In situ investigation of the CO 2 methanation on carbon/ceria-supported Ni catalysts using modulation-excitation DRIFTS

The development of novel cost-efficient, high-performing catalysts for CO 2 methanation that are active at low temperatures can be optimized through the understanding of the reaction mechanism on different materials. A series of Ni-based catalysts supported on CeO 2 and carbon/CeO 2 composites was investigated, showing that Ni nanoparticles supported on a carbon/CeO 2 composite with a 50:50 wt ratio and on pure CeO 2 have excellent low-temperature activity and achieve up to 87% CO 2 conversion with full selectivity towards CH 4 at 370 ◦ C. Importantly, meaningful insights on the reaction mechanism were gathered for the different types of materials by using the emerging ME (cid:0) PSD (cid:0) DRIFTS technique. The study of the rate of formation/consumption of the various intermediates showed that the CO 2 methanation reaction follows a combination of the CO and formate pathways in the case of Ni on pure CeO 2 ; however, in the case of Ni on the carbon/CeO 2 composite, it follows only the formate pathway.


Introduction
The CO 2 methanation reaction (CO 2 + 4 H 2 → CH 4 + 2 H 2 O, ΔH 298 K = − 165 kJ mol − 1 ) is the most thermodynamically favorable reaction of CO 2 hydrogenation, and is a promising way to convert anthropogenic CO 2 back to CH 4 fuel [1][2][3][4].The hydrogenation proceeds through an eight-electron reduction process with consequent significant kinetic barriers, thus requiring the action of a catalyst.From the mechanistic perspective, extensive research has been conducted on CO 2 methanation.Nevertheless, the reaction mechanism is still vigorously debated, since it depends on the catalytic material and the hydrogenation conditions [1,3].The most predominant mechanisms proposed in the literature can be grouped in two different categories: (i) the dissociative pathway, where CO 2 is dissociated into CO, which acts as the main intermediate, and (ii) the associative pathway, where CO 2 is molecularly adsorbed associatively with a H atom previously adsorbed on the surface, forming formate as an intermediate, i.e., without CO formation [3].
CO 2 methanation has played a significant role in understanding CO 2 hydrogenation and the associated development of catalysts.Importantly, researchers continue to exploit different catalysts to achieve high CO 2 conversion and complete selectivity towards CH 4 .For example, a number of catalysts based on transition metals (Ni, Co, Fe) or platinumgroup metals (Ru, Rh) dispersed on porous supporting materials have been developed [2,[5][6][7][8][9][10][11][12].Notably, Ni-based catalysts, compared to platinum-group metals, exhibit an optimal combination of activity, selectivity, and low price [2,5,6].Despite the advances in using Ni catalysts for CO 2 methanation, a way to reduce the reaction temperature from the current ≥ 400 • C while preserving high catalytic performance remains a challenge.
One interesting approach to address this challenge is to develop Ni catalysts supported on novel materials.Traditionally, alumina (Al 2 O 3 ) is used as supporting material for Ni methanation catalysts [13][14][15].In contrast, we and other researchers have focused on carbon supporting materials [2,5,16,17].An exciting feature of these high-surface-area porous supports are their versatile surface properties [18].Specifically, via modification of their surface chemistry, carbon materials can exhibit increased basicity, which improves the CO 2 coverage of the catalyst, since it is an acidic molecule, thus enhancing the CO 2 methanation performance at lower reaction temperatures [2].Compared to the more common Ni supported on metal oxides, the research on Ni supported on carbon materials is less advanced, motivating a fundamental investigation of such alternative catalytic systems [2,5].Among the metal-oxide supporting materials, despite being based on a rare-earth metal, ceria (CeO 2 ) has emerged as a promising alternative to Al 2 O 3 because it enhances the metal-support interactions, leading to high dispersion of the resultant Ni nanoparticles over the support.Moreover, owing to the enhanced surface coverage by CO 2 species, enabled by the CeO 2 surface basicity, it improves the low-temperature catalytic performance of the catalysts in CO 2 methanation [19][20][21][22][23][24][25][26].These positive effects of CeO 2 motivated us to develop high-performing carbon/ceria-supported Ni catalysts for CO 2 methanation, as well as to conduct a comparative mechanistic study of CO 2 hydrogenation over these catalysts.
In this work, we report the successful precipitation synthesis of a composite supporting material consisting of reduced activated carbon (ACR) and CeO 2 (ACR-CeO 2 ) as well as pure CeO 2 , followed by deposition of Ni nanoparticles via the incipient wetness impregnation route.An important feature of our findings is the remarkable reduction by 80 • C of the optimal temperature of the CO 2 methanation reaction, compared to that for a CeO 2 -free Ni/ACR catalyst developed in our previous study [2], thus significantly improving the sustainability of the methanation approach without jeopardizing the catalytic performance.Furthermore, we conducted a detailed in situ modulation-excitation phase-sensitive-detection diffuse-reflectance infrared Fourier-transform spectroscopy (ME-PSD-DRIFTS) investigation of the methanation over the newly synthesized high-performing Ni/ACR-CeO 2 and Ni/CeO 2 catalysts, to obtain key mechanistic insights.

Synthesis
In the present work, we synthesized reduced activated carbon (ACR) with Lewis base sites by heat-treatment of a commercial NORIT GAC 1240 PLUS activated carbon (CABOT) at 900 • C for 1 h under H 2 flow [2].Pure CeO 2 and composite ACR-CeO 2 supporting materials were prepared by a precipitation method adapted from elsewhere [27].Briefly, Ce(NO 3 ) 3 ⋅6H 2 O is precipitated by NaOH under vigorous stirring, along with the addition of calculated amounts of ACR for the composite ACR-CeO 2 synthesis.The 15% Ni nanoparticles were deposited onto as-synthesized supporting materials via incipient wetness impregnation using Ni(NO 3 ) 3 ⋅6H 2 O, followed by heat treatment to decompose nitrate, thus generating metallic Ni (Ni/CeO 2 and Ni/ACR¡CeO 2 ).

Catalytic testing
The activity, selectivity, and stability of the synthesized catalysts in CO 2 methanation were evaluated in a Microactivity XS15 system (PID Eng & Tech), using a fixed bed quartz reactor (Ø int = 1 cm).Outflow products were analyzed online using a GC 1000 gas chromatograph (DANI), equipped with a thermal conductivity detector (TCD) and a GS-CarbonPLOT capillary column, while using He as the carrier gas and a continuous flow of N 2 as the internal standard.Each catalyst (100 mg) was diluted by inert SiC material, and then was pre-treated in situ under H 2 flow (40 cm 3 min − 1 ) at its reduction temperature for 30 min, at 1 bar.After that, the reactor temperature was decreased to 100 • C under He (50 cm 3 min − 1 ), and then, the reactor was fed with 10% CO 2 , 40% H (the stoichiometric composition) and 50% He, with a total flowrate of 100 cm 3 min − 1 .A temperature ramping from 100 to 500 • C at a rate of 5 • C min − 1 was performed under reaction conditions to evaluate the catalytic properties of the catalysts at various temperatures.Additionally, isothermal time-on-stream (TOS) experiments were conducted under the same reaction conditions to evaluate the stability of the catalysts.

In situ spectroscopy
CO 2 adsorption and CO 2 methanation over the synthesized materials were probed using steady-state in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) using a Nicolet iS50 FTIR spectrometer (Thermo Scientific), equipped with a Mercury-Cadmium-Telluride (MCT) detector and a high temperature reaction chamber (Harrick Scientific).Additionally, the CO 2 methanation over the prepared materials was investigated by in situ modulation excitation phase-sensitive detection DRIFTS (ME-PSD-DRIFTS) using the same spectrometer.The ME-PSD analysis highlight the species that change with the modulation frequency (f).The collected spectra were converted into the socalled frequency domain by Fast Fourier Transform (FFT).The data was then filtered at the modulation frequency and converted into the phase-domain by Inverse Fast Fourier Transform (IFFT).Activated carbon absorbs most of the IR radiation, so it was not possible to perform DRIFTS analyses in the Ni/ACR sample.
Readers interested in the detailed experimental information are referred to the Supplementary material (SM).

CO 2 methanation over Ni/CeO 2 and Ni/ACR-CeO 2
In our previous study [2], a novel methanation catalyst with 15 wt% Ni loading over ACR having O-free Lewis basic sites (denoted as Ni/ACR) has been developed, exhibiting CO 2 conversion (X CO2 ) of 76% and selectivity towards CH 4 (S CH4 ) of 97%, but at a fairly high reaction temperature of 450 • C. To increase the conversion while lowering the reaction temperature, we sought to introduce CeO 2 , thus forming a composite supporting material with ACR.We began by optimizing the CeO 2 loading in the ACR-CeO 2 composite support.It was established that 15 wt% Ni catalyst supported on ACR-CeO 2 with ACR:CeO 2 weight ratio of 50:50 exhibits the optimal catalytic properties in CO 2 methanation (Fig. S1).Hence, we further conducted our studies with this catalyst, hereafter referred to as Ni/ACR-CeO 2 .
With the optimized composite catalyst in hand, we investigated its methanation performance and compared it to a control 15 wt% Ni catalyst supported on pure CeO 2 (denoted as Ni/CeO 2 ).Upon the temperature increase, the catalysts further achieve CO conversions that are very close to the thermodynamic equilibrium, exhibiting maximum X CO2 of 85.7% (Ni/CeO 2 ) and 87.4% (Ni/ ACR-CeO 2 ) at a notably low temperature of only 370 • C (Fig. 1a).In terms of S CH4 , the catalysts also demonstrate values comparable to the thermodynamic equilibrium over the range of the studied temperatures, with S CH4 ≈ 100% at maximum X CO2 (Fig. 1a,b), with the only byproduct being CO.We finally investigated the stability of the catalysts through a long time-on-stream (TOS) testing at the best performing temperature of 370 • C. It was established that both the control Ni/CeO 2 and composite Ni/ACR¡CeO 2 demonstrate excellent stability over 90 h on stream with no loss in either X CO2 or S CH4 (Fig. 1c,d).Importantly, a comparison of the catalytic properties between the new carbon/ceriacontaining Ni/ACR-CeO 2 and our previously reported carboncontaining Ni/ACR [2] (Fig. 1) evidently highlights that the compositing of ACR with CeO 2 not only leads to a significant improvement in the methanation performance but also simultaneously lowers the optimal reaction temperature by 80 • C (from 450 • C for Ni/ACR to 370 • C for Ni/ACR-CeO 2 ).Moreover, the compositing of ACR with CeO 2 leads to an improvement in stability at the respective optimal temperature of the resultant Ni/ACR-CeO 2 catalyst, relative to that of Ni/ACR [2] (Fig. 1).

Characterization of the catalysts
Our catalytic testing reveals that both Ni/CeO 2 and Ni/ACR-CeO 2 exhibit similarly excellent catalytic properties at reasonably low reaction temperatures, despite being based on different supporting materials.Therefore, we characterized the catalysts to determine the parameters that govern the methanation performance.
Even though Ni/CeO 2 and Ni/ACR-CeO 2 have similar catalytic properties (Fig. 1), the data presented in Table 1 demonstrate that the catalysts exhibit quite different textural properties.The textural properties of the synthesized materials were estimated using Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) methods.Pure CeO 2 support has a relatively low specific surface area (S BET ) of 104 m 2 g − 1 , while the pure ACR support has about 7-fold higher S BET = 784 m 2 g − 1 .As expected, the composite ACR-CeO 2 exhibits a S BET = 373 m 2 g − 1 that is higher than that of pure CeO 2 , but lower than that of pure ACR.This trend is sustained for the respective Ni catalysts supported on the two materials, exhibiting a decreasing order in S BET Ni/ACR-CeO 2 > Ni/CeO 2 (Table 1).The addition of Ni leads to a slight decrease in S BET compared to the respective supporting materials (Table 1).Both CeO 2 and ACR-CeO 2 supports, and the derivative catalysts present Type IV isotherms (Fig. S2); the pore-size-distribution data (Fig. S3) reveal that Ni/CeO 2 exhibits a higher number of pores having larger width, while Ni/ACR¡CeO 2 has a peak in the smaller width range, also containing pores with diameters higher than 5 nm.
Next, we probed the metal-support interactions and metallic surface areas of Ni/CeO 2 and Ni/ACR-CeO 2 .The comparison of hydrogen temperature-programmed reduction (H 2 -TPR) profiles of Ni/CeO 2 and Fig. 1.Comparison of X CO2 (a) and S CH4 (b) as a function of reaction temperature, together with the stability testing results in terms of X CO2 (c) and S CH4 (d) over 90 h TOS of the new Ni/CeO 2 and Ni/ACR-CeO catalysts in comparison with previously reported ceriafree Ni/ACR [2].The thermodynamic equilibrium curves were calculated using a model based on the minimization of Gibbs free-energy of the existing species.Methanation conditions: P = 1 bar; WHSV = 000 cm 3 g − 1 h − 1 ; CO 2 : H 2 (V:V) = 1: 4.

Table 1
Textural properties [specific surface area (S BET ), and total pore volume (V p )], average size of CeO 2 crystallites (d c,CeO2 ), and Ni crystallites (d c,Ni ), metallic Ni surface area (S M ), temperature at which X CO2 = 50% (T 50 ), and maximum X CO2 , attained at 450 • C for Ni/ACR [2] and 370 Ni/ACR-CeO 2 demonstrate different metal-support interactions between Ni nanoparticles and the supporting materials (Fig. S4a).Specifically, Ni/CeO 2 exhibits a large peak at 360 • C with a shoulder at 404 • C, as well as two small shoulders at 334 • C and 464 • C. In contrast, Ni/ACR-CeO 2 exhibits a large peak at a lower temperature of 320  1).The H 2 -TPR of the supporting materials was also assessed (Fig. S4b) and it was observed that the H 2 consumed for CeO 2 reduction is minimal when comparing to the H 2 consumed for the reduction of the Ni-based catalysts.Additionally, the analysis of the H 2 − TPR profiles indicate that the ACR− CeO 2 support is a homogeneous mixture that interacts in a way that the reduction profile of the Ni/ACR¡CeO 2 sample is different than that of the other two supports (CeO 2 and ACR), and not a combination of both Ni/CeO 2 and Ni/ACR [2] H 2 -TPR profiles.
To gain structural insight into the synthesized materials, we further investigated their phase composition using powder X-ray diffraction (XRD).According to the XRD patterns (Fig. S5), the as-synthesized pure and composite supports contain cubic CeO 2 phase (ICDD #04-006-1932).In contrast, the respective Ni catalysts additionally contain cubic metallic Ni (ICDD #04-016-4761), with no secondary phases being present.There is possibly a graphitic carbon peak, but since it should be present at 28 • , overlapping with a peak from CeO 2 , it is impossible to access its presence.The average size of Ni and CeO 2 crystallites was assessed from XRD data using the Scherrer equation (Table 1), and the size of Ni crystallites in Ni/CeO 2 (ca.9 nm) was estimated to be nearly 2 times larger than in Ni/ACR-CeO 2 (ca. 5 nm), which is consistent with the smaller S M of Ni/CeO 2 .
X-ray photoelectron spectroscopy (XPS) analysis revealed that both catalysts, Ni/CeO 2 and Ni/ACR-CeO 2 , exhibited two different Ni phases, observed from the deconvolution of the XPS data for Ni 2p 3/2 (Fig. 2).The lower energy component (853.0 ± 0.1 eV) is assigned to Ni 0 in metallic Ni species, while the components at 854.4 ± 0.2 eV and 856.4 ± 0.1 eV are consistent with Ni 2+ in nickel-oxide phase, and the ones at 861.3 ± 0.3 eV and 864.3 ± 0.1 eV are satellite peaks from Ni 2+ .The Ni 2p 3/2 spectral envelopes of the catalysts resemble the typical spectra for nanoscale oxidation layers on Ni and Ni compounds, being consistent with a thin, predominantly Ni 2+ , nickel oxide layer on metallic Ni [5,[28][29][30].The binding energies (BEs) of fitting components are presented in Table S1 and the relative atomic percentages (at%) of Ni 0 and Ni 2+ are presented in Table S2.The relative atomic percentage of Ni 0 is similar in Ni/ACR-CeO 2 and Ni/CeO 2 catalysts, at 4.2 at% and 4.4 at%, respectively (Table S2).
The Ce 3d spectra also revealed the presence of mixed Ce phases (Fig. 2) in both samples.The spectra can be represented as five pairs of doublets, identified in Fig. 2 as u and v for the Ce 3d 3/2 and Ce 3d 5/2 spin-orbit components, with spin-orbit splitting of 18.4 ± 0.5 eV for all the samples, consistent with those found in the literature [31,32].These spectra were fitted using a method similar to that reported by Zhang et al. [32] where the u''' peak is fixed and the remaining peaks are constrained according to the u''' position.
The v 0 , u 0 , v', and u' are attributed to Ce 3+ , while the highest biding energy peaks, v''' and u''', are attributed to the primary components of Ce 4+ , and the remaining v, u, v'', and u'', are identified as "shake-down" states, as reported by Mullins et al. [31,33] Table S2 presents the BEs of each of these components.Importantly, we note that there is a small overlap of the Ni 2p 1/2 components (orange lines and symbols below 881 eV in Fig. 2) and the low-BE components in the Ce 3d spectra.This overlap was accounted for by including in the Ce 3d fitting procedure Ni 2p 1/2 components (orange lines in Fig. 2), with fixed parameters based on the more reliable fits in the Ni 2p 3/2 region.
The Ce 3d peaks in Ni/ACR-CeO 2 are noticeably narrower than those in Ni/CeO 2 , which likely indicates that ceria nanoparticles are in a more uniform electrostatic environment in Ni/ACR-CeO 2 , where they are in a good electrical contact with the carbon substrate.
The amount of Ce 4+ and reduced Ce 3+ is quantified in Table S2.Interestingly, the Ni/ACR-CeO 2 catalyst exhibits a higher relative at% of Ce 3+ (20.5 at%) compared to Ni/ACR-CeO 2 that contains 16.4 at% Ce 3+ .
The C 1s spectra of Ni/ACR-CeO 2 were also analyzed, and the results are presented in Fig. S6 [35,36].
Interested in understanding the microstructural properties of the catalysts, we then conducted transmission electron microscopy (TEM) and energy-dispersive X-ray spectroscopy in scanning STEM mode (STEM-EDX) analyses.These studies reveal metallic Ni nanoparticles well-dispersed over nanocrystalline CeO 2 and amorphous-like ACR.S1.
to a gap between the Ni nanoparticles sizes from TEM and those estimated by XRD, which is more representative.
Although it is difficult to distinguish the smaller Ni particles on the CeO 2 support due to the low contrast on the TEM images, the respective STEM-EDX maps evidence that there are also small Ni particles dispersed on CeO 2 material in both Ni/CeO 2 and Ni/ACR-CeO 2 catalysts (Fig. 3d,h).A closer look at CeO 2 supporting material in Ni/CeO 2 (Fig. 3c) and Ni/ACR-CeO 2 (Fig. 3g) catalysts reveals that the ceria consists of agglomerated CeO 2 nanoparticles (ca.5-10 nm), preferentially faceted by the {110} plane.Even though there was no loss in X CO2 and S CH4 over 90 h of TOS testing (Fig. 1c,d), we decided to investigate the impact of the reaction conditions on the microstructure and chemical composition of the catalysts by means of electron microscopy and XPS.
The Ni 2p 3/2 spectra (Fig. 2) of the samples after TOS testing (Ni/ CeO 2 TOS and Ni/ACR-CeO 2 TOS ) include fitting components at BEs similar to those of the components in the spectra of fresh Ni/CeO 2 and Ni/ACR-CeO 2 catalysts.Considering the relative Ni 0 atomic content (Table S2), however, we observed that it decreases to 2.8 at% (cf.4.4%) in the case of Ni/CeO 2 TOS and, on the contrary, increases to 7.4 at% (cf.4.2 at%) in the case of Ni/ACR-CeO 2 TOS .In the Ce 3d spectra (Fig. 2), again the positions of the peaks remain similar for both catalysts after 90 h of TOS, but there are differences in the Ce 3+ /Ce 4+ ratio before and after TOS in both Ni/CeO 2 and Ni/ACR-CeO 2 .The Ce 3+ atomic content in Ni/CeO 2 increases to 21.3 at% (cf.16.4 at%), while in Ni/ ACR-CeO 2 , it remains approximately constant with 19.9 at% (cf.20.5 at%).
From Figs.S8a-d, one can see that the TOS-tested Ni/CeO 2 maintains the microstructure of the Ni nanoparticles dispersed over the ceria support composed by small CeO 2 nanoparticles (ca.5-10 nm), similar to that in pristine Ni/CeO 2 (Fig. 3a).In contrast, TOS-tested Ni/ACR-CeO 2 undergoes significant microstructural changes within CeO 2 support.Specifically, Fig. S8e evidences the appearance of CeO 2 particles with a needle-like shape after 90 h of methanation.Such nanoneedles were not observed in pristine Ni/ACR-CeO 2 (Fig. 3e) or pristine and TOS-tested Ni/CeO 2 (Fig. 3a, S8a).
Notably, a closer look at Ni particles in TOS-tested Ni/CeO 2 (Fig. S8c) and Ni/ACR-CeO 2 (Fig. 4c,d) indicates a preservation of Ni@NiO core-shell nanostructure of the catalyst after prolonged methanation.

Probing the mechanism of CO 2 methanation by in situ DRIFTS
Having in hand two well-characterized high-performing methanation catalysts, we further conducted in situ DRIFTS studies to elucidate the reaction mechanism over Ni/CeO 2 and Ni/ACR-CeO 2 .This technique is used to understand the reaction mechanism in heterogeneous catalysis, probing the adsorbed species on the catalysts under reaction conditions.There are some drawbacks of this technique.Specifically, it detects not only the intermediate species but also the sample background, noise, and spectator species, making it difficult to distinguish what are the real intermediates of the reaction [37,38].
Furthermore, the intermediate species have a short lifetime, making it even more difficult to differentiate them from the remaining species, which are present under steady-state conditions.Fortunately, the modulation-excitation (ME) technique offers an interesting solution for overcoming the aforementioned drawbacks.It is based on the introduction of a rapid periodic perturbation on the testing system, to which only the intermediate species of the reaction respond.The resultant spectroscopic signal is then analyzed by phase-sensitive detection (PSD) technique to enhance the signal-to-noise ratio and to extract the information about intermediate species that responded to the modulation signal, thus avoiding the signal from the spectator species [10,[37][38][39][40][41].
We started by performing steady-state in situ DRIFTS for CO adsorption and then for CO 2 methanation at 250, 300, and 350 • C. In this temperature range, the X CO2 goes from low to almost complete.Fig. S9 shows the full range spectra for each experiment, along with a discussion of the observed trends for hydroxyl groups, gaseous CO 2 , and gaseous CH 4 [42].Fig. 5 displays a zoom-in of the region below 2000 cm − 1 , which carries important information about the CO adsorption and methanation mechanisms, while Table 2 summarizes the respective band assignment.During CO 2 adsorption (Fig. 5a,c), two bands detected at 1500 and 1370 cm − 1 (ν as / ν s ) can be attributed to formate (HCOO) species, which are probably close to a reduced Ce site in CeO 2 , because they have a low Δ as-s = 130 / 150 cm − 1 [10,26].Two other bands are observed at 1600 and 1280 cm − 1 (ν as / ν s ), that can be assigned to bi-dentate carbonates (b-CO 3 ), along with the band at 1030 cm − 1 [10,26].Moreover, a small band at 1415 cm − 1 is also observed in the spectrum of Ni/CeO 2 at 250 • C, which can be attributed to poly-dentate (p-CO 3 ) carbonates Interestingly, a small amount of methoxy species (OCH 3 ) are present in the CO 2 adsorption spectra without H 2 being present, corresponding to the bands at 1080-1060 cm − 1 [24,25]; therefore, residual H 2 should be present on the Ni nanoparticles from the reduction pre-treatment which might lead to the methoxy formation during CO 2 adsorption step.
When H 2 is introduced into the gas flow to mimic the methanation conditions (Fig. 5b,d), the bands corresponding to b-CO 3 and HCOO remain at similar wavenumbers as during the CO 2 adsorption probing; however, the difference in intensity between them decreases.Furthermore, the methoxy bands increase and a new band appears at 1710 cm − 1 for both catalysts that can be attributed to the formation of formyl (CHO) groups [24].
To gain a deeper understanding of the reaction mechanism, we next conducted modulation-excitation phase-sensitive detection DRIFTS (ME-PSD-DRIFTS), allowing for a better signal-to-noise ratio plus distinguishing spectator and intermediate species.The modulated conditions applied in the experiments were a change in flowrate from A (CO 2 = 2.5 cm 3 min − 1 and 10%H 2 /N 2 = 100 cm 3 min − 1 ) to B (CO 2 = 2.5 cm 3 min − 1 and N 2 = 100 cm 3 min − 1 ).Fig. 6 shows the phase-resolved spectra of the modulation excitation in situ DRIFTS studies conducted for Ni/CeO 2 at 250, 300, and 350 • C. A comparison of Fig. 5 and Fig.
show that the data from the ME-PSD-DRIFTS experiments look quite different from those of the steady state experiments.In particular, three main differences can be identified: (i) the peak detected in the steadystate spectra at 1700 cm − 1 (Fig. 5b) is actually caused by a convolution of 2 peaks that are out-of-phase during modulation (Fig. 6); (ii) instead of 4 peaks observed in the steady-state spectra between and 1200 cm − 1 (Fig. 5b), the phase-resolved spectra show 7 peaks (Fig. 6), indicating the appearance of new groups not identified in the steady-state spectra, namely, p − CO 3 and carboxylate (CO 2 δ− ) groups (Table 2); and (iii) two from those 7 peaks are out-of-phase with the others (Fig. 6).
With regards to the aforementioned first difference, the band at 1700 cm − 1 is indicative of the formyl (CHO) group, while the band at 1760 cm − 1 is most likely related to bridged CO (br-CO) on Ni 0 sites [20,24].The appearance of br-CO on Ni 0 should be related to two reactions occurring simultaneously: it can be formed by direct dissociation of CO  over Ni, and it can be a product of HCOO dissociation, since the presence of CO2 δ− , b-CO 3 , p-CO 3 and HCOO bands indicate that CO 2 should also be adsorbed and hydrogenated on CeO 2 .Furthermore, there is a possibility that br-CO can be hydrogenated to CHO in the presence of H 2 , as previously reported [24].This is a likely scenario since the respective bands are out of phase with each other, indicating that br-CO is hydrogenated to CHO on the surface of CeO 2 in the presence of H 2 .A more detailed analysis of the phase delay φ (the time to reach maximum accumulation) for both br-CO and CHO species over Ni/CeO 2 at 300 • C, show that they differ by 180 • [φ(br-CO) = 108 • and φ(CHO) = 288 • ].This phase delay means that the CHO is formed at the same time as br-CO is consumed.Another possible scenario would be that this phase delay is dependent on that of the H 2 , which would mean that H 2 favors the net formation of b-CO and disfavors the formation of CHO, leading to the maximum coverage of b-CO being observed mostly in phase with H 2 , whereas CHO is mostly out of phase to H 2 .However, since the phase delays are not close to 0 • or 180 • , which would be close to the value where H 2 is maximum and minimum, respectively, due to low residence time, the first option is more probable: br-CO should be hydrogenated to CHO in the presence of H 2 .Notably, it is not possible to present the phase delay values for the remaining intermediates due to the accumulation of multiple maxima.
More detailed analysis of the differences between the steady-state and ME-PSD data reveal that while the bands from b-CO 3 appear at the same positions (Table 2) [10,26], the set of bands associated with HCOO in the steady-state spectra (Fig. 5b) are, in fact, a convolution of different species, namely, HCOO and p-CO 3 (Fig. 6, Table 2) [10].Notably, the later p-CO 3 assignment could also be ascribed to mono-dentate carbonates (m-CO 3 ), but the low Δ as-s suggests the presence of p-CO 3 rather than m-CO 3 [10].In that case, p-CO 3 appearance is out-of-phase with b-CO 3 appearance, which indicate that the coordination of the carbonates may be changing from b-CO 3 to p-CO 3 .Previously, this effect was explained as a change in carbonate adsorption configuration over CeO 2 as a consequence of its reduction state.Specifically, in the absence of H 2 (ϕ PSD = 180 • ), CeO 2 is partially oxidized, and when it is reduced in the presence of H 2 (ϕ PSD = 0 • ), the b-CO 3 reconfigures to either m-CO 3 or p-CO 3 [10].
When we compare the spectra at different temperatures (Fig. 6), we observe that the bands remain at the same positions in all spectra, although their relative intensities varied.The change in intensity is particularly clear from the relative difference between the two peaks assigned to br-CO on Ni 0 and CHO, respectively.In contrast, the peaks from b-CO 3 and p-CO 3 remain intense, as they are known to be relatively stable [10].
Motivated by this detailed insight, we then proceeded to investigate the composite Ni/ACR-CeO 2 catalyst by ME-PSD-DRIFTS.Due to the absorption of most of the IR radiation by the activated carbon, we observe a lower signal-to-noise ratio for Ni/ACR-CeO 2 (Fig. 7) compared to Ni/CeO 2 (Fig. 6).Nevertheless, we were still able to draw some conclusions from the phase-resolved spectra.First, the same species are detected in the ME-PSD-DRIFTS spectra of Ni/ACR-CeO 2 , as in the case of Ni/CeO 2 , moreover at nearly the same band positions (Figs.6,7).However, three main differences were identified between methanation behaviors of these two catalysts: (i) in Ni/ACR-CeO 2 , the relative intensity of br-CO band in comparison with CHO band (Fig. 7) is lower than in the case of Ni/CeO 2 (Fig. 6); (ii) as the reaction temperature increases the presence of CO is not observed for Ni/ACR-CeO 2 (Fig. 7); and (iii) the bands corresponding to p-CO 3 are significantly less pronounced for Ni/ACR-CeO 2 (Fig. 7) than for Ni/CeO 2 (Fig. 6).These differences show that, in sharp contrast to Ni/CeO 2 , there is no direct transformation between br-CO and CHO over the Ni/ACR-CeO 2 catalyst, which is further confirmed by the absence of CO at high temperatures (Fig. 7).
The full-range in situ ME-PSD-DRIFTS spectra may help to explain the general trends of the methanation over the synthesized catalysts and deviations between them.Fig. S10 shows that, with the Ni/CeO 2 catalyst, the band corresponding to gaseous CH 4 at 3017 cm − 1 is not present at 250 • C but it appears at 300 • C. Fig. S11 shows the same observations with the Ni/ACR¡CeO 2 catalyst.For both samples, the band corresponding to gaseous CO at 2200-2100 cm − 1 starts to be apparent at 300 • C and is even more pronounced at 350 • C (Figs. S10, S11) [25].For comparison, we also investigated the pure CeO 2 and composite ACR-CeO 2 supports without Ni being present and the obtained results are displayed in Figs.S12 and S13, respectively.

Discussion
It is well known that CeO 2 -supported Ni catalysts have high activity and selectivity for CO 2 methanation at low temperatures, typically ≤ 400 • C. Most likely, these properties are related to the improved metal-support interactions between Ni and CeO 2 and a high surface coverage of CO 2 , alternatively, they can be explained by the reducibility of CeO 2 [44].Based on our previous work [2], we hypothesized that an interesting approach to obtain high-performing methanation catalyst would be the preparation of Ni catalyst supported on composite supporting material consisting of reduced activated carbon and CeO 2 .
Furthermore, we wanted to study whether such compositing would generate an active, selective, and stable methanation catalyst that operates at a temperature below 400 • C. Thus, we synthesized a set of 15 wt% Ni catalysts supported on ACR-CeO 2 with different ratios of ACR:CeO 2 , as well as a 15 wt% Ni catalyst supported on pure CeO 2 (Ni/ CeO 2 ) for reference.Fig. S1 shows that Ni/ACR-CeO 2 with ACR:CeO 2 weight ratio of 50:50 afforded the best methanation performance.Besides improving the activity and selectivity compared to those for our previously reported Ni/ACR, the Ni/ACR-CeO 2 catalyst achieved improved stability over long TOS at the respective optimal temperature.This improvement can be explained by the different interactions between the Ni and the pure ACR and ACR-CeO 2 composite and also by the fact that lowering the reaction temperature to 370 • C (cf. 450 • C) decreases the possibility of Ni deactivation over long TOS.
Significantly, although half of the supporting material in Ni/ ACR-CeO 2 is reduced activated carbon, we observed that the catalytic properties of the composite catalyst are comparable or even slightly better than those of Ni/CeO 2 (Fig. 1).This result highlights that our compositing strategy is a useful tool for reducing the usage of the expensive rare-earth CeO 2 by replacing this oxide with a cost-effective carbon material.As explained in our previous work [2], the Lewis basic sites present on the ACR supporting material improve the Ni dispersion, and furthermore, its basicity is responsible for an improved CO 2 adsorption capacity over the catalyst, enabling CO 2 to be close to the Ni active sites.
The characterization of the samples allowed some propertyperformance relationships to be established.Even though the samples presented comparable catalytic performance, they are composed of materials with different textural properties.Furthermore, the Ni species on Ni/CeO 2 exhibit a stronger interaction with pure CeO 2 support, compared to that in composite Ni/ACR-CeO 2 , which can be a hypothesis for the fact that Ni/ACR-CeO 2 presented slightly better activity at lower temperature.The reducibility of CeO 2 could also be a possible reason for slight differences in catalytic performance [44]; however, it is difficult to compare the reducibility of CeO 2 in CeO 2 and ACR− CeO 2 supports by H 2 − TPR analysis because ACR can undergo decomposition above 500 • C, which influences with the TPR profile of the sample above this temperature.The microscopy analysis demonstrated that in both samples, the particles exhibit Ni@NiO core-shell appearance.
XPS analysis revealed that both catalysts, Ni/CeO 2 and Ni/ ACR-CeO 2 , exhibited two different Ni phases, corresponding to metallic Ni and NiO, with Ce 3d spectra also revealing the presence of mixed Ce phases in both samples.Interestingly, relative at% of Ce 3+ is different in both samples, with the Ni/ACR-CeO 2 catalyst exhibiting a higher value.It is known that the oxygen vacancies can be formed in the transformation process of Ce 4+ into Ce 3+ ; thus, the amount of oxygen vacancies is usually proportional to the Ce 3+ amount [34].Therefore, we can hypothesize that Ni/ACR-CeO 2 catalyst may contain a higher amount of oxygen vacancies, which are known to be important in the methanation reaction, than does Ni/CeO 2 .This hypothesis cannot be directly supported by the analysis of the O 1s spectra because the O 1s spectra of Ni/ACR-CeO 2 contain components from the organic phase of ACR, in addition to the components related to CeO 2 (Fig. S7) [31,33,35,45].
After the stability experiment over a long TOS, the major modification was observed in Ni/ACR-CeO 2 , in which the CeO 2 particles presented a needle-like shape.Most likely, the presence of steam, a subproduct of the reaction, leads to hydrothermal recrystallization of the initial CeO 2 nanoparticles into CeO 2 nanoneedles.This recrystallization is apparently templated by the presence of ACR, since carbon-free Ni/ CeO 2 did not exhibit the appearance of the CeO 2 needles after TOS testing (Fig. S8a).The carbon balance, that was close to zero, and the fact that our catalysts do not suffer deactivation over 90 h, indicate that carbon deposition should be minimal, if present at all, in these systems.
Our studies of CO 2 methanation over Ni/CeO 2 and Ni/ACR-CeO have established that both catalysts demonstrated outstanding catalytic activity, selectivity, and stability in CO 2 methanation (Fig. 1).Notably, Ni catalysts supported on CeO 2 are among the materials exhibiting the highest methanation performance reported in the literature.To our gratification, the observed catalytic properties for our Ni/CeO 2 and Ni/ ACR-CeO 2 catalysts compare favorably to the reported ones.Specifically, our catalysts achieved a maximum X CO2 of 86% and 87%, with S CH4 ≈ 100% at 370 • C, while having a T 50 of 280 • C; this T 50 is similar or even lower from most of the reported in the literature for similar catalytic systems (Table S3).For example, Rui et al. [20] prepared two different CeO 2 supports that were used for the preparation of two Ni-based catalysts, which could achieve X CO2 ≈ 85% with S CH4 ≈ 100% at 275 • C (T 50 ≈ 260 • C) and 350 • C (T 50 ≈ 280 • C).In another interesting report, which compares two differently synthesized Ni on CeO catalysts, Ye et al. [24] observed X CO2 = 82.5% and S CH4 = 94.8% at 250 • C with Ni/CeO 2 -SGM catalyst (synthesized via sol-gel method), as well as X CO2 = 69.8% and S CH4 = 84.8%,at 450 • C with Ni/CeO 2 -IM (prepared by a simple impregnation method).They used conditions similar to those in the present study, albeit at a lower (10,000 cf.60, 000 cm 3 g − 1 h − 1 ) weight hourly space velocity (WHSV); the catalysts exhibited a T 50 of ≈ 240 • C and ≈ 380 • C, respectively.In another study with two differently synthesized CeO 2 supports, with and without a g-C 3 N 4 template, Yu et al. [25] obtained two Ni-based catalysts exhibiting T 50 of 305 • C and 330 • C, with maximum X CO2 ≈ 75% and S CH4 ≈ 100%, at 350 • C and 400 • C, for the catalyst prepared with g-C 3 N 4 template and without, respectively, also at a lower WHSV (10,000 cf.60, 000 cm 3 g − 1 h − 1 ).Cárdenas-Arenas et al. [26] prepared a Ni/CeO 2 catalyst that reached ≈ 85% X CO2 with ≈ 100% S CH4 at 375 • C, with T 50 of ≈ 280 • C.
The mechanism of the CO 2 methanation over Ni/CeO 2 catalysts has been a subject of intensive research.Zhang et al. [19] investigated the reaction mechanism on Ni/CeO 2 by first-principles calculations and concluded that the reaction on this catalyst follows the CO pathway through reverse water gas shift, rather than the formate (HCOO) formation or direct C-O bond cleavage pathways.Based on in situ DRIFTS study, Lee et al. [23] compared the mechanism of CO 2 methanation for Ni-based catalysts supported on different materials.They concluded that the Ni/CeO 2 catalyst, dissociates CO 2 into CO, which then participates in the CO 2 methanation reaction at high temperature.In another study with a Ni/CeO 2 catalyst synthesized by the sol-gel method, Ye et al. [24] suggested that the reaction occurred through a formate (HCOO) pathway with formyl (CHO) intermediate.Furthermore, the authors suggested that the good performance of the catalysts was related to the effective dissociation of H 2 by Ni and the strong adsorption/activation of CO 2 by CeO 2 support.Based on a DRIFTS study, Rui et al. [20] determined that the reaction proceeded through the CO pathway over Ni/CeO 2 catalyst, namely, by direct C− O bond cleavage via reverse water gas shift reaction followed by the methanation of the resultant CO.
Although some studies have indicated that CO 2 methanation over Ni/CeO 2 proceeds through the CO pathway [19,20,23], our results show that the methanation over our Ni/CeO 2 follows a combination of CO pathway and the formate (HCOO) pathway, where CO forms through HCOO dissociation.
For Ni/CeO 2 , we observed from steady-state in situ DRIFTs experiments (Fig. 5) that the CO 2 adsorption on the catalyst resulted in the formation of b-CO 3 and HCOO.When H 2 was added to the gas flow generating methanation conditions, another band attributed to CHO appeared, indicating that CHO is an important intermediate for the methanation over Ni/CeO 2 catalyst.
The ME− PSD− DRIFT methodology improved the signal-to-noise ratio and allowed us to distinguish between the spectator and intermediate species.Now for Ni/CeO 2 , it was possible to observe that the peak attributed to CHO in the steady-state in situ DRIFTS was in fact composed of two peaks out-of-phase with each other, namely, br-CO on Ni 0 and CHO [20,24].
Based on the collected experimental in situ ME-PSD-DRIFTS evidence (Fig. 6, S10, S12 and Table 2), Scheme 1 outlines the envisioned mechanism of the CO 2 methanation over synthesized Ni/CeO 2 .There are two possibilities for the formation of br-CO on Ni 0 : by direct dissociation of CO 2 over Ni, or dissociation of HCOO.In the second case, specifically, CO 2 is adsorbed on the CeO 2 surface as CO 2 δ− , which then forms b-CO 3 /p-CO 3 species [43].These species are then hydrogenated by the hydrogen dissociated over Ni 0 to form HCOO, which is further dissociated forming br-CO on Ni 0 .Next, the br-CO is firstly hydrogenated to CHO over the CeO 2 [24] and then to OCH 3 , as observed in the respective in situ ME-PSD-DRIFTS spectra (Fig. 6, Table 2).It is difficult to access the phase delay of the methoxy species, due to a small signal being observed, but according to the literature, the presence of OCH indicate that CHO should be hydrogenated to this species.Eventually, the as-formed OCH 3 is sequentially hydrogenated to CH 4 and released as the end-product.
There is an alternative pathway in which br-CO could be only involved in CO formation and not in the formation of the desired CH 4 , and, accordingly, CHO would be formed through direct dissociation of HCOO.However, the fact that b-CO and CHO bands are 180 • out-ofphase with each other (Fig. 6) strongly suggests that their transformation is directly correlated, which rules out this pathway.
When The presence of CHO and especially the br-CO intermediates are less pronounced in Ni/ACR-CeO 2 (Fig. 7), while the relative intensity of the HCOO, compared to the other species, appears to be higher (Fig. 7) than in the case of Ni/CeO 2 (Fig. 6).Additionally, at higher temperature of 350 • C, the presence of the band from br-CO is not observed (Fig. 7).This could be explained by the fact that the consumption rate of CO on this catalyst is more favored by high temperatures than the formation rate, leading to less accumulation of this intermediate.However, if we consider that CO is hydrogenated to CHO, the instantaneous reaction (with lack of CO accumulation) would lead to a higher accumulation of CHO, if CHO is not hydrogenated at the same rate as CO, which is not observed.Therefore, the disappearance of the band from br-CO at high temperatures is related to desorption, instead of hydrogenation to CHO.
Thus, the data indicate that the br-CO does not play a significant role in the reaction pathway but may be related to the formation of the CO Scheme 1. Proposed mechanism for the CO 2 methanation over Ni/CeO 2 catalyst, displaying the formation/transformation of the reaction intermediates as elucidated by in situ ME-PSD-DRIFTS.The green boxes mark the differences in the mechanism between the two samples (Scheme 1 vs. Scheme 2).byproduct (Scheme 2 (b)).
Regarding Ni catalysts supported on carbon materials, Hu et al. [17] concluded that the methanation over Ni catalyst supported on graphene aerogel followed the formate pathway, with carbonate intermediates being present, i.e., without the formation of CO.Consistently with this report, the CO 2 methanation over our carbon-containing Ni/ACR-CeO 2 catalyst proceeds through the formate pathway, which is depicted on Scheme 2 (a).The methanation starts with the adsorption of CO 2 as CO 2 δ− , followed by the formation of b-CO 3 /p-CO 3 species [43].Those are then hydrogenated to HCOO, similarly to what we observed for the carbon-free Ni/CeO 2 .At the same time, in sharp contrast to Ni/CeO 2 , br-CO was not detected for Ni/ACR-CeO 2 at higher temperatures (350 • C).This fact indicates the significant difference between methanation over Ni/CeO 2 and Ni/ACR-CeO 2 , wherein over the composite catalyst, the formate (HCOO) is directly dissociated to formyl (CHO) [46], then following the same pathway as over the Ni/CeO 2 .In particular, CHO is further hydrogenated to methoxy and finally to methane end product.
With regards to the proposed mechanisms (Schemes 1, 2), it is important to emphasize that some of the H 2 can be partially dissociated on the CeO 2 surface (i.e., not only on the Ni 0 ).This is suggested by in situ ME-PSD-DRIFTS of pure CeO 2 and composite ACR-CeO 2 supporting materials, revealing the presence of partially hydrogenated HCOO and OCH 3 species (Figs.S12, S13).
Interestingly, our detailed electron microscopy investigation of the microstructural changes, together with the XPS analysis, within the catalysts after 90 h of time-on-stream testing evidences that even though the catalysts did not lose their catalytic performance over 90 h (Fig. 1c,  d), their fine microstructure and chemical composition were indeed affected.
The analysis of the Ni 2p 3/2 and Ce 3d spectra of the fresh and TOS tested Ni/CeO 2 and Ni/ACR-CeO 2 revealed alterations in the Ni 0 content and the Ce 3+ /Ce 4+ ratios.The Ni 0 relative at% increase in Ni/ ACR-CeO 2 TOS indicating that the Ni nanoparticles likely undergo a further reduction under the reductive conditions of the CO 2 methanation reaction, probably due to the higher reducibility and weaker metalsupport interactions in this sample (Fig. S4).On the contrary, an increase in the Ni 0 relative at% is observed in Ni/CeO 2 .Regarding the CeO 2 support, Ce 3d spectra indicate that the relative at% of Ce 3+ , and consequently the amount of oxygen vacancies remain similar in the Ni/ ACR-CeO 2 TOS , while, in contrast, increasing under reaction conditions in Ni/CeO 2 .
The microstructural changes are mostly reflected by the preservation of the shell around Ni particles in both TOS-tested Ni/CeO 2 and Ni/ ACR-CeO 2 materials (Fig. 4,S8).The catalytic functions of the observed NiO shells are not well understood and our future work underpinning the role of the shell in the CO 2 methanation over high-performing Ni catalysts is ongoing.

Conclusions
A set of Ni catalysts deposited either on pure CeO 2 or composite carbon-CeO 2 supporting materials have been synthesized and characterized.CO 2 methanation with excellent conversion, selectivity and stability at favorably low reaction temperature was achieved using the as-synthesized Ni/CeO 2 and Ni/ACR-CeO 2 catalysts.
Our work demonstrated that compositing carbon with CeO 2 within the supporting material afforded an effective reduction in the usage of the expensive rare-earth CeO 2 while providing access to not only a costeffective but also a high-performing methanation catalyst operating at low temperature.The mechanism of the CO 2 methanation over the newly synthesized catalysts was investigated by means of in situ spectroscopy, using the emerging ME− PSD− DRIFTS technique.The collected insights uncovered that the methanation over Ni/CeO 2 proceeded via a combination of the CO and formate pathways, while in Ni/ ACR-CeO 2 it proceeded exclusively via the formate pathway.The main underlying mechanistic difference between the catalysts is that, during methanation over carbon-free Ni/CeO 2 , the reaction is mediated by CO from both CO 2 and formate dissociation, while carbon-containing Ni/ ACR-CeO 2 catalyst is producing CH 4 without intermediate CO formation.Overall, these results illustrated that even partial modification of the supporting material for Ni methanation catalysts can influence the reaction mechanism.

CRediT authorship contribution
Fig. 1a,b show the obtained X CO2 and S CH4 results in the temperature range of 200-500 • C. Both catalysts start to convert CO 2 from ca. 200 • C reaching X CO2 = 50% at 280 • C, the so-called T 50 .
Fig. 2. High-resolution XPS data for the Ni 2p 3/2 and Ce 3d region collected from as-synthesized and TOS-tested Ni/CeO 2 and Ni/ACR− CeO 2 .Symbols: raw data; black lines: overall fits; colored lines: fits of individual components; dashed lines: background.The observed binding energies (BE) and assignment of the peak positions are summarized in TableS1.

Fig. 3 .
Fig. 3. TEM images of Ni nanoparticles dispersed on CeO 2 (a) and ACR-CeO 2 (e) supporting materials, together with the representative HRTEM images of Ni and CeO 2 nanoparticles in the Ni/CeO 2 (b,c) and Ni/ACR-CeO 2 (f,g) catalysts.Low-magnification STEM images of the Ni/CeO 2 (d) and Ni/ACR-CeO 2 (h) catalysts, together with the respective STEM-EDX maps for Ni, Ce, O and their mixture.

Fig. 4 .
Fig. 4. HRTEM imaging of the surface of a single Ni nanoparticle in the as-synthesized Ni/ACR-CeO 2 catalyst (a) and zoom-in (white rectangle) image (b) and in the same catalyst after its TOS testing for 90 h (c,d), revealing the core-shell appearance of the nanoparticles.The insets in (b,d) show the FT patterns from the corresponding regions.The white arrows mark the NiO shell and the white arrow heads depict the twinning within the Ni nanoparticles.
comparing the carbon-containing Ni/ACR-CeO 2 to the carbon-free Ni/CeO 2 , our characterization data indicate that the catalysts are quite different.Ni/ACR-CeO 2 has a significantly higher surface area, porosity, and metallic Ni surface area, with smaller Ni nanoparticles size in comparison with Ni/CeO 2 .At the same time, Ni/CeO 2 is expected to have stronger metal-support interactions.The observation of similar methanation properties for such different catalysts leads to the hypothesis that the CO 2 methanation over Ni/ACR-CeO 2 and Ni/CeO might proceed through different mechanisms.

Scheme 2 .
Scheme 2. Proposed mechanism for the CO 2 methanation over Ni/ARC-CeO 2 catalyst, displaying the formation/transformation of the reaction intermediates as elucidated by in situ ME-PSD-DRIFTS.(a) illustrates the mechanism of CH 4 formation while (b) depicts the mechanism of CO formation.The green boxes mark the differences in the mechanism between the two samples (Scheme 1 vs. Scheme 2).
b Estimated from XRD data using Scherrer equation.c Estimated by H 2 pulse chemisorption.