The Partial Oxidation of Methane Over Pd/Al₂O₃ Catalyst Nanoparticles Studied In-Situ by Near Ambient-Pressure X-ray Photoelectron Spectroscopy

Instead of flaring off large quantities of unused natural gas around the world, it could be used in a more environmentally friendly way as fuel for automotive engines. Natural gas vehicles (NGVs) operate under lean conditions with low methane concentrations (500–1000 ppm) and working temperatures typically under 823 K (550 °C). The beneficial outcome of these working conditions is that relatively ”clean” products are produced (CO\(_{2}\), H\(_{2}\), H\(_{2}\)O) in comparison to other fossil fuels producing harmful nitrogen and sulphur containing compounds. In NGVs, NO\(_{x}\) and SO\(_{x}\) exhaust emissions are greatly reduced [1‐3]. The challenge lies with the storage and transport of gases, and the conversion into a more useful, transportable product. The C–H bonds in aliphatic hydrocarbons have high dissociation energies (439.3 kJ mol\(^{-1}\) in CH\(_{4}\)) and the absence of functional groups leaves the molecule with zero polarity and no sites for either nucleophilic or electrophilic attack. This makes small hydrocarbons, such as methane, very difficult to oxidise without a catalyst at low temperatures [4].

The partial oxidation of methane (POM) uses an oxygen-deficient mixture to produce carbon monoxide and hydrogen:This CO/H\(_{2}\) mixture (commonly known as “syn-gas”) is a useful precursor for methanol synthesis. Methanol is a versatile, clean-burning liquid that is used in many applications, including in the production of animal feed, hydrogen storage and as a precursor for the octane-enhancing additive methyl–tertiary–butyl ether (MTBA) [5]. Moreover, methanol in itself has the potential to become a large-scale renewable alternative to fossil fuels [6]. It is produced by the hydrogenation of CO at 773–923 K commercially using a Cu/ZnO/Al\(_{2}\)O\(_{3}\) catalyst [7]:Syn-gas is not currently produced on a large scale by partial methane oxidation, but instead by steam-reforming, where methane reacts with water, over a nickel-based catalyst, at high temperatures (>1000 K) and pressures [8]:Although this is a high-yielding reaction, the conditions at which it operates are expensive and unsustainable in the long term. Therefore it is of great importance to find a viable catalyst to enable large-scale POM for syn-gas production at lower temperatures and atmospheric pressures. POM reaction conditions are generally leaner and more cost effective than those of steam reforming. It has previously been reported that the syn-gas yield and selectivity is optimised with temperatures above 650 K [9, 10] (a 90 % yield is observed at 1050 K with a variety of transition metal-based catalysts [11]). For both methane oxidation and steam-reforming, a catalyst is yet to be developed that can activate this reaction at sustainable and more cost effective, lower temperatures and pressures.

Platinum group metals are widely used as the basis for catalysts for a number of different applications [12]. In particular, Pd and Pt are used for methane activation reactions, in the form of supported nanoparticles. The mechanism by which they operate is still relatively unknown. It has been established that a change in oxidation state of the metal occurs, but in the case of palladium-based catalysts, it is unclear as to which state is the active species. Burch et al. [13] reported that for a Pd catalyst the fully oxidized surface is the optimum state, but more recently Yang et al discovered that a mixed phase of PdO and PdO\(_{x}\) (where \(0 < x \le 1\)) is the most active state for methane combustion [14]. For the Pd(111) surface, it has been reported that a 2D surface oxide, Pd\(_{5}\)O\(_{4}\), forms after exposure to oxygen at high temperatures [15].

X-ray photoelectron spectroscopy (XPS) is a useful, surface sensitive technique for identifying the surface chemical state and composition of solids. In order to avoid unintentionally measuring contaminants rather than the species of interest, these experiments are generally performed in an ultra-high vacuum (UHV). However, in catalysis, UHV is not a representative condition and a catalyst under UHV conditions may behave differently compared to reaction conditions. Near ambient-pressure (NAP-) XPS helps to bridge this gap, with pressures up to 100 Torr (130 mbar) now possible [16]. The differential pumping of NAP-XPS systems gradually reduces the pressure in the analysis chamber down to UHV at the analyser [17], and so it is possible to collect information about the chemical state of the surface of the catalyst in-situ, under specific close-to-real-world reaction conditions.

Particle size effects are observed in the near ambient-pressure XP spectra and mass spectrometry data. NAP-XPS in the Pd 3d region is most useful to monitor the chemical state of the metal and therefore the nature of the catalytically active oxide. Figures 1, 2, 3, 4, and 5 show the in-situ data for Catalysts A–E respectively: near ambient-pressure XPS in the Pd 3d region, and the corresponding reactant to product conversion as calculated from mass spectrometry data. High resolution in-situ XPS was taken at 50 K intervals between 500 and 700 K, under 0.33 mbar (240 mTorr) of CH\(_4\) + O\(_2\) ([CH\(_4\)] : [O\(_2\)] = 2).

The peaks at lower binding energies (\(335.0 \pm 0.2\) eV) are characteristic of metallic Pd (Pd\(^0\)) with PdO\(_x\) species appearing slightly higher at 336–337 eV. Due to the catalyst preparation method and storage (i.e. exposure to air), Pd shows some degree of oxidation for all samples at 500 K, but metallic Pd peaks at \(335.0 \pm 0.2\) eV are dominant at these lower temperatures. As the temperature is increased to 600 K and again to 650 K, an XPS signal characteristic of a PdO\(_x\) species at 336.0–336.2 eV becomes dominant for most particle sizes. The CH\(_4\) conversion plots associated with the NAP-XPS measurements in Figs. 1, 2, 3, 4, and 5, calculated from mass spectrometry data, indicate an increase in syn-gas production around these temperatures. Our activity studies have shown that under similar conditions CH\(_4\) begins conversion to POM products between 550 and 600 K (see Supporting Information). In addition, metallic Pd (\(335.0 \pm 0.2\) eV) also has a strong contribution in the XP spectra, and the data show that a mixture of both Pd metal and Pd oxide is needed to continue the reaction, after the oxide has initially been formed. At temperatures above 700 K the oxide species is reduced to metallic Pd and the product formation of CO and H\(_2\) slows down.

The AP-XPS data in Figs. 1, 2, 3, 4, and 5 also show that at 700 K the oxide species is almost fully reduced to metallic Pd. This is reflected in the CH\(_4\) conversion plots which indicate a slowing down of CO and H\(_2\) formation. It could be argued that for Catalyst E (10 nm, Fig. 5) the reaction rate does not slow at 700 K and this could be because a larger ratio of metallic Pd to oxide Pd is present than for other catalysts (see Table 2), further supporting the hypothesis that metallic Pd is key for continuing the reaction.

The CH\(_4\) conversion plots in the lower panels of Figs. 1, 2, 3, 4, and 5 show that complete oxidation of methane to CO\(_2\) and H\(_2\)O occurs at low temperatures. The selectivity of these catalysts changes towards partial oxidation to syn-gas at temperatures above 570 K. There is also a distinct difference between catalysts of different sizes with respect to the temperature at which CO and H\(_2\) are produced. Catalyst E (Fig. 5, 10 nm) produces CO and H\(_2\) at the lowest temperature (570 K) and, with the exception of Catalyst A (Fig. 1, 4 nm), the catalytic activity with respect to syn-gas production decreases with decreasing particle size (Fig. 6). The XPS data for Catalyst A (Fig. 1) show a prominent peak at 337.1 eV at 550 and 600 K, characteristic of a divalent Pd species (Pd\(^{2+}\)) [25]. This peak corresponds to the temperature at which POM products are starting to form according to the mass spectrometry data, suggesting PdO is the active phase of the catalyst. A slight hysteresis is observed, but the process is generally reversible on cooling down. Note, the modes of recording the mass spectrometry data were different: the temperature was increased in steps of 50 K and held constant for about 30 min while the NAP-XP spectra were recorded, after the experiments at the highest temperature (700–750 K) had been completed, the sample was allowed to cool down while recording the reactant and product partial pressures. The NAP-XPS data in Fig. 2 for Catalyst B (5 nm) are slightly different from Catalyst A as the peak at 337.1 eV is not observed. At 600 K, a peak at 336.5 eV appears in the NAP-XP spectrum, which shifts downwards in binding energy with an increase in temperature. Syn-gas (CO and H\(_2\)) production, as shown in the conversion plot in Fig. 2, does not occur until 700 K, suggesting the absence of the high binding energy peak (\(337.0 \pm 2\) eV) is reflective in the activity of Catalyst B.

In the CH\(_4\) conversion data in Fig. 1 for Catalyst A (4 nm) it can be observed that the increase in CO pressure begins just under 600 K. H\(_2\) is produced around 650 K, which can explain why the peak at 334.9 eV reappears as the oxide has been reduced back to Pd\(^0\). In comparing Catalysts C and D (Figs. 3, 6 nm, and 4, 7 nm, respectively), there is a significant temperature dependence at which point the metallic Pd oxidises to the PdO\(_x\) species and the reverse reduction. The XPS data shown in Fig. 3 for Catalyst C (6 nm) show the presence of a higher binding energy peak at 337.0 eV (PdO) slowly increasing from 550 to 650 K, whereas the spectra for Catalyst D (7 nm, Fig. 4) show a quick conversion from 336.9 to 336.0 eV between 550 and 600 K. The peak at 336.9 eV in Figure 4 can be attributed to PdO. Its downward shift to 336.0 eV at 600 K could suggest a subsurface oxygen species, a surface 2D oxide such as Pd\(_5\)O\(_4\) [26] or even a PdC species [27]. A large peak is observed in the C 1s spectra at 285 eV (see Fig. 6 in Supporting Information). This could be indicative of a PdC surface species, however adventitious carbon from impurities and the reactant gas CH\(_4\) would also appear at this binding energy. Therefore it is impossible to identify whether a surface PdC species is present. C 1s peaks around 292 eV (appearing at 600 K) and at 290 eV (appearing at 650 K) are assigned to gaseous CO\(_2\) and CO, respectively, by comparison with spectra of the gas phase under reaction conditions (see Fig. 7 in Supporting Information). The Pd appears predominately reduced again by 650 K, when the selectivity shifts towards partial oxidation. This can be correlated to the mass spectrometry data, which show that H\(_2\) formation occurs at a lower temperature with Catalyst C (650 K) than with Catalyst D (677 K). XPS data for all catalysts show the reduced species is favoured at 700 K, which is supported by mass spectrometry showing an increase in H\(_2\) formation at much higher temperatures, between 700 and 750 K. Pd/Al\(_2\)O\(_3\) nanoparticles with an average diameter of 10 nm (Catalyst E, Fig. 5) also exhibit a PdO species (337.2 eV) that, on heating, reduces to a PdO\(_x\) or a subsurface-O species (336.5 eV). This reduction and downward shift in the NAP-XP spectrum occurs between 550 and 600 K, which coincides with the initial formation of CO as the CH\(_4\) conversion plot, shown in Fig. 5, suggests.

These studies show that the catalytic activity, in terms of the temperature at which CH\(_4\) conversion to CO and H\(_2\) begins, increases with increasing particle size for Pd/Al\(_2\)O\(_3\) nanoparticles over a range of particle sizes (5–10 nm). Nanoparticles with an average diameter above 10 nm were not available in this study. Therefore, it is not clear whether there is an optimum particle size or a continuous trend. Pd single crystals and polycrystalline foils have been used as model catalysts in similar AP-XPS studies before [28, 29] however, and show an onset of CO production at higher temperatures. Figure 7 shows the relative CO\(_2\) production of polycrystalline Pd foil compared with Pd/Al\(_2\)O\(_3\) nanoparticles when exposed to an oxygen-rich CH\(_4\) environment ([O\(_2\)] : [CH\(_4\)] = 5). In comparison to supported metallic nanoparticles, the catalytic activity of polycrystalline Pd foil is insignificant. This would suggest that there is a volcano-type distribution of particle size as a function of activity, with an optimum size at some point between 10 nm and the grain size of the polycrystalline foil (\({>}1\,\upmu\)m). Other studies using Pd/Al\(_2\)O\(_3\) catalysts have also found activity increases with increasing particle size [30, 31], explained by differences in electronic structure of nanoparticles \(<\)3 nm that form stronger interactions with the reactant molecules [32]. Similar trends have been reported in the literature [33, 34] for palladium catalysts, in which there is an optimum or critical nanoparticle size for enhanced reactivity when testing a range of particle sizes. Semagina et al. [34] tested a size range of Pd nanoparticles from 6 to 14 nm and Pd black, and found that a significant increase in turnover frequency was observed with nanoparticles of 11–14 nm, but decreased with a Pd black catalyst in the micrometer range. It has also been previously reported that smaller particles have stronger metal-support interactions, which can render them less reactive with the reactant molecules because the palladium content on the surface decreases [35], and conversely, a weaker metal-support interaction makes the catalyst more active [3, 19].

The need for fundamental knowledge of how catalytic nanoparticles behave in-situ is becoming increasingly important. Model catalysts such as polycrystalline foils and single crystals cannot be used to study size effects and metal-support interactions of industrial catalysts. The NAP-XP spectra and mass spectrometry data shown in Fig. 7, are indicative that a polycrystalline Pd foil behaves differently from supported nanoparticles under catalytic reaction conditions. At 700 K, a sharp PdO peak at 336.1 eV in the Pd foil spectrum is consistent with the rise in CO\(_2\) formation shown in Fig. 7. The peaks in the spectra for the nanoparticles are much broader, suggesting a mixture of both oxide and metal, contributing to a rise in catalytic activity at lower temperatures. By 800 K, the Pd foil is further reduced (334.8 eV) [28], whereas the nanoparticles are still predominately in their oxide state (335.9 eV) at this temperature. It could be argued that reactant pressures of around 1 mbar are not representative of industrial catalysis conditions, as syn-gas is currently produced in a large scale reactor at 5–40 bar [36]. However, NAP conditions are considerably more realistic than ultra-high vacuum and reproduce the reaction yields achieved under industrial conditions.

The exact oxidation state of Pd during the reaction cannot be confirmed by XPS measurements alone. PdO\(_x\) species have binding energies between 335.5 and 337.0 eV [12, 25, 27]; therefore the exact nature of the oxide is not clear. It could be subsurface oxygen, bulk or surface PdO oxide, or a surface 2D oxide such as Pd\(_5\)O\(_4\). These peaks could also indicate a carbon build-up on the Pd surface from the cracking of the reactants and products in the gas phase [37]. The particle size may have a direct impact on the nature of PdO\(_x\), as smaller nanoparticles have increased metal-support interactions, which can in turn affect their thermal stability [38]. This would therefore lead to an increase in binding energy of the Pd 3d\(_{5/2}\) peak of the PdO\(_x\) species in smaller particles (\({<}5\) nm). It is probable that more than one oxidation state is present in the oxidised metal, as the fitted XPS peaks are still quite broad over this range and the increasing concentration of newly formed H\(_2\) is simultaneously reducing Pd. Despite this uncertainty in the exact nature of the oxidation state, in-situ XPS shows distinct changes in chemical state of the Pd surface as a function of temperature and a clear correlation with the catalytic activity.

A particle size effect has been observed in Pd/Al\(_2\)O\(_3\) nanoparticles catalysing the partial oxidation of methane to syn-gas (CO and H\(_2\)). In general, CH\(_4\) conversion to POM products occurs at lower temperatures for larger particles from 5 to 10 nm. NAP-XPS data show the appearance of a metal oxide species around 550–650 K for all particle sizes, whilst mass spectrometry indicates the start of syn-gas production during these temperatures. As H\(_2\) is produced, the oxide species on the Pd surface reduces to metallic Pd by 700 K, slowing down the production of CO. This suggests the formation of the oxide species is the driving force in initial product formation. A mixture of PdO and metallic Pd is needed to maintain the reaction as we observe an increase in intensity of the Pd metal NAP-XPS peaks (335 eV) at the same temperatures at which product formation increases, as shown by mass spectroscopy. A full reduction of the oxide to metal is detrimental to the catalytic process.