Stable and Selective Dehydrogenation of Methylcyclohexane using Supported Catalytically Active Liquid Metal Solutions – Ga52Pt/SiO2 SCALMS

The use of gallium‐rich, Supported Catalytically Active Liquid Metal Solution (SCALMS) is a promising new concept to achieve catalysis with atomically dispersed active metal atoms. Expanding our previous work on short alkane dehydrogenation, we present here the application of SCALMS for the dehydrogenation of methylcyclohexane (MCH) to toluene (TOL) using a Ga52Pt alloy (liquid under reaction conditions) supported on silica. Cycloalkane dehydrogenation catalysis has attracted great attention recently in the context of hydrogen storage concepts using liquid organic hydrogen carrier (LOHC) systems. The system under investigation showed high activity and stable conversion of MCH at 450 °C and atmospheric pressure for more than 75 h time‐on‐stream (XMCH=15 %) with stable toluene selectivity (STOL) of 85 %. Compared to commercially available Pt/SiO2, the SCALMS system resulted in higher yields and robustness. Baseline experiments with Pt‐free Ga/SiO2 under identical conditions revealed the decisive influence of Pt dissolved in the liquid Ga matrix.

Dehydrogenation reactions of cycloalkanes are important transformations in the chemical and refinery industries, e. g. to increase the research octane number of straight-run gasoline in the platforming process. [1] Furthermore, the reaction has recently gained very high interest in the context of hydrogen storage using reversible catalytic hydrogenation/dehydrogenation cycles in the context of the so-called Liquid Organic Hydrogen Carrier (LOHC) technology. [2] The endothermal cycloalkane dehydrogenation reaction is thermodynamically favored at high temperatures and low pressures. [3] Therefore, the applied catalysts operate under harsh conditions and are prone to deactivation by coke formation. [4] This makes the design of effective and robust catalysts and processes for cycloalkane dehydrogenation a rewarding research and developmental challenge. Dehydrogenation is typically carried out on noble metal (e. g. Pt) or oxide (e. g. Cr 2 O 3 ) catalysts. [3,5] Coke formation requires the presence of vicinal sites on the catalyst surface. [6] Regeneration of the catalyst after coking is possible but causes either reactor downtime or additional investment into stand-by or reserve reactor equipment.
Recently, we reported Supported Catalytically Active Liquid Metal Solutions (SCALMS) as a new class of liquid phase catalysts. [7] In these materials, a small amount of catalytically active metal is dissolved in a low melting metal matrix, such as liquid gallium. This alloy is supported on a porous support material to maximize the gas-liquid interface. Under reaction conditions, the supported alloy is present in the molten, liquid state. The highly dynamic liquid metal-gas interface of SCALMS enables the catalytically active metal to operate in an atomically dispersed manner at the gas-liquid interface within a matrix of liquid gallium. This particular nature of the active sites has been demonstrated by some of us together with collaboration partners using XPS, DFT, and operando DRIFTS investigations. [7][8] Concerning catalytic applications, we have applied these SCALMS systems for butane (GaPd on porous glass) and propane (GaRh on Al 2 O 3 , GaPt on Al 2 O 3 , GaPt on SiO 2 ) dehydrogenation reactions already. [7,9] In the present work, we extend the application of SCALMS systems to the dehydrogenation of cyclic alkanes. In particular, the dehydrogenation reaction of methylcyclohexane (MCH) to toluene (TOL) has been studied due to recent interest in this reaction in the context of reversible hydrogen storage (Liquid Organic Hydrogen Carrier, LOHC concept). [10] The results from our dehydrogenation experiments at 450°C and 1 bar are shown in Figure 1. The absence of blind activity was confirmed for the reactor (empty and filled with SiO 2 , see SI for details). The commercial Pt/SiO 2 catalysts showed an initial conversion of 13 %, which quickly decreased by half and eventually declined to values below 2 % after 12 h time-onstream (TOS). At the same time, the selectivity of the catalyst toward desired toluene declined from initially 90 % to values around 60 %. The selectivity of C7 reaction intermediates such as methylcyclohexene and methylcyclohexadiene increased to 23.3 % after 12 h TOS. The amount of other by-products from cracking or isomerization was below 5 %. In contrast, the Ga 52 Pt/SiO 2 SCALMS catalyst showed significantly better performance. A Ga/Pt ratio > 50 was chosen as for these Pt concentration GaÀ Pt alloys are fully liquid at the reaction temperature, i. e. 450°C. [11] After the preparation of catalysts via calcination, it can be assumed that Ga is present as bulk Ga 0 droplets coated with a thin passivation layer of Ga 2 O 3 which is known to be an active alkane dehydrogenation catalyst itself but leads also to the formation of cracking products. [3,12] For a material such as Ga 52 Pt/SiO 2 SCALMS, the presence of a hydrogen-activating metal enables the reduction of the oxidic passivation layer under hydrogen atmosphere at 450°C. [9b] Hence, we exposed Ga 52 Pt/SiO 2 SCALMS to a hydrogen pretreatment at 450°C before MCH dehydrogenation reaction. The initial conversion quickly exceeded 15 % and was maintained over 75 h TOS, after which the run was deliberately stopped (average conversion X MCH = 15.6 %). Processes, which led to the deactivation of the Pt/SiO 2 catalyst are obviously suppressed in the SCALMS system. Furthermore, the average selectivity toward toluene remained high around 82.4 % during the entire time-on-stream. Only a little amount of C7 intermediates (2.4 % selectivity) was found. Ring-opening/closing, condensation, or isomerization reaction products were observed in low amounts (< 5 %, see Table 1). Noteworthy, without co-feeding hydrogen, the conversion and toluene selectivity was significantly lowered (see supplementary information, Fig. S7).
In order to elucidate the role of Pt and Ga in the Ga 52 Pt/SiO 2 SCALMS catalyst, we also prepared a Pt-free Ga/SiO 2 material and exposed the Ga/SiO 2 to the same pretreatment and MCH dehydrogenation reaction conditions. Initially, the conversion was below 0.5 % and gradually increased to a low level of around 3 % after 18 h TOS, see Table 1. Interestingly, the selectivity towards toluene was comparable with the one of pure Pt/SiO 2 , however, it could not compete with the Pt-SCALMS system. Additionally, higher amounts of � C6 compounds were detected (6.2 %), which stem from the cracking of the MCH substrate. Formation of C7 paraffins/isoparaffins (e. g. methylhexene or ethylpentene) and C7 isomers (e. g. ethylated cyclopentenes or cyclopentanes) was higher than with the other catalysts. As mentioned above, the ability of oxidic Ga to catalyze alkane dehydrogenation has been already reported, [3,12] and the presence of Ga δ + species [9a,13] on the surface of the passivated gallium droplets may be held responsible for the observed catalytic activity. Conversely to the SCALMS system, Ga oxide cannot be fully reduced in the absence of H 2 -   activating metal atoms (such as Pt in the case of Ga 52 Pt/ SiO 2 ). [9b, 13,14] The dehydrogenation activity of gallium oxide is accompanied by a significant tendency to promote by-product formation, via cracking and/or isomerization reactions. Similar results have been previously observed for the dehydrogenation of propane. [9a] Clearly, the use of Pt together with Ga in the form of SCALMS catalyst brings significant advantages for the dehydrogenation of methylcyclohexane under the here-applied conditions. From the conversion and selectivity data, the productivity (in kg TOL kg Pt À 1 h À 1 ) of the different catalysts with respect to the desired toluene product can be calculated (see SI for details). Summing up of these values over time-on-stream results in a cumulative productivity value (in kg TOL kg Pt À 1 ) that enables a proper comparison of the different catalytic systems under investigation as shown in Figure 2. The deactivation and loss of selectivity of the commercial Pt/SiO 2 catalyst result in a leveling off of the cumulative productivity to a value of 20 kg TOL kg Pt À 1 after 18 h TOS. The stable conversion and selectivity of Ga 52 Pt/SiO 2 SCALMS catalyst, in contrast, leads to a linear increase in cumulative productivity reaching 140 kg TOL kg Pt À 1 after 18 h TOS and close to 700 kg TOL kg Pt À 1 after 75 h TOS. To the best of our knowledge, these results represent the most productive SCALMS system in the dehydrogenation of alkanes. [15] We have postulated in previous papers describing SCALMS systems for the catalytic dehydrogenation of short alkanes that the liquid nature of the gallium matrix under reaction conditions prevents to a large extent deactivation via coke formation. [9,16] First, Pt is present in an atomically dispersed form under the SCALMS conditions. Since vicinal sites are required for coking, coke formation cannot be triggered easily in SCALMS systems. Second, Pt comes into contact with the substrate MCH only when it appears dynamically at the alloy/ gas interface. [7][8][9] After product desorption, Pt dives back into statistically preferred Ga matrix, [7] while hydrocarbon compounds do not dissolve in liquid gallium. Lastly, the dynamics of the liquid gallium surface prevents the blocking of the active Pt atoms via agglomeration of coke precursors.
We were therefore interested to analyze the spent SCALMS after MCH dehydrogenation for carbon deposition by means of temperature-programmed oxidation (TPO). As shown in Figure 3, only minor amounts of coke were indeed found on the Ga 52 Pt/SiO 2 SCALMS system after MCH dehydrogenation experiments over 75 h time-on-stream as evidenced by a net weight loss of only 0.06 wt % in our TPO experiment. Taking into account the initial increase of the sample weight upon first exposure of the catalyst to 21 % O 2 /He the observed weight variation of the spent SCALMS in the TPO experiment increases to 0.18 wt %. The initial weight gain could be due to the oxidation of the SCALMS surface. However, the formation of metastable oxygen functionalities of the carbon deposits will also increase the weight during the heating ramp. [9b, [16][17] Simultaneous gasification of said intermediates resulting in CO 2 was observed at temperatures exceeding 200°C even though the weight was increasing until reaching 350°C due to the continuous formation of intermediates. Complete combustion of carbon deposits dominates the weight change at temperatures > 400°C as confirmed by means of mass spectrometry. The profile of CO 2 formation (m/z = 44) suggests two major species of coke. Both species are highly reactive excluding polyaromatic condensation as formation mechanism of coke, which would result in less reactive, graphitic carbon.
Interestingly, no detectable carbon deposits were identified on the commercial Pt/SiO 2 catalyst after catalytic application even though strong deactivation was observed. This is not unexpected as MCH conversion declines very rapidly and therefore little amount of intermediates that may lead to the formation of coke is present. We assume that the rapid deactivation of the classical heterogeneous Pt/SiO 2 could essentially be caused by 2 phenomena: 1. Small amount of carbon species may form in the early stage of the catalytic reaction and strongly adsorb on Pt surface and cause poisoning of the active sites. [6] 2. The Pt particles undergo sintering. Both phenomena seem to apply for the pure Pt catalyst as the EDX mapping showed less dispersed Pt on the surface of the spent catalyst, while CO-TPD experiment hint for a significant poisoning of the Pt-sites. (For details see supplementary information Figures S8, S9, and S10). In contrast, although significant CO 2 formation is observed in TPO of SCALMS, the catalytic activity in MCH dehydrogenation remained almost constant for 75 h time-on-stream. Thus, the SCALMS concept seems to effectively prevent deactivation. In fact, despite the formation of some reactive carbon deposits, they cannot strongly adsorb and poison the surface of the liquid alloy. Moreover, being in a liquidus regime, the active metal is always present atomically dispersed at the surface, [7-8,9a] de facto avoiding sintering phenomena.
In conclusion, we have demonstrated the first application of SCALMS in the dehydrogenation of cyclic alkanes. This was demonstrated for Ga 52 Pt/SiO 2 , which showed stable conversion and selectivity in the dehydrogenation of methylcyclohexane to toluene at 450°C and atmospheric pressure over 75 h time-onstream with a cumulative productivity of 700 kg TOL kg Pt À 1 . Under the applied conditions, this SCALMS system significantly outperformed the corresponding Pt/SiO 2 and Ga/SiO 2 systems with comparable amounts of Pt and Ga, respectively. Thus our results demonstrate the high potential of the SCALMS concept for aromatization reactions of cycloalkanes that are technically relevant for the refinery industry and also for future hydrogen storage applications. Further studies exploring other supports, liquid metals, variation of Ga/Pt ratio, and reaction conditions (temperature, partial pressure of MCH and H 2 , residence time, etc.) are currently ongoing.

Experimental Section
The SCALMS materials under investigation were prepared by the deposition of (Et 3 N)GaH 3 onto mesoporous silica followed by thermal decomposition to yield Ga decorated silica (Ga/SiO 2 ). [7] Pt deposition was achieved via partial galvanic displacement followed by calcination at 500°C overnight. The desired Ga to Pt atomic ratio of 52 was obtained by adjusting the amount of stock solution added (here: H 2 PtCl 6 ), see SI for details. The chemical composition of the SCALMS material was determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES, Ciros CCD instrument, Spectro Analytical Instruments GmbH). Additionally, commercially available Pt/SiO 2 was purchased from Sigma-Aldrich for benchmark experiments. The dehydrogenation of MCH was carried out in a fixed-bed tubular reactor at 450°C and atmospheric pressure (see SI for details). Prior to the dehydrogenation experiment, the catalyst was reduced in situ in an equimolar H 2 -He gas mixture (24.50 mL N min À 1 He and 24.50 mL N min À 1 H 2 ) at 450°C and atmospheric pressure for 2 hours. The reaction gas had a molar ratio of H 2 :MCH:He = 5 : 1 : 4 (19.60 mL N min À 1 He, 24.50 mL N min À 1 H 2 , 0.02 g min À 1 MCH). The product gas stream was analyzed by an online GC equipped with FID. Additionally, the condensed product stream was analyzed using offline GC-MS. The spent catalysts were subjected to temperature-programmed oxidation (TPO) using highresolution thermogravimetric analysis coupled with mass spectrometry (HRTGA-MS) in a XEMIS sorption analyzer (Hiden Isochema) to investigate the amount of coke formed. [16] Figure 3. Sample weight relative to the weight prior to exposure to 21 % O 2 /He at 100°C (filled symbols) and formation of CO 2 (open symbols) during temperature-programmed oxidation of spent Pt/SiO 2 (black) and Ga 52 Pt/SiO 2 SCALMS (red) catalysts after dehydrogenation of methylcyclohexane as monitored via high-resolution thermogravimetry coupled with mass spectrometry. Reaction conditions: 0.2 g spent catalyst, 1°C min À 1 , p total = 1 bar, O 2 : He = 0 (TOS < 0) and 0.26 (TOS > 0) mol mol À 1 , 100 mL N min À 1 .