Ni-Phosphide catalysts as versatile systems for gas-phase CO2 conversion: Impact of the support and evidences of structure-sensitivity

Fuel 323 (2022) 124301 Available online 5 May 2022 0016-2361/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Ni-Phosphide catalysts as versatile systems for gas-phase CO 2 conversion: Impact of the support and evidences of structure-sensitivity Q. Zhang a , * , L. Pastor-P´ erez a , c , J.J. Villora-Pico b , M. Joyce a , A. Sepúlveda-Escribano b , M.S. Duyar a , T.R. Reina a , c a Department of Chemical and Process Engineering, University of Surrey, Guildford GU2 7XH, United Kingdom b Laboratorio de Materiales Avanzados, Departamento de Química Inorg´ anica, Instituto Universitario de Materiales de Alicante, Universidad de Alicante, Apartado 99, E03080 Alicante, Spain c Department of Inorganic Chemistry and Material Sciences Institute of Seville, University of Seville-CSIC, 41092, Seville, Spain ARTICLE INFO Keywords: CO 2 conversion RWGS Methanation Switchable Catalysts Ni Phosphide ABSTRACT We report for the first time the support dependent activity and selectivity of Ni-rich nickel phosphide catalysts for CO 2 hydrogenation. New catalysts for CO 2 hydrogenation are needed to commercialise the reverse water–gas shift reaction (RWGS) which can feed captured carbon as feedstock for traditionally fossil fuel-based processes, as well as to develop flexible power-to-gas schemes that can synthesise chemicals on demand using surplus renewable energy and captured CO 2 . Here we show that Ni 2 P/SiO 2 is a highly selective catalyst for RWGS, producing over 80% CO in the full temperature range of 350–750 ◦C. This indicates a high degree of suppression of the methanation reaction by phosphide formation, as Ni catalysts are known for their high methanation activity. This is shown to not simply be a site blocking effect, but to arise from the formation of a new more active site for RWGS. When supported on Al 2 O 3 or CeAl, the dominant phase of as synthesized catalysts is Ni 12 P 5 . These Ni 12 P 5 catalysts behave very differently compared to Ni 2 P/SiO 2 , and show activity for methanation at low temperatures with a switchover to RWGS at higher temperatures (reaching or approaching thermodynamic equilibrium behaviour). This switchable activity is interesting for applications where flexibility in distributed chemicals production from captured CO 2 can be desirable. Both Ni 12 P 5 /Al 2 O 3 and Ni 12 P 5 /CeAl show excellent stability over 100 h on stream, where they switch between methanation and RWGS reactions at 50–70% conversion. Catalysts are characterized before and after reactions via X-ray Diffraction (XRD), X-ray Photoelectron Spectroscopy (XPS), temperature-programmed reduction and oxidation (TPR, TPO), Transmission Electron Microscopy (TEM), and BET surface area measurement. After reaction, Ni 2 P/SiO 2 shows the emergence of a crystalline Ni 12 P 5 phase while Ni 12 P 5 /Al 2 O 3 and Ni 12 P 5 /CeAl both show the crystalline Ni 3 P phase. While stable activity of the latter catalysts is demonstrated via extended testing, this Ni enrichment in all phosphide catalysts shows the dynamic nature of the catalysts during operation. Moreover, it demonstrates that both the support and the phosphide phase play a key role in determining selectivity towards CO or CH 4 . 1. Introduction Global warming caused by the accumulation of atmospheric CO 2 is considered as one of the greatest environmental threats in the 21st century [1,2]. Climate change is causing rise in sea levels, increasing the occurrence of hurricanes, wildfires, tornadoes, and floods, and having pronounced impacts on human health [3,4]. Since the consequences have the potential to bring serious unrest in the world, CO 2 utilization in the context of a circular economy has become a top priority in recent years. Among the CO 2 utilization strategies, the CO 2 methanation reaction (eq. (1)) and reverse water–gas shift (RWGS) reaction (eq. (2)) have been studied extensively [5,6,7,8] because their products (methane and carbon monoxide) can be used as fuel, in attempts at closing the combustion-based carbon cycle. Carbon monoxide (CO) from RWGS can be mixed with hydrogen (H 2 ), and this mixture is called synthesis gas (syngas). Syngas is used in the chemical industry as reactant to form methanol or liquid hydrocarbons. CO2+4H2↔ CH4+2H2OΔH0 298 = − 165⋅kJ/mol (1) * Corresponding author. E-mail address: [email protected] (Q. Zhang). Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel https://doi.org/10.1016/j.fuel.2022.124301 Received 4 January 2022; Received in revised form 6 April 2022; Accepted 19 April 2022 Fuel 323 (2022) 124301 2 CO2+H2↔ CO +H2OΔH0 298 =41⋅kJ/mol (2) The CO 2 methanation process, also called the Sabatier reaction, is an important catalytic reaction that is of interest for renewable energy storage applications based on power-to-gas (PtG) conversion schemes. This PtG approach begins with H 2 production by water electrolysis using renewable electricity, followed by H 2 conversion with an external (captured) CO 2 source, methane being the main product. On the other hand, the reverse water −gas shift reaction coupled to chemical looping cycles (RWGS −CL) for the intensified conversion of CO 2 is regarded as a highly effective utilisation route with promising potential in industrial applications [5]. Such a hybrid process is a combination of water electrolysis with RWGS reaction. Herein, water produced in the RWGS can be recycled to the water electrolysis part, and the combination of CO and additional H 2 can be used for liquid fuel production via FT or methanol synthesis. Fig. 1 depicts a switchable process, which we have investigated in recent years [7]. The switchable nature of the process ensures flexibility in CO 2 utilisation to be able to match the demand for chemicals with surplus electricity production. For example, methane can be used as heating fuel, which has a seasonal demand change (due to residential use) and the ability to produce syngas opens up possibilities in expansion of the role of captured CO 2 in the chemical industry. Clearly the success of both routes relies on the availability of green hydrogen from renewable sources (i.e. solar or wind) [6]. In order to manage the variable production of wind and solar energy, CO 2 is captured using mature carbon capture technologies, and then utilised at times of low electricity demand to ensure renewable power conversion into chemical energy and added value products [9]. As shown in eq. (1) and eq. (2), the CO 2 methanation is exothermic and the rWGS reaction is endothermic. The end product selectivity can be controlled by adjusting the reaction temperature and utilizing a suitable catalyst [7]. Since the RWGS reaction is favoured at high temperatures due to the endothermic nature of the reaction, a great deal of heat input is needed to produce CO. CO 2 methanation on the other hand is a highly exothermic process, which requires heat input to overcome kinetic barriers of the reaction [8]. Assuming that the designed catalyst is both active and selective to CO 2 methanation in the lower temperature range (between 300 ◦C and 500 ◦C) and RWGS in the higher temperature window (550 ◦C to 750 ◦C), the integration proposal depicted in Fig. 1 can be potentially achieved, and the desired product composition can be controlled by adjusting the reaction temperature. Under these premises, a switchable catalyst is needed. When seeking novel switchable catalyst formulations, Ni catalysts are commonly used for both RWGS and CO 2 methanation [10,11,12]. Nickel phosphide (Ni 2 P) has shown a promising catalytic performance for the dry reforming of methane (DRM) in both theoretical studies [13] and experimental research [14]. Previously, Ni 2 P has also shown to be active for the hydrodeoxygenation of guaiacol [15,16]. The extraordinary hydrogen dissociation and hydrogenation abilities shown by Ni 2 P has prompted us to investigate this catalyst for CO 2 hydrogenation reactions, such as the RWGS reaction and CO 2 methanation. As an additional advantage, Ni 2 P has previously shown some prevention of carbon deposition in DRM [13], which is a common deactivation mechanism in some CO 2 conversion reactions. Since highly effective catalysts are normally composed of well isolated and dispersed active phases supported on carrier materials, supports should be chosen in conjunction with nickel phosphide phases. Various supports have been used for both CO 2 methanation and RWGS, such as SiO 2 [17], Al 2 O 3 [18,19], CeO 2 [20] and CeAl mixed oxides [21]. In this regard, SiO 2 has been proven to inhibit agglomeration of metal crystals leading to enhanced CO 2 conversion levels in hydrogenation reactions [22]. Al 2 O 3 is also a widely investigated support that favours high metallic dispersion, resulting in active and stable supported metal hydrogenation catalysts [23]. However, coking and sintering were promoted by the inherent acidity of Al 2 O 3 during the reaction, what leads to catalyst deactivation [24]. As previously reported, the addition of ceria to alumina-based supports decreases the overall acidity thus helping to avoid carbon deposition [21]. Under this scenario, a series of nickel phosphides supported on SiO 2 , Al 2 O 3 or CeAl catalysts have been prepared in this work. The catalytic performances of the synthesized catalysts have been tested for both the CO 2 methanation and the RWGS reaction, targeting a switchable scheme allowing end-product flexibility. The physico-chemical properties of these catalysts have been explored, and the main reasons for their versatile behaviour are carefully addressed. 2. Experimental section 2.1. Catalysts preparation Catalysts were synthesised using a wet impregnation method. According to previous research [25], moderately excess P content has a positive effect on stability because the excess P content is lost during the reduction process. Hence, a larger P:Ni ratio than the target stoichiometric ratio (1:2) was chosen in this synthesis. Nickel nitrate hexahydrate [Ni(NO 3 ) 2 ⋅6H 2 O] (Merck) and diammonium hydrogen phosphate [(NH 4 ) 2 HPO 4 ] (Merck) were mixed to obtain a P:Ni ratio of 2.4:2. This mixture was dissolved in deionised water and added to the point of incipient wetness of the supports to achieve a loading of 15 wt%. The solids were then left to dry for 12 h at 80 ◦C and calcined at 500 ◦C for 2 h in air. The prepared materials were then reduced under the atmosphere of H 2 (60 mL/min) using a temperature-programmed reduction (TPR) protocol described as follows: the temperature was increased from room temperature to 650 ◦C at a rate of 2 ◦C/min, and held at 650 ◦C for 2 h. After cooling to room temperature in N 2 , the solids were passivated in 2.1% O 2 /N 2 for 12 h under the mixture of 5 mL/min of air and 45 mL/ min of N 2 . The same synthesis method was used for each of the three selected supports: Silica (SiO 2 , SASOL), γ-Al 2 O 3 (PURALOX SCFa-230, SASOL) and CeO 2 -Al 2 O 3 (PURALOX SCFa-160/Ce20, 80% Al 2 O 3 + 20% CeO 2 , SASOL). The catalysts prepared with different supports are referred to in this manuscript as Ni 2 P-SiO 2 , Ni 12 P 5 -CeAl and Ni 12 P 5 -Al 2 O 3 respectively, where the dominant phase identified by X-ray diffraction after synthesis was used in the name of each supported Ni x P catalyst. 2.2. Catalysts characterization X-ray photoelectron spectroscopy (XPS, K-ALPHA, Thermo Scientific) was used to detect the surface chemical valence states of the synthesized materials. The samples were exposed to air prior to analysis. A Al-K radiation (1486.6 eV), monochromatized by a twin crystal monochromator was used for the spectra collection, yielding a focused X-ray spot at 3 mA ×12 kV. The alpha hemispherical analyser was operated in the constant energy mode. The whole energy band was measured by Fig. 1. Simplified scheme of the methanation and RWGS processes integrated with water electrolysis. Q. Zhang et al. Fuel 323 (2022) 124301 3 200 eV survey scan pass energies. 50 eV in a narrow scan was selectively used for some particular elements. Avantage software was used for XPS data analysis. The C 1 s core level located at 284.6 eV was used as reference binding energy. The experimental backgrounds were corrected by smart background function and surface elemental composition measures were carried out by calculating background-subtracted peak areas. The system flood gun that provides low energy electrons and low energy argon ions from a single source was used to achieve charge compensation. X-Ray Diffraction (XRD) analysis was performed with an X’Pert Pro PANalytical using Cu-K 。 (40 mA, 45 kV) over a 2 theta range of 10◦-80◦ at room temperature. The N 2 adsorption studies at −196 ◦C with an AUTOSORB-6 equipment (QUANTACHROME INSTRUMENTS) were conducted to determine the Brunauer–Emmett–Teller (BET) specific surface area. Samples were previously out-gassed at 250 ◦C for 4 h under vacuum. The Barret–Joyner–Halenda (BJH) method was used to estimate the average pore size and pore volume. The result has been presented in the Support Information (SI). The hydrogen consumption of the catalyst precursors was determined by H 2 -Temperature programmed reduction (TPR) in this work. The experiment was conducted in a vertical fixed bed quartz reactor under the mixture of 5 mL/min H 2 and 20 mL/min Ar. The right amount of quartz wool was placed into the reactor and 50 mg of catalyst precursor was loaded on the quartz wool, and then heated up to 920 ◦C at a ramp of 10 ◦C/min. The hydrogen consumption and water production were recorded by an online mass spectrometer (Pfeiffer, OmniStar GSD 301). Temperature-programmed oxidation (TPO) experiments were carried out in the reactor coupled to an online mass spectrometer (Pfeiffer, OmniStar GSD 301) to measure the amount of carbon deposited on the tested catalysts. 3% O 2 in He (40 mL/min) was flowed through the sample while the temperature was increased from room temperature to 920 ◦C at 10 ◦C/min. Mass 44 (CO 2 ) was recorded, hence the TPO peak area could be converted to the amount of carbon deposition due to the calibrated CO 2 signal. Thermogravimetric analysis (TGA) was conducted on the post reaction catalysts in an SDT Q600 V8.3 Instrument from TA Instruments. A flow of 100 mL/min of air was used, while the temperature was increased from room temperature to 900 ◦C at 10 ◦C/min. 2.3. Catalytic behaviour The RWGS reactions were carried out and recorded in a vertical continuous fixed bed reactor coupled to an ABB AO2020 Advanced Optima Process Gas Analyser. The reactor was a 7 mm inner diameter quartz tube in which 0.25 g of catalyst was placed on the quartz wool in the middle of the reactor. The sample was heated in flowing N 2 from room temperature to 300 ◦C. Then, the N 2 was replaced by the feed gas mixture of H 2 :CO 2 =4:1 at a constant weight-hourly space velocity (WHSV) of 12000 mL/(g⋅h) and the products of the reaction were evaluated from 300 to 750 ◦C. At each temperature, the gas products were analysed after 20 min of steady-state reaction. Stability tests were conducted at a WHSV of 12000 mL/(g⋅h) with a H 2 :CO 2 ratio of 4:1 at 450 ◦C for 60 h. CO 2 conversion (Eq. (3)), CO selectivity (Eq. (4)) and CH 4 selectivity (Eq. (5)) could be measured based on the above tests. The relative experimental error in CO 2 conversion and CO/CH 4 selectivity in this work was given within ±0.5 %[6]. Where nCO2in is the initial molar flow (kmol/min) of CO 2 in the reactant mixture andnCOout,nCH4out,nCO2outare the outlet molar flows in the product stream of CO, CH 4 and CO 2 respectively. CO2conversion(%) = nCO2in −nCO2out nCO2in ⋅100 (3) COconversion(%) = nCOout nCO2in −nCO2out ⋅100 (4) CH4conversion(%) = nCH4out nCO2in −nCO2out ⋅100 (5) COyield(%) = nCOout nCO2in ⋅100 (6) 3. Results and discussion 3.1. Characterisation of as synthesised catalysts 3.1.1. XRD analysis Fig. 2 displays the XRD pattern of the fresh nickel phosphide catalysts. For the Ni 2 P-SiO 2 catalyst, the diffraction peaks at 40.8◦, 44.6◦, 47.3◦, 54.2◦, 54.9◦and 74.8◦are ascribed to Ni 2 P (JCPDS No. 03–0953) [26,27]. The broad scattering maximum centred at 22.5◦corresponds to amorphous SiO 2 [28,29]. For the Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl catalysts, all diffraction peaks labelled by blue dots should be aligned on Ni 12 P 5 (JCPDS No. 22–1190) [30,31,32]. The diffraction peak at 66.8◦ and the broad peak around 46◦are assigned to γ-Al 2 O 3 (JCPDS No. 29–0063) [31]. In addition, the peak at 2θ =28.65◦in the Ni 12 P 5 -CeAl sample matches well with the cubic fluorite-type CeO 2 structure (JCPDS No. 81–0792) [33,34]. Despite the use of a high P/Ni ratio of 1.2 during synthesis, all nickel phosphide active species are metal-rich and vary depending on the selected support. For the SiO 2 -supported catalyst, the dominant nickel phosphide phase is Ni 2 P, and there is one small peak belonging to Ni 12 P 5 that was detected. In Al 2 O 3 -containing catalytic formulations, Ni 12 P 5 is the only nickel phosphide phase observed. According to previous research, the phases of the as-synthesized nickel phosphide nanocrystals are mainly affected by the P/Ni precursor molar ratio, heating process, and time of reaction [35,36]. Once the initial P/Ni ratio and reduction condition are selected, the resulting NiP phases depend on the support type [37]. For an inert support such as SiO 2 , the Ni 2 P phase was obtained from a P/Ni ratio greater than 0.8 [38,39,40]. For acidic supports such as γ-Al 2 O 3 , a larger P/Ni (higher than 2) was required to obtain the Ni 2 P phase [38,41]. In this work, P/Ni =1.2 precursor was used in synthesis; hence, Ni 2 P was produced on SiO 2 while Ni 12 P 5 were formed on γ-Al 2 O 3 and CeAl. The result is accordance with previous research [42]. 10 20 30 40 50 60 70 80 CeO 2 Al 2 O 3 Ni12P5 Ni2P Ni 2 P-SiO 2 Ni x P-Al 2 O 3 Intensity (a.u.) 2θ ( o ) Ni x P-CeAl SiO2 Fig. 2. X-ray diffraction patterns for fresh Ni 2 P-SiO 2 , Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 - CeAl samples. Q. Zhang et al. Fuel 323 (2022) 124301 4 3.1.2. XPS study The chemical states and coordination environment of Ni and P in the studied samples were investigated by X-ray photoelectron spectroscopy (XPS) (Fig. 3 and Table 1)., Since same chemical information was carried by two splitted Ni 2p spectra (Ni 2p 1/2 and Ni 2p 3/2 ), only the Ni 2p 3/2 bands were curve fitted in this paper (Fig. 3 (A)). Concerning the Ni 2 PSiO 2 sample, the Ni peak at 853.9 eV was assigned to Ni δ+(0 <δ <2) in Ni 2 P [43]. Since the binding energy of metallic Ni vary from 852.3 to 852.9 eV in different reference, the peaks observed at 852.8 ±0.1 eV in the Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl catalysts can be attributed to Ni in Ni 12 P 5 or metallic Ni. In this case, very clear Ni 12 P 5 peaks were observed in Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples in XRD and negatively charged P δbelongs to Ni 12 P 5 was detected in the P spectra in same samples. Hence, it is reasonable to attribute the peaks located at 852.8 ±0.1 eV to reduced Ni δ+(0 <δ <2) in Ni 12 P 5 phase. The magnitude of δ followed the sequence: Ni 2 P >N 12 P 5 [44,45,46]. Meanwhile, the presence of metallic Ni cannot be excluded. Ni 12 P 5 and metallic Ni might coexistence on the surface the Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples. In higher binding energy region, peaks located around 856.5 eV were observed for Ni 2+ in the passivation layer formed on the surface of studied catalysts following synthesis [43,47]. The broad peaks at 861.5 eV are attributed to the shake-up satellite peak for Ni 2p 3/2 [47,48]. The P 2p scan is shown in Fig. 3 (B). The peaks located at 129.1 can be assigned to P belonging to nickel phosphide. The above binding energy is lower than that of elemental P (130.2 eV), indicating that the P species in nickel phosphide carries a partial negative charge (P δ- ) [47]. Since the corresponding Ni species have a very small positive charge, the results confirm the formation of nickel phosphide, which is also in good agreement with our XRD data. Meanwhile, the peaks located around 134 eV could be ascribed to residual oxidized P species as a consequence of passivation [47,49]. It is worth noting that although the Ni δ+ peak was shown in the Ni 2p spectra of Ni 2 P, there is no P δpeak belonging to Ni 2 P observed in the P 2p spectrum of Ni 2 P-SiO 2 , only oxidized P species were detected. This is because of the surface oxidation; according to density functional theory (DFT) calculations, unlike nickel, phosphorus in a Ni 2 P surface preferentially interacts with oxygen [50]. Compared to the alumina-containing supported catalysts, the binding energy of Ni in Ni 2 P-SiO 2 is higher than that of Ni 12 P 5 , indicating a lower electron density and a comparatively weaker Ni-P interaction. As for the two catalysts presenting Ni 12 P 5 as dominant nickel phosphide phase, the binding energy of Ni in Ni 12 P 5 -CeAl is lower than that of Ni 12 P 5 -Al 2 O 3. This is because of the n-type semiconductor property of CeO 2 . Indeed, when the CeO 2 is reduced, oxygen vacancies are produced, and free electrons migrate to Ni sites in nickel phosphide [38 51], leading to an increment of the electron charge density (ECD) around the Ni 12 P 5 . In addition, as can be seen from Table 1, the content of Ni 12 P 5 in Ni 12 P 5 -CeAl (38%) is higher than the value in Ni 12 P 5 -Al 2 O 3 (26%), indicating that the presence of CeO 2 in the support can help the electron transfer to Ni and stabilize Ni 12 P 5 in a reduced phase from surface oxidation. In addition, it is worth noting that the intensity of Ni 2 P-SiO 2 spectrum is clearly weaker than the other two catalysts. Since the intensity of XPS spectrum measures how much of a material is at the surface, the weaker intensity of Ni 2 P peaks shows that the content of Ni 2 P component presented on the Ni 2 P-SiO 2 surface is less than the contents of Ni 12 P 5 presented on the alumina containing catalysts. 3.1.3. H 2 –TPR results H 2 -TPR were conducted to gather further understanding of catalysts’ redox features and the interactions among the nickel phosphide phases and the different supports. For this experiment the transformation of the catalyst precursors (before reduction) to phosphides was tracked by H 2 - TPR. Fig. 4 shows hydrogen consumption profiles of the studied samples from room temperature to 920 ◦C. The precursor of Ni 2 P-SiO 2 presents the typical reduction peak around 400 ◦C corresponding to the reduction of bulk NiO. The maximum peak at 700 ◦C corresponds to the coreductions of the nickel species in phosphate and the P–O bond [38,52,53]. Normally, the reduction of PO x /SiO 2 is reported to begin at about 800 ◦C and was not complete until 1000 ◦C due to the high thermally stable P-O bond [54]. However, the corresponding reduction happened at lower temperature (700 ◦C) in this case, indicating that the PO x /SiO 2 phase was more easily reduced due to extra hydrogen atom availability from dissociation of H 2 on metallic Ni. Based on TPR findings, during the H 2 -TPR process, Ni species were first reduced to metallic Ni, which via dissociation of H 2 appear to promote the reduction of P-O bond, in good agreement with prior findings [55]. For the precursor of Ni 12 P 5 -Al 2 O 3 , the peak centred at 690 ◦C was assigned to the reduction of nickel phosphate species on Al 2 O 3 [38,56]. The reduction temperature of nickel species in Ni 12 P 5 -Al 2 O 3 precursor is lower than that of Ni 2 P/SiO 2 precursor because the strong interaction between Al 2 O 3 and PO 4 3reduced the interaction between nickel species and phosphate. Therefore, Al 2 O 3 support promoted the reduction of nickel phosphate to a different extent [38,57]. The temperature of the main reduction peak for Ni 12 P 5 -CeAl precursor is even lower than that for Ni 12 P 5 -Al 2 O 3 (670 ◦C) due to the enhanced redox properties ascribed to CeO 2 [58]. The surface oxygen species on CeO 2 combined with highly reactive hydrogen atoms (those from spillover) lead to the formation of OH − species. Such OH − species are known to facilitate the surface diffusion of hydrogen boosting the overall sample reducibility [59]. Therefore, our TPR results showcase a clear influence of the support on the reduction behaviour of the synthesised catalysts. The existence 867 864 861 858 855 852 849 846 Ni-P (Ni 2 P) Ni-P (Ni 12 P 5 ) Sat. Sat. Ni 2+ Ni 2+ Sat. Ni 2+ Ni-P (Ni 12 P 5 ) Ni 12 P 5 -CeAl Ni 12 P 5 -Al 2 O 3 Intensity (a.u.) Binding energy (eV) Ni 2 P-SiO 2 (A) Ni 2p 138 136 134 132 130 128 126 P 5+ P 5+ Ni 12 P 5 P 2p3/2 P 2p1/2 P 2p3/2 P 2p1/2 Ni 12 P 5 Ni 12 P 5 -Al 2 O 3 Ni 12 P 5 -CeAl Intensity (a.u.) Binding energy (eV) (B) P 2p Ni 2 P-SiO 2 P 5+ Fig. 3. X-ray photoelectron spectroscopy (XPS) Ni 2p spectra (A) and P 2p spectra (B) and the deconvoluted peaks of the Ni 2 P-SiO 2 , Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl. Q. Zhang et al. Fuel 323 (2022) 124301 5 states of nickel and phosphorus species vary according to different supports after reduction. 3.1.4. TEM results The nanostructure and morphology of the three fresh synthesized catalysts were investigated by TEM (as shown in Fig. 5 A, B, C) and the elements distribution of Ni 12 P 5 -Al 2 O 3 can be seen in Fig. 5(D). Observations from TEM shows that the morphology of the nickel phosphide is greatly affected by the different supports. For Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 - CeAl catalysts, large particles of Ni 12 P 5 phases are observed from the outside of the support structure. According to our BET results (details presented in SI), the pore sizes of the supports are around 6–10 nm, and the nano cube size of the Ni 12 P 5 structures is within the 20–50 nm range based on this TEM study, indicating that there is improvement room for the high temperature TPR synthesis process. If the precursors can be anchored inside the pores, much higher catalytic surface areas can be achieved for these catalysts. As for the Ni 2 P-SiO 2 catalyst, Ni 2 P attached the support as bigger chunks which have the nano size larger than 200 nm. In any case, it is very interesting to see once again that phosphides are attached but present on the surface of the supports and not in the pores. The corresponding element mappings of Ni 12 P 5 -Al 2 O 3 shown in Fig. 5(D) confirmed that the elements of Ni and P are uniformly present on the surface of catalyst support. 3.2. Catalytic performance 3.2.1. CO 2 -methanation-RWGS test The catalytic behaviour of the prepared catalysts in terms of CO 2 conversion as a function of temperature is shown in Fig. 6 (A), whereas CO and CH 4 selectivities are shown in Fig. 6(B) and (C) respectively. All the studied catalysts are active for CO 2 upgrading via RWGS in the temperature range 300–750 ◦C. However, Ni 2 P-SiO 2 is only active in RWGS reaction while Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl catalysts show high activities for both CO 2 methanation and RWGS reaction. The CO yield is shown in Fig. 6 (D); Ni 2 P-SiO 2 exhibited higher CO yield in the 300–600 ◦C range. Compared to Ni 12 P 5 -CeAl and Ni 12 P 5 -Al 2 O 3 , Ni 2 P/SiO 2 is a more selective catalyst for RWGS, producing 80% CO in the full temperature range of 350–750 ◦C. This indicates a high degree of suppression of the methanation reaction by Ni 2 P-SiO 2 catalyst, as Ni catalysts are known for their high methanation activity. This is shown to not simply be a site blocking effect; the nickel phosphide phase and SiO 2 support play key roles to explain the better selectivity. A similar CO selectivity towards RWGS reaction has been observed in the Ni 2 P-CeO 2 catalyst synthesized by Cui et al. Despite Ni catalysts have been demonstrated to be highly active to CO 2 methanation, Ni 2 P-CeO 2 presented high CO selectivity, better than that of the conventional Ni/CeO 2 catalyst [60]. Their results of the CO 2 -TPD and H 2 -TPD analysis indicated that Ni 2 P/CeO 2 had a moderate CO 2 adsorption strength and a strong adsorption ability for H species. The moderately adsorbed CO 2 species were easily desorbed, on the contrary, the hydrogen species were strongly adsorbed at low reaction temperatures, preventing the CO 2 from further hydrogenating to CH 4 , which lead to the remarkable CO selectivity. For the influence of SiO 2 support, Riani et al. explored the performance of Ni-SiO 2 /Al 2 O 3 catalyst for CO 2 methanation, and they found that the catalytic performance for CO 2 methanation decreased with the addition of SiO 2 to the Al 2 O 3 support due to the lowered CO 2 adsorption [61]. Therefore, the combination of Ni 2 P and SiO 2 support suppresses CO 2 methanation, this being an important finding for RWGS catalyst development. Ni 12 P 5 -CeAl and Ni 12 P 5 -Al 2 O 3 catalysts are also of interest as switchable catalysts due to their switchover from methanation at low temperatures to rWGS at higher temperatures. This behaviour in itself is not unique (it is predicted by thermodynamics), but the relatively low temperature light-off behaviour observed on these catalysts demonstrates good kinetics, which warrants further exploration of their use in a flexible CO and CH 4 production scenario. Ni 12 P 5 -CeAl exhibits higher CO 2 conversion in the whole temperature range, particularly in CO 2 methanation at 350–550 ◦C. The CO 2 conversion on Ni 12 P 5 -CeAl at 300 ◦C (52%) is almost double that of Ni 12 P 5 -Al 2 O 3 (27%). This support effect is similar to what has been observed in CO 2 methanation with Ni/ CeO 2 and Ni/Al 2 O 3 catalysts [62]. Ni/CeO 2 showed high CO 2 conversion especially at low temperatures compared to Ni/Al 2 O 3 . This is because the amount of CO 2 adsorbed onto Ni/CeO 2 was much larger than that on Ni/Al 2 O 3 , leading to the improvement of CO 2 methanation [54]. In our case, it is considered that the coverage of CO 2 increased with the addition of CeO 2 in the support, resulting in a higher CO 2 conversion over Ni 12 P 5 -CeAl. In addition, the oxygen vacancies created in the ceria lattice could also enhance the catalytic activity for CO 2 methanation, being considered as preferential sites for CO 2 activation [63]. Apart from the effect of supports, it is worth noting that the concentration of Ni 12 P 5 on Ni 12 P 5 -CeAl surface (38%) is larger than of Ni 12 P 5 -Al 2 O 3 (26%, as can be seen in table 1). Since the Ni 12 P 5 is considered as an active phase for CO 2 hydrogenation, the Ni 12 P 5 concentration could be another reason why the CO 2 conversion over Ni 12 P 5 -CeAl is higher than over Ni 12 P 5 - Al 2 O 3 . In terms of CO and CH 4 selectivities, both Ni 12 P 5 -CeAl and Ni 12 P 5 - Al 2 O 3 catalysts display 100% CH 4 selectivity from 300 ◦C to 450 ◦C. When we move to the RWGS range 600–750 ◦C, both catalysts also reached highly remarkable CO selectivity. Herein, a faster selectivity switch from CH 4 to CO is achieved by the Ni 12 P 5 -CeAl catalyst. As can be Table 1 Binding energies and atomic proportion of relevant species of the studied samples. Samples Ni 2p 3/2 (eV) P 2p (eV) Ni/P Ni x P content Ni-P (Ni 12 P 5 ) Ni-P (Ni 2 P) Ni 2+ P 5+ Ni-P(Ni 12 P 5 ) Ni 2 P-SiO 2 – 853.9 (40%) 856.8 (60%) 133.8 – 1.4 0.5% Ni 12 P 5 -Al 2 O 3 852.9 (26%) – 856.5 (74%) 134.2 (91%) 129.1 (9%) 0.67 5.15% Ni 12 P 5 -CeAl 852.7 (38%) – 856 (62%) 133.5 (85%) 129.1 (15%) 0.63 11.95% Fig.4. H 2 – Temperature-programmed reduction (TPR) results for the precursors of Ni 2 P-SiO 2 , Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl. Q. Zhang et al. Fuel 323 (2022) 124301 6 seen from Fig. 6 (B) (C): CO 2 methanation decreased from 500 ◦C and RWGS started to become the dominant reaction. The CO selectivity of Ni 12 P 5 -CeAl increased from 10% to 30% and the CH 4 selectivity decreased from 97% to 73% when the temperature changed from 500 ◦C to 550 ◦C, while the CO selectivity of Ni 12 P 5 -Al 2 O 3 only increased by 10% and CH 4 selectivity only decreased by 5%. In addition, it is worth noting that Ni 12 P 5 -CeAl shows higher CO selectivity than Ni 12 P 5 -Al 2 O 3 during the whole high temperature range (500 ◦C −750 ◦C). The better performance towards Ni 12 P 5 -CeAl for RWGS reaction could be attributed to the excellent redox properties of CeO 2 , which enhances the Ni reducibility. CeO 2 participates the RWGS reaction as it does in the forward shift process, being one of the most promising supports for this reaction thus yielding advanced catalytic performance [64–67]. The result is also in agreement with our XPS results. Overall, both the support and the phosphide structure appear to be intimately connected with methanation vs. RWGS activity. In order to Fig. 5. TEM micrographs of (A) Ni 2 P-SiO 2 ; (B) Ni 12 P 5 -Al 2 O 3 ; (C) Ni 12 P 5 -CeAl (D) SEM-EDX micrographs of Ni 12 P 5 -Al 2 O 3. Q. Zhang et al. Fuel 323 (2022) 124301 7 further investigate the features of each catalyst, H 2 -TPR, TPO, BET, TEM and stability tests were conducted and discussed in the following part. 3.2.2. Stability study for switchable catalysts Our findings indicate that Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl demonstrate good kinetics for both methanation and RWGS. We investigated the use of the Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl for switchable operation, in a scenario where renewable hydrogen would be used to synthesise a variety of chemicals via the syngas route in addition to the direct synthesis of methane as a fuel. For such applications long-term runs are essential in catalysts’ design and the stability is a major performance indicator, especially when novel formulations are developed. Ni 12 P 5 - CeAl and Ni 12 P 5 -Al 2 O 3 catalysts have been subjected to long-term stability tests under methanation and RWGS conditions, simulating a switchable operation. The stability tests were conducted over four consecutive 24 h cycles. The temperature was switched from 375 ◦C to 700 ◦C to simulate the change between a methanation process and a RWGS unit under the operating conditions. Fig. 7 (A) (B) show the conversion of CO 2 and associated products selectivity recorded during stability tests. It is clear that both catalysts exhibit very stable performance after 96 h of continuous operation in terms of CO 2 conversion and CO&CH 4 selectivities. For Ni 12 P 5 -Al 2 O 3 (Fig. 7 (A)), the CO 2 conversion was maintained of at 68% in the RWGS zone and increased from 52% to 56% in the CO 2 methanation zone. Meanwhile, the production of CO and CH 4 were very stable. <1% stability changes were witnessed between the two CO 2 methanation and rWGS reaction cycles. For Ni 12 P 5 -CeAl (Fig. 7 (B)), the CO 2 conversion in the RWGS zone (67%) is nearly identical to the value of Ni 12 P 5 -Al 2 O 3 , but the CO 2 conversion in CO 2 methanation zone towards Ni 12 P 5 -CeAl is clearly higher than that of Ni 12 P 5 -Al 2 O 3 , which is in agreement with our temperature screening test. The CO selectivity in the RWGS zone and the CH 4 selectivity in the CO 2 methanation zone were maintained constant (98% and 86%) respectively. It is also worth noting that a slight decrease of CO 2 conversion happened in the first CO 2 methanation cycle, but after one cycle RWGS test, the CO 2 conversion increased to an even higher value in the second CO 2 methanation cycle suggesting the catalyst had stabilised. Overall, our catalysts display excellent stability under both reaction conditions, validating their switchable behaviour – a very promising result allowing end-product flexibility. Considering about the performance during whole stability process, CeAl is a better support than Al 2 O 3 for Ni 12 P 5 catalysts because of the higher catalytic activity exhibited in the CO 2 methanation zone. The performances of some Ni catalysts reported recently in the literature are listed in the Table 2 below. As shown in Table 2, our catalysts show comparable performance as several nickel catalysts under the same temperature window and reaction mixture (H 2 :CO 2 =4). Our group has previously synthesized and explored the CO 2 hydrogenation activity of Ni/Al 2 O 3 and Ni/CeAl [11]. The trend of CO 2 conversion and 300 350 400 450 500 550 600 650 700 750 0 10 20 30 40 50 60 70 80 90 100 CO2 conversion (%) Temperature ( o C) Ni2P-SiO2 Ni12P5-Al2O3 Ni12P5-CeAl (A) 300 350 400 450 500 550 600 650 700 750 0 10 20 30 40 50 60 70 80 90 100 CO selectivity (%) Temperature (oC) Ni2P-SiO2 Ni12P5-Al2O3 Ni12P5-CeAl (B) Fig. 6. (A) CO 2 conversion (B) CO selectivity (C) CH 4 selectivity and (D) CO yield for Ni 2 P-SiO 2 , Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl. Condition: H 2 : CO 2 =4:1, WHSV = 12000 mL g −1 h −1 , T: 300–750 ◦C. Fig. 7. Stability test at 375 ◦C and 700 ◦C, WHSV of 12,000 mL/(g⋅h) with a H 2 : CO 2 ratio of 4:1 for (A) Ni 12 P 5 -Al 2 O 3 (B) Ni 12 P 5 -CeAl. Q. Zhang et al. Fuel 323 (2022) 124301 8 CO selectivity towards Ni/CeAl and Ni/Al catalysts is in line with this work. With the addition of CeO 2 in support, both the CO 2 conversion and CO selectivity increased. However, even for the better one (Ni/CeAl), the Ni 12 P 5 -CeAl catalysts in this work exhibited higher CO 2 activity for both CO 2 methanation and RWGS reaction compared to the Ni/CeAl catalyst in previous work. When considering selectivity, 1%NiCo@SiO 2 reported in ref [68] display competitive selectivity (greater than80%), which is also as the same level as the CO selectivity over Ni 2 P-SiO 2 in our work. 3.2.3. Stability study for Ni 2 P-SiO 2 Since Ni 2 P-SiO 2 is an excellent catalyst for RWGS leading to the total suppression of side reactions, a 48 h stability test was also conducted for Ni 2 P-SiO 2 with a H 2 : CO 2 ratio of 4:1 at WHSV of 12,000 mL g −1 h −1 . As the results show in Fig. 8 (A) (B), the CO 2 conversion dramatically declined from 48% to 32% in the first 5 h. After 5 h, the trend gradually flattened, but the decrease still continued. At the end of the 48 h test, the CO 2 conversion halved from 48% to 23%. Meanwhile, CO selectivity increased and CH 4 selectivity decreased during the process. 3.3. Post-reaction characterisation of catalysts 3.3.1. TPO results TPO experiments of the catalysts after stability tests were performed and the evolution of CO 2 signal (m/z =44) was analysed by MS, and the results are shown in Fig. 9. According to previous studies, certain temperature ranges of CO 2 peaks can be assigned to the different type of carbonaceous species. The peaks appeared lower than 380 ◦C are attributed to monoatomic and polymeric carbon [70,71], which are the active intermediates in the CO 2 -methanation-RWGS reaction. The second range peaks between 440 ◦C and 640 ◦C are assigned to whisker carbon formed on or close to Ni particles [72]. In general, the most stable carbon is the graphitic carbon formed on the support (temperature range: TPO greater than 650 ◦C), which does not appear in these Table 2 Catalyst performance comparison with materials reported in the literature. Catalysts H 2 :CO 2 WHSV (mL g −1 h −1 ) Temperature ( o C) CO 2 conversion CO selectivity CH 4 selectivity Ref Ni 2 P-SiO 2 4 12,000 650 58% 84% 6% This work Ni 12 P 5 -Al 2 O 3 61% 58% 40% Ni 12 P 5 -CeAl 68% 77% 15% Ni/Al 2 O 3 4 30,000 650 57% 64% 36% [11] Ni/CeAl 30,000 650 63% 75% 24% 1%NiCo@ SiO 2 4 15,000 650 65% 82% 6% [68] 2%NiCo@ SiO 2 65% 80% 11% NiCu-Saponite 4 15,000 650 68% 85% 8% [69] Ni-Saponite 62% 77% 12% NiCo-Saponite 61% 78% 11% Fig.8. Stability test at 550 ◦C, WHSV of 12,000 mL/(g⋅h) with a H 2 : CO 2 ratio of 4:1 for Ni 2 P-SiO 2 (A) CO 2 conversion for Ni 2 P/SiO 2 (B) CO&CH 4 selectivity for Ni 2 P/SiO 2 . Fig. 9. Temperature-programmed oxidation (TPO) results for post-stability Ni 2 P-SiO 2 , Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl. Q. Zhang et al. Fuel 323 (2022) 124301 9 three catalysts [73,74]. Since the carbon formed on and near the metal could be burned at lower temperature than the carbon formed on the support, the carbonaceous species in this case can be removed from the catalysts surface at a comparatively low temperature (below 600 ◦C). 3.3.2. TGA results In order to further quantify the carbon deposition, TGA tests were carried out and the results are shown in Fig. 10. All the samples used in this section are post stability samples. Ni 2 P-SiO 2 system was tested for 48 h at 550 ◦C, WHSV of 12,000 mL/(g⋅h) with a H 2 : CO 2 ratio of 4:1 while Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples were tested for 96 h at 375 ◦C and 700 ◦C (48 h for each temperature) under same inlet gas condition. As shown in Fig. 11, most of the carbon combustion happens in the temp range 0–300 ◦C. For the Ni 2 P-SiO 2 catalyst, we observe that the weight loss caused by coking is 5% (Fig. 10A). The same value for Ni 12 P 5 -Al 2 O 3 and N 12 P 5 -CeAl samples are around 7% and 5.5% (Fig. 10B, 10C). Since the Ni 2 P-SiO 2 sample has only been tested for 48 h, half of the stability time of Ni 12 P 5 -Al 2 O 3 and N 12 P 5 -CeAl samples, we conclude that Ni 12 P 5 -CeAl exhibits the best performance in terms of the carbon deposition resistance. When the temperature reached at 700 ◦C, there are small mass increases for all these three samples, which could be assigned to the oxidation of nickel phosphide phase. Overall, the carbon deposition happening on the surface of the synthesized catalysts is limited and the deposited carbon could be easily removed (<300 ◦C) indicating the suitability of these catalysts for longterm operations. 3.3.3. XRD patterns for post reaction and stability test samples To investigate the phase changes happened during the CO 2 -methanation-RWGS reaction and stability tests, the XRD patterns of the catalysts after CO 2 -methanation-RWGS reaction and stability tests are shown in Fig. 11 (A) and (B) respectively. Apart from the SiO 2 , Al 2 O 3 and Ni 12 P 5 compounds observed prior to reaction (Fig. 2), all diffraction peaks labelled by yellow dots in Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples should be attributed to Ni 3 P (JCPDS No. 34–0501) [75,76]. For the post reaction samples, very interestingly, despite the initial differences in their initial structure, catalytic testing induces a phase change in all studied catalysts. Ni 2 P evolves into Ni 12 P 5 in Ni 2 P-SiO 2 catalyst while most diffraction peaks of Ni 12 P 5 phase disappear and Ni 3 P became the main phase in spent Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples. The phase changes could be due to the increased temperature during the catalytic test (up to 750 ◦C in RWGS), which is higher than the reduction temperature in the synthesis (650 ◦C). According to the TPR results, the reduction of nickel phosphate continued to 750 ◦C, hence, these catalysts have been reduced further with the constant H 2 input and the elevated temperature of 750 ◦C. Also, it is worth noting that metal-rich phase appeared after CO 2 -methanation-RWGS reaction. According to the previous research [77], during the temperature-programmed reduction of nickel phosphate, PH 3 and P n+ species appeared in the gas flows from very low temperatures (200 ◦C) and increased significantly above 510 ◦C. PH 3 is formed by the reduction of phosphate precursors in H 2 , and the reaction of PH 3 with H 2 O yields P n+ species. PH 3 was involved in the formation of phosphide phases. Hence, the disproportionation of the phosphide can be attributed to the generation of PH 3 and P n+ species from the continuous H 2 reduction in RWGS. However, stabilised performance was observed during extended testing so it is worth noting that the loss of P may not be continuous. Compared with the XRD pattern of post reaction Ni 2 P-SiO 2 , there is no big difference observed in the XRD of post stability sample. However, the CO 2 conversion over Ni 2 P-SiO 2 decreased rapidly in the initial hours of the stability test. For the post-stability Ni 12 P 5 -Al 2 O 3 and Ni 12 P 5 -CeAl samples, more Ni 12 P 5 peaks disappeared compared to XRD patterns of the post reaction samples, indicating further reduction from Ni 12 P 5 phase to Ni 3 P happened during the long-term run. Such further reduction increases Ni electronic density favouring CO 2 activation by electron donation to the π * antibonding orbitals of CO 2 weaking the C-O bond and favouring both methanation and RWGS reactions. Indeed, for similar reasons, Ni 3 P has been shown to be highly active in the hydrodeoxygenation reaction where an electronically rich Ni catalyses the C-O and O–H bonds cleavage [78]. In addition, compared to the XRD of fresh Ni 12 P 5 -CeAl (Fig. 2), crystalline CeO 2 disappeared after the stability test, indicating that sintering happened during the stability test. Based on the XRD test of the post reaction and stability samples, it is worth pointing out that these catalysts are interesting candidates for operando spectroscopic investigations to better understand the key active phases associated with RWGS and methanation activity and for improved rational design by stabilising these active sites. Our group has explored the mechanism of RWGS towards Ni 2 P catalysts by using density functional theory (DFT) research previously. The results show that Ni 2 P favours the COOH-mediated path and the reaction step for the formation of COOH* intermediate may be considered as the main ratedetermining step on Ni 2 P (0001) in RWGS reaction. As for the formation of CH 4 , CO hydrogenates to form the CHO* intermediate, then dissociates to form CH* and O* species on the surface. Here the CH* species gets sequentially hydrogenated which leads to formation of CH 4 on the Ni 2 P (0001) surface [79]. In addition, it is widely accepted that the further hydrogenation and dissociation ability of CO rather than CO 2 on the catalyst affects the formation of the final product [80,81]. The mechanism of CO&CH 4 formation towards Ni 12 P 5 surface will be included in our ongoing projects. In addition, in order to achieve better control over phosphide phase and particle size irrespective of support, colloidal synthesis techniques can be used in the future work, where the phosphide nanoparticles are first synthesised in solution and then impregnated onto different supports [82]. This method has been used to successfully synthesize MoP catalysts for CO 2 hydrogenation [83]. 4. Conclusions This work showcases a strategy to design highly effective CO 2 conversion catalysts for combined CO 2 -methanation-RWGS schemes. Such strategy relies on the presence of an adequate switchable catalyst able to efficiently catalyse both reactions. Herein, we report innovative Ni phosphide-based systems with high CO 2 conversion activity and remarkable selectivity towards the desired end products (CH 4 and CO). We observe a strong influence of the support, which plays a role in both Fig. 10. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) for post stability (A) Ni 2 P-SiO 2 , (B) Ni 12 P 5 -Al 2 O 3 and (C) Ni 12 P 5 -CeAl. Q. Zhang et al.