Alkali Promotion in the Formation of CH4 from CO2 and Renewably Produced H2 over Supported Ni Catalysts

In Power‐to‐Methane (PtM) plants, the renewable electricity supply can be stabilized by using green electrons to produce H2 via H2O electrolysis, which is subsequently used to hydrogenate CO2 into CH4. In this work PtM is studied in a cascade fashion, from simulated solar light to methane production in an all‐in‐one setup, which was newly developed for this work. This setup was used to assess the effects of H2 stream purity on the activity of Ni/SiO2 catalysts in CO2 methanation. An activity effect in downstream methanation is shown to be onset by aerosols that evolve from the electrochemical splitting of water. Small amounts of K are shown to affect CH4 production positively, but only if they are deposited in situ, via KOH aerosols. K‐doped Ni/SiO2 catalysts prepared in an ex situ manner, by impregnation with a KOH solution, showed a decrease in activity, while the same amount of KOH was deposited. Operando FT‐IR spectroscopy reveals that increased back‐donation to CO‐containing intermediates and carbonates formation likely causes catalyst deactivation in ex situ samples as often reported in literature for Ni/SiO2 catalysts. The mechanism for in situ promotion is either an increased rate in the hydrogenation of CHx (X=0–3) fragments, or a more facile water formation or desorption as CO‐containing reaction intermediates are unaffected by in situ promotion. These results are relevant to PtM from a fundamental standpoint explaining the effect of potassium on nickel methanation, but also from a practical standpoint as the presented effect of in situ promotion is difficult to achieve via standard synthesis methods.


Introduction
The cascade synthesis of CH 4 from solar light (i. e., Power-to-Methane, PtM) is an attractive concept that will allow us to decrease CO 2 emissions, simultaneously demodulating the mismatch in renewable electricity demand and supply and stabilizing the electricity grid. [1,2] In the PtM concept first renewably produced electrons are used to split H 2 O into H 2 and O 2 . While this renewable H 2 can, in theory, be used as an energy buffer, storing H 2 is approximately an order of magnitude more costly than storing CH 4 , and thus particularly for long-term (seasonal) storage of electricity, is not necessarily costefficient. [3] Hence the conversion of this H 2 into CH 4 (also called e-gas), which can be stored safely in large quantities through infrastructure that already exists, becomes interesting from an economic standpoint. [3][4][5][6] Furthermore, methanation is already a crucial element of any engineering solution involving on-site H 2 production through low-temperature catalytic reforming of organic substrates (e. g. alcohols, formic acid). [7] To the best of our knowledge, at the time of writing, the Audi e-gas plant in Wertle, Germany is the largest commercial methanation plant worldwide and produces methane starting from CO 2 captured from close by produced biogas via amine scrubbing, and from H 2 generated by three 2 MW KOH-based alkaline electrolyzers powered by renewable energy. In this plant, the hydrogen produced from the electrolysis is filtered to eliminate KOH aerosols, and subsequently dehydrated and compressed to 10 bar to be stored in a buffer tank worth 1 h of methanator operation, decoupling the operation between the electrolyzer and the methanator units. The produced methane is dried and fed into the natural gas grid. Water is cycled back in the electrolyzers, while oxygen produced at the anode of the electrolyzer is vented out. The methanation reactor is cooled by molten salts and the heat is used to regenerate the amine scrubber, bringing the process efficiency from 54 % (without heat recovery) to 72 %. [3] Despite the high efficiency, the process is not yet profitable, mainly due to electricity costs and fees, showcasing how cheap electricity, but also policies can play a big role for the commercialization of PtM technologies. As described for the operating PtM plant above, several important steps occur for the application of this PtM process, which are illustrated separately in Figure 1A. Solar light is first converted into electrons via photovoltaics (step 1). Then, these green electrons are converted into H 2 via electrolysis of H 2 O (step 2, both the Hydrogen Evolution Reaction (HER) and Oxygen Evolution Reaction (OER) have to be considered here). [8,9] CO 2 streams are captured and purified (step 3), and can then be converted into CH 4 via renewably made H 2 (step 4 in Figure 1). This CH 4 is then stored and distributed (within a closed-cycle process, step 5). CO 2 reduction to methane may be regarded as an established process as French chemist Paul Sabatier (1845-1941) discovered the reaction over 100 years ago. [10,11] Nevertheless, there are novel questions that arise from the application of this process within the concept of electrification of the chemical industry described above. An obvious question is that of the effect of sequential process steps on one another. To study such questions in detail, a setup was built bringing together all elements of this process in a single, newly constructed experimental setup, combining simulated solar light with an electrolyzer to split water, and a methanation unit. Photographs of this setup can be seen in Figure S1.
The overall process efficiency of the PtM process is highlighted in Figure 1B,C. Figure 1B highlights the theoretical values of power storage (56 %) and subsequent power generation (36 %), as calculated for an optimized production plant. In Figure 1C we highlight the all-in-one setup, developed in our laboratory, that will be discussed in this work and is shown in Figure S1. The yields of each step in our (non-optimized) labscale setup are shown in Figure 1C. Obviously, the setup that we will discuss in this work is far from what an optimized industrial plant would be, mainly due to electrode size constraints and inefficient heating of the methanation reactor, but also because the kinetic conditions applied (i. e. low CO 2 conversion) were purposely chosen for the final catalytic methanation step to study activity trends. The applied setup has an overall efficiency of 0.018 %. As a comparison, the theoretical maximum efficiency for photosynthesis is 11 %. However, most plants yield an efficiency of approximately 0.1-3 % as plants do not absorb all incoming radiation. [12] Although the comparison between our setup and natural photosynthesis is not entirely fair, as plants convert CO 2 generally into much more useful or valuable, complex carbohydrates, it illustrates the efficiency of nature in a process that has been optimized over hundreds of millions of years.
In this work we examine the effect of hydrogen production (step 2 in Figure 1A) on methane production (step 4 in Figure 1A). More specifically, it is interesting to consider the possible presence of aerosols in the H 2 feed, which may evolve from the HER in alkaline medium. This should be done, as NASA literature from 1974 already reports the possible presence of aerosols in the feed of electrolysis systems, [13] and industrially often aerosol filters are applied as mentioned above.
The water splitting reaction is performed in alkaline media, as the kinetically limiting step in the electrolysis of water -the OER -benefits from a high pH. [14] Hence the aerosols that may be present in the gas feed from the split water may contain alkaline material. Literature shows both promoting, and deactivating effects for the addition of alkaline dopants to thermocatalytic methanation reactions, [15][16][17][18][19][20][21][22][23] and relevant reaction steps thereof, such as the Reverse Water-Gas Shift (RWGS) reaction. Table S1 lists some of this relevant literature, and their normalized promotional or deactivating effect on the turnover frequency of CO and CO 2 methanation. In general, alkali promotors (most commonly potassium) are added to methanation catalysts either to poison support acidity, or to catalyze coke removal via hydrogen or steam reactions. [15] Xu et al. show a "volcano-shape-curve" between the Mg/Al molar ratio and catalytic activity, explained by an optimal surface basicity induced by the Mg. [19] The activity and selectivity of K promoted nickel catalysts seem to mainly depend on the support. [20] On Ni/Al 2 O 3 a promotional effect is found, whereas on Ni/SiO 2 , an exponential decrease in total activity with increased promotor concentration is seen which is claimed not to be due to a decrease in the dispersion. [20] Campbell et al. also note an increase in activity for Al 2 O 3 -supported Ni, but a decrease in activity for SiO 2 supported Ni. [23] While Huang et al. find a promotional effect for Ni/SiO 2 . [15] Thus, fundamentally understanding the nature of this potential promotional effect is of direct practical and academic relevance.
In this work we will study the effect of K promotion on Ni/ SiO 2 catalysts of different mean nickel particle sizes, as well as different forms of addition of K to the Ni/SiO 2 catalysts via operando FT-IR spectroscopy, and the use of XRD, STEM-EDX and ICP-AES. To the best of our knowledge, no systematic study has been performed to distinguish the possible particle-size dependent effect of alkali promotion, while it seems a relevant variable in Table S1, and no study has considered the effect of in situ deposition of potassium onto a Ni-based methanation catalyst as a potential promotor. Table 1 lists an overview of the fresh, or "parent" Ni/SiO 2 catalyst samples that are studied in this work.

Catalyst materials
The Supporting information (e. g. Table S2), and previous literature [24] gives more information on the characterization of these materials. The effect of in situ doping with KOH-containing aerosols which evolve from hydrogen production on the catalyst samples listed in Table 1 was compared to an ex situ doping procedure where a 0.6 wt % K loading was achieved by impregnation, via suspension of the catalyst materials in a KOH solution and subsequent evaporation.

Hydrogen production via electrolysis and cascade methanation reaction
Pt mesh electrodes (MaTecK, 99.9 %, 25 × 50 mm, 3600 mesh/cm 2 , 0.04 mm wire diameter) interwoven with a Pt wire (MaTecK, 99.9 %, 1 mm diameter, 100 mm length) were suspended in a 1 M KOH electrolyte (ACS reagent, > 85 % pure,~15 % water) loaded into a homemade H-cell. This cell consists of GL14 glass tubes with an internal volume of 20 mL for the electrolyte on each side, resulting in 40 mL electrolyte in total. The HER and OER reactions are spatially separated by a Nafion 117 perfluorinated membrane (Aldrich, 0.007 in. thick). Prior to use the membrane is activated by immersion in nitric acid for several hours, while the Pt electrodes were cleaned by using a butane flame. The electrolyte was purged with 2 mL/min Ar, 0.1 mL/min Kr on the HER side and 2 mL/min N 2 , 0.1 mL/min Kr on the OER side for at least 2 h prior to the experiments.
The Ni/SiO 2 catalysts (3 mg, sieve fraction 75-150 μm) were loaded in a plug-flow reactor (borosilicate capillary, 1 mm diameter) and reduced at 550°C for 60 min, with a heating rate of 5°C/min under a flow of 1 mL/min of pure H 2 . Afterwards, the catalysts were cooled down to 400°C (5°C/min) under a flow of pure N 2 . This flow was maintained for 3 h to remove residual H 2 from the system. Electrochemistry (Ivium CompactStat) was then started potentiometrically at a current of 4 mA, requiring ca. 1.69 V potential (the Nafion membrane yields a high resistance in the system), 1.69 V also being the output of a small solar cell module of three cells in series. See also Figure S3 for chronopotentiometry of the water splitting with two Pt electrodes, and the GC response to the application of current. The flow of Ar/Kr and N 2 /Kr was constantly maintained as a carrier gas for the produced H 2 and O 2 . The feed from the HER was led over the plug flow reactor containing the Ni/ SiO 2 catalyst, optionally filtered by an aerosol filter.
After 15 min of flow from the H-cell an 800 ppm CO 2 feed was opened over the Plug Flow Reactor (PFR) at a rate of 1 mL/min and was mixed with the HER feed prior to injection on the PFR. The reactor output gas was again filtered by an aerosol filter before being injected on an InterScience CompactGC every 4 min and analyzed by FID. The flow was split over two column systems optimized for measure hydrocarbons on one end and CO/CO 2 /CH 4 on the other. The latter was equipped with a methanizer to visualize each component. On a separate channel the flow from the oxygen halve of the system was injected and measured by a TCD every 4 min. This state was maintained for several hours, up to 2 or 3 days depending on the experiment.

Aerosol measurement and formation
To measure the aerosol formation from the electrochemical cell, a Water-based Condensation Particle Counter (WCPC, TSI model 3785) was attached to the output of the cell, in place of the methanation reactor. This is shown in Figure S2. In a typical experiment, a 2 mL/min flow of N 2 was bubbled through the electrochemical cell filled with 1 M KOH, and the outlet stream was diluted with 1 L/min filtered air before being fed to the particle counter. The effect of flow, applied potential and KOH concentration was also studied, see Figure S4. Control experiments carried out bypassing the cell or using water instead of 1 M KOH didn't show any aerosol formation.

Operando FT-IR spectroscopy with on-line gas chromatography
Operando FT-IR spectroscopy was performed using a Bruker Tensor 37 FT-IR spectrometer and OPUS software with DTGS detector. [24] Spectra were recorded every 30 s. Gases were introduced with Brooks mass flow controllers. Self-supported catalyst wafers for operando transmission FT-IR spectroscopy were prepared using roughly 4 tons of pressure. The catalyst wafers (15 mg in weight) were loaded into a Specac high-temperature transmission FT-IR reaction cell, and subsequent in situ reduction was performed with 1 : 1 H 2 :N 2 flow of 20 mL/min each for 1 h at 550°C. Temperature increased at a rate of 5°C/min and atmospheric pressure was held throughout. The cell was cooled to 100°C before CO 2 hydrogenation commenced, after which temperature was increased at 5°C/min to 400°C where it was held for an hour. Gas rates used in CO 2 hydrogenation without water were 6.25 mL/min N 2 , 5.00 mL/ min H 2 , and 1.25 mL/min CO 2 . For experiments with water in the gas feed, water was fed through a stainless steel saturator, by bubbling inert gas. Milli-Q water was fed at 21°C, which at 6.00 mL/ min N 2 ensured 2.8 % (or 0.36 mL/min) was added to a feed of, 5.00 mL/min H 2 , and 1.25 mL/min CO 2 . Operando FT-IR experiments were performed at atmospheric pressure, with online gas chromatography. Global Analyst Solutions CompactGC 4.0 and Thermo Scientific Dionex Chromeleon 7 software were used for online gas chromatography, with FID and TCD detectors. Product and transients continued to be captured for 15 min after CO 2 and 5 min water saturator was closed until H 2 was closed, all while at 400°C.

KOH aerosol semi-quantitative characterization
The Ni/SiO 2 catalysts tested for up to 3 days in the methanation reaction using the H 2 produced by the HER showed a content of 0.6 � 0.2 wt.% K by ICP-AES analysis. Assuming that all K from the gas feed gets adsorbed on the catalyst, this would give a value of 4-11 μg/L KOH (4 mg of sample, 2 mL/min gas carrier flow).
Comparable to the KOH content reported by NASA (3-5 μg/L). [13] Considering pure KOH density (2.12 g/mL) we can calculate the volume (in mL) of KOH per mL:~2-5 × 10 À 9 . Which, divided by the observed number of particles (about 30 per mL in the flow analyzed in the counter, which is diluted 500 times with N 2 , corresponding to 15000 per mL in the starting 2 mL flow), gives a volume per particle of~0.3-0.6 × 10 À 12 mL, i. e.~0.3-0.6 μm 3 . Assuming a spherical shape, this corresponds to 0.4-0.5 μm radius,~1 μm diameter. [25] Setup efficiency calculation  Table 2, or Figure 1A for an overview.

Results and Discussion
We have studied a set of SiO 2 supported nickel catalysts, labelled from 1-6 in accordance with increasing particle size (see also Table 1). Previous literature [24] lists full details on the catalyst samples, which were characterized by several techniques such as H 2 chemisorption, N 2 physisorption, HAADF-STEM, temperature programmed reduction, X-ray absorption spectroscopy. A benchmark 1 M KOH alkaline solution is used for the production of renewable H 2 via H 2 O electrolysis, [14,26] and we focus on its effect on the catalytic hydrogenation of CO 2 into CH 4 over the mentioned catalyst samples. H 2 O splitting to H 2 and O 2 was performed with two Pt electrodes and done in a stable manner at 1.69 V at a current of 4 mA. O 2 production was used to control the water splitting efficiency, and was stable at a Faradaic efficiency of 99.6 % ( Figure S3). By use of a H 2 O-based condensation particle counter, which is shown in Figure S2, the presence of aerosols could be detected in the HER-produced H 2 feed. The concentration of aerosols formed is shown in Figure 2A. Figure 2B shows that the hydrogen produced by electrolysis has an effect on the activity of the sample shown (catalyst sample 5 in Table 1). More specifically, the H 2 produced via electrolysis, seemed to have a promoting effect on the activity of the Ni/ SiO 2 catalyst in CO 2 methanation at 400°C and ambient pressure, as shown in Figure 2B, when compared to clean hydrogen from the bottle under the same reaction conditions. In Figure 3, the effect of the H 2 from HER on different Ni/ SiO 2 catalyst samples with varying mean nickel particle sizes is shown, as tested in the all-in-one setup. The investigated Ni/ SiO 2 catalysts show stable methanation activity at 400°C, over a prolonged period of time when pure H 2 from a cylinder is used as a feedstock, but this changes when H 2 from the HER is used. The observed turnover frequency (TOF) values using "clean" hydrogen in this setup are comparable to previously shown results, which were measured in a different setup, an operando Table 2. Summary of energy input and output of each of the three steps with the energy efficiency per step as well as the total efficiency in the last column.
Step FT-IR spectroscopy setup. [24] Additional activity data is given in Figures S5-S8. Interestingly, when the reaction proceeds with H 2 produced via the HER, i. e. with alkaline aerosols, some trends are observed for catalysts with different mean nickel metal nanoparticle sizes. Over catalyst sample 5, which has 4.4 nm mean nickel particle size diameter supported on SiO 2 , a strong transient increase in methanation activity is observed, followed by a decay in activity over time, which nonetheless remains considerably higher than the values observed using pure H 2 (Figure 3 middle). On the other hand, over larger nickel metal nanoparticles (i. e., catalyst 6, 6.0 nm mean nickel particle size), no promotion is observed and the catalyst steadily deactivates over time (Figure 3, bottom). The smallest mean nickel metal nanoparticle size under study, catalyst 1, 1.2 nm Ni/SiO 2 , shows increased activity as well (Figure 3, top) but less than catalyst 5 sample. Figure 2A also shows that when a filter is placed before the methanation reactor, the aerosol can be effectively removed from the H 2 stream. In order to rule out a possible direct effect of the aerosol on the catalytic activity, e. g. by reaction with CO 2 in the gas-phase leading to carbonate formation (which is known to occur in mere seconds in the case of NaOH [27] ), an aerosol filter was installed on a by-pass between the electrochemical cell and the methanation unit (before the CO 2 feed), and the H 2 feed was switched from unfiltered to filtered during the reaction. Notably, the activity promotion on catalyst 5 (4.4 nm Ni/SiO 2 ) was maintained, which strongly suggests that the aerosol itself does not play an active role in the activity enhancement, which therefore will be solely due to K deposition on the catalyst. Strikingly, this also means that one can dose just the right amount of KOH to promote methanation activity and then remove the aerosol at will, to avoid further K accumulation, which will eventually be detrimental to the activity.  In these experiments, TOF values are proportional to activity as the particle size distribution does not change significantly for the spent catalysts using filtered ("clean") H 2 , and unfiltered H 2 (containing aerosols), as shown by TEM analysis in Figures S8-S10. Quite obviously, impurities in the H 2 feed from HER have an effect on the downstream catalytic methanation reaction. As a result, 0.6 � 0.2 wt.% K was observed by ICP-AES analysis of the Ni/SiO 2 catalysts after 2 days of reaction when using HERproduced H 2 , while on the undoped sample, and the catalyst tested with pure H 2 K amount was below the detection limit. These values correspond to expected KOH concentrations with respect to aerosol particle sizes in accordance with literature (4-11 μg/L KOH, 1 μm diameter particles, see the experimental section). [25,27] Furthermore, it should be noted that the particle count is proportional to the gas flow through the electrocatalytic cell, while it did not detectably vary with the potential that is applied to carry out the HER ( Figure S3). Based on ICP-AES and BET results, one can hereby calculate an average K + concentration on the surface of the catalysts of 0.2 K + /nm 2 . Considering a K + weight loading of 0.6 % and the measured BET surface area of the Ni/SiO 2 samples (482.7 m 2 /g), one can calculate the average distribution of K + on the catalyst surface. About 6 mg K + ions are deposited per g of catalyst, which corresponds to roughly 12.4 μg/m 2 , 0.3 μmol/m 2 , 0.2 K + atoms/nm 2 f or catalysts 1, 5 and 6. Assuming hemispherical nickel metal nanoparticles of 1.2, 4.4 and 6.0 nm in diameter, this gives a nickel surface area of about 2.5, 30 and 57 nm 2 per particle, i. e., 0.5 (implying about 1 K atom every two particles, on average), and 6 and 12 K + atoms/particle, respectively. Considering 10 nickel atoms/nm 2 , this gives a K : Ni surface atom ratio of 1/50. Accordingly, an effect on nickel activity at such low K/Ni ratios is consistent with results in literature on Ni(100) surfaces, for which CO and CO 2 methanation were affected starting from well-below Monolayer (ML) coverage (0.05 ML). [28,29] To benchmark the promotional effects that were reported in the literature, and those we have observed via in situ doping, K-doped Ni/SiO 2 catalyst samples 1-6 were prepared by impregnation using a KOH solution, to achieve a loading of 0.6 wt % K (i. e. comparable to the amount of K + deposited from the aerosol stream after 2 days). These samples will henceforth be called "ex situ doped" differentiated from the samples that are doped in situ via aerosol deposition. To prepare and simultaneously study in situ doped samples, a pure H 2 stream was bubbled through a saturator filled either with a 1 M KOH aqueous solution or with DI water.
To gain further insight into the nature of the (particle size dependent) effect of K on CO 2 methanation over nickel via both of these preparation methods, operando FT-IR spectroscopy was carried out at 400°C for the sets of undoped (catalyst 1-6), ex situ doped samples (ex situ doped catalysts 1-6), and in situ doped (in situ catalysts 1-6). Catalyst sample 6 has very high nickel weight loading (60 %) which makes the self-supported catalyst wafers of this sample needed to study FT-IR prone to breakage. The ex-situ catalyst preparation procedure apparently increased the fragility of this sample to an extent where it could not be studied in this setup. Figure 4A shows the TOF trends as measured for the different samples. Ex situ doping has a clear deactivating effect, while in situ doping has a positive effect on the activity of Ni/ SiO 2 catalysts in the methanation of CO 2 at 400°C. Furthermore, it can be seen in Figure 4A and 4B that the initial doping effect of the aerosol is higher than when it reaches steady state. This was also observed in the all-in-one setup as described above, and shown in e. g. Figure 2. It is important to note that no clear particle size trend to the in situ promotion effects can be observed. While this appeared to be the case for the three particle sizes shown in Figure 3, the full set of catalyst samples tested here shows less of a distinctive trend. The general differences in promotion/deactivation trends, however, that are shown in Figure 4 are reflected in the FT-IR spectra, of which examples are shown in Figure 5. The supporting information shows the full set of FT-IR spectra in Figures S11 and S12. Figure 5 shows the region from 1400-2200 cm À 1 of the FT-IR spectra recorded during CO 2 hydrogenation at 400°C for undoped, in situ, and ex situ doped catalyst samples 1, 4 and 5. In this region the CO stretching vibrations and also formate type species can be found, another important spectral feature which may be linked to the deactivating effect of ex situ doped catalysts. [24,30] Notably, ex situ doped samples show a lower relative intensity at around 2020 cm À 1 , which ascribes to linearly adsorbed CO atop a single nickel atom [31][32][33][34] and was shown to be an important descriptor for catalyst activity in Ni/SiO 2 in recent work. [30] Deactivation is accompanied by the disappearance of the CO (ads,top) signal with maximum at 2020 cm À 1 . CO is an important intermediate in the formation of methane via the RWGS reaction. It is then subsequently converted to methane either via H-assisted CO dissociation or intermediate carbide hydrogenation. The binding strength, and coverage of CO is thus crucial in the formation of methane. [24,30] A trend in intensity of CO ads-top can be observed in Figures 5AÀ C going from in-situ promoted > undoped > ex-situ doped catalysts. For the in situ promoted samples, this shift is much more subtle (if noticeable at all), while the change in activity is much more significant. Figures S13-S15 further show that no real trends can be ascribed to the effect of in situ doping on peak positions of CO ads species. Notably, a higher relative intensity of a broad feature around 1750 cm À 1 , ascribed to carbonate species, is observed for the ex situ samples. The detectable presence of these species only in the ex-situ doped samples, and the fact that they are not hydrogenated to formate species, suggests that their formation could be due to K 2 CO 3 formation over K deposits when exposed to air. In any case we can say that electronic effect of K on ex-situ doped Ni/ SiO 2 seems to be more significant, which results in more backdonation to CO, thus effectively shifting the signal towards lower wavenumbers, which is in line with recent literature. [21,22] The deactivating effect of K on Ni/SiO 2 is also shown in work by Campbell et al., in which K was only shown to deactivate Ni/ SiO 2 , even in very small amounts (0.05 wt.%). [20,23] Yet, also here, the catalysts were preimpregnated with a K-containing solution, or prepared via co-impregnation. In fact, each of the catalyst samples where a deactivating effect was observed (Table S1), was treated with hydrogen at high temperature in the activation procedure of the catalyst. Also in our case, the ex situ doped samples were prepared in such a way that the K and nickel underwent a reduction procedure together. This reduction procedure in the presence of K has an obvious effect already on the particle size of some catalysts as could be seen Figure 5. A) The TOF of Ni/SiO 2 catalysts of different mean nickel particle size in different states of KOH promotion (unpromoted, in situ, and ex situ doping) against the position of the peak maximum in operando FT-IR. The dotted lines are drawn as eye-guides, showing that with in-situ promotion the TOF increases but there is no significant change in the CO top position, and with ex-situ promotion there is no significant increase in the TOF, but there is a significant change in the CO top peak position. BÀ D) Operando FT-IR of the activity measurements shown in Figure 4, showing the catalysts of different mean nickel particle sizes B) catalyst 1, C) catalyst 4, D) catalyst 5) in undoped state, and doped in situ and ex situ with KOH. These spectra were baseline-subtracted and normalized to the peak at 1853 cm À 1 .
in the TEM analysis presented in Figure S9. This could be explained by an electronic interaction of K with Ni, or by the decomposition of KOH with temperature to yield K 2 O and water at around 400°C, [35] which is known to affect Ni sintering. [36] Another explanation could be the formation of bicarbonates after exposure to atmospheric CO 2 . Their decomposition to K 2 CO 3 , CO 2 and water proceeds at low temperature (100-200°C). [35] Poisoning via these strongly adsorbed species could inhibit the formation of product. During in-situ promotion, the reaction temperature is higher than the melting and decomposition temperature of KOH, which could explain why carbonates are not formed on the in situ doped catalysts (besides the low K concentration on the surface, as previously stated).
The effect of K on the activity could also be dependent on the K + distribution, as shown by recent observations in the coking resistance in catalytic methane steam reforming (the inverse reaction of CO 2 methanation) over Ni/Al 2 O 3 catalysts. [37] Yet no observable agglomeration could be found when the catalysts were examined by SEM-EDX ( Figure S16). When examined by higher resolution STEM-EDX ( Figure 6), there was also no observable difference in the comparison between the distribution of K throughout ex situ doped samples and in situ doped samples, which can be expected for such low amounts of K.
As no signs of agglomeration were detected by STEM-EDX and the operando FT-IR suggests a difference in electronic structure between in situ and ex situ doped samples, two plausible possibilities of the promotional effect of aerosols over ex situ promotion remain; first as already previously mentioned, alkali promotors are added to industrial methanation catalysts to suppress the formation of carbon deposits. The suppression of coke formation itself is not very relevant to our work as no aromatic carbon vibrations (as aromatics are often considered as precursor molecules for coke deposit formation) could be observed in FT-IR for promoted or unpromoted reactions, and the promotional effect is larger initially than at steady state. The second possibility explaining the promotional effect of the in situ deposition of KOH via aerosols is a mechanism where the KOH promotes the removal of oxidizing species on the nickel surface via the formation of water. Both of these mechanisms require further investigation.

Conclusions
By combining operando FT-IR spectroscopy and a well-defined set of Ni/SiO 2 catalysts, we have shown both alkali promotion and poisoning of carbon dioxide hydrogenation over Ni/SiO 2 catalysts by KOH. In situ deposition of KOH-containing aerosols, realized by constructing an all-in-one setup capable of combining the H 2 O electrolysis and CO 2 methanation steps, shows downstream promotion effects in the CO 2 methanation reaction. In contrast, the ex situ promotion of the Ni/SiO 2 catalysts with the same amount of KOH shows a deactivating effect for all catalyst samples under study. This deactivation can be rationalized in terms of the electronic effect of K in combination with the Sabatier principle, as ex situ promoted samples likely bind the reaction intermediate CO more strongly than is required. The mechanism for in situ promotion is either an increased rate in the hydrogenation of CH x (X = 0-3) fragments, or more facile water formation or desorption as the COcontaining reaction intermediates in FT-IR spectroscopy are unaffected by in situ promotion.
Practically this information translates to the application of a method of doping which is similar to dosing a certain amount of K via a H 2 feed coming directly from H 2 O electrolysis in KOH electrolytes. In this way, the catalytic activity of nickel metal nanoparticles supported on SiO 2 in the Sabatier reaction can be efficiently enhanced. Although the mechanism as to what this can be attributed to requires further investigation, the work shows how a small amount of alkali promoters can dramatically change catalyst performance in CO 2 methanation .  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57 Keywords: Power-to-methane · aerosol · alkali promotion · CO 2 hydrogenation · nickel