ARTICLE

https://doi.org/10.1038/s42004-023-01018-w OPEN

Dry reforming of methane over gallium-based

supported catalytically active liquid metal solutions
Moritz Wolf 1,2, Ana Luiza de Oliveira1,2, Nicola Taccardi1, Sven Maisel3, Martina Heller4,
Sharmin Khan Antara1, Alexander Søgaard1, Peter Felfer4, Andreas Görling 3, Marco Haumann 1 &
Peter Wasserscheid 1,2 ✉

Gallium-rich supported catalytically active liquid metal solutions (SCALMS) were recently
introduced as a new way towards heterogeneous single atom catalysis. SCALMS were
1234567890():,;

demonstrated to exhibit a certain resistance against coking during the dehydrogenation of

alkanes using Ga-rich alloys of noble metals. Here, the conceptual catalytic application of
SCALMS in dry reforming of methane (DRM) is tested with non-noble metal (Co, Cu, Fe, Ni)
atoms in the gallium-rich liquid alloy. This study introduces SCALMS to high-temperature
applications and an oxidative reaction environment. Most catalysts were shown to undergo
severe oxidation during DRM, while Ga-Ni SCALMS retained a certain level of activity. This
observation is explained by a kinetically controlled redox process, namely oxidation to gallium
oxide species and re-reduction via H2 activation over Ni. Consequentially, this redox process
can be shifted to the metallic side when using increasing concentrations of Ni in Ga, which
strongly suppresses coke formation. Density-functional theory (DFT) based ab initio mole-
cular dynamics (AIMD) simulations were performed to confirm the increased availability of
Ni at the liquid alloy-gas interface. However, leaching of gallium via the formation of volatile
oxidic species during the hypothesised redox cycles was identified indicating a critical
instability of Ga-Ni SCALMS for prolonged test durations.

1 Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Chemische Reaktionstechnik (CRT), Egerlandstr. 3, 91058 Erlangen, Germany.
2 Forschungszentrum Jülich, Helmholtz Institute Erlangen-Nürnberg for Renewable Energy (IEK 11), Cauerstr. 1, 91058 Erlangen, Germany. 3 Friedrich-
Alexander-Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Theoretische Chemie, Egerlandstr. 3, 91058 Erlangen, Germany. 4 Friedrich-Alexander-
Universität Erlangen-Nürnberg (FAU), Lehrstuhl für Werkstoffwissenschaften (Allgemeine Werkstoffeigenschaften), Martensstr. 5, 91058 Erlangen, Germany.
✉email: [email protected]

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 1

ARTICLE COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w

D
ry reforming of methane (DRM, Eq. 1) oxidatively con- products in liquid metals confines the catalytic reaction over
verts CH4 with CO2 to synthesis gas, which makes this SCALMS to the liquid metal/gas interface20,26,31–35. The active
process a promising alternative for a sustainable valor- sites in SCALMS have been described as single atoms of the
isation of methane. The integration of CO2 in the production of secondary metal in a Ga matrix20,26,33–35.
H2 and CO is ecologically highly attractive and a first step Herein, we report the first application of SCALMS in a high-
towards a circular economy1–5. The strong C–H bonds in CH4 in temperature process exceeding 700 °C. The performance of Ga-
combination with the mildly oxidising CO2 make DRM a highly rich SCALMS using mesoporous SiC as carrier material was
endothermic process. The operation temperatures for DRM are in conceptually evaluated during dry reforming of methane (DRM)
the range of 700–1000 °C3, which induces rapid catalyst deacti- in the temperature range of 700–1000 °C3. Furthermore, DRM
vation via coking, oxidation of the active metallic phase, and over SCALMS also represents the first application of this SLP
sintering4. In fact, long-term stability is one of the major chal- catalysis concept for oxidative reaction mixtures. This is an
lenges in the commercialisation of DRM. Hence, recent advances essential difference due to the high oxophilicity of Ga. In our
in catalyst research for DRM mostly focus on concepts to prevent study, the evaluation of the catalytic performance of Ga-rich
excessive sintering and carbon deposition to improve the stability SCALMS with secondary metals that are potential candidates for
and long-term performance5–9. catalysing DRM (Co, Cu, Fe, Ni) is accompanied by microscopic
characterisation of the materials prepared. Moreover, the
CH4ðgÞ þ CO2ðgÞ ! 2H2ðgÞ þ 2COðgÞ ð1Þ SCALMS materials are thoroughly analysed after their catalytic
application to understand the mechanisms at play.
Sintering of the active phase and coke formation are classical
deactivation pathways for heterogeneous catalysts10. Sintering of
Ni-based catalysts during DRM is often discussed in the Results and discussion
literature5,11–16. Several recent studies focus on novel reactor and Catalyst characterisation. Four abundant metals, namely Co, Cu,
catalyst concepts to suppress (excessive) carbon deposition in Fe and Ni, were selected as potential candidates for SCALMS dry
order to enable long-term operability5,17–19. Two innovative reforming of methane (DRM) catalysts. Gallium-based SCALMS
approaches employing liquid alloys as the catalytically active were prepared with these secondary metals via impregnation of
phase in a bubble column reactor demonstrated the successful pre-synthesised gallium-decorated mesoporous β-SiC (BET sur-
elimination of both deactivation mechanisms for DRM18. How- face area of 23 m2 g-1; Figs. S1–S3). Analysis of the metal content
ever, the approach with liquid alloys in a bubble column reactor resulted in Ga loadings of 4.5-4.9 wt.% and Ga/metal ratios close
suffers of practical issues, because of the highly corrosive nature to the targeted value of 50 (Table 1). According to the corre-
of the liquid metals and large required quantities resulting in high sponding bimetallic phase diagrams36–39, all of these Ga-rich
reactor costs. alloys are expected to be present in the liquid state at tempera-
Such limitations in the applicability may be overcome by tures exceeding 700 °C. Furthermore, passivation of the prepared
combining supported liquid phase (SLP) catalysis with liquid SCALMS by a thin layer of GaxO species has been previously
alloys. In general, SLP catalysis merges advantages from classical reported but may be easily reduced during catalysis by in situ
heterogeneous and homogeneous catalysis, such as easy handling formed reductants, such as H226,28,29. This observation has been
of the catalyst and well-defined active sites, respectively20. How- assigned to H2 activation by the secondary metal allowing
ever, the application of SLP catalysts with supported organic reduction of GaxO species at low temperatures.
liquids, ionic liquids, or molten salts is limited due to the The as-prepared SCALMS materials were characterised by
restricted thermal stability of these liquid phases21–24. Recently, means of scanning electron microscopy with elemental mapping
we introduced a new class of SLP catalysts, namely gallium-rich via energy-dispersive X-ray spectroscopy (SEM-EDX) to evaluate
supported catalytically active liquid metal solutions (SCALMS), the morphology of the Ga-rich droplets on the external surface of
which may extend the application of SLP catalysis to high- the β-SiC support material and the distribution of the active metal
temperature processes due to the low metal vapour pressures (Fig. 1; Figs. S4–S5). The Ga-droplet decoration in SCALMS was
(boiling point of Ga: 2400 °C) and high thermal stability of liquid previously achieved via impregnation followed by the thermal
metals20,25. Application of SCALMS materials in the dehy- decomposition of gallane complexes (Et3N)GaH320,26,28,29. When
drogenation of propane resulted in suppression of carbon using ultrasound emulsification, the Ga droplets are sonochemi-
deposition26–29, which enables extended dehydrogenation cycle cally produced from elemental Ga, dispersed in a solvent, and can
times in potential commercial applications29,30. SCALMS mate- be directly deposited on the substrate40. Even though the rough
rials are composed of supported droplets of a liquid alloy con- outer surface of the mesoporous support may enhance physical
sisting of a catalytically active metal and an excess of a low stabilisation, the droplets showed a certain degree of clustering. A
melting metal20,25,26,28,29. The insolubility of reactants and wide size distribution of the droplets was observed, ranging from

Table 1 Chemical composition of the supported catalytically active liquid metal solutions (SCALMS) prepared.

Catalyst Metal loading (as prepared)/wt.% Molar ratio Metal loading (spent)a/wt.%
Ga Secondary metal Ga Secondary metal
Ga47Co/SiC 4.85 0.09 47.2 3.68 0.08
Ga52Cu/SiC 4.91 0.09 51.6 4.76 0.08
Ga44Fe/SiC 4.53 0.08 43.7 3.46 0.08
Ga45Ni/SiC 4.52 0.08 45.3 4.25 0.08
Ga12Ni/SiC 5.44 0.38 12.1 4.44 0.39
3.70 0.37 (100 h TOS)
aLoadings of the spent catalysts were corrected for the amount of carbon as identified by means of TGA.

2 COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem

COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w ARTICLE

Fig. 1 Morphology of gallium-rich phase on SiC. a Scanning electron

micrograph, b backscattered micrograph and elemental mapping of
c gallium and d oxygen via energy-dispersive X-ray spectroscopy of
Ga45Ni/SiC with metal loadings of Ga and Ni of 4.52 and 0.08 wt.%,
respectively.
Fig. 2 Dry reforming of methane over supported catalytically active liquid
less than 0.1 up to 3 μm. The low concentration of the secondary metal solutions (SCALMS). a Conversion of CH4 and b CO2 using various
metals (<0.09 wt.%) resulted in low signal-to-noise ratios for the active metals in gallium-rich alloys employing a mesoporous SiC support.
SEM-EDX elemental maps. Nevertheless, the acquired maps Metal loadings: 4.85 wt.% Ga and 0.09 wt.% Co for Ga47Co/SiC, 4.91 wt.%
suggest a successful deposition of Ni onto the Ga phase (Fig. S5). Ga and 0.09 wt.% Cu for Ga52Cu/SiC, 4.53 wt.% Ga and 0.08 wt.% Fe for
Ga44Fe/SiC, 4.52 wt.% Ga and 0.08 wt.% Ni for Ga45Ni/SiC. Reaction
Catalyst testing. The performance of the four SCALMS systems conditions: 900 °C, 1 bar, 1 g SCALMS, CH4:CO2:Ar = 1:1:3, 3 LN gcat-1 h-1.
during DRM was evaluated at 900 °C and atmospheric pressure in
a quartz tube fixed-bed reactor with an equimolar feed of 3:1:1
Ar:CH4:CO2. The blank quartz tube reactor displayed negligible
conversion levels (<1% for CH4 and below detection limit for
CO2; Fig. S6). Of the four metals studied, the presence of Ni
within the liquid Ga matrix in the Ga45Ni/SiC SCALMS resulted
in the highest conversion of CH4 and CO2 (Fig. 2). No increase of
activity was observed when comparing the Ga44Fe/SiC SCALMS
catalyst with the Ga/SiC benchmark without any secondary metal
(Fig. S7). The initial conversion of CH4 and CO2 was the highest
for Ga52Cu/SiC SCALMS (Fig. 2). However, this catalyst rapidly
deactivated and displayed the activity level of the Ga/SiC refer-
ence after 2 h time on stream (TOS). Aside from the Ga-Ni
SCALMS, the Ga43Co/SiC SCALMS was the only other system
displaying a certain level of DRM activity. However, the con- Fig. 3 Proposed reaction network for dry reforming of methane over
version levels of this Co-based SCALMS system rapidly decreased gallium-based supported catalytically active liquid metal solutions
from an initial conversion of CH4 of 14% to 7% after 5 h TOS, (SCALMS). Reaction 1: Activation of methane; reaction 2: Boudouard
close to the conversion of the Ga/SiC reference. While the initial reaction; reaction 3: reverse water gas shift reaction; reaction 4: oxidation
conversion of CH4 was only marginally improved over the of Ga0 by carbon dioxide; reaction 5: reduction of GaxO by in situ formed
Ga45Ni/SiC SCALMS, an initial activation period of one hour was hydrogen; reaction 6: reduction of GaxO by in situ formed monoatomic
observed obtaining conversion levels of CH4 and CO2 of 24 and carbon.
34%, respectively (Fig. 2). Subsequently, the Ga-Ni SCALMS
slowly deactivated over 20 h TOS to 8 and 11% conversion of simplified reaction network, monoatomic carbon is then gasified
CH4 and CO2, respectively. via the Boudouard reaction yielding CO (Fig. 3, reaction 2). The
Interestingly, CO2 conversion in the Ga-Ni and Ga-Co formation of CO is typically increased to the sacrifice of H2 due
SCALMS catalysed DRM is 50% higher than CH4 conversion to the additional conversion of CO2 via the reverse water gas
throughout the experiment (Fig. 2; Fig. S8). The Ga-Cu and Ga- shift (RWGS) reaction (Fig. 3, reaction 3). Hence, synthesis gas
Fe SCALMS also result in an initially enhanced conversion of from DRM is typically rich in CO and the compositions
CO2 prior to deactivation within the first hours of TOS. This reported in the literature vary in H2/CO ratios from 0.8 to
distinct deviation from the equimolar conversion of the reactants equimolarity42,43. Herein, the high reaction temperature of
suggests an additional conversion pathway for CO2, which ceases 900 °C allows for enhanced conversion of CO2 via the RWGS
in the case of strongly deactivating Ga52Cu/SiC or barely active reaction when compared to lower temperatures (Fig. S9). This
Ga44Fe/SiC within 5 h TOS (Fig. S8). was confirmed in dedicated experiments studying the extent of
RWGS over Ga-Ni SCALMS and the SiC support material
Proposed reaction network. Activation of CH4 during thermo- (Fig. S10). While, the bare support material resulted in low
catalytic DRM can be described by the formation of H2 alongside conversion levels, almost equilibrium conversion was achieved
adsorbed CHx or monoatomic carbon (Fig. 3, reaction 1)41. In a for CO2 over the Ga45Ni/SiC SCALMS at 900 °C. Hence, the

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 3

ARTICLE COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w

RWGS reaction is expected to play an important role as a side Similar redox cycles of the active metal phase during DRM
reaction during DRM and increases the conversion of CO2. have been hypothesised in the literature. Ruckenstein et al.
However, additional conversion pathways for CO2 are at play as reported a solid solution of CoO in MgO forming small clusters,
the obtained H2/CO ratio during DRM over Ga45Ni/SiC stabi- which hindered sintering during DRM51. Further, the strong
lises at a value of 0.55 after 1 h TOS (Fig. S8). interaction suppressed excessive coking. The authors hypothe-
Gallium is highly oxophilic with a low standard Gibbs free sised that the active Co atoms are constantly oxidised by CO2 and
energy (ΔGrxn0) of the oxidation reaction with O2 forming Ga2O3 H2O and reduced by CH4 and H2 during DRM at 900 °C, which
of -1996.0 kJ mol-1 (Fig. S12a), which is more than twice as high was key to achieving high stability and catalytic activity for a 12
as for full oxidation of pyrophoric metallic Co to Co3O4 wt.% Co/MgO catalyst. More recently, McFarland et al. per-
(–802.1 kJ mol-1). Expectedly, oxidation of liquid Ga0 by mildly formed combined dry reforming and pyrolysis of CH4 in a
oxidative CO2 (Fig. 3, reaction 4) is also feasible with a Gibbs free molten metal bubble column reactor using a Ni1.86In alloy18.
energy at reaction temperature (900 °C) of –160.7 kJ mol-1 Continuous redox cycles were hypothesised for the metallic In
(Fig. S12b). In addition, partial oxidation of Ga0 is even more phase. In line with early studies by Otsuka et al.52, cyclic
likely than full oxidation to Ga2O3 in the present system. oxidation of In and reduction of In2O3 allowed for the conversion
Nevertheless, bulk oxidation of Ga0 is unlikely as passivation by a of CO2 in the molten metal bubble column reactor, while the
thin GaxO layer has been previously identified after exposure to oxidic intermediate serves as an oxygen shuttle between CO2 and
the ambient atmosphere during the preparation of CH418.
SCALMS26,28,29. Hence, we hypothesise additional conversion
of CO2 to CO during DRM via partial oxidation of metallic liquid Role of nickel concentration. A Ga12Ni/SiC SCALMS with an
Ga (Fig. 3, reaction 4). increased concentration of Ni (Table 1) was prepared to study the
The presence of H2 may enable the reduction of surface oxidic hypothesised promotional effect of active metal atoms on the
gallium species even though the direct reduction of pure Ga2O3 is kinetically controlled Ga0-GaxO redox reaction and con-
thermodynamically restricted at 900 °C. However, the reduction sequentially the activity and long-term stability during DRM.
of GaxO has recently been monitored during the dehydrogenation This alloy composition is expected to have a liquidus line at
of propane over Ga-Rh and Ga-Pt SCALMS at 500 °C by means approx. 600 °C39. The increased concentration of Ni in the
of in situ high-resolution thermogravimetric analysis coupled SCALMS enabled elemental mapping of the active metal using
with mass spectrometry (HRTGA-MS). This observation was SEM-EDX (Fig. 4; Fig. S13), while the general morphology of the
linked to the presence of the secondary metal for the activation of supported Ga phase resembles the as-prepared Ga45Ni/SiC
H2 and for the reduction of the gallium-oxygen binding SCALMS (Fig. 1). The analysis confirms co-location of Ni and Ga,
strength28,29. Hence, the presence of a dissolved metal at the which is a prerequisite for the formation of SCALMS upon in situ
GaxO interface, such as Ni, changing the gallium-oxygen binding reduction of the GaxO passivation layer and calcined Ni pre-
energies and being capable of H2 activation can be expected to cursors. Similar to the Ga-Ni SCALMS with a lower concentra-
allow for the reduction of adjacent GaxO species (Fig. 3, reaction tion of Ni, such an activation within the first hour of DRM at
5), which has recently been observed under mild conditions 900 °C was also observed for the Ga12Ni/SiC SCALMS (Fig. 5).
(300 °C) by our group for Ga-Ni SCALMS44. Reasonably, the The maximum CH4 conversion level after this activation period
mechanism may be similar to H2 spillover from promoters to the was significantly increased from 24 up to 59%. Subsequently, the
active metal in conventional heterogeneous catalysts45, i.e., H2 is Ni-enriched SCALMS also deactivated to a CH4 conversion of
adsorbed and dissociated on the active secondary metal providing 26%. However, contrary to the Ga45Ni/SiC SCALMS that
adjacent GaxO species with activated hydrogen. Hence, the underwent a steady deactivation, the conversion over the Ga12Ni/
presence of the secondary metal at the interface will also strongly SiC SCALMS stabilised after 15 h TOS and was demonstrated to
affect the physico-chemical properties, while the partially oxidised be stable over 100 h TOS in a dedicated long-term experiment
GaxO species are generally expected to deviate from the (Fig. S14). Note, that the specific activity of Ni of approx. 0.1
thermodynamically stable Ga2O3. Lastly, the reverse reaction, molCH4 h-1 gNi-1 is inferior when compared to classical catalyst
that is oxidation of liquid Ga0 by H2O, has been reported to be
kinetically hindered on some metals due to the high stability of
hydroxyl groups46–49.
Monoatomic carbon from CH4 activation may also serve as a
reducing agent for GaxO species (Fig. 3, reaction 6). Thermo-
dynamic calculations reveal a strong temperature dependency of
the reduction of Ga2O3 by graphite to Ga0 and CO indicating a
small Gibbs free energy at reaction conditions (900 °C) of
56.8 kJ mol–1 (Fig. S12b). Nevertheless, reduction via this path-
way may become feasible for thermodynamically less stable,
highly reactive monoatomic carbon in combination with partially
oxidised GaxO species (Fig. 3, reaction 6)50. Hence, the closed
redox cycle with reduction by monoatomic carbon potentially
allows for the Boudouard reaction via a Mars-van Krevelen-type
mechanism with oxidation of Ga by CO2 filling the oxygen
vacancies formed by reduction with carbon. However, the impact
of this reaction on the kinetically controlled Ga0-GaxO redox
process may be limited when compared to the reduction pathway
via activated H2 (Fig. 3, reaction 5). Furthermore, only oxidation Fig. 4 Morphology and metal distribution in with higher nickel
of Ga0 by CO2 in combination with a preferential reduction of concentration. a Backscattered scanning electron micrograph and
GaxO species via H2 activation on the secondary metal (Fig. 3, elemental mapping of b gallium, c oxygen, and d nickel via energy-
reaction 3) results in an enhanced conversion of CO2 when dispersive X-ray spectroscopy of Ga12Ni/SiC with metal loadings of Ga and
compared to CH4. Ni of 5.44 and 0.38 wt.%, respectively.

4 COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem

COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w ARTICLE

SCALMS20,26,33. However, owing to the high mobility of the

liquid system at elevated temperatures, the Ni atoms can easily
access the surface from time to time to act as single atoms
catalytic centres. To quantify the amount of Ni available at the
surface, the number of Ni atoms that are present in the first layer
of the density profile was evaluated (see Fig. 6a for definition of
the first layer). The calculations suggest that 2.5% and 0.7% of the
surface atoms are Ni atoms in the case of Ga12Ni and Ga45Ni,
respectively. Hence, the availability of Ni atoms at the surface of
Ga12Ni is 3.6 times higher than for Ga45Ni, which also
approximates the increase in the conversion of CO2 (Fig. 5).
The relative depletion of Ni atoms at the surface is nearly equal in
both cases. The higher Ni surface concentration in Ga12Ni simply
results from the higher total amount of Ni in Ga12Ni. This higher
surface concentration may explain the higher catalytic activity
going along with better long-term stability due to a higher rate of
H2 activation, which ultimately leads to a less oxidised state of the
Ga-Ni interface.
In line with the conversion of CH4, a stable conversion of CO2
over the Ga12Ni/SiC SCALMS was achieved after 15 h TOS
(Fig. 5). Once again, the level of CO2 conversion is 50% higher
than the conversion of CH4 (Fig. S8) suggesting additional
conversion of CO2 via the RWGS reaction and oxidation of liquid
Ga0 with subsequent reduction of GaxO species by H2 (Fig. 3,
reaction 4 & 5). However, the ratio in the conversion of both
reactants is constant over 100 h TOS (Fig. S15), while the ratio
slowly decreases for the Ga45Ni/SiC SCALMS indicating less
conversion of CO2 via oxidation of Ga0 due to a continuous shift
of the kinetically controlled Ga0-GaxO redox process to the oxidic
side. Contrary, the increased Ni content in the Ga12Ni/SiC
SCALMS allows for a kinetically controlled redox process on the
metallic side due to sufficient H2 activation. The enhanced
Fig. 5 Effect of nickel concentration during dry reforming of methane. consumption of CO2 and H2 in said oxidation-reduction-cycles
Conversion of a CH4 and b CO2, as well as c the obtained H2/CO ratio in also results in a lower H2/CO ratio in the produced synthesis gas,
the product gas during dry reforming of methane over Ga-Ni SCALMS with which stabilises at 0.45 (Fig. 5c). In contrast, the ratio obtained
different concentrations of the active metal Ni employing a mesoporous SiC over the Ga45Ni/SiC SCALMS was continuously increasing from
support. Metal loadings: 4.52 wt.% Ga and 0.08 wt.% Ni for Ga45Ni/SiC, 0.3 to >0.8 over 20 h TOS. This observation, once again, may be
5.44 wt.% Ga and 0.38 wt.% Ni for Ga12Ni/SiC. Reaction conditions: linked to the detrimental effect of low Ni concentrations on the
900 °C, 1 bar, 1 g SCALMS, CH4:CO2:Ar = 1:1:3, 3 LN gcat-1 h-1. stability of SCALMS during DRM causing steady deactivation.
The kinetically controlled redox process is far on the oxidic side
concepts, which may reach up to 100 times higher specific con- and not reached over Ga45Ni/SiC SCALMS within 24 h TOS.
version rates53. The stabilisation at higher conversion levels may
indeed be caused by enhanced activation of H2 over Ni atoms.
This promotes the reduction of GaxO species (Fig. 3, reaction 5) Post-run catalyst characterisation. Characterisation of the post-
and in consequence, shifts the kinetically controlled Ga0-GaxO run Ga-Ni SCALMS by means of XRD and comparison with the
redox process to the metallic side allowing for enhanced DRM as-prepared samples suggests an increased fraction of Ga2O3 after
over atomically dispersed Ni atoms in a matrix of Ga atoms. To catalysis (Fig. 7). The patterns of the fresh samples only feature
further investigate this kinetically controlled redox process, the diffractions for this metal oxide close to the detection limit. A
present Ga12Ni/SiC SCALMS was also tested with a 2:1 CH4:CO2 passivation layer after preparation of SCALMS has been
inlet ratio to compare the performance during DRM under a reported26,28,29 and the performance of the SCALMS during
more reducing environment (Fig. S15). As expected, the equili- DRM suggests in situ reduction within the first hours TOS
brated conversion of CH4 and CO2 lies above the one of DRM (Fig. 5). Contrary, the spent samples feature pronounced dif-
with equimolar feed as the concentration of H2 form is increased. fractions of Ga2O3 suggesting oxidation of the Ga-rich alloy
The generally increased conversion levels may be explained by during DRM. Potential passivation of the (bi-)metallic phase
the almost five times higher loading of Ni in the Ga12Ni/SiC upon exposure to ambient air at room temperature during
SCALMS (Table 1). In order to quantify the amount of Ni unloading of the catalyst from the reactor can be expected to form
available at the surface of both Ga-Ni SCALMS, ab initio less oxide than the exposure to 500 °C during calcination. In
molecular dynamics (AIMD) simulations were carried out using addition, the Ga2O3 diffractions are even more pronounced for
slab models (Fig. 6a) at 900 °C with Ga/Ni ratios of 45 and 12, the Ga45Ni/SiC catalyst when compared to Ga12Ni/SiC, which
respectively, equalling those in the experiments. The systems were supports the hypothesised link between kinetically controlled
proven to be in a liquid state at this temperature by computation redox process and Ni concentration.
of the mean square displacement of the atoms from their initial The Ga-Ni combination was shown to play an essential role as
positions, which showed a linear behaviour. Furthermore, density an oxygen shuttle between CO2 and CH4. In particular, the
profiles for Ga12Ni and Ga45Ni were calculated (Fig. 6b). reducibility of the GaxO species is crucial and triggered by H2
Expectedly, a surface depletion of Ni can be observed for both activation over the active metal. Here, the Ga-Ni SCALMS (and
cases. This is in line with previous calculations for Pd, Pt, and Rh to a minor extent the Co-Ga SCALMS) were found to allow for

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 5

ARTICLE COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w

Fig. 6 AIMD simulations of gallium-nickel supported catalytically active liquid metal solutions (SCALMS). a Snapshot of the slab system with a unit cell
of Ga165Ni15 corresponding to the experimental composition Ga12Ni. Similarly, a Ga176Ni4 unit cell was used in the simulations of a Ga45Ni composition; Ga
is shown in blue, Ni in red. b Density profiles for the compositions Ga12Ni (left) and Ga45Ni (right). The dashed lines mark the end of the first layer. The
density profiles of Ni were scaled by a factor of 3 and 9 in the case of Ga12Ni and Ga45Ni, respectively, for the sake of clarity.

Fig. 7 X-ray diffraction of fresh and spent catalysts. X-ray diffractograms (Cu K-alpha radiation with λ = 1.541 Å) with an inset of the magnified intensity
of Ga-Ni SCALMS employing a mesoporous SiC support, as well as the bare support material before and after application in dry reforming of methane
together with reference patterns for SiC (o), graphite (C), and β-Ga2O3 (*).

closing the DRM catalytic cycle. A nearly constant conversion activity during DRM suggesting the formation of highly reactive
ratio of CO2/CH4 of approx. 1.5 was observed (Fig. S8) along with carbon species during DRM over Ni atoms. This process seems to
a constant hydrogen yield (Fig. S11). On the contrary, the Ga-Cu be less efficient for Co and Cu dissolved in the liquid Ga matrix.
and Ga-Fe SCALMS underwent rapid deactivation resulting in a Such highly active species may react with CO2 via the Boudouard
decreased conversion of CO2 when compared to CH4. Further, a reaction (Fig. 3, reaction 2) or even initiate the reduction of
steadily increasing H2 yield indicates that these systems (after full oxidised GaxO species (Fig. 3, reaction 6). Another reason may be
oxidation) rather cause CH4 pyrolysis under the applied the observed RWGS activity resulting in the formation of the
conditions, which is expected to result in excessive coking. The oxidant H2O, which may oxidise the liquid metal and carbon
increase in H2 yield is also due to a lower WGS activity after deposits alike. A similar suppression of coke formation over
approx. 10 h TOS, as the conversion of CH4 remains constant SCALMS has already been reported for propane dehydrogenation
(Fig. 2a). at lower temperatures of 450–550 °C26,28,29. The increased coke
TGA of the spent catalysts was performed in order to quantify content of 12.4 wt.% in the Ga45Ni/SiC is most likely due to the
the deposition of carbon during DRM at 900 °C for 24 h. The detrimental transformation of the metallic Ga to oxidic species as
smallest weight loss during TPO in 21% O2/N2 was observed for the kinetically controlled redox process lies on the oxidic side and
the Ga-Ni SCALMS catalysts, while the catalysts with zero or low seemingly does not equilibrate over the studied TOS. In other
level of activity, as well as the SiC reference sample, displayed a words, coking may be reasonably suppressed by a liquid Ga-rich
high weight loss in the range of 16-19 wt.% (Fig. 8). In fact, TGA alloy, while its (partially) oxidised counterpart cannot restrict
suggests a minor coke content of 0.7 wt.% for the Ga12Ni/SiC, coking. This dependency was also observed by means of XRD
which may indicate a significant suppression of coking during (Fig. 7, inset). The typical (002) diffraction of carbon was
DRM over SCALMS when compared to catalysts in the observed for Ga45Ni/SiC and all less active catalysts (Fig. S17), but
literature17,54,55. This is also evidenced during TGA of the spent absent in the case of Ga12Ni/SiC. The final weight gain of all
Ga12Ni/SiC SCALMS after DRM at 900 °C for 100 h TOS, which samples during TGA (see range above 800 °C in Fig. 8) is a result
exhibited a remarkably low coke content of 2.9 wt.% (Fig. S16). of the enhanced SiO2 passivation layer of the SiC support material
This superior coke resistance underlines the stable performance at elevated temperatures40,56,57.
of this Ga-Ni SCALMS during DRM at 900 °C. In general, the The spent catalysts were also characterised by means of Raman
coke content in the SCALMS systems correlates with the observed spectroscopy (green laser with a wavelength of 532 nm) to obtain

6 COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem

COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w ARTICLE

Fig. 8 Thermogravimetric analysis via temperature-programmed oxidation. Characterisation of spent SCALMS with various active metals employing a
mesoporous SiC support, as well as the bare support material after application in dry reforming of methane.

additional information on coke deposits (Fig. S17). The bare SiC preferential evaporation of Ga species. Such a change in the
support material also features the characteristic D band SCALMS composition may shift the kinetically controlled redox
(~1350 cm-1) and the G band (~1600 cm-1) of carbon. However, process further to the metallic side, which results in increased
the intensity ratio of D over G band changes when comparing the activity of the catalyst. Hence, the optimum Ga/Ni ratio for a
spectra before and after catalytic application during DRM. The balanced kinetically controlled redox process during DRM over
ratio lies in the range of 1.00–1.10 for the bare SiC, as well as the SCALMS is expected to be below 8.4, which is the final
as-prepared SCALMS, and increases to 1.15–1.45 due to the composition of the catalyst after 100 h TOS (Table 1). The
coking of the catalysts (Table S1). Furthermore, said change was liquidus line of Ga-Ni at 900 °C is expected to have a Ga/Ni ratio
more pronounced for less active catalysts, which is in line with of 3.139, which leaves significant space for further improvements
the results from TGA. In fact, the ratio was the smallest for the of the SCALMS composition while complying with the funda-
Ga12Ni/SiC SCALMS, which displayed the highest activity during mental idea of fully liquid-supported alloys. Furthermore, we
DRM. In addition, the spectra of all spent SCALMS and the bare assume that the tendency for Ga2O evaporation reduces with
SiC support after DRM feature the G’ band at ~2700 cm-1 higher Ni content of the alloy. Nevertheless, an additional
(Fig. S18). No difference in carbon deposits may be identified experiment with a bimetallic catalyst with a molecular ratio of
by means of XRD (Fig. S17). Ga/Ni of 0.74 was performed (Fig. S20). While the catalyst
Potential loss of gallium from the SCALMS catalyst during initially outperformed the Ga45Ni/SiC SCALMS, rapid deactiva-
DRM at 900 °C was investigated by analysing the metal loadings tion within 5 h TOS and excessive coke formation (16.2 wt.%;
after catalytic testing using ICP-AES (Table 1). Noteworthy, the Fig. S21) were observed exhibiting the coking affinity of solid-
identified broad range in the coke content (Fig. 8) with an based catalysts. Furthermore, and in line with our expectations,
expected link to the hygroscopic behaviour of the spent samples28 no loss in Ga during DRM was detected when comparing the
lowers the accuracy of the analysis of the metal content after loading before and after catalysis by means of ICP-AES
catalytic application. Nevertheless, a certain loss in gallium was (Table S2).
identified for all catalysts, which apparently scales with the Finally, the Ga-Ni SCALMS materials were analysed by means
activity of the catalyst, i.e. the loss of approx. 18% Ga was found of SEM-EDX after application in DRM to study morphological
to be larger for the more active Ga12Ni SCALMS than for a molar changes and evaluate the metal distributions. Note, that artefacts
ratio of Ga/Ni of 45. Furthermore, a loss of approx. 30% was from solidification of the liquid alloy during cool-down and
identified for the long-term experiment of Ga12Ni after 100 h subsequent exposure to ambient air may affect this ex situ
TOS. This observation renders physical evaporation of liquid analysis. Nevertheless, the dispersion of gallium over the SiC
gallium unlikely, which is supported by the low vapour pressure support appears to be generally unchanged. However, the spent
of Ga0 at 900 °C of 6.2 ∙ 10-7 bar58. The dependency on the Ga12Ni/SiC SCALMS features needle-type structures of several
reaction atmosphere rather suggests a strong link to the micrometres in length and diameters below 100 nm (Fig. 9;
hypothesised redox cycles. In particular, the formation of more Fig. S22). These structures cannot be identified in the as-prepared
volatile Ga2O, a likely intermediate formed during oxidation to samples nor in the Ga45Ni/SiC SCALMS after 24 h TOS (Fig. 1;
Ga2O3 with CO2 or subsequent re-reduction, may result in loss of Figs. S4 & S23). According to elemental mapping via SEM-EDX,
gallium from the catalyst bed at reaction temperatures. H2- the needles consist of gallium and oxygen (Fig. 9). No nickel,
containing atmospheres may even increase the volatility of carbon, or silicon can be detected suggesting the formation of
gallium oxide59. The potential formation of volatile gallium GaxO needles during DRM, most likely as Ga2O3 phase (Fig. S24).
hydrides cannot be excluded but has been reported for In addition, the Ga45Ni/SiC SCALMS features some needles after
temperatures exceeding the herein applied conditions60. Our DRM for an extended duration of 100 h (Fig. S25 & S26),
thermodynamic calculations also indicate a low feasibility of the probably because a longer run time was needed to locate the
formation of GaHx (Fig. S19). Hence, the redox cycles may needles in post-run analysis. Hence, the formation of these
continuously result in slow leaching from the catalyst bed. structures depends on the extent and number of redox cycles as
However, this process does not heavily affect the catalytic the Ga12Ni SCALMS heavily oxidises and re-reduces due to the
performance in the studied time range. Contrary, a minor increased concentration of Ni in the Ga-rich supported alloys. As
increase in conversion between 60–100 h TOS (Fig. S15) was extended Ga leaching has been observed for both samples, a
observed during the long-term testing and may indicate a slow needle growth via the gas phase and volatile GaxO species is
and steady decrease of the Ga/Ni ratio below 12 due to the highly likely. The volatility of oxidised gallium species has already

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 7

ARTICLE COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w

degree of redistribution of the bimetallic phase, which as well may

also be an artefact from cool-down.

Summary and conclusion

Supported Ga-rich liquid Ga-Co, Ga-Cu, Ga-Fe, and Ga-Ni alloys
were prepared, characterised, and applied in DRM to assess the
suitability of SCALMS for high-temperature applications and to
push this supported liquid phase catalysis concept to boundaries.
While physical evaporation of the liquid metal phase could be
excluded, a certain degree of chemical leaching, most likely via
oxidic gallium species, was observed. In general, all SCALMS
were shown to oxidise during the applied DRM conditions. Only
the supported Ga-Ni and, to a smaller extent, the Ga-Co alloy
retained activity after a run-in phase. Our study indicates that this
is due to the capability of the supported liquid Ga-Ni alloy to
efficiently activate H2 (formed during DRM) for the re-reduction
of oxidised gallium species. For Ga-Ni SCALMS, the resulting
kinetically controlled redox process could be shifted to the
metallic side by increasing the concentration of Ni in the Ga-rich
alloy, which promotes the activation of H2 and enables stable
conversion levels during DRM. A long-term experiment with
Ga12Ni/SiC for 100 h time on stream demonstrated the general
suitability of this SCALMS system for DRM, but the leaching of
gallium species (in combination with the deposition of β-Ga2O3
needles) will affect the stability of Ga-Ni SCALMS with the here-
studied composition in even longer operation. Further, the esti-
mated intrinsic activity of Ni is inferior when compared to the
literature.
Several important findings for future applications of SCALMS
systems were made in this conceptual work:
– The SCALMS concept allows for the employment of less
noble metals, such as Co and Ni, as the secondary active
metal dissolved in Ga-rich liquid alloy phase. Ab initio
molecular dynamics simulations showed a surface depletion
of Ni accompanied by the dynamic temporary appearance of
single Ni atoms at the surface, in analogy to the situation of
SCALMS containing noble metals.
Fig. 9 Morphology of spent gallium-nickel supported catalytically active – Ga-based SCALMS are generally suitable for high-
liquid metal solutions (SCALMS). a Backscattered scanning electron temperature operation at 900 °C, but volatile Ga-oxide species
micrograph and elemental mapping of b gallium, c oxygen, and d nickel via can cause Ga losses due to evaporation under an oxidising
energy-dispersive X-ray spectroscopy of Ga12Ni/SiC with metal loadings of atmosphere.
Ga and Ni of 5.44 and 0.38 wt.%, respectively, after catalytic application in – A reduced amount of coke formation was observed for Ga-Ni
dry reforming of methane. Magnified insets show needle-type structures of SCALMS when compared to the other catalysts with low or
gallium oxide. zero levels of activity. In fact, the Ga12Ni/SiC SCALMS with a
high concentration of Ni in the supported liquid alloy almost
completely suppressed coking during DRM due to the high
been linked to the extent of redox cycles. Further, the preparation dynamics of its supported liquid alloy interface.
of β-Ga2O3 via chemical vapour deposition (CVD) using Ga and
H2O has been reported in literature61, while the presence of β-
Ga2O3 in the spent SCALMS has been identified by means of Methodology
XRD (Fig. 7). Other studies also describe the formation of needle- Materials. Gallium pellets (≥99.9999% purity) were purchased
type structures of gallium oxide with comparable dimensions62,63. from Alfa Aesar, cobalt(II) chloride (CoCl2; ≥ 98.0% purity),
Lastly, the formation of needle- or platelet-type structures during copper(II) chloride (CuCl2; ≥ 99.995% purity), iron(II) chloride
water-induced deactivation of metal-based catalysts has been (FeCl2; ≥ 98.0% purity), and nickel(II) chloride ethylene glycol
reported for other applications as well64,65. dimethyl ether (C4H10Cl2NiO2; ≥ 98.0% purity) were supplied
Aside from the needle-type structures, EDX mapping suggests from Sigma Aldrich, and isopropanol (≥99.3% purity) was pur-
a general co-location of nickel with the Ga-rich areas (Fig. 9), i.e., chased from Jäkle Chemie (Germany). All chemicals were used as
pronounced segregation during DRM is not observed. In contrast received. Mesoporous 2 mm pellets of β-SiC (SIC3-E3-M) were
to the as-prepared SCALMS with randomly shaped Ni enrich- purchased from SICAT (France), ground, and sieved to yield a
ments after impregnation and calcination (Fig. 4), the distribution particle size range of 500-630 μm.
of Ni after DRM suggests the formation of a liquid alloy under
reaction conditions (Fig. 9). Such supported liquid alloy droplets Preparation of catalysts. Ga-based SCALMS materials were
can be expected to solidify during cool-down forming the herein synthesised in a two-step procedure. The first part comprises the
observed spherical bimetallic phases. However, several Ni- deposition of liquid Ga on the SiC support material. Ga nuggets
enriched Ga-Ni structures may be observed indicating a certain of ~1.00 g were melted and dispersed in 100 mL isopropanol via

8 COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem

COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w ARTICLE

ultrasonication with a BRANSON-450D sonifier at 80% intensity Spent samples were analysed via temperature-programmed
(maximum power output: 450 W) for 10 min at 40 °C forming an oxidation (TPO) in 21% O2/N2 in order to quantify carbon
emulsion. Subsequently, a defined amount of SiC (~13.00 g) was deposits formed during DRM. The weight change was monitored
added to yield a targeted Ga loading. The mixture was stirred by means of thermogravimetric analysis (TGA) using a SETSYS
thoroughly for 1 h. Finally, the Ga-decorated SiC was dried with a Evolution high-performance modular TGA (Setaram). A total of
rotary evaporator at 50 °C and 150 mbar for 2 h and calcined 30–45 mg of the particular sample were placed in a quartz
overnight at 500 °C40. The second part of the preparation pro- crucible, heated to 120 °C (10 °C min-1) for 100 min to remove
cedure comprises the addition of the active metal via wetness adsorbed H2O. Subsequently, the temperature was increased with
impregnation of the Ga/SiC material. The corresponding a heating rate of 5 °C min-1 to 1000 °C for 10 min. The overall
amounts of metal precursors, namely metal chlorides (Co, Cu, Fe) flow rate was 50 mL min-1 throughout TPO.
and nickel(II) chloride ethylene glycol dimethyl ether (Ni), were
dissolved in distilled water and isopropanol in a proportion of Catalytic testing. Catalytic experiments were performed in a high-
1:10, added to the previously calcined Ga/SiC sample targeting a temperature set-up consisting of a tubular split furnace (Carbolite)
molar Ga/secondary metal ratio of 50 and mixed for 1 h. Subse- with three heating zones of 20 cm length each. Quartz tubes (L:
quently, the solvents were evaporated in a rotary evaporator at 100 cm, OD: 1.2 cm, ID 1.0 cm) were used as reactors, while three
50 °C and 150 mbar for 2 h and the obtained SCALMS materials pins at a height of 55.5 cm from the bottom end supported the
were calcined overnight at 500 °C. catalyst bed. A total of 1.0 g of catalyst was placed in between two
plugs of 0.1 g quartz wool. DRM was performed at 900 °C and
1 bar for 20 hours with an inlet gas composition of 20% CH4
Characterisation of fresh and spent catalysts. The metal load- (≥99.9995% purity) and 20% CO2 (≥99.9995% purity) in Ar
ings of prepared samples and post-run catalysts were determined (≥99.998% purity and further purified by an Agilent CP17974 gas
by means of inductively coupled plasma atomic emission spec- clean filter) and an overall flow rate of 50 mLN min-1 resulting in a
troscopy (ICP-AES) using a Ciros CCD (Spectro Analytical gas hourly space velocity (GHSV) of 3000 mLN h-1 gcat-1. The off-
Instruments GmbH). The solid samples were digested with con- gas was quantitatively analysed in a micro gas chromatograph (I-
centrated HCl:HNO:HF in a 3:1:1 volumetric ratio using micro- GRAPHX PR; Industrial Graph Xolutions, Germany) using Ar as
wave heating up to 220 °C for 20–40 min. The instrument was the internal standard.
calibrated with standard solutions of Ni, Co, Fe, Cu, Ga prior to
the measurements. Computational methods. Density-functional theory (DFT) based
The BET surface area and BJH pore volume of the SiC support ab initio molecular dynamics (AIMD) simulations were per-
material were analysed by N2 physisorption in a QUADROSORB formed employing the Vienna Ab Initio Simulation Package
SI Surface Area and Pore Size analyser (Quantachrome Instru- (VASP) using the projector augmented wave (PAW) method to
ments) with a degassing temperature of 200 °C. represent the ion cores and a plane wave basis set with a kinetic
Scanning electron microscopy (SEM) was carried out using a energy cutoff of 300 eV70–72. The exchange-correlation function
Phenom Desktop SEM (BSD detector, 15 kV voltage). The of Perdew, Burke and Ernzerhof (PBE) was employed73. The
samples were deposited directly onto a conductive sticky carbon GaNi systems were simulated using periodic slab models with a
pad. In addition, scanning electron microscopy with elemental tetragonal unit cell (12.69 × 12.69 × 40 Å) comprising 180 Ga
mapping via energy-dispersive X-ray spectroscopy (SEM-EDX) atoms with 15 and 4 Ni atoms in case of Ga12Ni and Ga45Ni,
was performed on a ZEISS Cross Beam 540 Gemini II scanning respectively. A vacuum layer of 12 Å was added in the direction
electron microscope equipped with an EDX detector from Oxford perpendicular to the surface to ensure that periodic images do not
Instruments Group. Silver paste (Acheson Silver DAG 1415) was interact. To integrate the equations of motion a Verlet algorithm
used to enhance the electrical conductivity. Additionally, the was used with a time step of 5 fs. After sufficient equilibration, a
samples were PVD carbon coated prior to the investigations. SEM Nosé-Hoover thermostat was applied in order to simulate a
images were obtained at 3 kV and 2 nA. For images acquired with canonical ensemble at 900 °C74. To sample the first Brillouin zone
a backscatter electron detector (BSD) a voltage of 20 kV and a a Γ-containing 2x2x1 k-point mesh was used. The SCF con-
current of 2 nA was used. EDX analysis was performed at 20 kV vergence criterion was set to 10-7 eV. A total of 10 independent
and 2 nA. trajectories were sampled for each composition, every trajectory
X-ray diffraction (XRD) was conducted using an X’Pert PRO running for at least 150 ps of production time and results were
(Philipps) equipped with a Cu anode (λKα1 = 1.54056 Å). The averaged over all trajectories.
samples were placed in a sample holder and analysed in a
continuous scan mode in the 2θ range of 1.992° to 80.000° with a
step size of 0.0167113° and a scan time of 1.11 s per step. XRD Data availability
patterns were compared to references from the Crystallography All relevant data are available from the authors upon request. Please contact the
Open Database (COD)66, namely graphite (COD ID 1200017)67, corresponding author.
β-gallium(III) oxide (COD ID 2004987)68, and β-silicon carbide
(COD ID 1010995)69. Received: 20 January 2022; Accepted: 2 October 2023;
Raman spectroscopy was conducted using an AvaRaman-PRB-
532 (Avantes) probe with an AvaRaman-532HERO-EVO
(Avantes) system. The Raman solution consists of a 532 nm
(green) solid-state laser (Cobolt) and an AvaSpec-HERO
(Avantes) spectrometer with a grating set of 1200 lines mm-1 References
(HSC1200-0.75). The spectrometer is equipped with a 50 µm slit 1. Pakhare, D. & Spivey, J. A review of dry (CO2) reforming of methane over
and the detected wavelength range is 534-696 nm. If not noble metal catalysts. Chem. Soc. Rev. 43, 7813–7837 (2014).
2. Bode, A. et al. in 20th World Hydrogen Energy Conference 2014. Held 15–20
otherwise stated, Raman spectra were collected in 10 repetitions
June 2014, Gwangju, South Korea (2014).
at 15 mW laser power with an exposure time of 20 s. Three 3. Aramouni, N. A. K., Touma, J. G., Tarboush, B. A., Zeaiter, J. & Ahmad, M. N.
different spots of the sample were analysed and averaged. Catalyst design for dry reforming of methane: analysis review. Renew. Sustain.
Typically, all three spectra of the samples coincided. Energy Rev. 82, 2570–2585 (2018).

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 9

ARTICLE COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w

4. Wittich, K., Krämer, M., Bottke, N. & Schunk, S. A. Catalytic dry reforming of 33. Bauer, T. et al. Operando DRIFTS and DFT study of propane
methane: insights from model systems. ChemCatChem 12, 2130–2147 (2020). dehydrogenation over solid- and liquid-supported GaxPty catalysts. ACS
5. Song, Y. et al. Dry reforming of methane by stable Ni-Mo nanocatalysts on Catal. 9, 2842–2853 (2019).
single-crystalline MgO. Science 367, 777–781 (2020). 34. Hohner, C. et al. Pt–Ga model SCALMS on modified HOPG: growth and
6. Schwab, E., Milanov, A., Schunk, S. A., Behrens, A. & Schödel, N. Dry adsorption properties. Top. Catal. 62, 849–858 (2019).
reforming and reverse water gas shift: alternatives for syngas production. 35. Kettner, M. et al. Pd-Ga model SCALMS: characterization and stability of Pd
Chem. Ing. Tech. 87, 347–353 (2015). single atom sites. J. Catal. 369, 33–46 (2019).
7. Moradi, G. R., Rahmanzadeh, M. & Khosravian, F. The effects of partial 36. Nishizawa, T. & Ishida, K. in Binary Alloy Phase Diagrams (eds. Massalski, T.
substitution of Ni by Zn in LaNiO3 perovskite catalyst for methane dry B., Okamoto, H., Subramanian, P. R. & Kacprzak, L.) 1186 (ASM
reforming. J. CO2 Util. 6, 7–11 (2014). International, 1990).
8. Royer, S. et al. Perovskites as substitutes of noble metals for heterogeneous 37. Subramanian, P. R. & Laughlin, D. E. in Binary Alloy Phase Diagrams (eds.
catalysis: dream or reality. Chem. Rev. 114, 10292–10368 (2014). Massalski, T. B., Okamoto, H., Subramanian, P. R. & Kacprzak, L.) 1410–1411
9. Mao, X., Foucher, A. C., Stach, E. A. & Gorte, R. J. Changes in Ni-NiO (ASM International, 1990).
equilibrium due to LaFeO3 and the effect on dry reforming of CH4. J. Catal. 38. Okamoto, H. in Binary Alloy Phase Diagrams (eds. Massalski, T. B., Okamoto,
381, 561–569 (2020). H., Subramanian, P. R. & Kacprzak, L.) 1702–1703 (ASM International, 1990).
10. Bartholomew, C. H. Mechanisms of catalyst deactivation. Appl. Catal. A 212, 39. Lee, S. Y. & Nash, P. in Binary Alloy Phase Diagrams (eds. Massalski, T. B.,
17–60 (2001). Okamoto, H., Subramanian, P. R. & Kacprzak, L.) 1829–1833 (ASM
11. Zuo, Z.-J., Shen, C.-F., Tan, P.-J. & Huang, W. Ni based on dual-support Mg- International, Russell Township, USA, 1990).
Al mixed oxides and SBA-15 catalysts for dry reforming of methane. Catal. 40. Raman, N. et al. GaPt supported catalytically active liquid metal solution
Commun. 41, 132–135 (2013). catalysis for propane dehydrogenation–support influence and coking studies.
12. Han, J. W., Kim, C., Park, J. S. & Lee, H. Highly coke-resistant ni nanoparticle ACS Catal. 11, 13423–13433 (2021).
catalysts with minimal sintering in dry reforming of methane. ChemSusChem. 41. Papadopoulou, C., Matralis, H. & Verykios, X. in Catalysis for Alternative
7, 451–456 (2014). Energy Generation (eds L. Guczi & A. Erdôhelyi) Ch. 3, (Springer, 2012).
13. Huang, F. et al. Catalytic performances of Ni/mesoporous SiO2 catalysts for 42. Shah, Y. T. Water for Energy and Fuel Production. (CRC Press, 2014).
dry reforming of methane to hydrogen. J. Energy Chem. 25, 709–719 (2016). 43. Wang, Y., Yao, L., Wang, S., Mao, D. & Hu, C. Low-temperature catalytic CO2
14. Zhang, Q. et al. A sintering and carbon-resistant Ni-SBA-15 catalyst prepared dry reforming of methane on Ni-based catalysts: a review. Fuel Process.
by solid-state grinding method for dry reforming of methane. J. CO2 Util. 17, Technol. 169, 199–206 (2018).
10–19 (2017). 44. Søgaard, A., de Oliveira, A. L., Taccardi, N., Haumann, M. & Wasserscheid, P.
15. Yang, E. et al. Al2O3-coated Ni/CeO2 nanoparticles as coke-resistant catalyst Ga–Ni supported catalytically active liquid metal solutions (SCALMS) for
for dry reforming of methane. Catal. Sci. Technol. 10, 8283–8294 (2020). selective ethylene oligomerization. Catal. Sci. Technol. 11, 7535–7539 (2021).
16. Wolf, M. Thermodynamic assessment of the stability of bulk and 45. Nabaho, D., Niemantsverdriet, J. W., Claeys, M. & van Steen, E. Hydrogen
nanoparticulate cobalt and nickel during dry and steam reforming of methane. spillover in the Fischer-Tropsch synthesis: an analysis of platinum as a
RSC Adv. 11, 18187–18197 (2021). promoter for cobalt-alumina catalysts. Catal. Today 261, 17–27 (2015).
17. Ha, Q. L. M. et al. Development of highly stable low Ni content catalyst for dry 46. Wolf, M., Mutuma, B. K., Coville, N. J., Fischer, N. & Claeys, M. Role of CO in
reforming of CH4‐rich feedstocks. ChemCatChem. 12, 1562–1568 (2020). the water-induced formation of cobalt oxide in a high conversion Fischer-
18. Palmer, C. et al. Dry reforming of methane catalysed by molten metal alloys. Tropsch environment. ACS Catal. 8, 3985–3989 (2018).
Nat. Catal. 3, 83–89 (2020). 47. Kocić, S., Corral Valero, M., Schweitzer, J.-M. & Raybaud, P. Surface
19. Yang, Q., Zhou, C., Ni, J. & Guan, X. Methane dry reforming in a coking- and speciation of Co based Fischer-Tropsch catalyst under reaction conditions:
sintering-free liquid alloy-salt catalytic system. Sustain. Energy Fuels 4, Deactivation by coke or by oxidation? Appl. Catal. A 590, 117332 (2020).
2768–2774 (2020). 48. Wolf, M., Fischer, N. & Claeys, M. Water-induced deactivation of cobalt-
20. Taccardi, N. et al. Gallium-rich Pd-Ga phases as supported liquid metal based Fischer–Tropsch catalysts. Nat. Catal. 3, 962–965 (2020).
catalysts. Nat. Chem. 9, 862–867 (2017). 49. Wolf, M., Fischer, N. & Claeys, M. Formation of metal-support compounds in
21. Herman, J. M., Rocourt, A. P. A. F., Van den Berg, P. J., Van Krugten, P. J. & cobalt-based Fischer-Tropsch synthesis: a review. Chem. Catal. 1, 1014–1041
Scholten, J. J. F. The industrial hydroformylation of olefins with rhodium- (2021).
based supported liquid phase catalyst SLPC. Chem. Eng. J. 35, 83–103 (1987).. 50. Mars, P. & van Krevelen, D. W. Oxidations carried out by means of vanadium
22. Riisager, A., Fehrmann, R., Haumann, M. & Wasserscheid, P. Supported ionic oxide catalysts. Chem. Eng. Sci. 3, 41–59 (1954).
liquid phase (SILP) catalysis: an innovative concept for homogeneous catalysis 51. Ruckenstein, E. & Wang, H. Y. Carbon dioxide reforming of methane to synthesis
in continuous fixed-bed reactors. Eur. J. Inorg. Chem. 2006, 695–706 (2006). gas over supported cobalt catalysts. Appl. Catal. A: Gen. 204, 257–263 (2000).
23. Lijewski, M., Hogg, J. M., Swadźba-Kwaśny, M., Wasserscheid, P. & 52. Otsuka, K., Yasui, T. & Morikawa, A. Production of CO from CO2 by reduced
Haumann, M. Coating of Pd/C catalysts with Lewis-acidic ionic liquids and indium oxide. J. Chem. Soc. Faraday Trans. 1, 78, 3281–3286 (1982).
liquid coordination complexes – SCILL induced activity enhancement in 53. Shang, Z., Li, S., Li, L., Liu, G. & Liang, X. Highly active and stable alumina
arene hydrogenation. RSC Adv. 7, 27558–27563 (2017). supported nickel nanoparticle catalysts for dry reforming of methane. Appl.
24. Marinkovic, J. M., Riisager, A., Franke, R., Wasserscheid, P. & Haumann, M. Catal. B: Environ. 201, 302–309 (2017).
Fifteen years of supported ionic liquid phase-catalyzed hydroformylation: 54. Horváth, É. et al. Dry reforming of CH4 on Co/Al2O3 catalysts reduced at
material and process developments. Ind. Eng. Chem. Res. 58, 2409–2420 different temperatures. Catal. Today 281, 233–240 (2017).
(2018). 55. Siew, K. W., Lee, H. C., Gimbun, J. & Cheng, C. K. Production of CO-rich
25. Rupprechter, G. Popping up to the surface. Nat. Chem. 9, 833–834 (2017). hydrogen gas from glycerol dry reforming over La-promoted Ni/Al2O3
26. Raman, N. et al. Highly effective propane dehydrogenation using Ga-Rh catalyst. Int J. Hydrog. Energ. 39, 6927–6936 (2014).
supported catalytically active liquid metal solutions. ACS Catal. 9, 9499–9507 56. Roy, J., Chandra, S., Das, S. & Maitra, S. Oxidation behaviour of silicon
(2019). carbide-a review. Rev. Adv. Mater. Sci. 38, 29–39 (2014).
27. Sebastian, O. et al. Stable and selective dehydrogenation of methylcyclohexane 57. Berton, B., Bacos, M. P., Demange, D. & Lahaye, J. High-temperature
using supported catalytically active liquid metal solutions–Ga52Pt/SiO2 oxidation of silicon carbide in simulated atmospheric re-entry conditions. J.
SCALMS. ChemCatChem. 12, 4533–4537 (2020). Mater. Sci. 27, 3206–3210 (1992).
28. Wolf, M., Raman, N., Taccardi, N., Haumann, M. & Wasserscheid, P. Coke 58. Alcock, C. B. Vapor pressure of the metallic elments, in CRC Handbook of
formation during propane dehydrogenation over Ga−Rh supported Chemistry and Physics (ed. D. R. Lide) Ch. Section 4, (CRC Press, 2003).
catalytically active liquid metal solutions. ChemCatChem. 12, 1085–1094 59. Butt, D. P., Park, Y. & Taylor, T. N. Thermal vaporization and deposition of
(2020). gallium oxide in hydrogen. J. Nucl. Mater. 264, 71–77 (1999).
29. Wolf, M. et al. Capturing spatially resolved kinetic data and coking of Ga-Pt 60. Breisacher, P. & Siegel, B. Comparative stabilities of gaseous alane, gallane,
supported catalytically active liquid metal solutions during propane and indane. J. Am. Chem. Soc. 87, 4255–4258 (1965).
dehydrogenation in situ. Faraday Discuss. 229, 359–377 (2021). 61. Terasako, T., Kawasaki, Y. & Yagi, M. Growth and morphology control of β-
30. Anand, M. et al. Advanced approaches: general discussion. Faraday Discuss. Ga2O3 nanostructures by atmospheric-pressure CVD. Thin Solid Films 620,
229, 378–421 (2021). 23–29 (2016).
31. Grabau, M. et al. Surface enrichment of Pt in Ga2O3 films grown on liquid Pt/ 62. Kakoria, A. et al. Gallium oxide nanofibers for hydrogen evolution and oxygen
Ga alloys. Surf. Sci. 651, 16–21 (2016). reduction. ACS Appl. Nano Mater. 2, 64–74 (2018).
32. Grabau, M. et al. Spectroscopic observation and molecular dynamics 63. López, I., Nogales, E., Hidalgo, P., Méndez, B. & Piqueras, J. Field emission
simulation of Ga surface segregation in liquid Pd-Ga alloys. Chem. A Eur. J. properties of gallium oxide micro- and nanostructures in the scanning
23, 17701–17706 (2017). electron microscope. Phys. Status Solidi A 209, 113–117 (2012).

10 COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem

COMMUNICATIONS CHEMISTRY | https://doi.org/10.1038/s42004-023-01018-w ARTICLE

64. Kiss, G., Kliewer, C. E., DeMartin, G. J., Culross, C. C. & Baumgartner, J. E. supervision; writing—review & editing. M. Haumann: conceptualisation; project
Hydrothermal deactivation of silica-supported cobalt catalysts in Fischer- administration; resources; validation; writing—review & editing. PW: conceptualisation;
Tropsch synthesis. J. Catal. 217, 127–140 (2003). funding acquisition; project administration; resources; validation; writing—review &
65. Wolf, M. et al. In-depth characterisation of metal-support compounds in editing.
spent Co/SiO2 Fischer-Tropsch model catalysts. Catal. Today 342, 71–78
(2020).
66. Downs, R. T. & Hall-Wallace, M. The American mineralogist crystal structure
Funding
Open Access funding enabled and organized by Projekt DEAL.
database. Am. Mineralogist 88, 247–250 (2003).
67. Hassel, O. Ueber die Kristallstruktur des Graphits. Z. fuer Phys. 25, 317–337
(1924). Competing interests
68. Åhman, J., Svensson, G. & Albertsson, J. A reinvestigation of β-gallium oxide. The authors declare no competing interests.
Acta Crystallogr. Sect. C. 52, 1336–1338 (1996).
69. Braekken, H. Zur Kristallstruktur des kubischen Karborunds. Z. Kristallogr.
Kristallgeomet. Kristallphys. Kristallchem. 75, 572–573 (1930). Additional information
70. Kresse, G. & Furthmuller, J. Efficient iterative schemes for ab initio total- Supplementary information The online version contains supplementary material
energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169–11186 available at https://doi.org/10.1038/s42004-023-01018-w.
(1996).
71. Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations for Correspondence and requests for materials should be addressed to Peter Wasserscheid.
metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci.
6, 15–50 (1996). Peer review information Communications Chemistry thanks Yu Tang, Yasin Orooji and
72. Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector the other, anonymous, reviewers for their contribution to the peer review of this work. A
augmented-wave method. Phys. Rev. B 59, 1758–1775 (1999). peer review file is available.
73. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation
made simple. Phys. Rev. Lett. 77, 3865–3868 (1996). Reprints and permission information is available at http://www.nature.com/reprints
74. Nosé, S. A unified formulation of the constant temperature molecular
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
dynamics methods. J. Chem. Phys. 81, 511–519 (1984).
published maps and institutional affiliations.

Acknowledgements
Financial support by the European Research Council (Project 786475: Engineering of Open Access This article is licensed under a Creative Commons
Supported Catalytically Active Liquid Metal Solutions) and the German Research Attribution 4.0 International License, which permits use, sharing,
Foundation (DFG) in the frame of the Collaborative Research Centre 1452 Catalysis at adaptation, distribution and reproduction in any medium or format, as long as you give
Liquid Interfaces (CLINT) is gratefully acknowledged. The authors thank Narayanan appropriate credit to the original author(s) and the source, provide a link to the Creative
Raman and Susanne Pachaly for conducting the N2 physisorption and XRD measure- Commons licence, and indicate if changes were made. The images or other third party
ments, respectively. material in this article are included in the article’s Creative Commons licence, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons licence and your intended use is not permitted by statutory
Author contributions
regulation or exceeds the permitted use, you will need to obtain permission directly from
M.W.: conceptualisation; investigation; data curation; formal analysis; validation;
visualisation; writing—original draft. A.L.d.O.: Investigation; data curation; formal ana- the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
lysis; writing—review & editing. N.T.: investigation; formal analysis; validation; writing— licenses/by/4.0/.
review & editing. S.M.: investigation; data curation; formal analysis; visualisation. M.
Heller: Investigation; data curation; formal analysis; visualisation. SKA: investigation; © The Author(s) 2023
data curation; formal analysis. AS: resources. PF: resources; supervision. AG: resources;

COMMUNICATIONS CHEMISTRY | (2023)6:224 | https://doi.org/10.1038/s42004-023-01018-w | www.nature.com/commschem 11