Maciej G.
Walerowski
a,
Matthew E.
Potter
bcd,
Elizabeth S.
Burke
a,
Stylianos
Kyrimis
e,
Lindsay-Marie
Armstrong
e and
Robert
Raja
*a
aSchool of Chemistry, University of Southampton, Southampton, SO17 1BJ, UK. E-mail: R.Raja@soton.ac.uk
bChemistry Department, University College London, London, WC1E 6BT, UK
cUK Catalysis Hub, Harwell Research Complex, Harwell, OX11 0FA, UK
dDepartment of Chemistry, University of Bath, Bath, BA2 7AY, UK
eSchool of Engineering, University of Southampton, Southampton, SO17 1BJ, UK
First published on 2nd April 2024
Meeting ambitious net-zero targets will require the replacement of marine fossil fuels with sustainable alternatives such as dimethyl ether (DME). DME is non-toxic, can be fully produced via a circular carbon economy and can be rapidly deployed due to its compatibility with existing liquid petroleum gas infrastructure. One-pot production of DME from CO2via a methanol intermediate is achieved by combining redox and Brønsted or Lewis acid sites. Herein, we have synthesised, characterised and tested a variety of bifunctional CuZnO/silicoaluminophosphate catalysts for the one-pot production of DME. A range of synthetic approaches were employed to combine the redox and acidic functionalities in order to derive synthesis–structure–property correlations to guide the design of improved catalysts. We found that a CuZnO/SAPO-34 catalyst made via impregnation and drying can achieve 80% DME selectivity with no detectable toxic CO by-product formation. High acid site abundance resulted in extensive dehydration of the intermediate methanol, which increased localised water production, suppressing the CO-forming reverse water gas shift reaction and thus yielding exceptional DME selectivity that is amongst the highest in literature.
2CO2 + 6H2 ↔ CH3OCH3 + 3H2O (ΔH298K = −123 kJ mol−1) | (1) |
CO + 2H2 ↔ CH3OH (ΔH298K = −90.7 kJ mol−1) | (2) |
2CH3OH ↔ CH3OCH3 + H2O (ΔH298K = −23.4 kJ mol−1) | (3) |
In a one-pot approach, both the MeOH synthesis and subsequent MeOH dehydration reactions occur in one reactor using either a mixture of two distinct catalysts, or with a singular bifunctional catalyst which contains both redox and acid sites.5,7 The one-pot approach has a number of benefits over the traditional two-pot method. Firstly, there are large financial and engineering advantages as a singular reactor is required and the intermediate MeOH purification and separation steps are removed.9 The initial MeOH synthesis reaction itself is challenging as usually it involves two distinct reactions: CO2 hydrogenation to MeOH (eqn (4))13 and the reverse water gas shift (RWGS) reaction (eqn (5))7,12 to form CO with their equilibrium determining the final MeOH product yield.
CO2 + 3H2 ↔ CH3OH + H2O (ΔH298K = −49.5 kJ mol−1) | (4) |
CO2 + H2 ↔ CO + H2O (ΔH298K = 41.2 kJ mol−1) | (5) |
As discussed previously, SAPO based bifunctional catalyst could give high DME selectivity due to their weaker strength acid sites. Yoo et al. have created a series of physical mixture (PM) catalysts by mixing Cu/ZnO/Al2O3 catalyst with different SAPO solid acid catalysts for the one-pot conversion of syngas to DME.28 They found that solid acids with stronger acid sites such as SAPO-18 and SAPO-34 had higher initial activity for conversion of syngas to DME but these deactivate rapidly due to the formation of pore plugging coke whereas the contrary was true for SAPO-5 and SAPO-11 which had weaker acid sites. Navarro-Jaén et al. have also developed a series of PM catalysts by mixing Cu–ZnO–MxOy (where M = Al, Zr or Ce) with SAPO-34, but the catalyst was used for the conversion of CO2, rather than syngas, to DME.29 Although the Cu–ZnO–Al2O3/SAPO-34 PM catalyst had high CO2 conversion of 19.8%, its CO selectivity was very high (70.9%) emphasising the difficulty in using PM catalysts for the one-pot production of DME. Yao et al. investigated whether this high CO selectivity could be suppressed by combining SAPO-34 and Cu–In–Zr–O catalysts in different integration manners to give catalyst beds with varying proximity between the two functionalities. The best catalyst achieved 4.3% CO2 conversion and 65.1% DME selectivity at 250 °C and 30 bars.26 It was shown that more intimate contact between the two catalysts (e.g. powder vs. pellet/granule mixing) gave higher CO2 conversions and DME selectivity, which suggested that bringing redox and acid active sites closer might suppress the RWGS reaction. Core–shell catalysts such as Cu–ZnO–Al2O3@SiO2–Al2O3 have been developed as a single bifunctional catalyst for the one-pot conversion of CO2 to DME.30 By enveloping the Cu-based catalyst in a mesoporous acidic shell, it was possible to retain the active Cu0–ZnO interface which increased MeOH + DME selectivity (from 9.1 to 63.3 mol%) and significantly tempered, although not fully eliminated, the formation of CO. This suggested that controlling active site proximity can be used to optimise product selectivity, something which not possible with a PM system or in a two-pot approach. Bahruji et al.27 and Tariq et al.31 have used chemical vapour impregnation (CVI) to deposit PdZn and CuZnO nanoparticles onto the zeolite ZSM-5, respectively, to create integrated bifunctional catalysts for conversion of CO2 to DME. Through electron microscopy, they found that by using CVI, it is possible to decorate the outer crystal faces of the solid acid with uniformly distributed metallic nanoparticles, thus achieving high proximity between redox and acid active sites. Here a low DME selectivity was achieved and this was attributed to blocking of the Brønsted acid sites by the metallic nanoparticles. Bahruji et al. did however find that subsequent dehydration of MeOH can suppress the parasitic RWGS reaction and decrease the selectivity for CO during one-pot conversion of CO2 to DME.27 Based on these works, a highly active and selective bifunctional catalyst for the one-pot conversion of CO2 to DME requires a weakly acidic functionality, proximity between the redox and acid active sites and a high dispersion and anchorage of the redox functionality.
A diverse array of catalytic candidates have been synthesised using various methods for the one-pot synthesis of DME from either syngas or CO2. A number of works explored the impact of the synthesis method or precursor identity on the structure and performance of bifunctional catalysts.5,32–35 However, these studies have focused on the one-pot conversion of syngas to DME rather than CO2 and H2 to DME and mostly utilised commercially available zeolite ZSM-5 or γ-Al2O3 as the acidic functionality which, as already discussed, may not be the most suitable acidic functionality. Herein, we have developed a series of bifunctional CuZnO/SAPO-11 and CuZnO/SAPO-34 catalysts using different preparation methods to develop holistic synthesis–structure–property correlations (SSPCs). SAPO-11 and SAPO-34 were chosen as they have contrasting acid strengths, textural properties, morphologies and topologies. A range of wet chemistry methods were used to combine CuZnO with SAPO-11 or SAPO-34 to understand the interplay between synthesis, structure and performance. We have demonstrated that by tailoring the synthesis method, it is possible to obtain synergy between Cu and ZnO sites as well as interaction between these species and the solid acid support. Through careful choice of synthesis approach we were able to deposit the CuZnO nanoparticles directly onto the SAPO support surface which resulted in instantaneous dehydration of MeOH giving CO2 conversions of 8%, high DME selectivity of 80% and no detectable CO due to significant suppression of the parasitic RWGS reaction.
Characterisation protocols are outlined in section 3 of the ESI.†
Synthesis method | Cu loading (wt%) | Zn loading (wt%) | Cu/Zn mass ratio | |||
---|---|---|---|---|---|---|
−11 | −34 | −11 | −34 | −11 | −34 | |
ID | 13.1 | 12.5 | 5.6 | 5.6 | 2.3 | 2.2 |
OG | 13.1 | 14.8 | 5.5 | 5.0 | 2.4 | 3.0 |
DP | 10.7 | 9.3 | 5.9 | 4.3 | 1.8 | 2.2 |
Intended | 15.4 | 7.7 | 2 |
Powder X-ray diffraction (PXRD) confirmed the phase purity of SAPO-11 and SAPO-34 solid acid catalysts as only the AEL and CHA crystalline phases were observed, respectively (Fig. 2).15,39,40 Impregnation of Cu and Zn onto the SAPO supports gave new peaks at 43 and 51° (111 and 200 planes of metallic Cu)41,42 and at 32, 34.5, 36.5, 48 and 56.5° (100, 002, 101, 102 and 110 planes of wurtzite ZnO),43,44 respectively (Fig. S5†). The ZnO peak intensity was noticeably weaker than the Cu, suggesting small ZnO particles were present, in agreement with Tariq et al.31
![]() | ||
Fig. 2 PXRD patterns of a) CuZnO/SAPO-11 and b) CuZnO/SAPO-34 bifunctional catalysts prepared via different methods showing the appearance of Cu and ZnO reflections. |
Total surface area of the undoped SAPOs was within the expected range (Table S2†).15,36,37,45 SAPO-11 showed a type IV isotherm with a H4 hysteresis loop indicating micro-mesoporosity (Fig. S6a†) due to interparticle voids caused by an aggregation of SAPO-11 particles rather than due to ordered mesopores within the SAPO-11 framework itself.46 Lack of significant peaks in the pore size distribution plot has indeed verified that SAPO-11 contains no ordered mesopores. Type I isotherm typical of microporous frameworks with small external surface areas was obtained for SAPO-34 (Fig. S6b†).46 The overall surface area decreases upon introduction of Cu and Zn (Table S2†), suggesting either the porous SAPO is being diluted by the dense Cu and Zn species, or these species are blocking access to the internal framework of the SAPO support. The synthesised bifunctional catalysts have similar metal loadings, but the decrease in surface area varies significantly between the different synthetic approaches and the solid acid, which suggests that the preparation method and the support itself can impact the size, morphology and distribution of the CuZnO species. The synthetic approach also leads to a decrease in the surface area as is seen for the undoped SAPO-11 ID and SAPO-34 ID catalysts which have undergone the impregnation and drying approach without the deposition of metal (Table S3†). This is due to a combined effect of the subsequent hydrothermal, calcination and reduction steps which have a mild detrimental effect on the crystallinity of the solid acid.
Scanning electron microscopy (SEM) images for the different CuZnO/SAPO systems (Fig. S7†) show spherical SAPO-11 and cubic SAPO-34 crystals with an average size of 10 and 2 μm, respectively. Here the heavier Cu and Zn species appear brighter compared to the SAPO support. As can be seen in Fig. S7,† the synthetic approach has a profound impact on the resulting shape, size, and location of the CuZnO nanoparticles whereas the choice of SAPO support had limited effect. The ID method results in a uniform distribution of irregular nanoparticles across outer crystal faces (Fig. S7a and d†) while the OG approach leads to two distinct types of metallic species: nanoparticles which adhere directly to crystal surfaces and larger particles which appears loosely associated with the crystals (Fig. S7b and e†). The DP method results in coating of the crystals with flocculent metallic species (Fig. S7c and f†).
Transmission electron microscopy (TEM) was performed on both CuZnO/SAPO-11 and CuZnO/SAPO-34 catalysts but only CuZnO/SAPO-34 catalysts will be discussed as the irregular SAPO-11 crystals made it challenging to accurately discern nanoparticles and the support. TEM imaging (Fig. S8†) has verified our SEM findings (Fig. S7†) that the synthesis method plays a key role in the morphology and distribution of the CuZnO nanoparticles. The ID method gives rise to nanoparticles which adhere to the outer crystal faces of SAPO-34 (Fig. S8a†). Previously using SEM for the CuZnO/SAPO-34 OG catalyst, large (∼100 nm) CuZnO particles were observed associated with the SAPO-34 and SAPO-11 crystals (Fig. S7b and e†). Using high resolution TEM, it was found that these large particles are in actual fact agglomerates of smaller nanoparticles. OG therefore gives a mixture of nanoparticles which reside on the SAPO-34 crystals as well as agglomerates of nanoparticles that are loosely associated with the SAPO support (Fig. S8b†). The DP method forms large agglomerates of nanoparticles which are loosely affiliated with the cubic SAPO-34 crystals (Fig. S8c†). In terms of nanoparticle size, the OG, ID and DP methods gave an average size of 13, 15 and 20 nm respectively with all methods giving a narrow nanoparticle size distribution (Fig. 3 and S9†).
![]() | ||
Fig. 3 Nanoparticle size distribution of CuZnO/SAPO-34 catalysts prepared via different methods. Results are based on a measurement of 300 nanoparticles. |
Ammonia temperature programmed desorption (NH3-TPD) probed the influence of the synthetic method on acid strength and quantity (Fig. S10 and Table S4†). As expected, the bare SAPO-34 system shows, on average, stronger acid sites, and a greater quantity of acid sites than the SAPO-11. The SAPO-34 appears to have multiple acid sites, as shown by a shoulder peak at 300 °C, and a main peak at 450 °C. In contrast SAPO-11 shows a single feature, centred at 320 °C. The choice of synthetic method has a pronounced effect on acid site strength and quantity of the CuZnO/SAPO-X catalysts, and this is independent of the SAPO itself (Fig. S10 and Table S4†). The ID approach creates a range of acid sites and leads to a significant increase in the total quantity of acid sites which is highly desirable for the acid catalysed MeOH dehydration reaction. Baek et al. have also reported that the choice of copper acetate precursors used in the ID method leads to a higher number of acid sites in a bifunctional Cu–ZnO/γ-Al2O3 catalysts.33 Navarro-Jaén et al. have also reported the appearance of a strongly acidic peak in NH3-TPD following introduction of CuZnO to SAPO-34 to make a bifunctional catalyst.29 The OG method creates new weakly acidic sites at the behest of stronger acid sites but the total amount of acid sites remains constant for CuZnO/SAPO-34 OG and increases for CuZnO/SAPO-11 OG. The DP method reduces both the strength and quantity of acid sites, and this is attributed to the degradation of the solid acid framework by the alkaline precipitant which has also been noted by Wang et al.47
Cu and Zn K edge X-ray absorption spectroscopy (XAS) spectra for CuZnO/SAPO-11 and CuZnO/SAPO-34 ID, OG and DP catalysts are presented in Fig. S11 and S12 of the ESI.† Linear combination fitting of the X-ray absorption near edge structure (XANES) region (Fig. S11†) has indicated that the Cu in CuZnO/SAPO-11 and CuZnO/SAPO-34 catalysts prepared via the ID and DP methods exists mostly as Cu0 with some being oxidised to Cu+/Cu2+ (Table S5†), whereas the OG catalysts existed in a more oxidised form highlighting once more the impact of synthesis choice. SAPO-34 based catalysts have a marginally lower Cu oxidation states. Based on in situ XAS studies of various Cu-based MeOH synthesis catalysts,48–51 it is highly likely that all the Cu sites are reduced to Cu0 during the pre-reaction in situ reduction in pure H2 at 300 °C. The Zn oxidation state for all the catalysts was determined as +2 and neither the synthetic method nor the solid acid appeared to have any impact on the Zn centre (Fig. S12 and Table S5†).
Extended X-ray absorption fine structure (EXAFS) fitting has shown that the Cu sites in catalysts prepared via ID and DP methods are mostly Cu0, whereas those prepared via OG method are a mixture of Cu0 and Cu2O/CuO (Table 3 and Fig. S13†). The OG catalysts are more sensitive to oxidation and this is especially evident for the CuZnO/SAPO-11 OG catalyst in which the Cu exists mostly as Cu2O/CuO rather than Cu0. These catalysts have Cu–Cu coordination number significantly lower than 12 which would be expected for bulk Cu and this phenomenon has also been witnessed by Sun et al. for ultrafine Cu/ZnO/Al2O3 catalysts prepared via the OG method.52 This is due to oxidation of Cu centres as well as the possibility that some Cu exist in the amorphous state with poorly defined Cu–Cu coordination. The Zn exists solely as ZnO and there are only minor differences between the Zn–O coordination number with no clear trend witnessed (Table S6 and Fig. S14†). The synthetic method nor the solid acid have any appreciable impact on the Zn site. Wavelet transformation of the EXAFS region was performed to resolve any possible contributions of backscattering atoms which could overlap in the R space due to similar bond lengths. No additional backscattering atoms were identified and only hotspots consistent with the anticipated Cu–O, Cu–Cu, Zn–O and Zn–Zn were observed (Fig. S15–S18†). Wavelet analysis has shown no sign of alloying between Cu and Zn.
Synthesis method | Scattering path | N | 2σ2 (Å2) | R (Å) | E f (eV) | R factor | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
−11 | −34 | −11 | −34 | −11 | −34 | −11 | −34 | −11 | −34 | ||
Amplitude reduction factor of 0.886 was determined from fitting of Cu0 with fixed CN of 12. A fit employing only the Cu–O path was attempted for the CuZnO/SAPO-11 OG catalyst but this provided an inferior fit compared to fitting both Cu–Cu and Cu–O paths. Values in brackets indicate error (±) in the last reported significant figure. | |||||||||||
ID | Cu–Cu | 6.2(3) | 6.7(2) | 0.0085(4) | 0.0091(3) | 2.543(9) | 2.542(5) | 5(1) | 4(1) | 0.009 | 0.003 |
Cu–O | 1.4(2) | 1.0(1) | 0.006(2) | 0.006(2) | 1.89(2) | 1.87(1) | |||||
OG | Cu–Cu | 0.4(2) | 4.5(2) | 0.004(4) | 0.0089(4) | 2.55(3) | 2.549(8) | 10(2) | 5(1) | 0.035 | 0.007 |
Cu–O | 3.0(3) | 1.4(1) | 0.005(1) | 0.007(2) | 1.94(2) | 1.87(1) | |||||
DP | Cu–Cu | 6.2(2) | 6.2(2) | 0.0095(3) | 0.0091(3) | 2.543(5) | 2.545(6) | 4(1) | 4(1) | 0.003 | 0.004 |
Cu–O | 1.0(1) | 1.1(1) | 0.005(2) | 0.006(2) | 1.87(1) | 1.86(1) |
X-ray photoelectron spectroscopy (XPS) spectra of all the bifunctional catalysts showed shake up satellites (peaks 3–5) in the Cu 2p3/2 region (Fig. S19†). These are indicative of ground state d9 Cu2+ species.53,54 Fitting of the main Cu 2p3/2 peak has shown that the peak is caused by Cu2+ (933.5 eV, peak 2) and Cu+/Cu0 species (932.4 eV, peak 1). Due to spectral overlap of the Cu 2p3/2 peak, it is not possible to discern whether peak 1 is caused by Cu+ or Cu0 species but through examination of the Cu L3M4,5M4,5 spectrum it is usually possible to resolve the two.53,54 Unfortunately, for all catalysts other than the DP catalysts, the Cu L3M4,5M4,5 spectrum showed no discernible peaks (Fig. S20†) and as such it is not possible to conclusively verify whether peak 1 is caused by Cu+ or Cu0 species. Based on the XPS and XAFS results it can thus be stated that the Cu exists as a mixture of Cu0 and Cu2O/CuO. In line with previous reports,50,55 it is possible that the Cu exists as a metallic core – oxidised shell structure. This explains why the surface-sensitive XPS technique predicts higher Cu oxidation compared to the bulk-sensitive XAFS technique. The CuZnO/SAPO-11 catalysts showed on average higher peak area percentages of peaks 2–5 compared to peak 1 (Fig. S21 and Table S7†) which would indicate that the CuZnO/SAPO-11 catalysts suffered from greater Cu oxidation (Fig. 4) which is consistent with the XANES analysis (Table S5†). Examination of the Zn 2p region provided limited information due to a statistically insignificant difference between the binding energy of Zn0 and Zn2+ species. However, it is possible to resolve the two species via interrogation of the Zn L3M4,5M4,5 spectrum (Fig. S22†).53 For the OG and DP catalysts the Zn L3M4,5M4,5 Auger peaks were centred around 988 eV which is consistent with ZnO. Interestingly, catalysts prepared via the ID method had the Zn Auger peak shifted to slightly lower kinetic energy and appeared at approximately 987 eV. No peak was witnessed at 992 eV for any of the catalysts which indicates a lack of Zn metal. This is consistent with XAFS analysis and shows that Zn exists solely as ZnO.
![]() | ||
Fig. 4 Curved-fitted Cu 2p3/2 XPS spectra of top) CuZnO/SAPO-11 ID and bottom) CuZnO/SAPO-34 ID catalysts. Peak 1 is due to Cu+/Cu0 and peaks 2–5 are due to Cu2+. |
Catalyst | CO2 conversion (%) | CO selectivity (%) | MeOH selectivity (%) | DME selectivity (%) | CO yield (%) | DME yield (%) | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
−11 | −34 | −11 | −34 | −11 | −34 | −11 | −34 | −11 | −34 | −11 | −34 | |
DB is dual bed, GM is granule/pellet physical mixture, PM is powder physical mixture. Low DME selectivity of DP catalysts is attributed to degradation of the solid acid framework. Average carbon mass balance of 97%. | ||||||||||||
CuZnO/SAPO-X ID | 4.7 | 8.1 | 0 | 0 | 32 | 20 | 68 | 80 | 0.0 | 0.0 | 3.2 | 6.5 |
CuZnO/SAPO-X OG | 14 | 12 | 14 | 30 | 19 | 15 | 67 | 55 | 2.0 | 3.6 | 9.4 | 6.6 |
CuZnO/SAPO-X DP | 17 | 13 | 40 | 40 | 36 | 35 | 24 | 25 | 6.8 | 5.2 | 4.1 | 3.3 |
Cu/ZnO/Al2O3 + SAPO-X PM | 23 | 18 | 42 | 39 | 14 | 13 | 44 | 48 | 9.7 | 7.0 | 10 | 8.6 |
Cu/ZnO/Al2O3 + SAPO-X GM | 26 | 21 | 44 | 43 | 13 | 12 | 43 | 45 | 11 | 9.0 | 11 | 9.5 |
Cu/ZnO/Al2O3 + SAPO-X DB | 18 | 11 | 42 | 46 | 13 | 11 | 45 | 43 | 7.6 | 5.1 | 8.1 | 4.7 |
Of the three active catalyst sets, catalysts produced via the ID method had the highest DME selectivity and no CO was detected (Table 4) during the reaction showing a significant suppression of the RWGS reaction. The most striking difference between the catalysts produced using the three different methods was the acid site strength and quantity as well as the location and distribution of the CuZnO nanoparticles. A strong negative correlation was found between acid site quantity and CO selectivity, with the CuZnO/SAPO-X ID catalysts showing the lowest CO selectivity and highest acid site quantity. A linear correlation was obtained by plotting the total integrated area of the NH3-TPD signal (acid site quantity) for the CuZnO/SAPO-X catalysts against their CO selectivity (Fig. S23†). The DME selectivity showed a positive correlation to acid site quantity. This demonstrates that by tailoring the acid site quantity of bifunctional catalysts it is possible to optimise the DME and CO selectivity during one-pot conversion of CO2. We hereby propose a water inhibition effect as a possible hypothesis to explain the relationship between acid site abundance and CO-formation inhibition. Catalysts with stronger and more abundant acid sites catalyse the dehydration of the intermediate MeOH to a greater extent which leads to significant production of water. This increase in localised water concentration can help to shift the thermodynamic equilibrium towards the conversion of CO into CO2via the WGS reaction (eqn (5)). This therefore helps to significantly suppress the RWGS reaction which leads to low CO levels that are below the detectable level of our mass spectrometer instrument. The CuZnO/SAPO-X ID catalysts contain the most abundant acid sites which led to significant water production (Table S10†) which resulted in the lowest CO selectivity observed. Further experimental validation through operando studies, coupled with computational calculations are required to ascertain whether this proposed localised water inhibition hypothesis vindicates the relationship between active site abundance and CO-formation suppression. Unfortunately, an increase in localised water concentration also shifts the equilibrium away from MeOH formation which reduces the CO2 conversion and explains the lower CO2 conversions for the ID catalysts. As such, there is the commonly encountered trade-off between activity and selectivity. Nevertheless, from an industrial perspective, high selectivity to DME is preferred as this significantly reduces toxic CO by-product formation and it is possible to increase overall conversions by using approaches such as recycling loops.
The secondary difference between the catalysts produced via the three methods was the proximity between the CuZnO nanoparticles and the SAPO supports. Although electron microscopy enables a visual assessment of this, it is not possible to use it to calculate a bulk-representative average distance between the nanoparticles and the support. To overcome this challenge and understand whether there is a link between site proximity and selectivity, a follow-up study was conducted whereby the distance between the redox and acid components was varied on the macroscopic scale. An optimised Cu/ZnO/Al2O3 MeOH synthesis catalyst and the SAPO-X solid acid catalysts were mixed in different manners to obtain varying degrees of proximity between the two functionalities. A dual bed (DB), granule mixing (GM) and powder mixing (PM) were used with the dual bed having the largest spatial separation between the two and powder mixing the smallest spatial site separation. Table 4 shows that CO selectivity decreases marginally with an increase in proximity between the two components which is in line with previous reports.25,26 The water production also increases with an increase in proximity (Table S10†) which is in line with our proposed water inhibition effect and explains the lower CO selectivity. However, the CO selectivity is still high and the influence of proximity on CO selectivity is minor. This demonstrates that the water inhibition effect is a very localised phenomenon and requires direct contact between the CuZnO nanoparticles and the SAPO support which demonstrates the advantage of bifunctional catalysts over a physical mixture of two distinct catalysts. The CuZnO/SAPO-34 ID catalyst has both a high abundance of acid sites and the CuZnO nanoparticles are deposited directly onto the SAPO support surface which results in highly efficient suppression of the RWGS reaction.
Comparing SAPO-11 and SAPO-34, it can be seen that on average, the CuZnO/SAPO-34 catalysts produced via all methods tend to give higher DME selectivity compared to their SAPO-11 equivalents. This is attributed to the stronger and more abundant acid sites of SAPO-34 compared to SAPO-11 (Fig. S10†) which results in rapid MeOH dehydration. This is further shown by the fact that the CuZnO/SAPO-11 catalysts tend to give higher MeOH selectivity compared to their SAPO-34 equivalents as the MeOH is clearly not as rapidly dehydrated (Table 4).
Further characterisation of spent catalysts used for the above time-on-stream stability was performed to better understand structural changes triggered by the catalysis. XRD characterisation has shown that the OG and DP catalysts suffer from Cu sintering due to an increase in the intensity of the Cu 111 peak (Fig. S29†). The ID catalysts on the other hand do not appear to suffer from sintering to the same degree as the intensity and line width of the Cu 111 peak appears to remain unchanged during the reaction. Indeed, TEM analysis of spent CuZnO/SAPO-34 OG and DP catalysts has shown that the average nanoparticle size has increased from 13 & 20 to 15 and 23 nm, respectively. The average nanoparticle size of the CuZnO/SAPO-34 ID catalyst on the other hand has decreased from 15 to 11 nm during the reaction (Fig. S30†). BET analysis of spent samples has curiously shown that the surface area of the spent pelletised CuZnO/SAPO-34 ID, OG and DP catalyst has increased marginally during the reaction (Table S11†). This could possibly be explained by the agglomeration of nanoparticles which leads to a reduction in the surface coverage of the SAPO-34 support which improves the access of guest species to the internal cage network of the solid acid. The CuZnO/SAPO-11 catalysts on the other hand appear to suffer from a marginal reduction in their surface area during the reaction. CHN analysis has found that both catalyst sets suffer from very marginal coking (<0.1 wt%) during the reaction which can contribute to deactivation (Table S1†). This level of coking is minor and significantly less than what is usually seen for SAPO-11 and SAPO-34 during MeOH dehydration, where the catalysts usually gain 1 & 10 wt% of carbon during the reaction, respectively.15 On average, the CuZnO/SAPO-11 catalysts gained more coke during the reaction compared to their SAPO-34 analogues. The higher level of coking of SAPO-11 would explain why its surface area decreased more significantly compared to SAPO-34. The elevated carbon gain could also indicate why SAPO-11 catalysts appear to deactivate more rapidly compared to their SAPO-34 analogues. There is no clear trend between coking and the catalyst synthesis method.
In this work, the CuZnO/SAPO-34 catalyst produced via the impregnation and drying method is the most promising for the selective one-pot conversion of CO2 to DME. CuZnO nanoparticles are anchored directly onto the SAPO-34 solid acid support using the ID approach which affords synergy between the Cu and ZnO species. Furthermore, the ID synthesis method yields a bifunctional catalyst with both a high surface area and high number of moderate strength Brønsted acid sites. Overall, these beneficial structural features obtained via the ID method result in modest CO2 conversions, high stability, exceptional DME selectivity and most importantly significant suppression of the CO-producing RWGS reaction due to the theorised localised water inhibition effect. Compared to other similar catalytic systems used for the one-pot synthesis of DME, our catalyst shows amongst the highest DME selectivity and lowest CO formation (Table S13†). This is highly desirable for industrial applications as on large scale this not only reduces overall waste, but also the formation of toxic by-products which is in line with green chemistry principles. Although the CO2 conversions are modest for the ID catalysts, there exist scope in both laboratory and industrial practice to take advantage of the high DME selectivity and boost CO2 conversions through further catalytic modification and reactor engineering.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy00020j |
This journal is © The Royal Society of Chemistry 2024 |