Alice H. M.
Leung
a,
Andrés
García-Trenco
a,
Andreas
Phanopoulos
a,
Anna
Regoutz
b,
Manfred E.
Schuster
c,
Sebastian D.
Pike
a,
Milo S. P.
Shaffer
*de and
Charlotte K.
Williams
*a
aChemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford, OX1 3TA, UK. E-mail: charlotte.williams@chem.ox.ac.uk
bDepartment of Chemistry, University College London, 20 Gordon Street, London WC1H 0AJ, UK
cJohnson Matthey Technology Centre, Blount's Court, Sonning Common, RG4 9NH, UK
dDepartment of Chemistry, Imperial College London, London, SW7 2AZ, UK. E-mail: m.shaffer@imperial.ac.uk
eDepartment of Materials, Imperial College London, London, SW7 2AZ, UK
First published on 27th May 2020
Doped-ZnO nanoparticles, capped with dioctylphosphinate ligands, are synthesised by the controlled hydrolysis of a mixture of organometallic precursors. Substitutional doping of the wurtzite ZnO nanoparticles with 5 mol% Mg(II), Al(III) and Cu(I) is achieved by the addition of sub-stoichiometric amounts of the appropriate dopant [(n-butyl)(sec-butyl)magnesium, triethylaluminium or mesitylcopper] to diethylzinc in the precursor mixture. After hydrolysis, the resulting colloidal nanoparticles (sizes of 2–3 nm) are characterised by powder X-ray crystallography, transmission electron microscopy, inductively-coupled plasma optical emission spectrometry and X-ray photoelectron spectroscopy. A solution of the doped-ZnO nanoparticles and colloidal Cu(0) nanoparticles [M:ZnO : Cu = 1 : 1] are applied as catalysts for the hydrogenation of CO2 to methanol in a liquid-phase continuous flow stirred tank reactor [210 °C, 50 bar, CO2:H2 = 1:3, 150 mL min−1, mesitylene, 20 h]. All the catalyst systems display higher rates of methanol production and better stability than a benchmark heterogeneous catalyst, Cu–ZnO–Al2O3 [480 μmol mmolmetal−1 h−1], with approximately twice the activity for the Al(III)-doped nanocatalyst. Despite outperforming the benchmark catalyst, Mg(II) doping is detrimental towards methanol production in comparison to undoped ZnO. X-Ray photoelectron spectroscopy and transmission electron microscopy analysis of the most active post-catalysis samples implicate the migration of Al(III) to the catalyst surface, and this surface enrichment is proposed to facilitate stabilisation of the catalytic ZnO/Cu interfaces.
Beyond semiconducting applications, ZnO is also one component of the industrial heterogeneous three-component catalyst Cu–ZnO–Al2O3, which is used to produce methanol on a 50 million-tonne scale in conjunction with a syn-gas feedstock.21,22 Methanol is a clean burning liquid fuel that can be blended with petrol providing an attractive ‘drop-in’ substitute with existing fuels, and moreover can be generated from renewable raw materials (i.e. CO2/H2).23–26 Doping the ZnO component can enhance the catalytic performance using either syn-gas or CO2/H2 gas mixtures.27–29 The range of dopants includes alkaline metals (e.g. Cs(I) or Mg(II)),30 transition (e.g. Zr(II) or Mn(II)/(III))27,31 or main group metals (e.g. Al(III) or Ga(III)).32,33 These dopants are proposed to increase activity by acting as electronic and/or structural promoters, i.e. by altering the electronic structure through addition of energy levels to reduce activation energy, or by modifying the catalyst nanostructure to increase the number of active sites, respectively.28,34 In particular, there is an enhancement in activity when n-type dopants, such as Al(III) and Ga(III) are incorporated into Cu/ZnO catalysts.28,29 It is proposed that dopants facilitate formation of Cu/ZnO interfaces by labilising the ZnOx (x < 1) component.35 These interfaces, often observed at step-defects in the Cu phase,36 are the putative catalytic active sites.29,37 Conversely, by lowering the density of donor states, e.g. by doping Cu/ZnO with Mg(II), an increase in optical band-gap and reduction potential were observed.28 The lower efficiency for methanol production was attributed to the converse effects limiting the formation of critical Cu/ZnOx interfaces.28,29
Most methanol production occurs using syn-gas in fixed bed reactor configurations but slurry phase methods, where the catalyst is suspended in solution, are attractive alternatives. This slurry process can improve reaction kinetics and may increase the intrinsic activity. As long as high enough pressures and dilute conditions are implemented, the mass transport effects can be managed, though for productivity reasons, reactors may, in fact, be operated near the mass transport limit.38 Additionally, liquid phase synthesis may mitigate local hot spots, accelerate catalyst activation and facilitate continuous product removal. These benefits have led to the development of a commercial liquid-phase methanol synthesis process (LPMeOH™).39 One method to optimize liquid phase processes is to develop colloidal (quasi-homogeneous) catalysts. Our group,8,9,40–43 and others,44–46 have previously outlined some of the advantages of such species including easy control over catalyst composition (Cu:Zn ratio), access to small, well-defined, high surface area nanoparticles, and control over surface chemistry. These colloidal catalysts have been applied either to the synthesis of methanol from syn-gas21,47 or by carbon dioxide hydrogenation (CO2/H2).9,40,41 They can show high activity and selectivity over the reverse water–gas shift (rWGS) reaction.28
To make these colloidal catalysts, we have developed a synthesis using the hydrolysis of organozinc reagents. The method generates small (2–4 nm) and monodispersed ZnO nanoparticles, which are soluble in a range of organic solvents.48–50 The hydrolysis of organozinc compounds (e.g. ZnPh2, ZnCy2, ZnEt2), in the presence of an aqueous stable, anionic coordinating ligand (e.g. carboxylate or phosphinate), readily generates soluble ZnO nanoparticles,48,50–57 with a high density of surface defects, proposed to be beneficial for catalysis.9,44,58 Hydrolytic syntheses provide straightforward control of ligand coverage and kinetic control over the growth of particles, allowing access to small ZnO nanoparticles (2–4 nm).4,50,54,57 The methodology is potentially attractive for dopant incorporation since a second organometallic reagent can simply be added to the one-pot hydrolysis reaction. The synthesis of doped-ZnO nanoparticles via co-precipitation methods has allowed doping levels >10% (e.g. 18% Mg-,14 13% Al-29 or 12% Cu-doping59) with a range of metals (e.g. Group 1,15,17 Group 2,14 Group 13,28,29 and Group 1116,59−61); however, there are only a few examples of successful doping using organometallic routes.4,62,63 Chaudret and co-workers demonstrated that the addition of Li-amides during the hydrolysis of ZnCy2, in the presence of octylamine, led to the formation of spherical Li-doped-ZnO nanoparticles (2.4–4.3 nm, Li content 1–10 mol%). The nanoparticle growth was controlled by the localisation of LiOH on the surface of ZnO.62 Alternatively, our group prepared Mg-doped ZnO nanoparticles by mixing ZnEt2 with MgBu2 before hydrolysis, in the presence of dioctylphosphinic acid (DOPA–H, (C8H17)2PO2H). The resultant particles showed up to 10% Mg incorporation into the wurtzite ZnO structure.63
This work aims to investigate the preparation of other doped ZnO materials, e.g. with Al(III), Mg(II) and Cu(I). These colloidal nanoparticles will be tested as components in liquid phase catalysts for the reduction of CO2 to methanol.
Elemental analyses were performed by Mr Stephen Boyer at London Metropolitan University and inductively-coupled plasma optical emission spectrometry (ICP-OES) was performed by Mr Alan Dickerson at the University of Cambridge.
Powder X-ray diffraction (XRD) experiments were performed using a PANalytical X'pert PRO diffractometer in the theta/theta reflection mode, fitted with a nickel filter, 0.04 rad Soller slit, 10 mm mask, 1/4° fixed divergence slit, and 1/2° fixed antiscatter slit. An operating voltage of 40 kV was used with a step size of 0.033° 2θ, scan step time of 70 s and scan range of 10–80° 2θ. Air sensitive samples were prepared in a glovebox and sealed with polyimide tape. The diffraction patterns were analysed using Fityk (version 0.9.0; Marcin Wojdyr, 2010). The peaks were fitted to a SplitPearson7 function, and the crystallite size was calculated from the full-width half-maximum of the fitted curve, using the Scherrer equation, to the most intense and not overlapped reflections; Cu: 43.5° (hkl = 111), ZnO: 47.5° (hkl = 102), 57.2° (hkl = 110). For the detailed XRD study of the (101) ZnO nanoparticle peak, the diffraction scans were performed using the same diffractometer with a step size of 0.0167° 2θ, scan step time of 120 s and scan range of 32–38° 2θ. The expansion/contraction of the ZnO lattice was determined, using Bragg's equation, from the position of the (101) peak which was determined by Gaussian fit of the diffraction peak. UV-Vis absorption spectra of ZnO nanoparticles were recorded using a PerkinElmer Lambda 950 spectrometer, in transmission mode, from 300–500 nm with a scan speed of 1 nm s−1. Fourier transform infrared (FT-IR) spectra were recorded using a PerkinElmer Spectrum 100 spectrometer fitted with an ATR attachment and each spectrum was collected with 32 scans, at a resolution of 4 cm−1. Thermogravimetric analysis (TGA) was carried out using a Mettler Toledo TGA/DSC 1, under a flow of dry N2 at 60 mL min−1, from 100–700 °C, at a heating rate of 5 °C min−1.
Bright field (BF) TEM and HAADF-STEM images were acquired with Johnson Matthey's probe corrected ARM200F, at the ePSIC facility (Diamond Light Source), with an acceleration voltage of 200 keV. Measurement conditions were a CL aperture of 30 μm, convergence semiangle of 24.3 mrad, beam current of 12 pA, and scattering angles of 0–10 and 35–110 mrad, for BF and HAADF-STEM respectively.
XPS was performed on a Thermo Scientific Kα+ X-ray photoelectron spectrometer system operating at 2 × 10−9 mbar base pressure. This system incorporates a monochromated, microfocused Al Kα X-ray source (hν = 1486.6 eV) and a 180° double focusing hemispherical analyser with a 2D detector. The X-ray source was operated at 6 mA emission current and 12 kV anode bias and a flood gun was used to minimize sample charging. Samples were mounted using conductive carbon tape and transferred to the spectrometer, using a special glovebox module, which ensured that samples were transferred without exposure to air. Data were collected at 20 eV pass energy for core levels and Auger lines. Data were collected over six measurement positions using an X-ray spot size of 400 μm to avoid beam damage affecting the results. Data were then averaged across the different measurement positions. All data were analysed using the Avantage software package.
The commercial CuO–ZnO–Al2O3 catalyst was obtained from Alfa Aesar (45776, pellet size 5.4 × 3.6 mm, mass composition CuO, 63.5%; ZnO, 25.1%; Al2O3, 10.1%; MgO, 1.3%). It was ground to a fine powder and tested as a reference material. The choice of catalyst loading was dictated by the normalisation considerations, making sure that the total molar concentration of Cu and Zn was the same as in the Cu/ZnO colloids. The catalyst is pre-activated using a diluted H2 stream (5 vol% H2/N2) at 4.5 bar and 240 °C (ramp 2 °C min−1) for 3 h, according to a standard activation protocol for the ternary Cu–ZnO–Al2O3 methanol synthesis catalyst in a slurry reactor.65 The catalytic runs were performed under the conditions stated above.
The reaction products and unreacted material were monitored by an online gas chromatograph (Bruker 450 GC), equipped with a thermal conductivity detector (TCD) and a flame ionisation detector (FID). TCD was used to quantify CO, CO2 and Ar, whilst FID was used to quantify methanol and other oxygenates or hydrocarbons. The lines from the reactor to the GC were heated to 180 °C to avoid the condensation of any products. All experiments were conducted under differential conditions, with an overall CO2 conversion <2%, avoiding mass transport limitations (see ESI† for details). Selectivity values are given on a carbon basis, where the only two products were methanol and CO.
In each case, a pre-hydrolysis mixture was obtained by reacting 0.2 equiv. of DOPA–H with commercially available diethylzinc (0.95 equiv.) and 0.05 equiv. of commercially available (n-butyl)(sec-butyl)magnesium, triethylaluminium or mesitylcopper (CuMes) in toluene. These precursors were selected not only due to their availability, but also high solubility in organic solvents and rapid reactivity with water to deter phase separation between zinc and dopant. The overall metal:ligand ratio was maintained at 5:1 as these conditions yielded particles fully saturated with ligand. Upon reaction with water, the organometallic reagents produce only inert hydrocarbons as byproducts (i.e. ethane, butane and/or mesitylene, respectively) which are easily removed from the final product in vacuo. The post-hydrolysis mixture, consisting of a colloidal suspension of ZnO nanoparticles, was purified by reducing the solvent volume and via centrifugation, yielding Mg:ZnO@DOPA and Al:ZnO@DOPA as white powders and Cu:ZnO@DOPA as a dark green powder.
Analysis of the as-synthesised particles by X-ray diffraction (XRD) showed the formation of small crystalline particles (2–3 nm) with sizing determined via peak analysis using the Scherrer equation (Table 1). In each case, only the wurtzite ZnO phase was observed (Fig. 1a–d), suggesting incorporation of any dopants into the crystalline structure rather than co-crystallisation of other phases. High-angle angular dark-field scanning tunnelling electron microscopy (HAADF-STEM) images (Fig. 1e–g) revealed spherical nanoparticles with narrow size distributions (2.6–2.8 nm, Fig. S1†) and no obvious amorphous phases were observed by electron microscopy.
M:ZnO samples | Size (nm) | (101) reflectionc | |||
---|---|---|---|---|---|
XRDa | TEMb | 2θ° | d Å−1 | Δd Å−1 | |
a Average particle size determined by application of the Scherrer equation to XRD (102) and (110) reflections (Fig. 1a–d). b Average particle size ± standard error (standard deviation) determined by size analysis of TEM images (Fig. 1e–f). The standard error of the mean is defined as standard deviation/(no. of measurements). c The positions of (101) peaks were determined using Gaussian fits to the XRD reflection in the range of 30–35° 2θ (Fig. S2). | |||||
ZnO@DOPA | 2.3 | 2.8 ± 0.02 (0.6) | 34.62 | 2.59 | — |
Mg:ZnO@DOPA | 2.0 | 2.7 ± 0.04 (0.5) | 34.31 | 2.61 | +0.02 |
Al:ZnO@DOPA | 2.3 | 2.7 ± 0.06 (0.9) | 35.75 | 2.51 | −0.08 |
Cu:ZnO@DOPA | 2.2 | 2.6 ± 0.03 (0.4) | 34.82 | 2.58 | −0.01 |
Incorporation of the second metal was assessed further, by considering changes to the lattice parameters, evidenced in the strongest (101) reflection (Table 1, Fig. S2†). A significant contraction was observed in Al:ZnO@DOPA, indicative of successful substitution of tetrahedral Zn(II) sites (rion[Zn(II)] = 0.74 Å) with Al(III) (rion[Al(III)] = 0.53 Å).76,77 No clear lattice changes were observed for Mg:ZnO@DOPA and Cu:ZnO@DOPA due to the similar ionic radii of the metals (rion[Mg(II)] = 0.71 Å and rion[Cu(I)] = 0.74 Å) and the peak breadth resulting from the small particle sizes.14,78
Bulk compositional analysis by ICP-OES confirmed and quantified the presence of the dopants. For Al:ZnO@DOPA and Cu:ZnO@DOPA, the heteroatom concentration of 5 mol% is in line with starting reagent stoichiometry, but the value for Mg:ZnO@DOPA is slightly lower than expected (Table 2). The solid-state27 Al-MAS-NMR spectrum of Al:ZnO@DOPA (Fig. S3†) confirms the replacement of tetrahedral Zn(II) with Al(III), indicated by the sharp signal at δ 55 ppm which corresponds to tetrahedral Al(III) environments. In addition, pentahedral (δAl: 47 ppm) and octahedral (δAl: −17 and 16 ppm) Al environments are also present and are tentatively assigned as surface sites.28,66,67
M:ZnO@DOPA | Dopant quantity (rel. at%) | Optical band-gap (eV) | |
---|---|---|---|
ICP-OES | XPS | UV-Visc | |
a Derived from comparison of the peak area from peak fit analysis of Zn 2p3/2 with Cu 2p3/2 and Mg 1s (Fig. 2). Error of ±0.5%. b Derived from comparison of the peak area from peak fit analysis of Zn 3s with Al 2p (Fig. 2). Error of ±0.5%. c Error of ±0.04 eV, derived from uncertainty of linear fits (coefficient of determination, R2 > 0.99) (Fig. S4).28 | |||
ZnO@DOPA | — | — | 3.71 |
Mg:ZnO@DOPA | 3.77 ± 0.05 | 4.5a | 3.84 |
Al:ZnO@DOPA | 4.96 ± 0.07 | 5.0b | 3.65 |
Cu:ZnO@DOPA | 5.05 ± 0.07 | 3.3a | 3.67 |
The chemical environments and oxidation states for all samples were analysed using XPS (Fig. 2). All samples show a binding energy (BE) for Zn 2p3/2 at 1021.7 eV, however from the Zn core level alone, it is not possible to determine its oxidation state, as the BEs of Zn metal and ZnO are identical.79 However, the Zn L3M4,5M4,5 Auger lines clearly show the typical structure of ZnO, with two features at 987.9 eV and 991.1 eV kinetic energy (KE). The P 2p and C 1s lines show BEs expected for the DOPA ligand. The O 1s core level has two contributions, one from ZnO at 530.4 eV and one at 531.5 eV from DOPA.80 The Mg 1s and Al 2p core level appear at BEs typical for Mg(II) and Al(III), 1304.2 eV and 73.9 eV, respectively. The Cu 2p3/2 core level is at a BE of 932.3 eV, which can either stem from Cu(0) or Cu(I).81 Usually, the Cu LMM Auger lines can be used to distinguish between these two oxidation states, but in these samples, the lines overlap with the much stronger Zn LMM lines, preventing further analysis. The core level also exhibits a small shoulder towards higher BE, at 933.7 eV, indicating a small amount of Cu(II). From peak fit analysis of the Zn 2p3/2 and the metal core levels, relative atomic ratios were determined (Table 2). Comparing the dopant quantities obtained by XPS and ICP-OES highlights the likely surface speciation of Mg, compared to Al which appears to be distributed throughout the particles.
The optical band-gaps were determined via Tauc plots from the on-set absorption in solid-state UV-Vis spectra (Table 2, Fig. S4†). The Al-doped sample showed a smaller band-gap than ZnO, presumably due to the introduction of Al(III) additional donor levels.11,82–84 In contrast, Mg(II) doping widens the band-gap, as observed by others.14,28,78,85–87 A small decrease in the band-gap of Cu:ZnO@DOPA (3.67 eV) suggests partial incorporation of Cu(I)/Cu(II) into the lattice, where the dopant provides additional shallow acceptor levels,88,89 as a larger decrease would be expected for 5% doping.90 It is possible that some of the CuMes is also reduced to Cu(0), rather than doping the ZnO. It was previously reported that CuMes reacts with diethylzinc to produce Cu(0) nanoparticles.41 In this work, however, there was no direct evidence for such small Cu(0) nanoparticles by TEM or XRD.
Analysis of all the new nanoparticles by TGA showed a mass loss of 25–30% between 150–500 °C for all M:ZnO@DOPA, assigned to the thermal decomposition of the [DOPA]− ligand (Table S2, Fig. S5†). Along with elemental analysis (Table S2†), ratios close to M:ZnO : [DOPA]− = 6 : 1 were found for M:ZnO@DOPA. The ratio differs slightly to the starting reagent stoichiometry and it is proposed that low quantities of soluble molecular clusters e.g. [Zn4O(DOPA)6], are removed during nanoparticle purification and account for the lower ligand loading.57
From IR spectroscopy (Fig. S6†), the presence of C–H stretches (2800–3000 cm−1), along with phosphinate, PO2−, stretches (∼1047 cm−1, asymm; ∼1126 cm−1, symm) associated with the ligand were detected. The separation of 79–81 cm−1 between the asymmetric and symmetric phosphinate stretches indicates κ2-ligand coordination (see Fig. S6† inset). The presence of O–H stretches centred at 3380 cm−1, is indicative of hydroxyl and water coordinated to the nanoparticle surface. Overall, all characterisation data strongly suggest the successful incorporation of Mg(II) and Al(III) into the lattice of ZnO nanoparticles, whilst Cu(I) appears to have been partially incorporated.
Fig. 3 Overview of M:ZnO@DOPA syntheses and the subsequent combination with Cu@DOPA for the formation of colloidal Cu/ZnO nanocatalysts in hydrogenation of CO2 to methanol. |
All catalysts reached peak methanol production rate after 3–4 h time-on-stream and remained stable for over 15 h (Fig. S7 and S8†). High methanol selectivity was observed in all cases and CO was the only by-product (from the rWGS reaction). Both doped and undoped colloidal catalysts displayed higher activities than the commercial heterogeneous catalyst (Fig. 4). Mg-doping was somewhat detrimental to activity in comparison with undoped ZnO@DOPA; Cu/Mg:ZnO@DOPA displays 30–40% lower activity than the other three colloidal catalysts. Cu/Al:ZnO@DOPA showed a small but noticeable improvement when compared to Cu/ZnO@DOPA (+10%), but no clear difference was observed for the Cu-doped sample.
All post-catalysis colloids were characterised using air-sensitive techniques to prevent the oxidation of Cu(0). By XRD, all samples displayed typical diffraction peaks attributed to crystalline ZnO and metallic Cu(0), with no additional crystalline phases observed (Fig. S9–S11†). Some ripening of the original M:ZnO@DOPA and Cu(0) nanoparticles was indicated by XRD and TEM (Table 3).
Catalyst | ZnO (nm) | Cuf (nm) | ||||
---|---|---|---|---|---|---|
Pre-catalysis | Post-catalysis | Post-catalysis | ||||
XRDa | TEMb | XRDa | TEMb | XRDa | TEMb | |
a Average particle size determined by application of the Scherrer equation to XRD (102) and (110) reflections (Fig. 1a–d and S9–S11). b Average particle size ± standard error (standard deviation) determined by size analysis of TEM images (Fig. 1e–f and S14, S15). The standard error of the mean is defined as standard deviation/(no. of measurements). c The sizing measurement is taken from previously published system.40 d Measurement taken from 40 h time-on-stream study. e Crystallite size of catalyst after 20 h time-on-stream. f For pre-catalysis Cu@DOPA, sizes of ∼2 nm were reported using TEM.64 | ||||||
Cu/ZnO@DOPAc | 2.3 | 2.8 ± 0.02 (0.6) | 3.2 | 4.9 ± 0.1 (1.1) | 6.4 | 7.2 ± 0.2 (1.9) |
Cu/Mg:ZnO@DOPA | 2.0 | 2.7 ± 0.04 (0.5) | 2.3 | 3.5 ± 0.04 (0.4) | 4.4 | 4.3 ± 0.06 (0.6) |
Cu/Al:ZnO@DOPAd | 2.3 | 2.7 ± 0.06 (0.9) | 3.6 (3.1)e | 4.5 ± 0.07 (1.3) | 3.0 (2.4)e | 4.6 ± 0.02 (1.1) |
Cu/Cu:ZnO@DOPA | 2.2 | 2.6 ± 0.03 (0.4) | 3.3 | — | 2.0 | — |
The difference in activity between the highest (Al-doped) and lowest (Mg-doped) samples is tentatively attributed to the quantity of Cu/ZnO interfaces. These interfaces, formed under the reductive catalytic conditions, are commonly suggested to be the active site for methanol synthesis. They can be identified using electron microscopy, through analysis of the degree of contact between different identifiable crystalline phases.36,37,46,68,69 To this end, the nanoscale structures of post-catalysis samples of Cu/Mg:ZnO@DOPA and Cu/Al:ZnO@DOPA were investigated using BF-TEM, and were found to differ significantly.
Cu/Al:ZnO@DOPA showed contact between Cu and Al:ZnO nanoparticles (Fig. 5a & S13†). The nanocrystalline particles were additionally surrounded by an amorphous network that is rich in Al(III) (identified by EDX) and likely to be alumina as suggested by the emergence of a second, higher BE species (76.8 eV) in the Al 2p core line by XPS (Fig. S12 & S13†). The formation of surface alumina under reducing conditions has also been found in the heterogeneous Cu–ZnO–Al2O3 catalyst. In this case it appears the Al(III) that remains within the lattice acts as an electronic promoter, while the in situ formed alumina matrix acts as a structural promoter.28,29,67 It is proposed that Al-doping of ZnO increases the density of donor states (n-type) and contributes to the formation of oxygen vacancies which accelerates migration of Zn onto Cu(0) and hence increases the formation of Cu/ZnO interfaces.71–73
More isolated nanoparticles and fewer Cu/ZnO interfaces were observed for Cu/Mg:ZnO@DOPA (Fig. 5b & S14†). Additionally, the Mg-rich surface of Mg:ZnO@DOPA (as suggested by XPS) might hinder ZnO and Cu(0) interface formation. EDX of the post-catalysis sample shows that some Mg is distributed in between Cu(0) and ZnO nanoparticles (Fig. S14†). It appears Mg(II) is detrimental to activity as both a structural and electronic dopant. This is in line with a previous study of Mg(II) doping into the heterogeneous Cu–ZnO–Al2O3 methanol catalysts from syn-gas (230 °C, 30 bar, CO:CO2:H2: inert gas = 6:8:59:27).28
Fig. 6 Methanol rates of Cu/Al:ZnO@DOPA and commercial heterogeneous Cu–ZnO–Al2O3, over 40 h time-on-stream. |
The doped ZnO nanoparticles were combined with Cu(0) nanoparticles (<2 nm), capped by phosphinate, to form a series of active colloidal nanocatalysts for the hydrogenation of CO2 to methanol in the liquid phase (210 °C, 50 bar, H2:CO2 molar ratio of 3:1, 150 mL min−1). All catalysts showed methanol production rates higher than a benchmark heterogeneous catalyst Cu–ZnO–Al2O3, although the Mg-doped catalyst showed ∼40% lower activity than the other colloidal catalysts. It is proposed that the Mg rich surface layer prevents formation of catalytically important Cu/ZnO interfaces. In contrast, Al-doping improved activity compared to ZnO nanoparticles. An amorphous surface alumina network is indicated in these post-catalysis samples and may hinder nanoparticle ripening and improve catalyst stability over a 40 h time-on-stream.
The organometallic synthesis of doped colloidal ZnO nanoparticles is a useful means to tune properties and appears to be somewhat generalizable. Future studies should target mixed metal oxides and dopant levels. In particular, transition metal and Group 13 n-type ZnO dopants should be explored, as these materials show promising performances in syn-gas liquid phase methanol synthesis.26
Footnote |
† Electronic supplementary information (ESI) available: Tables outlining amount of dopant precursor used and mass analysis by EA and TGA. Additional figures showing nanocatalyst characterisation, methanol production rates and selectivity as well as post-catalysts characterisation. Rationale for catalytic reaction conditions used. See DOI: 10.1039/d0ta00509f |
This journal is © The Royal Society of Chemistry 2020 |