐DNB]in − [1, 3‐DNB]out [1, 3‐DNB]in

(1)

where [1,3-DNB] is the concentration of 1,3-DNB; the subscripts in and out refer to the inlet and outlet streams. Selectivity, taking 1,3-phenylenediamine (S1,3‑PDM) production from 1,3-DNB, is defined by S1,3 ‐ PDM =

[1, 3‐PDM]out [1, 3‐DNB]in − [1, 3‐DNB]out

(2)

Repeated catalytic runs with different samples from the same batch of catalyst delivered product compositions that were reproducible to within ±5%. TPR, H2 Chemisorption, and BET Measurements. BET surface area, temperature programmed reduction (TPR), and H2 chemisorption were determined using the commercial CHEM-BET 3000 (Quantachrome) unit. The samples were loaded into a U-shaped quartz cell (100 mm × 3.76 mm i.d.) and heated in 17 cm3 min−1 5% v/v H2/N2 (Brooks mass flow controlled) at 2 K min−1 to a final temperature of 723 K (Ni/ Al2O3) or 603 K (Au/Al2O3 and Au−Ni/Al2O3). The effluent gas passed through a liquid N2 trap, and changes in H2 consumption were monitored by a thermal conductivity detector (TCD), with data acquisition/manipulation using the TPR Win software. The reduced samples were maintained at the final temperature until the signal returned to baseline, swept with 65 cm3 min−1 N2 for 1.5 h, cooled to room temperature, and subjected to H2 chemisorption using a pulse (10−50 μL) titration procedure. BET areas were recorded with a 30% v/v N2/He flow; pure N2 (99.9%) served as the internal standard. At least two cycles of N2 adsorption−desorption in the flow mode were employed to determine total surface area using the standard single point BET method. Hydrogen chemisorption and BET surface area values were reproducible to within ±4%. UV−vis, XRD, HRTEM-EDX, and XPS Measurements. A high resolution UV−vis spectrum of the aqueous HAuCl4 solution used in catalyst preparation was obtained employing a Perkin-Elmer Lambda 35 UV−vis spectrometer. Powder Xray diffractograms were recorded on a Bruker/Siemens D500 incident X-ray diffractometer using Cu Kα radiation. The 11167

dx.doi.org/10.1021/jp3025528 | J. Phys. Chem. C 2012, 116, 11166−11180

The Journal of Physical Chemistry C

Article

samples were scanned at a rate of 0.02° step−1 over the range 20° ≤ 2θ ≤ 90°. Diffractograms were identified using the JCPDS-ICDD reference standards, i.e., γ-Al2O3 (10-0425), Au (04-0784), and Ni (04-0850). Transmission electron microscopy analysis was conducted using a JEOL JEM 2011 HRTEM unit with a UTW energy dispersive X-ray (EDX) detector (Oxford Instruments) operated at an accelerating voltage of 200 kV and employing a Gatan DigitalMicrograph 3.4 for data acquisition/manipulation. The specimens were prepared by dispersion in acetone and deposited on a holey carbon/Cu grid (300 mesh). Up to 350 individual metal particles were counted for each catalyst, and the surface area-weighted metal diameter (dTEM) was calculated from d TEM =

Atoms,35 FEFF,35 and FEFFIT36 programs. The k- and Rranges were chosen to be 3.0−12.0 Å−1 and 1.4−3.4 Å for the Au LIII-edge data, while the ranges were 3.0−12.0 Å−1 and 1.5− 2.5 Å in the case of the Ni K-edge data. For each XANES analysis, a linear combination of reference spectra was used to fit the data using the WinXAS software.

■

RESULTS AND DISCUSSION Hydrogenation of 1,3-Dinitrobenzene (1,3-DNB). The catalytic properties of the three catalysts were evaluated using the gas phase hydrogenation of 1,3-DNB as a model reaction. The associated hydrogenation pathways are shown in Figure 1,

∑i nidi 3 ∑i nidi 2

(3)

where ni is the number of particles of diameter di. X-ray photoelectron spectroscopy (XPS) analyses were conducted using an Axis Ultra instrument (Kratos Analytical) under ultrahigh vacuum conditions ( 723 K. Moreover, H2 consumption during TPR (to 723 K) exceeded (by a factor of 1.3) the amount required for the reduction of the supported metal precursor. This “overconsumption” can be associated with hydrogen spillover during TPR, i.e., migration of H atoms from the metal to the support following the dissociative chemisorption of H2 on Ni, as has been previously reported for Ni/Al2O3.64 Post-TPR, Ni/Al2O3 was subjected to a room temperature passivation which served to introduce an oxide overlayer on the supported Ni phase. Removal of the passivation layer was far more facile than the reduction of supported NiO,8 as demonstrated by the subsequent TPR step (Figure 6(II)), which is characterized by a single lower temperature reduction peak at 490 K. The passivated sample was then contacted with a solution of chloroauric acid in order to prepare the bimetallic. The UV−vis spectrum recorded for the aqueous AuCl4− solution (Au precursor) is presented in Figure 6(III) where the band at ca. 383 nm can be attributed to Au3+.65 In this redox preparation, i.e., introducing Au to prereduced Ni/Al2O3, Ni with a lower electrochemical potential (ECP = −0.27) acts to reduce the gold precursor (HAuCl4, ECP = 1.00), resulting in a reductive deposition of Au onto Ni.66 The TPR profile for the as prepared Au−Ni/Al2O3 (see Figure 6(IV)) exhibits low intensity H2 consumption with two temperature maxima at 446 and 603 K. The occurrence of two reduction peaks during TPR has been reported elsewhere67 for bimetallic Au−Ni and ascribed to the reduction of gold and nickel oxidic species. Suo et al.22 have reported that the inclusion of Au with Ni (on SiO2) impacted the reduction of NiO, modifying the requisite temperature. A difference in the TPR response for Au containing bimetallic systems relative to the monometallic counterparts has been linked to interaction between both metals.68 Ou et al.69 reported a significant difference in the reduction temperature of Au−Cu/TiO2 (354 K) and Cu/TiO2 (415−425 K), which was ascribed to a weakening of the Cu−O bond and/or an increase in CuO

functionalities) over Au/Al2O3. In contrast, reaction over Ni/ Al2O3 generated 1,3,5-TM as the major product and the product composition reveals a significantly greater degree of nitro-group reduction over Ni. Au−Ni/Al2O3 delivered a selectivity response that was intermediate between the two monometallic systems, producing a combination of partially and fully hydrogenated products. These results supplement the trends established in the case of 1,3-DNB hydrogenation and further confirm the clear differences in terms of catalytic hydrogenation for alumina supported Au, Ni, and Au−Ni that we set out (below) to correlate with catalyst structure. Catalyst Characterization. The BET areas of the activated catalysts (Table 3) were lower than that of the starting Al2O3 support (190 m2 g−1), which can be attributed to a partial pore blockage by the supported metal(s). Table 3. Metal Loading, BET Surface Areas, H2 Chemisorption, Metal Particle Size (Range and Surface Area Weighted Mean (dTEM) Based on TEM Analysis), and XPS Binding Energies Associated with Activated Au/Al2O3, Ni/ Al2O3, and Au−Ni/Al2O3

metal loading (mol %) BET area (m2 g−1) H2 chemisorption (μmol g−1) metal particle size range (nm) dTEM (nm) XPS Ni binding energies (eV) 2p1/2 2p3/2 XPS Au binding energies (eV) 4f5/2 4f7/2 a

Au/ Al2O3a

Ni/ Al2O3b

Au−Ni/Al2O3a

1 161 0.4 1−20 9

10 151 3.6 5−50 31

Au (1), Ni (10) 143 1.1 1−50 20

872.0 854.2

872.3 854.5

87.6 84.0

87.7 84.1

Activated at 603 K. bActivated at 723 K.

Au/Al2O3: TPR/H2 Chemisorption. The TPR profile for the Au/Al 2O3 precursor, presented in Figure 5(I), is characterized by a single H2 consumption peak at 434 K that can be attributed to the reduction of Au3+ to Au0.26 Hydrogen chemisorption, post-TPR, delivered a low uptake (Table 3), confirming the limited capacity of Au to chemisorb H2.15 This result is in line with the literature,57 indicating a high activation energy barrier for H2 adsorption. Hydrogen uptake has been proposed58 to exhibit a dependence on Au coordination number, where dissociative chemisorption is favored on smaller Au particles at edge and corner sites.15,59 Au/Al2O3: XRD/TEM. XRD analysis of activated Au/Al2O3 generated the diffractogram pattern shown in Figure 5(IA). The profile exhibits four peaks over the 2θ range 30−70° that correspond to the (311), (222), (400), and (440) planes of cubic γ-Al2O3 (JCPDS-ICDD 10-0425). In addition to peaks due to the support, reflections at 2θ = 38.1, 44.4, 64.7, and 77.5° are consistent with the (111), (200), (220), and (311) planes of metallic Au (JCPDS-ICDD 04-0784), confirming the presence of zero valent Au post-TPR at 603 K. TEM analysis was performed in order to assess Au size and morphology; the particles exhibit a quasi-spherical shape with diameters ≤20 nm (Figure 5(II)). A representative high magnification micrograph of an isolated Au particle is shown in Figure 5(III), where the intensity profile across the particle yields an average d-spacing of 0.20 nm that matches the (200) plane of metallic Au. The 11171

dx.doi.org/10.1021/jp3025528 | J. Phys. Chem. C 2012, 116, 11166−11180

The Journal of Physical Chemistry C

Article

Figure 5. Au/Al2O3: (I) TPR profile with (IA) XRD pattern. Note: peak assignments based on JCPDS-ICDD reference data for (□) γ-Al2O3 (100425) and (●) Au (04-0784). Representative (II) medium and (III) high magnification TEM images with (IIIA) the intensity profile revealing the distances between the planes of the atomic lattice in the 12 nm segment marked on the isolated Au particle in (III); (IV) Au particle size distribution.

decrease in metal dispersion71 have been proposed for bimetallic catalysts where Au was incorporated as the second metal. Nonetheless, the opposite effect has also been noted72 for Ir−Au/γ-Al2O3 and attributed to hydrogen spillover (from

dispersion promoted by gold. With respect to H2 chemisorption, the addition of Au to Ni/Al2O3 resulted in an appreciably lower uptake (by over a factor of 3, see Table 3), suggesting surface Au−Ni interaction that inhibits H 2 chemisorption. An adsorption site “blocking effect”70 and 11172

dx.doi.org/10.1021/jp3025528 | J. Phys. Chem. C 2012, 116, 11166−11180

The Journal of Physical Chemistry C

Article

Figure 6. (I) TPR profiles generated for the reduction of the Ni/Al2O3 precursor; (IA) XRD pattern for Ni/Al2O3; (IB) representative TEM image of Ni/Al2O3; (IC) diffractogram pattern and (ID) inverse fast Fourier transfrom (IFFT) for the black framed section in the isolated Ni particle. Note: the “faceted” morphology of isolated Ni particle is represented by the white framed configuration; (II) TPR profile generated for passivated Ni/Al2O3; (III) UV−vis spectrum of HAuCl4 aqueous solution; (IV) TPR profile generated for Au−Ni/Al2O3; (IVA) XRD pattern for Au−Ni/ Al2O3; (IVB) representative TEM image of Au−Ni/Al2O3. Note: XRD peak assignments based on JCPDS-ICDD reference data for (□) γ-Al2O3 (10-0425), (●) Au (04-0784), and (■) Ni (04-0850).

Ir) onto adjacent Au sites and/or electronic modifications induced by Au, which influenced H2 adsorption. Ni/Al2O3 and Au−Ni/Al2O3: XRD/TEM. The XRD pattern of Ni/Al2O3 post-TPR (Figure 6(IA)) shows signals characteristic of γ-Al2O3 with an extra peak at 2θ = 44.5° that represents the (111) plane of metallic Ni (JCPDS-ICDD 04-0850). The XRD profile for Au−Ni/Al2O3 (see Figure 6(IVA)) presents, in addition, only characteristic peaks for metallic Au. This XRD response suggests the presence of two discrete metallic phases

in Au−Ni/Al2O3. The TEM image of an isolated Ni particle in Ni/Al2O3 is shown in Figure 6(IB) and provides evidence of particle faceting (see white frame that serves to illustrate Ni particle shape). Similar metal particle morphologies have been reported elsewhere for oxide supported Ni73 and taken to be a feature of metal/support interactions. The associated diffractogram pattern is included as an inset in Figure 6(IC) with a dspacing of 0.20 nm (Ni(111) plane). In order to analyze the distribution of Ni atoms in more detail, an IFFT (inverse fast 11173

dx.doi.org/10.1021/jp3025528 | J. Phys. Chem. C 2012, 116, 11166−11180

The Journal of Physical Chemistry C

Article

Figure 7. (A) Representative TEM image and (B) metal particle size distributions associated with (I) Ni/Al2O3 (solid bars) and (II) Au−Ni/Al2O3 (hatched bars).

Ni/Al2O3 and Au−Ni/Al2O3: STEM-EDX/XPS. Surface composition and metal distribution in Au−Ni/Al2O3 were assessed by STEM/EDX elemental mapping. A representative measurement is shown in Figure 8(I), which includes the bright field image of the sample area (see white frame) that was mapped with the associated distribution of (a) Al, (b) O, (c) Ni, and (d) Au. The average surface Ni/Au ratio (=10) obtained from repeated EDX measurements of multiple areas (4 × 105 nm2) matches the bulk composition. EDX mapping revealed a close proximity of Au and Ni on the Al2O3 carrier. XPS analysis was conducted to probe possible surface metal interactions, and the spectra over the Ni 2p (Ni 2p1/2 and Ni 2p3/2) and Au 4f (Au 4f5/2 and 4f7/2) binding energy (BE) regions are presented in Figure 8(II) and (III), respectively. The XPS profile for Ni/Al2O3 exhibits a BE maximum at 854.2 eV (with a satellite peak at higher BE) for Ni 2p3/2 and 872.0 eV for Ni 2p1/2. The reference BE values for zero valent Ni are 852.7 and 869.9 eV,76 but it has been noted in the literature77 that Ni−Al2O3 interactions can serve to shift the BE to higher values and the displacement in the BE of Ni 2p is consistent with the formation of electron deficient Ni particles via electron transfer to Al2O3.78 Moreover, overlapping of Ni 3d and O 2p orbitals can lead to the formation of a covalent polar bond

Fourier transform) was applied. The transformed image in Figure 6(ID), corresponding to the black framed section in the isolated particle in Figure 6(IB), shows the arrangement of Ni atoms in the crystal structure. A representative TEM image of Ni/Al2O3 post-TPR is given in Figure 7(IA), and the Ni particle size histogram presented in Figure 7(IB) exhibits a bimodal distribution with a surface-area-weighted mean Ni size of 31 nm. The medium (Figure 7(IIA)) and higher magnification (Figure 6(IVB)) TEM images for Au−Ni/Al2O3 show metal particles in the size range 1−60 nm. The overall particle size distribution is presented in Figure 7(IIB), where it can be seen that there is a preponderance of metal particles