CO2 Conversion via Reverse Water Gas Shift Reaction Using Fully Selective Mo-P Multicomponent Catalysts.

The reverse water gas shift reaction (RWGS) has attracted much attention as a potential means to widespread utilization of CO2 through the production of synthesis gas. However, for commercial implementation of RWGS at the scales needed to replace fossil feedstocks with CO2, new catalysts must be developed using earth abundant materials, and these catalysts must suppress the competing methanation reaction completely while maintaining stable performance at elevated temperatures and high conversions producing large quantities of water. Herein we identify molybdenum phosphide (MoP) as a nonprecious metal catalyst that satisfies these requirements. Supported MoP catalysts completely suppress methanation while undergoing minimal deactivation, opening up possibilities for their use in CO2 utilization.

Introduction

The global warming caused by excessive greenhouse gases (GHGs) has become one of the greatest environmental threats in the world. Among these different GHG emissions, such as water vapor, CH4, and CO2, CO2 is an important one which is mainly emitted from oil refineries, power plants, cement production, and steel and iron industries.[1] Due to the greenhouse effect, several CO2 conversion technologies are proposed. Among the different CO2 upgrading processes, the reverse water gas shift (RWGS) reaction represents a viable route to convert CO2 and H2 into CO and water (eq ), and the product CO could be used in downstream Fischer–Tropsch (FT) or MeOH synthesis processes.[2,3] However, due to the endothermic nature of the process, the RWGS reaction requires high temperatures to achieve equilibrium CO2 conversions. In addition, the CO2 methanation is a side reaction (eq ) which must be suppressed by using a selective catalyst. Therefore, considerable efforts have been made to develop thermally stable catalysts with high activities and selectivities toward carbon monoxide.[4]Normally, the RWGS catalysts consist of well dispersed metal active sites on high surface area metal oxide supports.[5] In terms of metal sites, copper[6] and some noble metals (Pt,[7] Pd,[8] and Rh[9]) have been studied extensively. Concerning the support, CeO2 is one of the most widely used for the RWGS reaction because of its excellent redox properties.[6] In addition to the metal oxide supports, transition metal carbides (TMCs) have been identified as desirable materials for the RWGS reaction as their properties are similar to Pt-group precious metals.[10]

Although transition metal phosphides (TMPs) have been investigated in the energy industry,[10−12] in the past decades, the research dealing with TMPs catalysts for CO2 upgrading reactions are still relatively scarce compared to the materials listed above. Among the TMPs, molybdenum phosphide (MoP) catalysts exhibit stable performance toward methanol synthesis from CO2 and CO.[13,14] During high pressure CO2 hydrogenation experiments for methanol synthesis, MoP catalysts have been observed to catalyze some CO formation as a byproduct.[13,14] The molybdenum phosphide phase is theoretically expected to remain stable under hydrogenating conditions[13] and has been shown experimentally to retain its chemical structure up to 950 °C in hydrogen,[15] making it a suitable catalyst for the RWGS reaction. Our group has previously used a DFT-based mechanistic study to explore the potential activity of MoP (0001) for the RWGS reaction and found that this surface is an active phase for the RWGS reaction.[2] This theoretical work and previously reported activity and stability of MoP for CO2 reduction leads us to investigate the performance of MoP catalysts toward the RWGS reaction experimentally.

Among these widely used metal oxide supports, the combination of MoP and SiO2 has already been shown to result in high activity for methanol synthesis from CO2.[13,14] In addition, SiO2 shows the potential to prevent the agglomeration of metal sites leading to enhanced catalytic activity levels in the hydrogenation reactions. Al2O3 is also a widely investigated support for RWGS which could facilitate the dispersion of the active phase and boost oxygen mobility.[16,17] However, the acidity of Al2O3 can induce coking. When seeking for a fair balance acid–base properties and coking mitigation solution, the addition of ceria to alumina-based supports could decrease the overall acidity thus helping to avoid carbon deposition due to enhanced oxygen mobility ascribed to CeO2-based systems.[18,19] Herein we investigate a series of molybdenum phosphide catalysts supported on SiO2, Al2O3, and CeAl for the RWGS reaction.

Experimental Section

Experimental methods are summarized here, with more detailed descriptions available in the Supporting Information (SI).

Catalysts Preparation

Catalysts were synthesized using a wet impregnation method. Ammonium heptamolybdate [(NH4)6Mo7O24] (Sigma-Aldrich) and diammonium hydrogen phosphate [(NH4)2HPO4] (Sigma-Aldrich) were mixed to obtain a P/Mo atomic ratio of 1.2:1, as a slightly phosphorus rich synthesis was shown previously to be beneficial for the formation of the MoP phase.[20] This mixture was dissolved in deionized water and added to the point of incipient wetness of the supports (Sigma-Aldrich). The weight loading of MoP was 15 wt % for all supports. The solution was dried in an oven for 12 h at 80 °C before calcining for 5 h at 500 °C. The precursor was reduced in a fixed bed reactor, where the sample was heated from room temperature to 650 °C using a ramp of 2 min–1 followed by holding at this temperature for 2 h. Reduction took place with a flow of 60 mL min–1 H2 before being cooled to room temperature in N2. The sample was passivated at room temperature in a flow of 40 mL min–1 of 1.5% O2/argon for 12 h. This method was repeated for each of the three selected supports: silica (SiO2, Sigma-Aldrich), alumina (Al2O3, Sigma-Aldrich), and ceria-alumina (CeO2–Al2O3, Sigma-Aldrich).

The catalysts prepared with different supports are referred as Mo–P–SiO2, Mo–P–CeAl, and Mo–P–Al2O3 in this manuscript.

Catalysts Characterization

X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), temperature-programmed oxidation (TPO), transmission electron microscopy (TEM), H2–temperature-programmed reduction (TPR) and BET surface area measurement are used in this work to characterize the prepared catalysts.

Catalytic Testing

The RWGS tests were evaluated within a temperature range of 400 to 750 °C at a constant weight hourly space velocity (WHSV) of 12 000 mL g–1 h–1 for all synthesized catalysts. Stability tests were conducted at a space velocity of 12 000 mL g–1 h–1 with a H2/CO2 ratio of 4:1 at 550 °C for 24 h. The continuous temperature-programmed RWGS reaction was conducted within a temperature range of 300 to 750 °C using the mass spectrum for product analysis at a space velocity of 12 000 mL g–1 h–1 with a H2/CO2 ratio of 4:1.

Performance of the catalysts are reported in terms of CO2 conversion (eq ), CO selectivity (eq ), and CH4 selectivity (eq ). Where nCO is the initial molar flow rate (kmol/min) of CO2 in the reactant mixture and nCOout, nCH, and nCO are the outlet molar flow rates in the product stream of CO, CH4, and CO2, respectively.

Results and Discussion

Characterization of as-Synthesized Catalysts

Figure A displays the XRD pattern of the as-synthesized molybdenum phosphide catalysts. The crystalline MoP phase cannot be detected on any of the catalysts via XRD, indicating this phase is highly dispersed as nanoparticles, present as an amorphous phase or a mixture of well dispersed phases (phosphide and phosphate). For the Mo–P–SiO2 catalyst, the broad scattering maximum centered at 22.5° is ascribed to amorphous SiO2.[21,22] For Mo–P–Al2O3 and Mo–P–CeAl catalysts, the peaks labeled by purple dots are assigned to γ-Al2O3 (JCPDS No. 29-0063).[23,24] In addition, the peak at 2θ = 28.7° in the Mo–P–CeAl sample is attributed to the cubic fluorite-type CeO2 structure (JCPDS No. 81-0792).[25,26] Molybdenum oxide peaks were not detected on any catalyst.

Figure 1

(A) X-ray diffraction patterns; (B) X-ray photoelectron spectroscopy Mo 3d spectra; (C) P 2p spectra and the deconvoluted peaks for fresh Mo–P–SiO2, Mo–P–Al2O3, and Mo–P–CeAl samples; (D,E) H2–temperature-programmed reduction (TPR) results for the precursors of Mo–P–SiO2, Mo–P–Al2O3, and Mo–P–CeAl.

The surface chemistry and the electronic properties of these prepared samples were studied by XPS. Mo 3d and P 2p XPS spectra were collected (Figure B,C and Table ). Mo 3d spectra are split into 3d5/2 and 3d3/2 peaks due to the spin–orbital coupling effect.[27] For the Mo–P–Al2O3 and Mo–P–CeAl catalysts, it is found that there are two different Mo valence states species on the surface. The one with Mo 3d5/2 binding energy of 231 ± 0.3 eV is identified as Mo5+ species involved in Mo2O5.[28−31] Doublets with Mo 3d5/2 peaks at 233 eV ± 0.2 eV should be assigned to Moε+(V<ε28,30,32,33] or Mo6+ in molybdenum phosphate.[34] The P 2p scan is shown in Figure C. The peaks located around 134 eV can be ascribed to molybdenum phosphate species as a consequence of passivation.[35,36]

Table 1

Mo 3d5/2 and P 2p3/2 Binding Energies of All Samples

	Mo 3d5/2(eV)			P 2p3/2 (eV)
sample	Mo5+	Moε+(5<ε<6)	Mo6+	P5+	P/Mo
Mo–P–SiO2		232.9 (86.7%)	234.3 (16.3%)	134.5	1.56
Mo–P–Al2O3	231.3 (28.6%)	233.2(71.4%)		134.3	1.41
Mo–P–CeAl	230.8 (17.6%)	232.8 (82.4%)		133.6	1.72

The XPS analysis results indicated that the surfaces of these synthesized catalysts have been fully oxidized, which is expected from the passivation and air exposure of the catalysts after synthesis. Peaks corresponding to MoP (which would be in the range 227.1–227.7 eV) could not be detected.[20] We have previously shown via XAS and XRD that the MoP phase formation is heavily dependent on support-precursor interactions, and exposure to air results in surface oxidation which is reversed upon treatment in hydrogen.[13] Our results are consistent with previous work reporting that MoP on silica cannot be observed at low loadings (<25 wt %) via XPS and XRD.[34] In the presence of CeO2 on the Al2O3 support (Mo–P–CeAl), the binding energy of Mo shifts to a lower valence state than in Mo–P–Al2O3. Since CeO2 has excellent reducibility,[18,37,38] we proposed that the n-type semiconductor property of CeO2 plays a key role in this process and promotes the reduction of surface phosphate to a larger extent.

To test this hypothesis and gather further understanding of catalysts’ reduction features and the interactions among the molybdenum phosphide/phosphate phases and the different supports, H2-TPR was conducted on the catalyst precursors (before reduction). Figure D,E shows hydrogen consumption and water generation profiles of the studied samples from room temperature to 900 °C. The precursor of Mo–P–SiO2 presents the typical reduction peak around 450 °C corresponding to the reduction of Mo6+ (MoO3) species to Mo4+ (MoO2). The maximum peak at 650 °C corresponds to the co-reductions of Mo4+ to Mo0 and of P5+ to P0. The water generation peak of Mo–P–SiO2 precursor matched well with the H2 consumption peak (Figure E), the extra peak located at ∼150 °C should be assigned to physically adsorbed water.[39] For the precursor of alumina-containing Mo–P, molybdenum precursor should be reduced to the metallic state first and then react with P to form phosphide according to the previous report from the Oyama group.[40] In this work, peaks around 450 °C were detected, consistent with some degree of MoO reduction. But there was no main peak detected while heating from 600 to 800 °C. It is likely due to the formation of aluminum phosphates; the reduction of aluminum phosphates are reported to occur at T > 850 °C.[40] While all the phosphate is not reduced, an excess P (P/Mo = 1.2:1) ratio was used in our synthesis which is known to improve MoP formation.[20] Therefore, while the XPS results leads us to believe better reducibility on CeAl, TPR shows this is a surface effect and the bulk reduction of the catalyst is not affected because of the different MoP formation mechanisms on alumina supported MoP due to the presence of aluminum phosphates. In addition, the TPR results suggest that MoP formation occurs on SiO2 supported catalysts at the temperature we employed in the synthesis, but on alumina supported catalysts the reduction of phosphates (likely bound to aluminum) is not complete.

The P/Mo ratio shown in Table indicates that the surfaces of all prepared catalysts are rich in P. Although the P/Mo ratio used in synthesis is 1.2, all the composition values of P/Mo shown in Table are higher than 1.4. A similar phenomenon has been observed in MoP-K-SiO2 catalysts. In that case, even though the synthesis P/Mo ratio was equal to 1.5, a P/Mo ratio higher than 2 was observed for all catalysts. The higher P/Mo ratio might be attributed to the formation of a P-rich phosphate shell over MoP that is later reduced to a P-rich MoP species.[13]

Catalytic Performance

Figure A shows the CO2 conversion trends over the prepared catalysts as a function of temperature. The CO and CH4 selectivities are displayed in Figure (B). All the synthesized catalysts are active for RWGS in the temperature range 400–750 °C and more importantly the Sabatier reaction is completely suppressed, despite the high H2/CO2 ratio used. Mo–P catalysts are highly selective toward RWGS at ambient pressure.

Figure 2

(A) CO2 conversion (B) CO and CH4 selectivity for Mo–P–SiO2, Mo–P–Al2O3, and Mo–P–CeAl. Mass spectrum for (C) Mo–P–SiO2, (D) Mo–P–Al2O3, and (E) Mo–P–CeAl. Condition: H2/CO2 = 4:1, WHSV = 12 000 mL g–1 h–1.

In terms of the CO2 conversion, the performance of Mo–P–SiO2 is slightly better than that of Mo–P–Al2O3 and Mo–P–CeAl in the high temperature range (650–750 °C). In the 450–600 °C range, the CO2 conversion toward Mo–P–Al2O3 shows the best CO2 activity than the other two. But in general, the performances of these three studied catalysts are similar.

All the synthesized catalysts exhibit high CO selectivity (>80%) in the temperature range 450–750 °C (Figure B). Mo–P–SiO2 is the most selective catalyst, especially in the temperature range of 450–550 °C, producing nearly 100% CO. As shown in the TPR section, the temperature we employed in the synthesis is suitable to produce silica-supported MoP, but for alumina-supported MoP catalysts, there are phosphates remaining on the surface under our synthesis condition, hence the different phosphorus compounds are likely to be the reason for different CO selectivity. In addition, our group has previously used systematic DFT (density functional theory) study on MoP (0001) to explore its potential for applications in chemical CO2 recycling via the RWGS reaction. Mechanistic investigation using potential energy surface (PES) profiles in this work showcased that MoP was active toward the RWGS reaction with the direct path (CO2* → CO* + O*) favorable on MoP (0001). Furthermore, it was observed that CH4* formation relative to CO* on the MoP (0001) surface requires higher energy from the PES profile thermodynamically, hence the MoP (0001) surface was more selective toward CO than CH4 generation.[2] In our case, the Mo–P–SiO2 catalyst with more MoP present on the surface exhibited higher CO selectivity than alumina-supported Mo–P catalysts, consistent with the DFT calculation. Therefore, we attribute the high CO selectivity toward the Mo–P–SiO2 catalyst to the MoP phase generated on the surface of the SiO2 support.

As can be seen from Table , the carbon balance did not reach 100% toward the tested catalysts for most of the temperatures. For the Mo–P–SiO2 catalyst, the carbon balance is ∼100% in the 450–550 °C range and decreased gradually with increasing temperature. Since no methane was detected, this indicates that there are either other gas phase products (other than CO&CH4) and/or the deposition of carbon species on the catalysts.

Table 2

CO2 Conversion, CO Selectivity, and Carbon Balance Calculation toward Synthesized Catalysts

To measure if there are other gas phase species present, a continuous temperature-programmed RWGS reaction was conducted using the mass spectrum for product analysis. In our previous work, CH4, CO, and methanol as well as C2+ oxygenates and hydrocarbons were detected as gas phase products when MoP/Al2O3 and MoP/CeO2 were tested for CO2 hydrogenation reaction at 40 bar.[14] Hence, we monitored C2H4, C2H6, CH3OH, and C2H5OH as possible products along with CH4 and CO. No change in ion current was detected for C2H4, C2H6, CH3OH, and C2H5OH. The signals for CO, CH4, H2O, CO2, and H2 are shown in Figure C,D,E and agree with our conversion and selectivity data shown in Figure A,B. This is indicative of the missing carbon being deposited as solid carbon on the catalysts. The carbon deposition is investigated further in the next section by temperature-programmed oxidation (TPO) and thermogravimetric analysis (TGA).

Post-reaction Characterization

Figure shows the TPO, XRD, and TGA results of the post-reaction samples. All the samples used in this section are post-temperature-screening samples that have been tested under RWGS reaction conditions (H2/CO2 = 4:1, WHSV = 12 000 mL g–1 h–1) from 400 to 750 °C, one hour for each temperature.

Figure 3

(A) Temperature-programmed oxidation (TPO); (B) X-ray diffraction patterns result for post-reaction Mo–P–SiO2, Mo–P–Al2O3, and Mo–P–CeAl; thermogravimetric analysis (TGA) for post-reaction (C) Mo–P–SiO2, (D) Mo–P–Al2O3, and (E) Mo–P–CeAl.

O2-TPO experiments of the post-reaction catalysts were carried out, and the results are shown in Figure A. Certain temperature ranges of CO2 peaks can be attributed to the different types of carbonaceous species. The peaks corresponding to the active intermediates in the RWGS reaction appeared lower than 380 °C.[41,42] The second range peaks between 440 °C and 640 °C are assigned to whisker carbon formed on or close to Mo oxides.[43,44] In general, the most refractory carbon is the graphitic carbon formed on the support (temperature range: TPO > 650 °C), which does not appear in these three catalysts.[44,45] The first two fractions of coke were classified as soft coke which can be removed at lower temperatures, in this case, below 600 °C.[46] As can be seen in the Figure A, the carbon deposited on Mo–P–SiO2 and Mo–P–CeAl can be more easily removed by treatment in hydrogen at mild conditions than on Mo–P–Al2O3. In addition, the TPO result confirms that carbon deposition happened during the RWGS reaction, which can explain the less than 100% carbon balance at certain temperatures.

Figure B displays the XRD patterns for post-RWGS reaction samples. All the SiO2 and Al2O3 peaks are observed in fresh samples (Figure A), with no new phases. Only crystalline CeO2 disappeared after the RWGS test in Mo–P–CeAl, indicating the reduction of CeO2 to an amorphous Ce (3+) species during the RWGS performance test.

In order to further quantify the carbon deposition, TGA tests were conducted for all the post-reaction samples. Generally speaking, most of the carbon combustion happens below 400 °C, and the heat flows show broad positive curves indicating an exothermic process, consistent with oxidation. For the Mo–P–SiO2 catalyst, it was observed that the weight loss caused by coking is 12.6% (Figure C), hence the carbon formation on the 250 mg catalyst is 36.1 mg. Based on the reaction conditions used in the RWGS test (5 mL/min inlet CO2 flow, 1 h test for each temperature) and the catalytic performance shown in Figure , the missing carbon during the performance test is 48.3 mg (the detailed calculation can be seen in the SI). Therefore, around 75% of the missing carbon became the coke formation deposited on the surface of the Mo–P–SiO2 catalysts. For the Mo–P–Al2O3 catalyst, the weight loss caused by coking is around 8.5% (Figure D) and the corresponding carbon formation is 23.1 mg. However, the missing carbon during the RWGS test toward Mo–P–Al2O3 is around 66.3 mg, indicating that there are some other gas phase products have not been detected. For the Mo–P–CeAl catalyst the plot trend is different than for the other two catalysts (Figure E). The weight decreased at the beginning, but when the temperature reached 300 °C, it started to increase. The first decrease should be attributed to the carbon combustion like that for the other two catalysts, the further mass increase could be assigned to the oxidation of the CeO phase. As can be seen in the post-reaction XRD pattern, crystalline CeO2 disappeared in Mo–P–CeAl after the RWGS test, indicating that the reduction of CeO2 happened during the RWGS reaction. Here the amorphous Ce3+ species might have been fully oxidized to CeO2 again during the TGA test; hence, a 4% weight gain shows in the TGA plot. The TPO results showcase that the carbon deposition is not the determining factor of the catalytic performance, despite the higher amount of carbon deposition on Mo–P–SiO2, it still shows higher CO selectivity than Mo–P–Al2O3. Since MoP is proposed to be very selective toward CO generation in our previous theoretical study,[2] the greater presence of MoP on the surface of Mo–P–SiO2 is likely to be the reason for the CO selectivity difference.

Stability Test

Since all three catalysts exhibit similar CO2 conversions, the one showing the best CO selectivity (Mo–P–SiO2) was chosen to assess 24 h stability during the RWGS. Normally, the RWGS reaction is combined with a Fischer–Tropsch process aiming for an integrated process of CO2 to fuels. The Fischer–Tropsch process is generally operated in the temperature range of 150–300 °C, while the endothermic nature of the RWGS imposes high operational temperatures. In this sense, the successful implementation of a medium/low-temperature RWGS catalyst would represent a step ahead in this technology, facilitating energy and process integration. Thus, 550 °C was selected as reaction temperature in here to bridge the RWGS-FTS gap.

As the results show in Figure A, the CO2 conversion declined from 27% to 18% in the first 2 h of testing, and the CO selectivity increased from 80% to 97% in the first hour and reached at 100% at 2 h. After 2 h, both the CO2 conversion and CO selectivity remained stable in the remaining 22 h, indicating carbon deposition occurs initially, after which catalytic activity is stabilized. Overall, our catalysts exhibit a stable performance once the steady state is reached showcasing full selectivity to CO at intermediate temperatures where CO2 methanation is typically an issue.[47]

Figure 4

(A) Stability test at 550 °C, WHSV of 12 000 mL g–1 h–1 with a H2/CO2 ratio of 4:1 for Mo–P–SiO2. (B) TEM micrographs of Mo–P–SiO2. (C) EDX micrographs of Mo–P–SiO2.

TEM characterization was used to study the nanostructure of as synthesized Mo–P–SiO2 (Figure B). Spherical MoP nanoparticles can be seen in Figure B, similar to MoP/SiO2 catalysts reported previously.[13] The corresponding element mappings of Mo–P–SiO2 shown in Figure C demonstrate that the elements of Mo and P are uniformly co-located on the entire nanoparticles of the SiO2 support. For the silicon-supported MoP catalysts, our previous works show that the catalyst synthesized in this same technique yields a mixture of phosphate and phosphide,[20] which might be the reason that the MoP peaks have not been observed in the XRD pattern. Since we have proven in the TPR section that the reduction temperature we used in synthesis is suitable for silicon-supported MoP production, and spherical MoP nanoparticles detected in Figure B are similar to the MoP/SiO2 catalysts reported in previous work,[13] the catalysts we synthesized in here are likely to be the mixture of phosphide/phosphate.

Our results show that supported Mo–P catalysts are robust materials that can run satisfactorily for continuous operations displaying complete RWGS selectivity. The suppression of the Sabatier reaction is particularly significant for the efficient use of hydrogen; for a net CO2 consuming RWGS process, H2 should have a low carbon footprint and currently green H2 is expensive as well.[3] Moreover, the complete RWGS selectivity across the full range of temperatures and conversions studied herein make it possible to explore tandem catalysis schemes where MoP catalysts producing CO could be coupled with CO consuming Fischer–Tropsch active catalysts. This area of tandem catalysis for CO2 utilization has gathered considerable interest and requires the development of fully selective RWGS catalysts.[48]Table presents the comparative performance of MoP catalysts in this work with prior investigations. Although MoP has been reported to be used in some reactions such as alcohol synthesis, to the best of our knowledge, no other paper has reported MoP as a catalyst for the RWGS reaction. Therefore, we have compared the performance to molybdenum carbides as well as our recent work on nickel phosphide catalysts (Table ). We have previously shown the activity of nickel phosphide toward the RWGS reaction, and it exhibited higher CO2 activity at the same temperature as MoP-SiO2 reported here.[49] However, unlike the MoP catalysts presented herein, nickel phosphides are also active for methanation, especially at the low temperature range (300–600 °C). We have also studied previously the performance of molybdenum carbides toward the RWGS reaction. The β-Mo2C shows higher CO2 activity than the MoP catalyst in this work, and with the addition of Cs or Cu, the CO selectivity reached 95–98%. However, for the 0.25 g Mo2C catalyst used in our previous test, it contains 100 wt % Mo2C (or 99 wt % Mo2C for Cu–Mo2C and Cs–Mo2C) in the catalyst. For the 0.25 g MoP catalysts used in this work, there is only 15 wt % MoP in the catalyst. Therefore, in terms of the mass activity, MoP is still a promising catalyst for the RWGS reaction.[50]

Table 3

Catalyst Performance Comparison with Materials Reported in the Literature

catalysts	temp (°C)	H2/CO2 ratio	CO2 conversion (%)	CO selectivity (%)	WHSV (mL/gcal h)	ref
1%NiCo@SiO2	500	4	50	47	15000	(51)
β-Mo2C	550	4	60	85
Cu–Mo2C	550	4	58	95	12000	(50)
Cs–Mo2C	550	4	56	98
Ni2P–-SiO2	550	4	43	79	12000	(49)
Mo–P–SiO2	550	4	18	100	12000	this work

Conclusions

In this work, we have synthesized supported MoP catalysts to investigate their activity in the RWGS reaction, which demands a stable and fully selective catalyst capable of operating at increased temperatures. Silica, alumina, and ceria-alumina supported MoP catalysts are all shown to be active for the RWGS reaction and demonstrate a complete suppression of the methanation side reaction. Mo–P–SiO2 showed limited deactivation in the first 2 h of the test due to carbon deposition, followed by stable performance for 22 h on stream. This high selectivity of MoP catalysts to CO is a significant advancement toward developing robust RWGS catalysts that make efficient use of green hydrogen, which is needed to develop net CO2 consuming processes. Moreover, MoP catalysts provide a step forward in developing tandem catalysts that can synthesize coupled carbon products from CO. The discovery of new catalysts for RWGS opens up opportunities for chemical CO2 recycling which are urgently needed in the context of a circular economy.