Fernando
Cárdenas-Lizana
and
Mark A.
Keane
*
Chemical Engineering, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh, EH14 4AS, Scotland. E-mail: M.A.Keane@hw.ac.uk; Tel: +44 (0)1314514719
First published on 2nd March 2015
The chemoselective continuous gas phase (T = 573 K; P = 1 atm) hydrogenation of nitroarenes (p-chloronitrobenzene (p-CNB) and m-dinitrobenzene (m-DNB)) has been investigated over a series of oxide (Al2O3 and TiO2) supported Au and Ni–Au (1:10 mol ratio; 0.1–1 mol% Au) catalysts. Monometallic supported Au with mean particle size 3–9 nm promoted exclusive formation of p-chloroaniline (p-CAN) and m-nitroaniline (m-NAN). Selective hydrogenation rate was higher over smaller Au particles and can be attributed to increased surface hydrogen (from TPD measurements) at higher metal dispersion. (S)TEM analysis has confirmed an equivalent metal particle size for the supported bimetallics at the same Au loading where TPR indicates Ni–Au interaction and EDX surface mapping established Ni in close proximity to Au on isolated nanoparticles with a composition (Au/Ni) close to the bulk value (= 10). Increased spillover hydrogen due to the incorporation of Ni in the bimetallics resulted in elevated –NO2 group reduction rate. Full selectivity to p-CAN was maintained over all the bimetallic catalysts. Conversion of m-DNB over the lower loaded Ni–Au/Al2O3 generated m-NAN as sole product. An increase in Ni content (0.01 → 0.1 mol%) or a switch from Al2O3 to TiO2 as support resulted in full –NO2 reduction (to m-phenylenediamine). Our results demonstrate the viability of Ni-promotion of Au in the continuous production of functionalised anilines.
The hydrogenation response over bimetallic catalysts is determined by the structure and physico-chemical properties of the supported nanoparticles, which are dependent on the distribution of both metals within the crystal nanostructure. In order to achieve any degree of catalytic synergy, the two metals must be in close proximity (if not in direct contact) on the support. The size, metal ratio and nature of the support are critical variables. Bimetallic particles at the nano-scale (∼3 nm) can form solid solutions (homogeneous alloy particles) regardless of the miscibility gap between the two metallic elements.7 The incorporation of a second metal with Au (X–Au, where X = Pd,8 Pt9) has been shown to increase hydrogenation activity but at low Au/X ratios the selectivity response is affected or even governed by the second metal.8,9 The redox character and acid–base properties of the (oxide) carrier can act to stabilize small Au nanoparticles as a result of electron transfer across the metal–support interface and induce geometric and electronic modifications that impact on catalysis.5
Prior research has focused on binary metals that form (ordered or random) bulk alloys with no miscibility gap in the corresponding bulk phase diagram. This is the case with Pd–Au as the most widely studied Au-containing bimetallic.10 Interest has shifted to systems such as Ni–Au that do not mix in the bulk but can form stable alloys in the outermost surface layers.11,12 The incorporation of Au onto partially oxidised Ni resulted in partial oxidation of Au atoms and the formation of Aun–O–Nim ensembles (from XPS),13 which were suggested as active sites in the isomerisation of methylstyrenes over Ni–Au/SiO2.14 Nishikawa et al.,15 studying the promotional effect of Au on Ni, demonstrated (by XRD, TEM, XAFS and 197Au Mössbauer) the formation of an Ni–Au alloy, following reduction (to 673 K in H2) of Au/NiO co-precipitates. These were active in the hydrogenolysis of benzylic alcohols with superior catalytic activity compared with RANEY® Ni. Chin et al.16 demonstrated by EXAFS/XANES analysis of Ni–Au/MgAl2O4 (prepared by reductive deposition of Au on Ni/MgAl2O4) Au → Ni electron transfer with surface alloy formation that served to inhibit carbon deposition during n-butane steam reforming. The addition of Au to Ni (1:4) resulted in the formation of nano-crystals with a Au-core and Ni-enriched shell structure (based on EXAFS, XPS and UV-vis) with increased hydrogenolysis activity in the conversion of 2-phenoxy-1-phenylethanol.17
We have previously examined the catalytic action of supported Ni–Au with Ni/Au mol/mol = 10 in promoting gas phase nitroarene hydrogenation.18,19 That work was directed at the continuous selective production of industrially important functionalised amines under mild reactions conditions. Supported Ni–Au delivered higher hydrogenation activities but was less selective than supported Au. We have refocused our attention and have considered the possible role of Ni (introduced as the minor component; Ni/Au mol/mol = 0.1) to enhance the chemoselective hydrogenation action of Au to deliver enhanced rates to the target amine product. We assess the feasibility of controlling hydrogenation rate by varying Au particle size and the nature of the oxide support.
Temperature programmed reduction (TPR), H2 chemisorption and temperature programmed desorption (TPD) were determined using the CHEM-BET 3000 (Quantachrome) unit. The samples were loaded into a U-shaped Quartz cell (3.76 mm i.d.) and heated in 17 cm3 min−1 5% v/v H2/N2 (Brooks mass flow controlled) to 603 ± 1 K (Au/Al2O3, Au/TiO2, Ni–Au/Al2O3 and Ni–Au/TiO2) or 723 ± 1 K (Ni/Al2O3 and Ni/TiO2) at 2 K min−1. The effluent gas passed through a liquid N2 trap and H2 consumption was 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 in H2 until return to baseline. After the reduction step, the samples were cooled in a He flow and subjected to H2 chemisorption (at ambient temperature) using a pulse (10 μl) titration procedure, as described elsewhere.23 TPD was conducted in a N2 flow (65 cm3 min−1) at 50 K min−1 to 873 K with an isothermal hold until the signal returned to the baseline. The support alone was subjected to an equivalent TPR and subsequent TPD, which was used to correct for H2 desorption from Al2O3.24
Metal particle morphology, size distribution and surface composition were determined by (scanning) transmission electron microscopy using a JEOL JEM-2100F unit operating at an accelerating voltage of 200 kV with resolution to 0.14 nm and an EDAX Genesis XM 4 system 60. Samples were dispersed in 1-butanol by ultrasonic vibration, deposited on a lacey-carbon/Cu grid (200 Mesh) and dried at 383 K. Up to 300 individual metal particles were counted for each catalyst and the surface area-weighted metal diameter (d) was calculated from:
(1) |
(2) |
(3) |
Catalysta | SSAb (m2 g−1) | Pore volume × 10−3c (cm3 g−1) | TPR Tmax (K) | TPR H2d,e (μmol gcatalyst−1) | H2 TPD (mmol molAu−1) | d (nm) |
---|---|---|---|---|---|---|
a Au metal content (mol%): LL = 0.1, HL = 1.0, 1:10 Ni:Au molar ratio in bi-metallics. b Specific surface area (SSA), Al2O3 = 191 m2 g−1, TiO2 = 52 m2 g−1. c Pore volume, Al2O3 = 450 × 10−3 cm3 g−1, TiO2 = 120 × 10−3 cm3 g−1. d H2 required for the reduction of the metal precursor. e Experimentally determined H2 consumption. f Surface area weighted mean metal particle size from TEM/STEM analysis. | ||||||
Au/Al2O3-HL | 161 | 427 | 434 | 102d/104e | 473 | 9 |
Ni–Au/Al2O3-HL | 189 | 444 | 470 | 151d/145e | 3118 | 9 |
Au/Al2O3-LL | 169 | 425 | 434 | 9d/16e | 1101 | 4 |
Ni–Au/Al2O3-LL | 190 | 447 | 470 | 9d/8e | 2471 | 4 |
Au/TiO2-LL | 49 | 117 | 383 | 11d/15e | 1233 | 3 |
Ni–Au/TiO2-LL | 47 | 115 | 414 | 12d/15e | 1765 | 3 |
The TPR profiles generated for Au/Al2O3-HL (I), Au/Al2O3-LL (II) and Au/TiO2-LL (III) can be compared in Fig. 1. The three samples present a principal H2 consumption peak with an associated temperature maximum (Tmax) over 383–434 K, which is within the range recorded elsewhere (328–465 K) for the activation of Au/Al2O3 and Au/TiO2 and attributed to precursor reduction.28 The lower Tmax recorded for Au/TiO2-LL (relative to Au/Al2O3) is consistent with the literature.29 Moreover, Delannoy et al.30 have demonstrated (by in situ XAFS and DRIFTS) greater reducibility of cationic gold on TiO2 compared with Al2O3. Hydrogen consumption during catalyst activation was close to that required for Au3+ reduction (Table 1). Supported bimetallic catalyst preparation involved reductive deposition where Ni with a lower electrochemical potential (ECP = −0.27) acts to reduce the Au precursor (HAuCl4, ECP = 1.00).31 The TPR profiles generated for Al2O3 and TiO2 supported Ni (not shown) exhibited a single broad H2 consumption peak at the final isothermal hold (723 K) that can be associated with a combined decomposition-reduction of the precursor to metallic nickel, as noted elsewhere.1 A single stage reduction was observed for the supported Ni–Au systems at a higher Tmax (414–470 K) than that recorded for the corresponding monometallic Au catalysts. Modification to the TPR response for Au containing bimetallic catalysts with respect to monometallic counterparts has been reported and linked to interaction between both metals.32 Pu et al.33 have recently described a stabilisation of Au species (on activated carbon spheres) with the introduction of Ni by co-impregnation that inhibited reduction, displacing the temperature for AuIII → Au0 (from 583 K to 588 K) and AuI → Au0 (from 619 K to 659 K) to higher values.
Fig. 1 TPR profiles generated for: (I) Au/Al2O3-HL, (II) Au/Al2O3-LL, (III) Au/TiO2-LL, (IV) (1:10) Ni–Au/Al2O3-HL, (V) (1:10) Ni–Au/Al2O3-LL and (VI) (1:10) Ni–Au/TiO2-LL. |
Gold particle size, from TEM analysis (Fig. 2 and 3 and Table 1), is dependent on metal loading and the nature of the support. The particle size in Au/Al2O3-HL was in the range 1–20 nm (Fig. 3(I)) with a surface area weighted mean diameter of 9 nm. The lower Au loaded sample (Au/Al2O3-LL) exhibited a narrower size distribution (1–8 nm, Fig. 3(II)) and smaller mean (4 nm). This effect is consistent with literature that has shown wider size range and the formation of larger nanoparticles with increasing Au content.34 This can be linked to mobility of the chloride precursor, resulting in Au agglomeration during thermal treatment.35 Enhanced Au dispersion on TiO2 (Fig. 3(III)) is a consequence of a difference in metal/support interaction relative to Au/Al2O3.36 Loss of lattice oxygen with the formation of surface vacancies is a feature of thermal treatment in H2 for reducible oxides such as TiO2 at T ≥ 573 K.37 These surface defects act as Au nucleation sites with electronic interactions that limit particle growth, leading to enhanced dispersion.38,39
Fig. 3 Gold particle size distributions associated with (I) Au/Al2O3-HL (open bars), (II) Au/Al2O3-LL (solid bars) and (III) Au/TiO2-LL (hatched bars). |
Representative (S)TEM images of Ni–Au/Al2O3-HL (I), Ni–Au/Al2O3-LL (III) and Ni–Au/TiO2-LL (IV) can be compared in Fig. 4. The metal size range in Ni–Au/Al2O3-HL (1–20 nm) coincided with Au/Al2O3-HL and there was no obvious alteration due the inclusion of Ni. This response extended to Ni–Au/Al2O3-LL and Ni–Au/TiO2-LL with an equivalent mean metal size relative to the monometallic system (see Table 1). This can be attributed to the low Ni content in the bimetallic samples. An unchanged metal particle size has been reported previously for TiO2 supported Au and Pt–Au with low Pt content (0.01–0.03 wt%).9 EDX mapping (see representative analysis for Ni–Au/Al2O3-HL in Fig. 4(II)) demonstrated that individual particles contained both Au and Ni in proportions close to the bulk value (Au/Ni = 10). EDX analysis cannot establish unequivocally the exact nature of Ni–Au interaction or rule out the occurrence of bimetallic clusters40 or surface alloy.41 Both possibilities have been suggested for Ni–Au catalysts prepared in an analogous manner and with a similar metal dispersion to that in this study.16,42,43 Deghedi and co-workers42 proposed the formation of Ni particles covered with Au adatoms (on the basis of EXAFS and TEM-EDX measurements) for Ni–Au/SiO2 synthesised by redox deposition that contained metal ensembles of 5.1 nm. Maniecki and co-workers44 reported an alloy phase (on the basis of XRD) for Ni–Au/Al2O3 prepared by co-impregnation (with Ni(NO3)2 and HAuCl4) and reduction in H2. Surface alloy formation was suggested by EXAFS/XANES for Au–Ni/MgAlO4 (10.1 nm) prepared by redox deposition.16 Molenbroek and Nørskov43 also demonstrated the occurrence of a surface alloy in Ni–Au/SiO2 (2.5–6.0 nm) and Ni–Au/MgAl2O4 (3.0–15.0 nm) by Monte Carlo simulations, which was experimentally verified by EXAFS, TEM and in situ XRD.
The hydrogenation of nitroarenes is enhanced by contributions due to spillover hydrogen.19,36 The generation of spillover species involves dissociative adsorption of H2 on a (donor) supported metal site with migration of the atomic hydrogen generated to the (acceptor) support.45 Hydrogen TPD is a practical approach that can serve to quantify the degree of spillover. Ambient temperature H2 chemisorption following TPR on all the catalysts was low (≤15 mmol molAu−1) and close to the detection limits. Low uptake on supported Au has been demonstrated elsewhere and attributed to the filled d-band and high activation energy barrier for dissociative adsorption.6,46 There was no measurable difference in ambient temperature H2 uptake on the bimetallics, which may be due to the low Ni content. The H2 TPD profiles are presented in Fig. 5. In all cases, the profiles show H2 release with maxima at T ≥ 770 K where the total amount desorbed (per molAu, Table 1) was appreciably greater (by two orders of magnitude) than that taken up in the chemisorption step. This response suggests the release of spillover hydrogen generated during catalyst activation by TPR.45 The shift in the TPD peak for TiO2 systems to a lower temperature (by up to a 100 K) is indicative of a less energetically demanding desorption of H2. This is consistent with literature47,48 that has reported a down-shift (by ca. 60–70 K) in the H2-TPD peak for Ni/TiO2 and Pd–Au/Al2O3-TiO2 compared with Ni/Al2O3 and Pd–Au/Al2O3, respectively. Taking the three monometallic Au catalysts presented in Fig. 5, H2 desorbed (Table 1) increased in the order: Au/Al2O3-HL (I) < Au/Al2O3-LL (II) < Au/TiO2-LL (III) where the values obtained are close (54–932 mmol molAu−1) to those reported elsewhere for Au with similar metal size (3–5 nm) supported on Fe2O3.36 This sequence reflects that of decreasing mean Au nanoparticle size (Table 1). There is evidence in the literature23 that metal size is a crucial variable controlling spillover where a greater (donor) surface area for smaller metal nanoparticles extends the donor/acceptor interface and facilitates spillover transfer. The profiles generated for the bimetallic Ni–Au catalysts (Fig. 5, profiles IV–VI) reveals a significant increase in H2 desorbed relative to Au samples with the same loading (see Table 1). This suggests a surface Ni/Au synergism that impacts on H2 adsorption/desorption dynamics. The existing literature on Au containing bimetallics for hydrogenation applications has been from the perspective of Au as a diluent with low H2 uptake capacity and a resultant decrease in surface hydrogen associated with the bimetallic.49 As a result, there is no directly comparable H2 TPD analysis for supported bimetallic Au-containing systems where the second metal was present in small amounts. We should flag the work of Wojcieszak et al.50 who observed an increase in H2 released (0.22 → 0.38 mol g−1) during TPD of Ni–Ag/SiO2 with increasing Ni content (0.50 → 0.75 wt%). It is important to note that Ni–Au/Al2O3-HL with lower metal dispersion showed significantly greater H2 desorption (Table 1) relative to Ni–Au/Al2O3-LL. STEM-EDX analyses have demonstrated the presence of both metals on individual nanoparticles. The formation of bimetallic nano-crystals (i.e. Au decorating Ni nanoparticles) has been demonstrated (EXAFS and TEM-EDS) for Ni–Au bimetallics prepared and activated using a similar methodology.42 It is known that H2 interaction is distinct for monometallic, bimetallic and alloy nanoparticles.51,52 The greater Ni loading in Ni–Au/Al2O3-HL (0.1 mol% vs. 0.01 mol%) must facilitate H2 dissociative adsorption and spillover during TPR, which is reflected in greater H2 release during TPD. In the case of the low loading bimetallics, H2 desorption from Ni–Au/Al2O3-LL exceeded Ni–Au/TiO2-LL (Table 1). This can be tentatively linked to modifications induced by the support that impact on H2 uptake on the bimetallic phase. Likewise, Gu et al.47 observed a greater H2 TPD from Pd–Au/Al2O3 relative to Pd–Au/Al2O3-TiO2. The characterisation analysis has established the formation of nano-scale (mean = 3–9 nm) Au particles in the monometallics with greater surface hydrogen for catalysts with increasing metal dispersion. TPR measurements suggest interaction of Ni with Au in the bimetallics. STEM-EDX analysis has demonstrated the presence of Au and Ni in individual nanoparticles and the TPD results are consistent with increased spillover resulting from Au/Ni synergy that increases the available surface hydrogen.
Fig. 5 Hydrogen TPD profiles generated for: (I) Au/Al2O3-HL, (II) Au/Al2O3-LL, (III) Au/TiO2-LL, (IV) (1:10) Ni–Au/Al2O3-HL, (V) (1:10) Ni–Au/Al2O3-LL and (VI) (1:10) Ni–Au/TiO2-LL. |
The applicability of pseudo-first order kinetics for the gas phase hydrogenation of p-CNB1,60 and m-DNB36 has been established elsewhere. The extracted specific (per molAu) pseudo-first order rate constants (k) are given in Table 2. The promoting effect of Ni is demonstrated by the significantly increased hydrogenation rate when compared with the monometallic Au catalysts with the same loading. Under the same reaction conditions, Ni/Al2O3 and Ni/TiO2 delivered negligible activity. Duan and co-workers61 using unsupported nano-crystals reported increased rate (molproduct molmetal−1) in the high pressure (40 atm) liquid phase hydrogenation of phenol (283 → 363) and benzene (164 → 261) over Ni–Rh relative to Rh, which was attributed to the bimetallic character of the nanoparticles (based on HRTEM-EDX and XRD) and possible modifications in active site electron density. Lu et al.62,63 recorded higher activity in the liquid phase hydrogenation of a series of p-substituted nitro-derivates over Ni–Pd colloids, ascribed to the formation of a Ni enriched alloy (from XRD, EXAFS, XPS). The elevated rate observed in this study can be linked to surface available reactive hydrogen as shown in Fig. 6 where H2-TPD is related to the rate constant. The increase in surface hydrogen that results from incorporation of Ni with Au was accompanied by a higher hydrogenation rate. Moreover, reaction of m-DNB uniformly generated greater rates, a result that can be attributed to the activating effect of the second –NO2 group, which is consistent with a nucleophilic mechanism, as demonstrated previously.64
Catalyst | p-CNB | m-DNB | ||
---|---|---|---|---|
k | Product(s) (Sproduct (%)) | k | Product(s) (Sproduct (%)) | |
a mol–NO2 molAu−1 h−1. | ||||
Au/Al2O3-HL | 4 | p-CAN (100) | 6 | m-NAN (100) |
Ni–Au/Al2O3-HL | 24 | p-CAN (100) | 40 | m-NAN (60) |
m-PDM (40) | ||||
Au/Al2O3-LL | 12 | p-CAN (100) | 16 | m-NAN (100) |
Ni–Au/Al2O3-LL | 21 | p-CAN (100) | 35 | m-NAN (100) |
Au/TiO2-LL | 18 | p-CAN (100) | 22 | m-NAN (100) |
Ni–Au/TiO2-LL | 22 | p-CAN (100) | 25 | m-NAN (80) |
m-PDM (20) |
Reaction selectivity is compared at the same degree of conversion in Table 2. p-CNB was converted solely to the target (p-chloroaniline) p-CAN over all the catalysts, which is consistent with reaction exclusivity in continuous gas phase p-CNB → p-CAN over oxide supported Au.49 In contrast, nitrobenzene,1 aniline1 and benzene65 have been observed in the gas phase hydrogenation of CNB over Ru,65 and Pd1 catalysts. Variations in product distribution are, however, evident in the hydrogenation of m-DNB and sensitive to the nature of the catalyst. Full selectivity to partially reduced (m-nitroaniline) m-NAN was obtained over monometallic Au systems. In batch liquid phase operation, high pressures (26–34 atm) have been deemed essential to achieve high selectivity (84–98%) to m-NAN.48,66 Preferential –NO2 activation on oxide supported Au in the presence of other reactive (CC, carbonyl, amide and ester) functional groups has been demonstrated by FTIR analysis.67 The incorporation of Ni at low loading (Ni–Au/Al2O3-LL) served to increase rate while retaining 100% selective to m-NAN. Deghedi et al.42 reported that the combination of Au and Ni on SiO2 (Ni:Au molar ratio = 4:1) lowered the rate of CC bond and aromatic ring hydrogenation relative to Ni in the conversion of styrene, which was accounted for in terms of geometric and electronic effects. At higher loadings (Ni–Au/Al2O3-HL) complete hydrogenation to m-PDM was observed, which is a characteristic of catalysis by Ni.18,19 This may be a consequence of increased surface reactive hydrogen which facilitates reduction of both –NO2 substituents. Alternatively, m-DNB activation on Ni–Au/Al2O3-HL may be distinct from Ni–Au/Al2O3-LL with a planar adsorption on the former through the aromatic ring, resulting in the formation of a resonance structure with two positive localised charges where both –NO2 groups are activated for nucleophilic attack. It is known that substituted aromatics interact with supported Ni catalysts via the π electron-delocalised aromatic ring.68 Lonergan et al.69 have observed n-butane selectivity effects in 1,3-butadiene hydrogenation over Ni–Pt/Al2O3 with increasing Ni content (1:3 → 1:10). The formation of m-PDM was also promoted over Ni–Au/TiO2-LL, which suggests a support effect in terms of m-DNB adsorption/activation that acts in tandem with differences in available reactive hydrogen. Wang et al.70 reported a modified catalytic response in the hydrogenation of 1,3-budatiene for Pt–Ni on Al2O3 and TiO2. DFT calculations revealed stronger metal interaction with Al2O3 (binding energy ca. −50 kcal mol−1) than TiO2 (−17 kcal mol−1) with a preferential Ni–Al2O3 interaction at the metal–support interface leading to a Pt-terminated bimetallic configuration. A weaker Ni interaction in the case of the TiO2 carrier can then result in some segregation of the Ni component to the nanoparticle surface.
This journal is © the Owner Societies 2015 |