Dry Reforming of Methane on NiCu and NiPd Model Systems: Optimization of Carbon Chemistry

Abstract

A series of ultra-clean, unsupported Cu-doped and Pd-doped Ni model catalysts was investigated to develop the fundamental concept of metal doping impact on the carbon tolerance and catalytic activity in the dry reforming of methane (DRM). Wet etching with concentrated HNO₃ and a subsequent single sputter–anneal cycle resulted in the full removal of an already existing oxidic passivation layer and segregated and/or ambient-deposited surface and bulk impurities to yield ultra-clean Ni substrates. Carbon solubility, support effects, segregation processes, cyclic operation temperatures, and electronic and ensemble effects were all found to play a crucial role in the catalytic activity and stability of these systems, as verified by X-ray photoelectron spectroscopy (XPS) surface and bulk characterization. Minor Cu promotion showed the almost complete suppression of coking with a moderate reduction in catalytic activity, while high Cu loadings facilitated carbon growth alongside severe catalytic deactivation. The improved carbon resistance stems from an increased CH₄ dissociation barrier, decreased carbon solubility in the bulk, good prevailing CO₂ activation properties and enhanced CO desorption. Cyclic DRM operation on surfaces with Cu content that is too high leads to impaired carbon oxidation kinetics by CO₂ and causes irreversible carbon deposition. Thus, an optimal and stable NiCu composition was found in the region of 70–90 atomic % Ni, which allows an appropriate high syngas production rate to be retained alongside a total coking suppression during DRM. In contrast, the more Cu-rich NiCu systems showed a limited stability under reaction conditions, leading to undesired surface and bulk segregation processes of Cu. The much higher carbon deposition rate and solubility of unsupported NiPd and Pd model catalysts results in severe carbon deposition and catalytic deactivation. To achieve enhanced carbon conversion and de-coking, an active metal oxide boundary is required, allowing for the increased clean-off of re-segregated carbon via the inverse Boudouard reaction. The carbon bulk diffusion on the investigated systems depends strongly on the composition and decreases in the following order: Pd > NiPd > Ni > NiCu > Cu.

1. Introduction

The steady increase in anthropogenic greenhouse gas (GHG) emissions and the growing energy demand has created a strong interest in CO₂ capture and its utilization as a feedstock for valuable chemical and fuel production to mitigate climate change [1,2]. An attractive route to utilize captured CO₂ is the production of syngas, a mixture of H₂ and CO, via the dry reforming of methane (DRM), which uses two abundant and harmful greenhouse gases: carbon dioxide and methane [3]. As part of the reforming technologies, the dry reforming of methane is the most promising and economical way to produce syngas (Equation (1)) [4,5], although steam reforming (SMR), partial oxidation (PO) and autothermal reforming (ATR) of methane currently dominate syngas production (Equations (2)–(4), respectively). While these processes yield more hydrogen-rich syngas with a H₂/CO ratio ≥ 2, DRM ideally approaches a hydrogen-to-carbon-monoxide ratio of unity [4]:

CH₄ + CO₂ → 2 CO + 2 H₂ ΔH⁰ = 247 kJ mol⁻¹

(1)

CH₄ + H₂O → CO + 3 H₂ ΔH⁰ = 206 kJ mol⁻¹

(2)

$${\mathrm{CH}}_4+\frac{1}{2}{\mathrm{O}}_2\to \mathrm{CO}+2{\mathrm{H}}_2\hspace{1em}\Delta {\mathrm{H}}^0=-36\mathrm{kJ}{\mathrm{mol}}^{-1}$$

(3)

$${\mathrm{CH}}_4+\frac{1}{2}{\mathrm{H}}_2\mathrm{O}+\frac{1}{4}{\mathrm{O}}_2\to \mathrm{CO}+\frac{5}{2}{\mathrm{H}}_2\hspace{1em}\Delta {\mathrm{H}}^0\approx 0\mathrm{kJ}{\mathrm{mol}}^{-1}$$

(4)

The utilization of CO:H₂ variable syngas offers the potential for a number of synthetic downstream processes, including the production of oxygenates and higher alkanes through Fischer–Tropsch synthesis [6,7]. Furthermore, it is well-established that syngas can be efficiently used as a fuel for, e.g., solid oxide fuel cells (SOFCs). Hydrocarbons, including CH₄, can be directly converted to syngas at the anode of a SOFC via internal reforming, and then simultaneously oxidised at the anode to generate electricity [8]. DRM was first studied by Fischer and Tropsch in 1928 over Ni and Co catalysts [9]. Extensive investigations revealed that all group VIII transition metal catalysts, except osmium, exhibited catalytic activity towards DRM [10,11]. Supported noble metals, such as Rh, Ru and Pt, can provide a high catalytic performance and stability, but base metals are preferred in industrial applications on a large scale, considering their low cost and wider availability. Ni is the most widely studied base metal for this reaction. Nevertheless, many Ni-based supported catalysts largely tend to deactivate during DRM due to severe coke formation and subsequent activity loss [12].

Mainly, three pathways are relevant for coke formation: Boudouard reaction (Equation (5)), reduction of CO by H₂ (Equation (6)) and methane decomposition (Equation (7)). At temperatures below 700 °C, CO disproportionation via the Boudouard reaction is generally more favourable. Direct CO reduction is also preferred at low temperatures as well. Below 700 °C, the free enthalpy trends suggest a stronger contribution of the reaction in Equation (5) relative to Equation (6). In turn, methane decomposition, via catalytic CH₄ cracking, becomes thermodynamically favoured at higher temperatures [13]:

2 CO → CO₂ + C ΔH⁰ = −171 kJ mol⁻¹

(5)

CO + H₂ → H₂O + C ΔH⁰ = −131 kJ mol⁻¹

(6)

CH₄ → 2 H₂ + C ΔH⁰ = 75 kJ mol⁻¹

(7)

For syngas production, at economically meaningful conversions, high temperatures of 800 °C are necessary, and feed ratios of CH₄/CO₂ close to 1 are required to achieve CO-rich syngas with an H₂/CO ratio of unity [14]. Hence, these conditions facilitate carbon growth, especially for Ni catalysts, which critically limit their industrial applications in the DRM process. Several approaches were developed to enhance the coking behaviour and long-term activity/stability of Ni-based catalysts, but catalyst deactivation remains a critical factor [15,16,17]. The most convenient approach to coke reduction is the addition of promoters, which modify the active metal domains quasi-electronically and/or ensemble-wise [18,19,20,21]. Pure Ni surfaces display a good capability of CH₄ and CO₂ activation [22,23]. Therefore, the alteration of the catalyst’s activation properties, and thus the strength of reactant adsorption, could counteract carbon deposition [24].

The addition of Cu or Pd was widely investigated to develop appropriate coking-resistant and highly active Ni-based catalysts for applications in methane reforming processes. Nevertheless, catalyst deactivation through severe coke formation remained a crucial factor [21,25,26,27]. Cu is suitable both as an electronic and a surface structure-modifying dopant, since it exhibits practically no activity in DRM as a pure metal. Hence, Cu addition could lead to minor coke deposition and enhanced catalytic stability [25]. Several studies investigated the effects of promoting Ni with Cu on several supports, although some critical points, such as activity loss upon coking, persisted, and the exact optimum NiCu composition for DRM cannot yet be identified [25,26,27]. Pd shows excellent CH₄ activation properties, while it is unable to activate CO₂ [28,29]. Pd doping on supported Ni catalysts led to an improved catalytic activity and coke removal properties during DRM with an overall better performance compared to catalysts containing pure Ni or Pd [21,30,31]. Supports such as ZrO₂ were also reported to play a direct role in DRM, which improves CO₂ activation and reduces carbon deposition simultaneously [32,33,34].

In a study with alumina-supported Ni catalysts, a 10 wt.% Ni/Al₂O₃ catalyst with a copper loading of 1 wt.% showed an enhanced stability and activity with reduced carbon deposition during DRM (750 °C, CH₄:CO₂ = 1:1), as compared to an undoped Ni/Al₂O₃ reference catalyst. However, Cu loadings of above 30 at.% led to a more rapid deactivation and facilitated carbon growth [35]. Similar effects of Cu addition were reported on silica-supported Ni catalysts, where a 1 wt.% Cu/8 wt.% Ni/SiO₂ catalyst exhibited excellent long-term stability. If the Cu amounts exceeded 20 at.%, carbon deposition increased dramatically during the CO₂ reforming of methane (800 °C, CH₄:CO₂ = 1:1) [36]. Bernardo et al. confirmed that the rate of carbon deposition in CH₄/H₂ mixtures on silica-supported Ni catalysts is reduced with the increasing Cu amount [37]. According to Tavares et al., NiCu/SiO₂ catalysts with Cu loadings below 25 at.% showed lowered carbon deposition when treated in a CO-CO₂ mixture. Nevertheless, enlarged Cu amounts on these catalysts instead revealed a decrease in their free energy deviation from the graphite equilibrium ΔG_C when treated in a CO-CO₂ atmosphere. A small ΔG_C value describes a higher probability of carbon deposition than gasification, which suggests facilitated carbon growth [38]. Mechanistically, this may be interpreted in terms of the balance between methane and CO₂ activation kinetics, becoming shifted towards increased carbon deposition. We note that the above-described studies of supported bimetallic NiCu catalysts only investigated Cu contents up to 50 at.% [35,36,37,38].

Singha et al. reported that the addition of only 0.2 wt.% Pd to a 5 wt.% Ni/MgO catalyst improved the catalytic activity and coke removal properties during DRM, and that higher Pd loadings likewise retained a conversion improvement [31]. Several supported bimetallic NiPd catalysts with different NiPd ratios showed only small differences in their catalytic behaviour. Their performance was strongly dependent on the support used, as indicated by Steinhauer et al. All bimetallic catalysts studied by this group exhibited a better performance than either pure Ni or Pd catalysts, with the NiPd (80:20)/ZrO₂-La₂O₃ catalyst having the best DRM conversion at temperatures of ~700 °C [30].

The main objective of our study was to characterize and quantify the “isolated” electronic structure effect caused by a variable doping of a bulk Ni host, with Cu and Pd as guest metals, on methane dry reforming (DRM) activity and carbon tolerance, in comparison to monometallic Ni, Cu and Pd model surfaces. For this purpose, purely (bi)metallic, ultra-clean and chemically uniform surfaces without contact with supporting and/or promoting oxides were prepared via Cu or Pd deposition from the gas phase and subsequent thermal annealing. The aim was to create a novel body of kinetic and spectroscopic evidence which is otherwise inaccessible from studies on the supported catalysts, eventually distinguishing the electronic structure effects from superimposed particle size and/or support interface effects, which potentially prevail on the “real” oxide-supported nanoparticulate NiCu and NiPd catalysts. Directional electronic structure knowledge is particularly important in view of future quantum chemistry approaches towards more coking-resistant and active catalytic sites or surface/interface structures. The underlying hypothesis is that Cu-Ni and Pd-Ni interactions at and near the surface distinctly influence the reactive sticking probabilities of CH₄ and CO₂, the adsorption energies of carbon, CO, hydrogen and oxygen, as well as the surface-near solubility and the subsurface and bulk diffusion properties of carbon. We anticipated that an optimized catalyst composition should be based on a compromise between sufficient oxophilicity, which is required for efficient CO₂ activation and de-coking properties, and modulated CH₄ activation properties. If a catalyst is capable of converting carbon faster than being deposited, the level of intermediate carbon deposits should remain low.

Experimentally, our measurements require a highly specialized approach involving the ultra-clean preparation of (bi)metallic model catalyst surfaces under ultra-high vacuum (UHV) conditions. Our dedicated setup allows for a direct transfer of the catalysts from the UHV chamber to a completely inert, high-temperature-compatible all-quartz reaction cell operating under realistic pressure and temperature conditions for DRM (pressures in the bar regime and temperatures up to 1100 °C). Catalytic testing via temperature-programmed DRM experiments was employed and accompanied by characterization techniques that involved a combination of X-ray photoelectron and Auger spectroscopies (XPS/AES) before and after catalyst operations in the high-pressure/high-temperature batch reactor.

2. Results and Discussion

2.1. Model Catalyst Characterization

2.1.1. Characterization of the NiCu Bulk-Intermetallic Catalysts

An already existing oxidic/passivation layer was observed on all initial Ni foils. This layer is induced by storage in ambient air and grows in thickness with increasing sample age. If this layer is already very thick, it contains a large amount of oxygen that becomes partially dissolved in the underlying metal bulk upon initial thermal treatment [39]. Therefore, heating without the initial removal of this ambient-induced oxide layer by etching leads to the enhancement of this unwanted effect, especially on “aged” samples. In terms of any subsequent catalytic measurements, this dissolved oxygen turned out to be a constant source of surface deactivation, due to the re-formation of a passivating oxide layer, even under strongly reducing conditions. Thus, the most convenient method to prepare oxygen- and carbon-free Ni foil substrates, which do not exhibit irreversible self-passivation during DRM, is the initial removal of this already existing oxidic/passivation layer via wet etching with concentrated HNO₃.

After the following thermal evaporation physical vapor deposition (PVD, see Section 3.2) process of metallic Cu, a single cycle of Ar⁺ sputter cleaning and thermal annealing in UHV conditions (base pressure ≈ 1 × 10⁻⁹ mbar) was sufficient for preparing oxide-free bimetallic surfaces. The chosen preparations resulted in the following nominal compositions of the bimetallic NiCu model catalysts (

Table 1. Overview of the molar fractions of Ni and labelling of the bimetallic NiCu model catalyst samples discussed below.

	Molar Fraction
Sample	Ni
NiCu Ia	0.1376
NiCu Ib	0.1628
NiCu Ic	0.3074
NiCu II	0.3188
NiCu III	0.8280
NiCu IV	0.9257

).

As an example of the effectiveness of a single sputter–anneal cycle, leading to completely clean and reduced NiCu sample surfaces, the preparation of NiCu Ia is outlined in Figure 1 via the XP and Auger spectra obtained during its cleaning process. All other bimetallic catalysts behaved accordingly.

Directly after thermal evaporation, oxygen/carbon species and completely oxidized copper cover the sample surface. Copper is present as CuO, as indicated by the shifted, broader peak and shake-up satellite features (Cu²⁺ sat.) in Figure 1d. Ar⁺ sputter cleaning leads to partial sample reduction (Figure 1k). Therefore, the exclusion of Cu(I) oxide by the acquisition of Cu LMM Auger spectra is necessary since its peak is not shifted and only slightly broader than that of metallic copper (Figure 1e). The obtained Cu LMM Auger spectrum in Figure 1h shows evidence of Cu₂O at a KE of 916.8 eV, apart from metallic copper at a KE of 918.6 eV, after a partial sample reduction. Thermal annealing ultimately achieved complete reduction, as no traces of oxidized species were detected spectroscopically. Until the last step, the corresponding Ni 2p_3/2 region showed no nickel at all due to plenty of deposited copper on top of the Ni foil substrate, whereas after the full cleaning/thermal annealing process, a well-resolved metallic Ni peak arose (Figure 1c).

The investigation of the NiCu system towards DRM requires an entirely clean and homogeneous catalyst, and thus sputter depth profile studies were performed to characterize the purity and surface-near/bulk compositions of selected catalyst systems. The depth profile of NiCu II obtained on the Cu-exposed foil after thermal evaporation is shown in Figure 2a, which reveals that a wide interdiffusion zone emerged underneath a pure copper film, finally approaching the pure nickel bulk. This zone is formed through metallic interdiffusion due to a sufficiently high substrate temperature during copper deposition (≥210 °C). As shown in Figure 2b, a subsequent heat treatment in a continuous flow tube furnace leads to further strong copper and nickel interdiffusion and results in a homogenous, Ni-rich NiCu alloy throughout a wide depth range (~70 at.% Ni). The usual cycle of Ar⁺ sputter cleaning and thermal annealing followed, and after two cycles of temperature-programmed DRM reactions, depth profiling indicates an even more Ni-rich NiCu bulk alloy (Figure 2c). The investigated NiCu II catalyst features an entirely clean, carbon- and oxygen-free bulk system after every treatment stage.

2.1.2. Characterization of the NiPd Bulk-Intermetallic Catalyst

In contrast to the NiCu bulk-intermetallic catalysts, NiPd alloys need higher temperatures for intermixing, and thus NiPd I was pre-heated in 10% H₂/He atmosphere at 1200 °C for 1 h after the thermal evaporation PVD process of Pd. The usual cycle of Ar⁺ sputter cleaning and thermal annealing followed and also led to a completely clean and reduced NiPd foil as well, as shown in Figure 3.

Prior to UHV annealing, no Pd was detected in the corresponding Pd 3d region (Figure 3d,e), which originates from strong palladium interdiffusion via the high alloying temperature (1200 °C) and in due course from plenty of oxidized Ni species on the surface, as seen in Figure 3a,b. Figure 3h shows that Ar⁺ sputter cleaning leads to partial sample reduction and complete carbon removal. However, as indicated by the presence of a pure Ni metal peak at a BE of 852.8 eV and a metallic Pd peak at a BE of 335.8 eV in Figure 3c,f, respectively, the residual oxygen species are easily reduced after thermal annealing under UHV conditions.

In addition, the depth profile of NiPd I was investigated after one DRM cycle (plotted in Figure 4) to determine the catalyst’s homogeneity compared to NiCu II (Figure 3c).

2.2. Catalytic Testing via Temperature-Programmed DRM Experiments

The freshly prepared, ultra-clean catalyst foil samples were investigated for their catalytic performance under DRM conditions by continuous online mass spectrometric reactant/product quantification in the ambient pressure-recirculating batch reaction cell, as described in further detail in Section 3.3. DRM experiments were carried out with feed ratios of CH₄/CO₂ close to 1 to approach a syngas (H₂/CO) ratio of approximately 1 and with high temperatures at ~800 °C to render typical technical DRM process conditions [14]. For continuous partial pressure detection of the gas composition, the respective pressure-drop normalized EID-MS signals were calibrated using pure gases with quantitative consideration of fragmentation. On this basis, the CO₂ conversion in % as a function of time is calculated via the measured CO₂ partial pressures and plotted in Figure 5 for pure Ni, NiPd I and all NiCu samples. From the CO₂ partial pressures, molar product formations were derived and normalized to the geometrically estimated total amount of surface Ni atoms of the respective catalyst, in order to obtain the turnover number (TON) values.

In addition, the turnover frequency (TOF) values of the four most active catalysts were calculated in the isothermal region of 800 °C, at three different reaction times, namely at 100 min, 110 min and 115 min (shown in

Table 2. Overview of the TOF values for the four most active catalysts in the isothermal region of 800 °C for different timestamps. Details of the TON and TOF calculations are provided in Section 3.3.

	TOF/s−1
Sample	at 100 min	at 110 min	at 115 min
NiCu III	8.2	5.2	3.9
NiCu IV	4.6	1.9	1.8
Pure Ni	2.5	1.3	1.0
NiPd I	5.1	2.1	1.8

), via differentiation of the molar product formations and subsequent normalization of the obtained molar rates, which is again based on the total amount of surface Ni atoms of the respective catalyst, compared to the TON calculation.

A brief examination of the different catalytic performances indicates that the integral CO₂ conversion during the first catalytic DRM cycles on the NiCu samples is generally diminished with the increasing Cu dopant amount, as compared to pure Ni. In contrast, the TOF values calculated from the isothermal slopes of the turnover curves denote a distinct behaviour, whereby the catalysts with low Cu doping levels possess generally higher turnover frequencies than pure Ni at every calculated point, as highlighted in Table 2. The most pronounced decrease in the TOF with the isothermal contact time is clear for NiCu III, which indicates a higher residual reaction rate both during and at the end of the isothermal region. It seems clear that this behaviour is characteristic of a more coking-resistant sample. On this basis, the influence of the surface molar fraction of Ni on the catalytic activity and the coking behaviour of the catalyst foils must be assessed further.

The group of the more active catalysts (pure Ni, NiCu IV and III, NiPd I in Figure 5) exhibited a sudden, stepwise onset of catalytic activity followed by rapid CO₂ conversion. Onset temperatures range from 649 °C (NiCu IV) to 677 °C (NiCu III). This apparently autocatalytic self-activation process occurred for the pure Ni catalyst within the described temperature range at 660 °C, as highlighted in Figure 5, and for NiPd I at 676 °C, which is very similar to NiCu III. However, bimetallic NiCu catalysts with moderate or low activities (NiCu Ia-Ic, NiCu II) possess hardly any self-activation step. For these samples, the determination of the exact catalytic onset temperature is not straightforward and can only be estimated at around 630–650 °C, i.e., at clearly lower values as compared to the distinctly “self-activating” catalysts. The observed rapid onset indicates a fast modification process on the respective catalyst’s surfaces occurring at a specific temperature, which may depend not only on their bulk composition. We propose that this onset emerges most likely from the sudden breakdown of an inactive/passivating oxide layer that was most likely formed on the catalyst’s surface during exposure to the reaction mixture in the sub-onset temperature regime. This breakdown/self-activation phenomenon is currently investigated in more detail both by XPS analysis in the combined UHV/high-pressure catalysis setup (i.e., by stopping the catalytic runs before and after the step).

2.3. Characterization after DRM Reaction

The suppression of carbon formation, as a function of bimetallic composition with the concomitant preservation of sufficient catalytic activity, represents the central focus of this study, since it is well-known that purely Ni-based catalysts are prone to pronounced and partially irreversible deactivation upon coking [4,15,16,17]. Therefore, an overview of the XPS-derived molar fractions of Ni in Cu and of Pd in Ni before DRM/after UHV cleaning vs. the amount of formed carbon monolayers after reaction, and the CO₂ conversion after 120 min total reaction time, is given in

Table 3. XPS-derived molar fractions of Ni in Cu (upper part) and of Pd in Ni (lower part) before DRM/after UHV cleaning vs. the amount of formed carbon monolayers after the reaction and the CO₂ conversion after 120 min total reaction time for all prepared catalysts. Additionally, the TOF values of the four most active catalysts in the isothermal region at 800 °C and 110 min are listed.

	Molar Fraction
Sample	Ni	Carbon ML	CO2 Conversion/%	TOF/s−1
Pure Cu	0.0000	0.0000	0.06	-
NiCu Ia	0.1376	0.0137	8.00	-
NiCu Ib	0.1628	0.0978	3.65	-
NiCu Ic	0.3074	0.5841	5.88	-
NiCu II	0.3188	0.3006	15.32	-
NiCu III	0.8280	0.0000	70.07	5.2
NiCu IV	0.9257	0.1584	79.70	1.9
Pure Ni	1.0000	0.6771	88.36	1.3
	Molar Fraction
Sample	Pd	Carbon ML	CO2 Conversion/%	TOF/s−1
NiPd I	0.0335	0.9496	69.57	2.1
Pure Pd	1.0000	17.84	2.92	-

for all prepared catalysts. For a quantitative analysis, the XPS peak areas are converted to atomic concentrations, which allows the surface molar fractions to be extracted. The carbon film thickness was estimated in monolayers from an XPS overlayer attenuation model. Further details of these calculations and of the standard quantification procedure are provided in Section 3.4 and Section 3.5.

Figure 6 illustrates the catalytic and coking behaviour as a function of the surface molar fraction of Ni (left) and Pd (right) of all prepared catalysts, using the values from Table 3. It is evident that the pure Ni catalyst exhibits the highest first-cycle activity towards DRM, alongside the highest amount of deposited carbon as compared to any bimetallic NiCu catalysts, whereas pure Cu is entirely inactive, but remains coke-free. While the CO₂ conversion is gradually diminished with the increasing copper amount (as also clearly visible in Figure 5), the coking susceptibility of the bimetallic NiCu catalysts appears more complex, particularly for those catalysts with moderate or low activities, upon which, at a Ni surface molar fraction of around 0.3, a rapid increase in carbon formation alongside a pronounced catalytic deactivation could be observed.

If the active metal is only doped with a minority amount of Cu, i.e., in the case of NiCu IV, the carbon deposition is markedly lowered, and only a moderate reduction in the catalytic activity is observed. As for NiCu III, further substitution of the active Ni metal by Cu to a surface molar fraction of nearly 0.83 leads to total coking suppression throughout the DRM process, despite the prevailing, slightly reduced activity with a CO₂ conversion of ~70%. By this means, the higher isothermal TOF values on NiCu III, shown in Table 3, can be rationalized.

Clearly, our NiCu bulk alloy model catalyst experiments confirm the observations from the supported NiCu systems reported previously in the introduction, Section 1. Moreover, they indicate that an optimized NiCu composition is suggested in the region of 85 at.% Ni, and most importantly, that the specific electronic and surface ensemble structure of this bimetallic state allows for an optimized combination of carbon reactivity and CO₂ activation kinetics. On the one hand, a sufficient oxophilicity of the bimetal is retained, which is required to keep the dissociation barrier for CO₂ thermally accessible. On the other hand, the blocking of carbon bulk diffusion/resegregation, on which the graphene/graphite growth process is based, can be minimized, and the modulation of bond strength (and thus, reactivity) of surface C atoms appears to be suitable. This combination allows an appropriate high syngas production rate to be retained combined with the apparent total suppression of carbon deposition.

Regarding theory approaches towards combined methane and CO₂ activation kinetics, recent density functional theory (DFT) studies of DRM on different bimetallic NiCu surfaces from Boualouache and Bouacenna showed that the carbon resistance of surfaces with low Cu content relates to the reduced interaction of the CH 1π orbital with the 3d states of the active sites [40]. Hence, the dissociation barrier of CH decomposition rises through the modification of Ni(111) by Cu, but the presence of small enough Cu amounts does not affect the dissociation reactions of CH₄ and CO₂ too negatively due to sufficient available Ni sites, which ensure the sequence of these two reactions. In other words, the combined electronic structure/ensemble effect still allows for the presence of a sufficient density of Ni-rich surface regions, which are to some extent electronically modified, but in a positive sense: both the adsorption and bulk diffusion rate of surface carbon resulting from methane are lowered to an extent, which allows for a virtually immediate carbon clean-off by simultaneously adsorbed CO₂.

A thermodynamic view suggests that under DRM conditions, deactivation upon coking is usually attributed to the successive dehydrogenation of methane, since methane thermal cracking is an endothermic process promoted at high temperatures. CO disproportionation through the Boudouard reaction, as an exothermic process, takes place below 700 °C (“coking window”). Nevertheless, any desirable coke-resistant catalyst is required to inhibit this reaction since the feed will cross the “coking window” region throughout the start-ups and shutdowns of the reforming process. For Ni(111) surfaces with a low Cu content, (5σ−3d) interactions dominate over (1π, 2π*−3d) interactions, which lead to reduced interactions of the CO molecule with the catalytic surfaces, thus favouring CO desorption relative to its dissociative activation. Therefore, the catalytic surface may be additionally protected from excessive carbon formation at low temperatures, enhancing CO anti-poisoning [40].

With respect to the reported beneficial role of Pd as a bimetallic Ni dopant (see Section 1), the dry reforming of methane was carried out over a Ni catalyst containing ~3 at.% Pd and a single-component Pd reference catalyst foil to assess their relative coking sensitivities and deactivation/activation behaviour, in particular in comparison to the above-described effects of NiCu alloy formation.

In contrast to the expectations derived from the literature (Section 1), the unsupported NiPd and Pd bulk model catalysts behaved very differently from NiCu in DRM testing (Figure 6, right side). Upon the doping of the Ni foil with a relatively small amount of Pd, as for NiPd I, not only the CO₂ conversion decreased to a similar amount to that of NiCu III, but even carbon was deposited to a higher degree than on the pure Ni foil. For the ultra-clean Pd foil, DRM led to a totally coked surface with 99.00 at. % (17.84 ML) carbon, in combination with a very poor activity (2.92% conversion at maximum). The observed detrimental surface- and bulk-near processes may result from a combination of faster carbon accumulation/solubility due to enhanced methane decomposition and C subsurface saturation, together with diminished CO₂ activation kinetics on the Pd-doped and Pd-based catalysts. Pure Pd, in particular, is known to be a very good activator for CH₄, but not for CO₂ [28,29]. Nevertheless, Pd-based catalysts using efficiently CO₂-activating supports are shown to perform very well in DRM. Enhanced kinetic carbon solubility, in combination with a geometrically maximized (bi)metal–support interface can improve de-coking via the faster re-dissolution of graphitic deposits and carbon diffusion to an active metal oxide phase boundary, which plays the role of a supplier of active oxygen species from CO₂ dissociation. In this specific case, improved carbon clean-off through the CO-forming inverse Boudouard reaction requires the above-mentioned (bi)metallic properties in combination with short diffusion pathways to the phase boundary, which is only matched for (bi)metallic nanoparticles with extended metal–support interfaces [41,42]. Hence, it appears unavoidable that Pd addition to unsupported Ni bulk model catalyst foils even enhances surface carbon deposition, as well as its subsurface/bulk saturation, in our model systems. In combination with poor intrinsic CO₂ activation properties of the “isolated” (bi)metallic surfaces, no result other than catalytic deactivation and enhanced coking can be expected.

As indicated above, the carbon solubility of the investigated model systems is expected to play a substantial role in the deactivation processes upon coking towards DRM, and thus sputter depth profile studies were performed after catalysis to determine the variation of the carbon bulk diffusion properties through doping. The depth profiles of pure Ni, NiPd I and pure Pd were obtained after one cycle, that of NiCu III after three cycles of temperature-programmed DRM reaction, providing the respective C 1s intensity trends as a function of sputtering time, as shown in Figure 7. The results of NiPd I (with only ~3 at% Pd content) are not included in Figure 7, as they hardly deviate from the pure Ni data.

Adventitious carbon is present on the surface of all catalysts without sputtering due to sample transport through air. After the first etching step, the pure Ni, NiPd I and Pd foils exhibit only graphitic carbon in the surface-near and bulk region, whereas NiCu III contains no carbon at all after 100 s of sputtering. While a large amount of carbon accumulates deep into the Pd bulk regions and is detectable up to 10,000 s sputter time, the carbon solubility is substantially lower for pure Ni and NiPd I, which already show almost no more carbon after around 1000 s. The higher carbon solubility of pure Pd model catalysts renders them as efficient carbon sinks, in which decomposed methane leads to inactive carbon species embedded in the bulk. This actually holds only for bulk metal samples without contact with an active metal oxide phase boundary promoting carbon clean-off. Corresponding to Köpfle et al., the zirconium-assisted activation of Pd enhances carbon clean-off and CO formation, whereby a Pd⁰/t-ZrO₂ interface represents the active phase boundary. In this mechanistic scenario, the improved carbon diffusion in Pd⁰ becomes an advantage, as it leads to a fast supply of the embedded carbon species to the phase boundary, where they react with lattice oxygen derived from the zirconia-assisted CO₂ activation. Thus, this mechanism allows for the inverse Boudouard reaction to exceed the carbon deposition rate at sufficiently high temperatures [41].

Concerning the literature, a considerably higher carbon diffusivity in Pd than in Ni was reported, which can be assigned to the ~10% larger Pd lattice parameter [43,44]. Cinquini et al. studied carbon diffusion on Ni and Ni₃Pd alloys utilizing DFT and observed an increased stability of subsurface carbon in Ni₃Pd(111) as compared to Ni(111), which suggests higher C-concentrations and less activated diffusion kinetics within the subsurface interlayers of the intermetallic. Therefore, a sufficient palladium addition to Ni expands the lattice parameter and, consequently, changes the thermodynamic stability of carbon, eventually leading to a higher carbon solubility, at least in Pd-richer NiPd alloys [45]. As the NiPd I data show hardly any significant change of C-antisegregation/solubility trends in comparison to pure Ni, we assign this to a still-too-low Pd concentration to induce measurable effects.

As a matter of fact, the carbon solubility on Ni(111) is perfectly suitable for efficient graphene growth, as confirmed by Rameshan et al., who formed a fully surface-covering epitaxial graphene overlayer through the thermal decomposition of a strongly dissolved carbon-supersaturated/surface-carbidic precursor state accumulated in the surface-near regions. The respective transformation was accompanied by slow carbon loss to deeper bulk regions alongside the efficient C resegregation from these supersaturated regions upon cooling. The long-range epitaxial growth of graphene on Ni(111) arises from a near-perfect lattice match [46]. For Cu, in contrast, the carbon diffusion compared with Ni can be neglected since the carbon solubilities in copper and nickel are ~0.0005 at.% and ~0.9 at.% at 900 °C, respectively [47,48].

As a consequence, the carbon bulk diffusion in selected catalytic model systems during DRM, studied by sputter depth profiling, depends strongly on the bimetallic composition used and can be assumed to decrease in the following order: Pd > NiPd > Ni > NiCu > Cu. As the C solubility will only be enhanced by the sufficient addition of Pd to Ni, we tentatively blame improved methane adsorption properties of the NiPd I surface along with still “too good” graphene nucleation/growth properties for its even poorer coking/deactivation properties (in comparison to pure Ni). Nevertheless, once an active metal oxide boundary is present on any NiPd catalyst, e.g., supported on ZrO₂, one may expect that enhanced C diffusion can give rise to increased carbon clean-off through the inverse Boudouard reaction, in analogy to ZrO₂-supported pure Pd.

Instead, unsupported Ni model catalyst foils doped with Cu, such as NiCu III, suppress the C solubility, and thus the nucleation kinetics of graphene/graphite on the surface, resulting in an improved “bimetal-intrinsic” carbon tolerance and reactivity throughout the DRM process.

Since the DRM performance of NiCu catalysts depends critically on the surface composition, they need to retain specific stability under these conditions. It is known that segregation processes occur in NiCu solid solutions, on which the surface is enriched with Cu due to the lower surface energy of Cu compared to Ni [49,50]. Furthermore, NiCu systems possess a wide miscibility gap at low temperatures, leading to the expectation of coexisting Ni-rich and Cu-rich bimetallic phases before and after specific heat treatments [51]. Wolfbeisser et al. reported that NiCu alloys show only a limited stability during methane decomposition, during which the carbon-induced segregation of Ni species occurred [52]. Therefore, the compositional variation of the investigated NiCu catalysts, in particular of their surfaces, was examined throughout DRM testing. Figure 8 illustrates the surface molar fraction of Ni before and after the DRM of all prepared NiCu catalysts, with values taken from

Table 4. Overview of the surface molar fraction of Ni before and after the temperature-programmed DRM reaction of each NiCu catalyst.

	Molar Fraction of Ni
Sample	Prior DRM	Post-DRM 1	Post-DRM 2
NiCu Ia	0.1376	0.1628	-
NiCu Ib	0.1628	0.1770	-
NiCu Ic	0.3074	0.4287	-
NiCu II	0.3188	0.3565	0.5309
NiCu III	0.8280	0.6846	0.7645
NiCu IV	0.9257	0.9556	-

, and indicates an overall Ni surface enrichment, except for NiCu III.

Additional DFT studies on NiCu nanoparticles by Wolfbeisser et al. outline that the adsorption of CH_x groups induces Ni segregation, whereby more dehydrogenated CH_x groups adsorb more strongly and lead to enhanced Ni segregation and stabilization. These authors prepared NiCu nanoparticles which revealed Cu-rich surfaces after synthesis and reduction, whereas Ni-rich surfaces were obtained after methane decomposition [53]. Analogously, on the herein studied NiCu catalysts, Cu segregated on the surface after thermal annealing in UHV (not shown) and DRM instead induced Ni segregation, as shown in Figure 8.

Comparing Figure 8 with Figure 6 (left side), a Ni surface enrichment on the more Cu-rich samples appears to be coupled with carbon enrichment. This can be explained by the suppressed carbon clean-off reaction by CO₂ leading to increased coking. While the deposited carbon layers appear to induce some extra Ni segregation to the surface, these Ni segregated states hardly exceeded a Ni concentration of ~50%. This may explain why these catalysts never became active in repeated cycles of DRM. On NiCu III, instead the surface region even becomes enriched in Cu after the first DRM cycle, but on average, the Ni concentrations stayed in the region between ~70 and ~90 at.%, as seen in Table 4. The NiCu III surface concentrations after the third DRM cycle (91.35 at.% Ni) correspond very well to the related bulk concentrations (Figure 9), which indicate a homogeneous NiCu distribution between the surface and subsurface region. Taken together, strong preferential segregation phenomena (although minor effects are observable) seem to be rather suppressed in NiCu III, which ensures a stable Ni-rich state of the surface, and thus the apparently total suppression of coking without sacrificing still-high CO₂ conversion.

Although NiCu II shows an increased Ni segregation on the surface with each DRM cycle (53.09 at.% Ni after second cycle), it shows a poor activity in the first cycle and practically no activity in the second cycle. This means that the catalyst surface stays too Cu-rich in any case when becoming active. In contrast, the bulk region exhibits only 24 at.% Cu after the second DRM cycle (Figure 2c), which can be assigned to a further loss of already dissolved Cu into the effectively infinite Ni bulk through the prolonged temperature treatments. Hence, only sufficiently diluted NiCu bulk alloys allow for an overall homogenous NiCu distribution, thus preventing or reducing undesired bulk and surface segregation processes. As a consequence, they appear more likely to remain in an invariant state regarding surface and bulk compositions during the dry reforming of methane.

Nevertheless, for the modulation of the surface and bulk carbon chemistry via Cu promotion of the Ni bulk catalysts, a qualitative kinetic scenario of the CH₄ and CO₂ activation kinetics towards the DRM process is proposed in Figure 10, which provides a simplified schematic representation of three types of Cu promotion, namely the pure Ni bulk, the Ni-rich NiCu bulk-intermetallic and the Cu-rich NiCu bulk-intermetallic catalysts, and their reaction behaviour under DRM conditions. According to the observed coking susceptibility, pure Ni bulk systems render a high C-dissolution via enhanced methane decomposition, as indicated by the broader reaction arrow in Figure 10 (left side), leading to an extended graphene/graphite growth. In turn, an overly Cu-rich surface is expected to exhibit insufficient CO₂ dissociation sites (Figure 10, right side), where methane decomposition could induce an unwanted kinetic prevalence in increased C deposition, even though the nucleation kinetics of graphene/graphite on the surface are suppressed through limited C-dissolution. Hence, reducing the adsorption and bulk diffusion rate of the surface carbon to an extent should allow for a virtually immediate clean-off by simultaneously adsorbed CO₂ and retain an appropriate high syngas production rate, alongside the almost total suppression of carbon deposition, as indicated by the Ni-rich NiCu bulk-intermetallic catalysts.

3. Materials and Methods

3.1. Model Catalyst Preparation

As reference catalyst and substrate for NiCu and NiPd alloy formation, ultra-clean Ni foil (Alfa Aesar, 99.994%, 0.1 mm thick, 18 × 20 mm², Thermo Fisher Scientific Inc., Waltham, MA, USA) was used. The cleaning process for pure Ni catalysts involved wet etching, Ar⁺ sputtering and thermal annealing steps to remove surface and bulk contaminations until XPS spectra showed no evidence of impurity traces. For wet etching, the samples were treated in 68% concentrated nitric acid (AnalaR Normapur^® analytical reagent, VWR LLC, Radnor, PA, USA) at 50 °C for 4 min and, afterwards, cleaned in distilled H₂O and an EtOH ultrasonic bath for 15 min. The remaining treatments, i.e., Ar⁺ sputter cleaning (2 keV, 1 µA, 15 min) and thermal annealing (800 °C, 10 min), were conducted with subsequent surface analysis in the combined state-of-the-art UHV/high-pressure catalysis setup (base pressure ≈ 1 × 10⁻⁹ mbar), as described in Section 3.3.

For the NiCu and NiPd alloy formation, Cu (Alfa Aesar, 99.9999%, 0.1 mm thick, Thermo Fisher Scientific Inc., Waltham, MA, USA) and Pd foil (99.95%, 0.1 mm thick, Goodfellow GmbH, Hamburg, Germany) were evaporated in a modified modular high vacuum chamber, described in Section 3.2, onto the ultra-clean Ni foil through thermal evaporation, respectively. Whereas the substrate was kept at a constant temperature of 211 °C throughout the copper deposition, in order to attain the improved initial wetting of the ambient-contaminated Ni surface by the evaporated Cu metal, and subsequently, cooled down. It was only heated once for 5 min at 211 °C prior to the experiment, and substrate heating was switched off during the Pd deposition process, as the surface wetting by Pd was found to be superior. The coating thickness of the thin films was monitored by a quartz crystal microbalance.

A brief overview of the different bimetallic catalyst treatments is given below since the catalysts were not treated entirely equally after the deposition process. NiCu I initially was Ar⁺ sputter cleaned (2 keV, 1 µA, 15 min) and then thermally annealed (800 °C, 10 min) in UHV, which led to the removal of the remaining oxygen/carbon species and annealing of the multi-layered CuNi alloy state. Consequently, its catalytic testing was conducted in the ambient pressure recirculating batch reaction cell (Section 3.3). DRM tests were performed for three different surface states of the same sample and described as separate experiments. The first cycle of the temperature-programmed DRM reaction, which is denoted as NiCu Ia, led only to marginal carbon deposition but to a slightly lowered Cu content on the surface, and thus an additional DRM step (NiCu Ib) was carried out. After the reaction, the sample was covered by some graphitic carbon and nearly no change in the Ni:Cu ratio occurred. Therefore, Ar⁺ sputter cleaning (2 keV, 1 µA, 15 min) and thermal annealing (800 °C, 10 min) in UHV were conducted with subsequent heat treatment in a continuous flow tube furnace in a 10% H₂/He atmosphere at 1000 °C for 30 min (all gases including Ar in 5.0 quality, obtained from Messer SE & Co. KGaA, Bad Soden, Germany), to obtain a higher Ni:Cu ratio. Since the heat treatment was accomplished ex situ, the usual cycle of Ar⁺ sputter cleaning and thermal annealing was required to remove ambient contaminants, in order to perform a third independent DRM cycle (NiCu Ic). NiCu II and NiCu III were first treated in 10% H₂/He atmosphere at 1000 °C for 30 min to achieve enhanced interdiffusion of their components. Afterwards, the usual cycle of Ar⁺ sputter cleaning and thermal annealing led to the removal of ambient contaminants. Subsequently, several cycles of temperature-programmed DRM reactions followed. NiCu IV was treated analogously to NiCu II and NiCu III, with the only difference that the preheating treatment in 10% H₂/He atmosphere was omitted due to the minor amount of deposited Cu, which could be largely dissolved in the Ni substrate during the final UHV annealing step. In contrast, NiPd alloys need higher temperatures for intermixing, and thus NiPd I was pre-heated in 10% H₂/He atmosphere at 1200 °C for 1 h, followed by the usual cycle of Ar⁺ sputter cleaning and thermal annealing and by subsequent temperature-programmed DRM reaction.

As single-component reference catalysts, ultra-clean Cu (Alfa Aesar, 99.9999%, 0.1 mm thick, Thermo Fisher Scientific Inc., Waltham, MA, USA) and Pd foil (99.95%, 0.1 mm thick, Goodfellow GmbH, Hamburg, Germany) were used. The cleaning process for pure Cu and Pd catalysts involved only the usual cycle of Ar⁺ sputter cleaning and thermal annealing to remove surface and bulk contaminations till XPS spectra showed no evidence for impurity traces. Afterwards, a temperature-programmed DRM reaction followed.

3.2. Modular High Vacuum Chamber for Physical Vapor Deposition (PVD)

The thin film deposition on the studied model catalysts was performed through thermal evaporation in a highly flexible modified modular high-vacuum chamber. Its multiple applications and an in-depth description are given in more detail in [54]. The chamber is based on the Schott Duran^® (Schott AG, Mainz, Germany) flat flange system and glass recipients. This system consists of home-built stainless steel ring modules fitting the standardized ground glass flat flanges from Schott DN100 (inner diameter: 90 mm). To arrange two metal rings over each other, fluorocarbon O-rings (diameter: 99 mm), lubricated with high-vacuum grease, were used as a seal. Instead of multiple viewports and to retain simplicity, a glass recipient is used. A quartz crystal microbalance monitors the coating thickness of the thin film, where QS 008 Ag quartz crystals (Umicore S.A., Brussels, Belgium) with a resonance frequency of 5 MHz were employed. Below the glass recipient, there is a custom-built substrate mount for a foil sample with a width of 20 mm, which is the desired size for the used Ni foil substrates of 18 × 20 mm². The boat suspension for the thermal-resistive tungsten boat, in which the source material is placed, is located underneath the substrate mount. Pressure control is provided by a combined PKR 361 Pirani/cold cathode gauge (Pfeiffer Vacuum AG, Aßlar, Germany) placed in the segment below, and subsequently, a liquid-nitrogen-filled copper vessel, used as a cooling trap, is placed there. The chamber is pumped by a turbo-molecular pump (300 L/s) and a two-stage rotary vane vacuum pump located below the operating desk level of the system, providing a base pressure in the low 10⁻⁷ mbar range. The substrate mount and the boat suspension are connected separately through electrical feedthroughs to a low-voltage transformer. As source materials, the above-described Cu and Pd foils were used to prepare the desired NiCu and NiPd catalyst films, respectively. The tungsten boat was heated resistively by applying a current via the transformer in dependency on the used source material. A current of above 220 A for Cu and 260 A for Pd is needed.

During the copper deposition, the substrate mount is additionally heated to a temperature of 211 °C by applying 78 A, which increases the sticking probability and adhesive growth properties of the deposited Cu. The quartz crystal microbalance, the boat suspension, and the transformer connectors are all water-cooled through their feedthroughs. Additional water cooling of the substrate is possible, which is necessary for a subsequent cooldown of the Cu-deposited Ni foil substrate and for maintaining a cooled substrate during Pd deposition. In terms of Pd deposition, the substrate mount is heated once for 5 min at 211 °C in a prior evaporation process, followed by subsequent cooling. The adhesion of Pd on Ni is better than that of Cu; hence, no improvement in its adhesive growth properties is needed.

3.3. Combined XPS/AES/LEIS-High Pressure Batch Reactor Setup

For the surface and near-surface spectroscopic characterization and catalytic testing of model catalysts, a combined state-of-the-art UHV/high-pressure catalysis setup is used. For an in-depth description, the work of Mayr et al. is suggested [55]. This UHV system is capable of XPS, AES (Auger Electron Spectroscopy) and LEIS (Low Energy Ion Spectroscopy) investigation and is equipped with a transfer system to shift the samples into an attached differentially pumped high-pressure all-quartz reactor cell for ambient-pressure gas and catalytic treatments. The sample annealing step is carried out via a home-built e-bombardment setup with four helical thoriated tungsten filaments, where the filament emission current controls their heating power. An Ar⁺ ion sputter gun with ions accelerated to 2 keV is used to clean the sample’s surface. An Alpha 110 hemispherical electron energy analyser (Thermo Fisher Scientific Inc., Waltham, MA, USA), a XR 50 twin Al/Mg X-ray gun (SPECS Surface Nano Analysis GmbH, Berlin, Germany) and an EFG-7 electron gun (Kimball Physics Inc., Wilton, NH, USA) are attached to the UHV chamber to analyse the sample spectroscopically. The herein obtained XP spectra are recorded through a non-monochromatic Mg X-ray source with an incident photon energy of 1253.6 eV at 250 W, a base pressure of ~1 × 10⁻⁹ mbar and an analyser angle of 54.7° (magic angle). The analyser is operated at a constant pass energy E_pass of 50 eV for the survey spectra and 20 eV for the high-resolution spectra. Cu LMM Auger spectra were recorded in direct mode to distinguish metallic copper from its oxides in the studied catalysts. Therefore, the electron gun was operated with a primary electron beam energy of 3.0 keV.

Catalytic DRM experiments of the prepared catalysts were performed in the attached ambient pressure recirculating batch reaction cell. The reaction cell consists of a fused-quartz glass tube surrounded by a cylindrical furnace (P330, Nabertherm GmbH, Lilienthal, Germany), with a total circulation volume of 296 mL and an accessible maximum temperature of 1150 °C. For online reactant/product quantification, an adjustable capillary leak leads directly to the mass spectrometer of the gas chromatograph (G1800A GCD System, HP Inc., Palo Alto, CA, USA). As a reference gas, argon (Ar 5.0, Messer SE & Co. KGaA, Bad Soden, Germany) is added up to 30 mbar to the gas mixture to normalize the electron impact detection mass spectrometry (EID-MS) partial pressure signals to the temperature-dependent effective reactor volume and the initial amounts of reactants (local pressure changes due to gas temperature change and continuous decrease in total pressure due to continuous gas withdrawal in the mass spectrometer). The normalized MS signals are calibrated using pure substances with quantitative consideration of fragmentation. All DRM experiments were carried out in an atmosphere of 50 mbar CH₄, 50 mbar CO₂, 30 mbar Ar and 870 mbar He (all gases in 5.0 quality, obtained from Messer SE & Co. KGaA, Bad Soden, Germany) leading to a total pressure of 1 bar. The reactor was equilibrated and premixed at 30 °C for 10 min. Consequently, a linear temperature ramp (heating rate: ~10 °C/min) up to 800 °C within 80 min was applied, followed by a subsequent isothermal period at 800 °C for 30 min. The catalytically measured molar product formations and molar rates were normalized to the geometrically estimated total amount of surface Ni atoms of the respective catalyst to obtain the TON and TOF values, respectively, by assuming a mean value of the Ni metal surface atom density of 1.74 × 10¹⁵ cm⁻² for every catalyst. The cleaned surface area of the catalysts amounts to 2 × 2 cm x 1.8 cm = 7.2 cm², corresponding to a total number of 1.25 × 10¹⁶ surface Ni atoms, on which all TON and TOF estimations are based.

3.4. XPS Setup for Depth Profile Studies

XPS depth profile studies were performed with a Multilab 2000 instrument equipped with an Alpha 110 hemispherical channel analyser, an XR5 X-ray monochromator with an integrated aluminium anode and an EX05 ion gun (complete instrumentation from Thermo Fisher Scientific Inc., Waltham, MA, USA). Furthermore, this system contains an optional XR3 twin Al/Mg X-ray gun. The ion sputter gun accelerated the Ar⁺ ions to 3 keV across 2.0 mm² on the sample with an ion current of 1.0 µA. For charge compensation on non-conducting samples, an electron flood gun providing electrons with a kinetic energy of 6 eV was employed. The XP spectra were recorded through a monochromatized Al X-ray source with an incident photon energy of 1486.6 eV at 150 W, a base pressure of ~1 × 10⁻⁹ mbar and an analyser angle of 54.7°. The analyser was operated at a constant pass energy E_pass of 100 eV for the survey spectra and 25 eV for the high-resolution spectra.

Depth profiling by ion sputter etching led to material removal, which destroyed the surface of the catalyst. By increasing the Ni foil size used for the preparation routine and subsequent sample cutting, it was possible to obtain separate pieces of foil for characterization and DRM testing. Thus, stepwise characterization before/after the distinct treatments was possible and yielded a consistent comparison. A Ni foil substrate of 26 × 20 mm² was initially used for the NiCu II catalyst preparation. Two 4 × 20 mm² pieces cut away after thermal evaporation and after heat treatment, respectively, results in the final desired catalyst size of 18 × 20 mm² for the subsequent DRM step.

3.5. Details of XPS Analysis

The XPS and AES data were analysed using the CasaXPS software program, version 2.3.16 pre-rel. 1.6 (Casa Software Ltd., Teignmouth, UK) [56]. A Shirley background was applied to all spectra, and the associated Scofield relative sensitivity factors [57] were used for quantitative analysis. All binding energies were referenced to the Fermi edge. The carbon film thickness was estimated in monolayers from an XPS overlayer attenuation model using the program XPS Thickness Solver [58]. Therefore, electron attenuation lengths were taken from the NIST database SRD-82 [59], and the orbital asymmetric parameters were taken from the ELETTRA online database [60].

4. Conclusions

A series of ultra-clean NiCu and NiPd near-surface alloys on Ni foil substrates were prepared by thermal evaporation PVD and compared to monometallic Ni, Cu and Pd model catalysts with respect to their carbon tolerance and catalytic activity in dry reforming of methane. It was found that wet etching with concentrated HNO₃ and a subsequent single sputter–anneal cycle provided the complete removal of an already existing oxidic/passivation layer and segregated and/or ambient-deposited surface/bulk impurities on all these systems, respectively.

The catalytic performance with respect to the DRM process on Ni-based catalysts, especially on NiCu catalysts, generally exhibited a high sensitivity to doping, which became gradually diminished with increased dopant amount. According to the observed coking susceptibility, minor Cu addition in the range around 15 at.% led to reduced or even total coking suppression, whereas higher amounts of Cu again facilitated carbon growth. The carbon resistance of surfaces with low Cu content can be related to changed electronic properties, involving the reduced interaction of, e.g., the CH 1π orbital with the 3d states of the active sites, leading to a slightly increased dissociation barrier of CH decomposition, which nevertheless allows for the required and well-balanced sequence of both dissociation reactions, namely of CH₄ and CO₂, due to sufficiently available Ni ensembles with optimized electronic structure. At higher Cu loadings, carbon supply to the surface, e.g., via CO disproportionation via the Boudouard reaction and/or methane decomposition, may eventually start to override CO₂ activation, which is expected to become progressively blocked by a Cu-richer surface with insufficient CO₂ dissociation sites. This, in turn, could induce an unwanted kinetic prevalence for increased carbon deposition, especially at temperatures around 700 °C, despite the fact that a Cu-richer surface prevails. On the other hand, minor Cu addition can protect the catalytic surface from this form of carbon formation due to enhanced clean-off of reactive atomic C species via nearby activated CO₂ towards more weakly bonded CO molecules, leading in due course to enhanced CO desorption. Hence, an optimal NiCu composition can be suggested in the region of 70–90 at.% Ni, which should allow an appropriate high syngas production rate to be retained and an almost total suppression of carbon deposition.

In contrast, NiPd and Pd model catalysts showed a severe deactivation towards DRM due to enhanced carbon solubility coupled with a high decomposition rate of methane on Pd-doped and pure Pd-based systems, as our depth profile studies evidenced. The observed carbon bulk diffusion on the investigated systems depends strongly on the composition and follows the trend: Pd > NiPd > Ni > NiCu > Cu. Nevertheless, if an active metal oxide boundary is present on Pd-doped/Pd-based catalysts, e.g., if sufficiently small nanoparticles are in close contact with ZrO₂, the enhanced C diffusion becomes rather beneficial, leading to increased carbon clean-off via the inverse Boudouard reaction [41,42]. The catalytic function of Cu addition to Ni catalysts is totally different, as it instead allows for the suppression of carbon solubility, and thus the nucleation kinetics of graphene/graphite on the surface, resulting in an improved carbon tolerance and, provided that the Cu content is optimized for sufficient CO₂ activation, stable carbon conversion and reactivity throughout the entire DRM process.

Furthermore, an activation step with a rapid conversion onset in the temperature region between 649 and 677 °C occurred for all highly active catalysts, which emerges most likely due to the removal of a passivating oxide layer that formed on the catalyst’s surface during the exposure to the reaction mixture at lower temperatures.

Moreover, the more Cu-rich “inverse” NiCu model catalyst systems showed a limited long-term stability under reaction conditions. Undesired segregation processes involved the carbon-induced segregation of eventually inactive Ni-CH_x surface species and/or Cu diffusion into the infinite Ni bulk through long-term temperature treatments. Therefore, the catalytic performance under DRM cyclization conditions should be interpreted with care.

Consequently, NiCu bulk alloys, as catalytic films with an overall homogenous NiCu distribution, can prevent or reduce segregation processes within the suggested optimized NiCu compositional region, and allow a specific stability and an appropriate high syngas production rate to be retained, alongside an almost total coking suppression during the dry reforming of methane. The implications for optimized supported NiCu catalyst systems require the transfer of this knowledge to technologically more relevant catalyst preparations.