The Effect of Mg and Zn Dopants on Pt/Al 2 O 3 for the Dehydrogenation of Perhydrodibenzyltoluene

: The effects of Mg and Zn dopants on the catalytic performance of Pt/Al 2 O 3 catalyst were investigated for dehydrogenation of perhydrodibenzyltoluene (H18-DBT) as a liquid organic hydrogen carrier. Al 2 O 3 supports were modiﬁed with Mg and Zn to produce Mg-Al 2 O 3 and Zn-Al 2 O 3 with a target loading of 3.8 wt.% for dopants. The modiﬁed supports were impregnated with chloroplatinic acid solution to produce the catalysts Pt/Al 2 O 3 , Pt/Mg-Al 2 O 3 and Pt/Zn-Al 2 O 3 of 0.5 wt.% Pt loading. Thereafter, the catalysts were characterised using inductively coupled plasma-optical emission spectrometry, scanning electron microscopy-energy dispersive X-ray spectroscopy, hydrogen temperature-programmed reduction, carbon-monoxide pulse chemisorption, ammonia temperature-programmed desorption, X-ray diffraction and transmission electron microscopy. The dehydrogenation experiments were performed using a horizontal plug ﬂow reactor system and the catalyst time-on-stream was 22 h. Pt/Mg-Al 2 O 3 showed the highest average hydrogen ﬂowrate of 29 nL/h, while an average of 27 nL/h was obtained for both Pt/Al 2 O 3 and Pt/Zn-Al 2 O 3 . This has resulted in a hydrogen yield of 80% for Pt/Mg-Al 2 O 3 , 71% for Pt/Zn-Al 2 O 3 and 73% for Pt/Al 2 O 3 . In addition, the conversion of H18-DBT ranges from 99% to 92%, Pt 97–90% and 96–90% for Pt/Mg-Al 2 O 3 , Pt/Zn-Al 2 O 3 and Pt/Al 2 O 3 , respectively. Following the latter catalyst order, the selectivity to dibenzyltoluene (H0-DBT) ranges from 78% to 57%, 75–51% and 71–45%. Therefore, Pt/Mg-Al 2 O 3 showed improved catalytic performance towards dehydrogenation of H18-DBT.


Introduction
Dibenzyltoluene (H0-DBT) is known as one of the most promising liquid organic hydrogen carrier (LOHC) molecules due to its competitive hydrogen storage properties reported in the literature [1][2][3][4]. Remarkably, the infrastructure for fuel can be repurposed for storage and transportation of dibenzyltoluene-based LOHC at ambient conditions, because it has diesel-like properties. Hence, the application of dibenzyltoluene for reversible hydrogen storage has reached the commercial stage. However, it is important to further develop the highly active and stable catalysts with minimum by-product formation for the recyclability of the LOHC molecule. It is for this reason that we first discuss literature work, where the catalytic performance was improved by modification of the catalyst with Mg and Zn. Arguably, Pt/Al 2 O 3 is the catalyst of choice for dehydrogenation of LOHC molecules. Modification of the Al 2 O 3 support using Mg and Zn has been reported to enhance catalytic performance, strengthen metal-support interaction, change support acidity and improve metal dispersion [5][6][7][8]. For example, Zn modification on both Pt/Al 2 O 3 and Pt/ZSM-5 catalysts showed improved catalytic performance for propane dehydrogenation [7][8][9][10].
Mg-doped Al 2 O 3 also has excellent attributes, such as the ability to promote surface diffusion and prevent clustering/sintering by suppressing grain growth [6,11]. Some patents filed in the U.S. also confirm the catalyst improvement upon modification of alumina support with Mg [12][13][14]. In this case, Mg-modified alumina resisted sintering and agglomeration of the catalyst particles during production of syngas from light hydrocarbons [12]. The challenge is that optimum quantity of Mg is required because too much or too little amounts could have little effect, no effect or negative effect on the catalytic performance [15][16][17]. For example, Zhang et al. [18] investigated the dehydrogenation of iso-butane using PtSnMgK/Al 2 O 3 with different Mg loadings (0.2-0.8 wt.%). The amount of 0.4 wt.% Mg decreased carbon deposition on catalysts, with increased reaction activity and stability. A further increase to 0.8 wt.% Mg drastically decreased both Pt dispersion and catalytic activity. Moreover, Arnby et al. [19] and Zhan et al. [20] also reported the same phenomena for Pt/Al 2 O 3 and Pd/Al 2 O 3 catalysts, respectively. Auer et al. [21] investigated the effect of trace amounts of sulphur on Pt/Al 2 O 3 in the dehydrogenation of H18-DBT and found that 0.25 wt.% S improved the catalytic activity and reduced the formation of high boiling point by-products.
It is reported that Mg does not only reduce the acidity of Al 2 O 3 support, but it also lowers the carbon deposition, which is known for catalyst deactivation [22][23][24]. Since the acidity of Al 2 O 3 support could contribute to C-C cleavage of the hydrocarbon molecule [25], therefore the by-products produced from dehydrogenation of H18-DBT [26] could be a result of this. Based on the improvements reported in the literature, modification of the Al 2 O 3 support is expected to change the chemical properties of the catalyst and to improve the catalytic performance. This could be a rewarding approach because the commonly used Pt/Al 2 O 3 catalyst suffers from deactivation, which results in low catalytic activity for dehydrogenation of H18-DBT. In this work, we investigate the effect of Mg and Zn dopants on Pt/Al 2 O 3 catalyst for the dehydrogenation of H18-DBT. This is done by first modifying the Al 2 O 3 support using Mg and Zn precursors, followed by impregnation of Pt on the support. The catalysts are further characterised to determine the effect of modification on the physical and chemical properties.

Catalytic Dehydrogenation Activity
It has been previously reported that dehydrogenation of H18-DBT follows the order H18-DBT-3H 2 → H12-DBT-3H 2 → H6-DBT-3H 2 → H0-DBT [25] Figure 2, where catalyst productivity is plotted against time-on-stream. A productivity of 0.1 gH 2 /g Pt /min was observed for all catalysts. Furthermore, all catalysts here showed a declining trend in hydrogen yield over a period of 22 h, see Figure 2. The Pt/Mg-Al 2 O 3 hydrogen yield decreased from 80% to 60%, whereas the Pt/Al 2 O 3 and Pt/Zn-Al 2 O 3 hydrogen yields decreased from 72% to 50% and 73-51%, respectively. Therefore, Pt/Mg-Al 2 O 3 improved the activity of un-doped Pt/Al 2 O 3 .   Figure 3 shows the H18-DBT conversion over Pt/Al 2 O 3 , Pt/Zn-Al 2 O 3 and Pt/Mg-Al 2 O 3 , determined using single quadrupole mass spectrometer (GC-SQ-MS). During the dehydrogenation of H18-DBT, the intermediates H12-DBT and H6-DBT are formed, and they then also become dehydrogenated, until H0-DBT is the only product present. The conversion vs. time-on-stream plot in Figure 3 is based on the amount of H18-DBT present after the reaction. The catalyst Pt/Mg-Al 2 O 3 had the highest initial conversion of 99%, followed by the Pt/Zn-Al 2 O 3 and Pt/Al 2 O 3 catalysts, with 97% and 96% conversions, respectively. After 22 h, the Pt/Mg-Al 2 O 3 catalyst still exhibited a stable conversion of 92%, whilst the other catalysts exhibited slightly lower conversions. It is a fact that Mg improves reduction of Pt-oxide and this has led to higher metallic Pt concentration (active sites) on Pt/Mg-Al 2 O 3 than that on Pt/Al 2 O 3 and Pt/Zn-Al 2 O 3 [18]. It is also known that Mg could increase the rate of Al 2 O 3 grain densification by raising the diffusion coefficient [11]. Therefore, the larger the grain size, the larger the pores. This has an effect on minimizing the pore diffusion limitation. Hence, improved H18-DBT conversion is observed. As shown in Figure 3, the conversion difference between Pt/Al 2 O 3 and Pt/Zn-Al 2 O 3 catalyst is not that significant. Further optimisation of Zn loading could result in improved H18-DBT conversion.   Figure 4 shows the selectivity of catalysts towards H0-DBT, determined using GC-SQ-MS. The initial selectivity of Pt/Mg-Al 2 O 3 , Pt/Zn-Al 2 O 3 and Pt/Al 2 O 3 towards H0-DBT was 78%, 75% and 71%, respectively. Pt/Al 2 O 3 was the least selective catalyst here, with a final selectivity of 45%, whereas Pt/Mg-Al 2 O 3 was the most selective catalyst, with a final selectivity of 57%, after 22 h. It appeared that a less acidic catalyst favoured dehydrogenation and increased the selectivity, compared with highly acidic supports. Acidic sites of the support facilitate C-C breaking, and as a result, carbon becomes deposited on the catalyst surface [18]. Therefore, lowering the acidity of the support minimizes side reactions (C-C cracking). Thus, the high selectivity could be attributed to an effect of Mg to the adsorption of H0-DBT on the Pt surface and promoting the H0-DBT desorption. Figure 4 also shows the selectivity of catalysts towards the intermediates: H12-DBT and H6-DBT. All catalysts here showed a decline in selectivity towards H0-DBT, but an increase towards the H6-DBT and H12-DBT intermediates. The decline in selectivity over time is due to the catalyst instability. Pt/Al 2 O 3 selectivity towards H6-DBT increased from 24% to 39%, while that of H12-DBT increased from 5% to 10%. This is an indication that catalyst active sites are compromised by deactivation and over time, H6-DBT and H12-DBT become difficult to be dehydrogenated. Figure 5 shows the total by-products formed during the dehydrogenation of H18-DBT. In the dehydrogenation of H18-DBT, the higher the catalytic activity, the more byproducts are produced. These by-products are a combination of high-and low-boiling-point compounds [26]. Pt/Mg-Al 2 O 3 showed the highest initial concentration of by-products (6.8 mol%), followed by Pt/Zn-Al 2 O 3 (5.3 mol%) and Pt/Al 2 O 3 (4.2 mol%). After 22 h, Pt/Mg-Al 2 O 3 still produced the highest concentration of by-products (3.2 mol%). Despite the fact that Pt/Mg-Al 2 O 3 is less acidic, a highly active catalyst produces more by-products because the Pt active sites promote C-C breaking by hydrogenolysis. Bocanegra et al. [27] have reported that Pt/MgAl 2 O 4 has higher hydrogenolytic capacity than Pt/Zn-Al 2 O 4 .
The deactivation parameter, ∆X = (X i − X f )/X i × 100, is calculated to determine the % deactivation over a reaction period of 22 h. Table 1 summarises the performance of the catalysts for dehydrogenation of H18-DBT. Pt/Mg-Al 2 O 3 catalyst has the lowest deactivation parameter (∆X = 6.1%) compared to Pt/Al 2 O 3 (∆X = 7.2%) and Pt/Zn-Al 2 O 3 (∆X = 7.3%). This is an indication of improvement as Mg-doped catalyst deactivation is lesser. Increasing the Mg loading could lower the deactivation parameter and thus improve the catalyst stability further. Bai and co-workers reported the same phenomena for Mg-doped PtSnNa/ZSM-5, where the deactivation parameter decreased from 10% to 7% [22].

Characterisation
The subsections below describe the results of experiments and procedures carried out to determine the following: Pt loading, catalyst composition, metal dispersion, metal reducibility, catalyst acidity, particle size distribution and the source of LOHC by-products.

Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES) and Scanning Electron Microscopy-Energy Dispersive X-ray Spectroscopy (SEM-EDS)
The target Pt loading for all catalysts was 0.5 wt. % Pt. The actual loadings confirmed by ICP-OES for Pt/Al 2 O 3 , Pt/Mg-Al 2 O 3 and Pt/Zn-Al 2 O 3 were 0.48, 0.49 and 0.48 wt.%, respectively (see Table 2). The target loading of Mg and Zn was 3.8 wt.%-the actual loadings were 3.89 and 3.44 wt.%, respectively. SEM-EDS confirmed the presence of Pt, Mg and Zn in the reduced catalysts, as shown in Figure 6. Elemental line analysis indicated that the Mg-doped catalyst has the greatest amount of Pt distributed throughout the pellet. The Zn-doped and undoped catalysts exhibited an eggshell type of catalyst, where there is much Pt on the edges and little inside (<0.1% Pt). It is possible that Mg promotes the diffusion of the precursor solution during impregnation, hence Pt is distributed throughout the support. This is also consistent with the image in Figure 6, where the cross-sections of the pellets shows Pt distribution.

Hydrogen Temperature-Programmed Reduction (H 2 -TPR)
H 2 -TPR profiles showed broad and intense peaks at temperatures between 100 and 200 • C and small peaks at temperatures between 300 and 500 • C (see Figure 7). The first broad peaks were due to the reduction of PtO and PtO 2 species weakly bonded to the surface of the support [28]. The second peak indicated strong support interaction with Pt, possibly leading to the formation of PtAl 2 O 4 spinel species (formed during impregnation and calcination), which are difficult to reduce. It is interesting to note that the second peaks of the Mg-and Zn-doped Pt/Al 2 O 3 shifted to higher temperatures, indicating the presence of stronger interactions with the support. Therefore, doped catalysts have stronger metal support interactions than non-doped Pt/Al 2 O 3 . This is known to prevent sintering of particles [6,7,11]. Furthermore, Kumar et al. [17] ascribed the increase or shift in reduction temperature of the Mg-doped catalyst to the weak electronegative bond between Mg and O 2 in MgO, which causes the oxygen to preferably bond with Pt. Another possible reason could be the formation of Mg and Zn aluminates, which increase the reduction temperature. The following hydrogen consumptions for Pt/Al 2 O 3 , Pt/Mg-Al 2 O 3 and Pt/Zn-Al 2 O 3 were observed: 34, 64 and 55 µmol/g, respectively. The hydrogen consumption of 34 µmol/g for Pt/Al 2 O 3 is close to the theoretical value (30.3 µmol/g) reported for the complete reduction of Pt 4+ to Pt 0 [29]. However, the consumption of Pt/Mg-Al 2 O 3 and Pt/Zn-Al 2 O 3 is higher than the stoichiometric value for reducing all platinum species. This increment could be possibly due to the reduction of impurities on the support [15,29].

CO Pulse Chemisorption
The addition of Mg and Zn dopants decreased the dispersion of Pt particles (see Table 3). A decrease in the metallic surface area and the formation of strongly bonded Pt complexes with doped Mg and Zn supports may have contributed to this. Strong metalsupport interactions make it difficult for CO to absorb onto Pt due to fewer exposed active sites. There are, however, several reports that describe an increase in dispersion upon Mg doping [6,17]. Low dispersion of 34% for Pt/Mg-Al 2 O 3 compared to 38% for Pt/Al 2 O 3 can be ascribed to alteration of the adsorptive properties of Pt/Mg-Al 2 O 3 when Pt reacts with basic sites of Mg [15,30]. Pt/Zn-Al 2 O 3 had the lowest Pt dispersion, 23%, relatively close to values reported in the literature [17,31,32]. This could be due to the formation of Pt-Zn alloys and strong interactions that may not be active in CO chemisorption [29]. The elemental line scan in Figure 6 shows high Pt distribution throughout the Mg-doped catalyst (not an eggshell type) as opposed to the surface dispersion of Pt shown in Table 3.

NH 3 -TPD
NH 3 -TPD results are shown in Figure 8, including desorption profiles (dotted lines) and their respective deconvoluted peaks (solid lines) for the Al 2 O 3 supports and Pt/Al 2 O 3 catalysts. The first peak was at 200-250 • C and the second peak at 550-700 • C. The peaks are attributed to desorption of NH 3 due to weak/intermediate acidic sites and strong acidic sites, which appear in the region of 200-400 • C and 400-550 • C respectively, for γ-Al 2 O 3 [33][34][35][36]. Values for the acidity of supports and catalysts, determined by NH 3 -TPD, are tabulated in Table 4. Results show that Mg doping decreased the total acidity of Al 2 O 3 from 0.63 to 0.40 mmol NH 3 /g, while Zn doping slightly increased it to 0.68 mmol NH 3 /g. The order of the total acidity of supports was Zn-Al 2 O 3 > Al 2 O 3 > Mg-Al 2 O 3 . The addition of Mg decreased the total acidity of alumina by 37%, whereas Zn increased it by 8%. Mg is inherently basic and leads to a decrease in the strength of strong acidic sites. According to Karnjanakom et al., increased acidity in Zn-doped alumina is due to the substitution of proton sites on the Al 2 O 3 framework with the doping Zn species [37]. This substitution results in the regeneration of new proton sites on the support.

Catalyst/Support
Temperature, °C Total Acidity, mmol NH3/gcat      The particle size diameter was obtained by measuring about 300 particles for each catalyst and fitting the plotted histograms using a log-normal distribution curve. The parameters of log-normal distribution (µ-lognormal mean and σ-lognormal standard deviation obtained after fitting the 300 particles) are used to calculate the actual mean (m) and mode (mo) of each catalyst using the equations below. The results are shown in Table 6.
The obtained mean particle size is in the range from 1.54 to 1.72 nm for fresh catalysts, and from 2.14 to 2.80 nm for catalysts which sustained 22 h of dehydrogenation, see Figure 10 and Table 6. However, Pt particles were not observed for the Pt/Zn-Al 2 O 3 sample and it could be that Zn affects nucleation and growth of particles, as reported by Belskaya [10]. The slight increase in particle size after 22 h of dehydrogenation reaction is negligible and cannot be linked to Pt sintering since it occurs at temperatures above 300 • C. The discrepancy in the particle size obtained from CO chemisorption and TEM have been reported elsewhere [2,39]. ---- Figure 10. Platinum nanoparticles' size distribution before and after dehydrogenation, experiments carried out at 300 • C for 22 h using a fixed bed reactor.

X-ray Diffraction (XRD)
Broad XRD peaks of pure alumina were detected at 32.5 • , 37.0 • , 38.9 • , 45.0 • and 67.5 • . There was no difference between doped, undoped or Pt-loaded catalysts. This is because the Pt, Mg and Zn particles are small, and highly dispersed in the bulk alumina phase, and as such, they were undetectable by XRD. The positions at which the Pt peaks should be observed are shown as vertical lines in Figure 11. The Pt loading was too low to be detected by XRD. For the Mg-and Zn-doped catalysts, there was no clear evidence of the existence of Mg and Zn oxides or aluminates (MgAl 2 O 4 /ZnAl 2 O 4 ) because their main peaks overlapped with those of alumina [8,40,41].

Characterisation of the Liquid Carrier Molecules
Here, we determined whether the by-products formed as a result of thermal cracking or by the nature of the support material or Pt catalyst. First, H18-DBT was heated without any catalyst or support material for 1.5 h at 300 • C. Second, in separate experiments, each support material (Al 2 O 3 , Mg-Al 2 O 3 and Zn-Al 2 O 3 ) was added to fresh H18-DBT and treated under the same conditions as mentioned above. Pt/Al 2 O 3 catalyst was added at a later stage. The results did not indicate the presence of any by-products in the low and high boiler window of the spectrum when H18-DBT and all the supports were used (see Figure 12). However, by-products appeared when the catalyst was used. These by-products can be seen in Figure 12 as peaks in the low and high boiler region of the Pt/Mg-Al 2 O 3 catalysts. This suggests that the operating conditions and support materials alone do not contribute to the formation of by-products, but the catalyst does. The aromatic molecules (H0-DBT) produced during dehydrogenation are susceptible to the acid sites of the support, which provides the cracking function. Hence, by-products were not seen in test of blank supports.

Materials
Dibenzyltoluene was purchased from Sasol (Marl, Germany) under the trade name Marlotherm-SH and further hydrogenated, using a nickel-based catalyst in a packed fixed bed reactor system, to obtain H18-DBT with a degree of hydrogenation (doh) of 99%. The spherically shaped catalyst support pellets (Al 2 O 3 , Mg-Al 2 O 3 , Zn-Al 2 O 3 ) of diameter in the range of 1-3 mm were supplied by Sasol, Johannesburg, South Africa. The Mg-Al 2 O 3 and Zn-Al 2 O 3 were prepared by modification of parent unmodified γ-Al 2 O 3 used in this study. Hexachloroplatinic acid hydrate (38% Pt basis) was purchased from Sigma Aldrich (Schnelldorf, Germany). The following gases were supplied by Afrox, Johannesburg, South Africa: H 2 (99.999%), N 2 (99.999%), He (99.999%), Ar (99.999%), CO/He (10%), NH 3 /He (10%) and H 2 /Ar (10% H 2 ). These gases were mainly used to supply the furnace setup and chemisorption equipment for catalyst processing and characterisation, respectively.

Catalyst Preparation
Chloroplatinic acid solution (0.025 M) was prepared by adding hexachloroplatinic acid hydrate (powder form) to deionised water. The solution was added to each alumina support to produce 0.5 wt.% Pt loading. A Büchi Rotavapor ® R300 was used to evaporate water from pellets. The heating bath was set at 50 • C while the evaporating flask was under 46 mbar vacuum pressure and rotating at a speed of 40 rpm. An oven (Espec SU-221) was set at 120 • C to dry the pellets for a period of 4 h. Thereafter, pellets were calcined at 350 • C (ramp rate = 5 • C/min) in a stream of air (150 mL/min) for 5 h, using a furnace from Carbolite, UK. The calcined catalysts were reduced in a stream of 10% H 2 /Ar (50 mL/min) over periods of 4 h at 350, 400 and 450 • C for Pt/Al 2 O 3 , Pt/Mg-Al 2 O 3 and Pt/Zn-Al 2 O 3 , respectively.

Catalyst
The catalyst loading was confirmed by inductively coupled plasma-optical emission spectrometry (ICP-OES), using an Agilent 5110 ICP-OES instrument and the ICP Expert software. Catalysts were crushed and leached for 24 h, using a mixture of H 2 O 2 and HCl (ratio 1:10) at 60 • C. The leachate was then analysed for the presence of Pt, Mg and Zn.
An ammonia temperature-programmed desorption (NH 3 -TPD) experiment was performed to measure the acidity of the support and catalyst samples using an Auto-Chem II 2910 chemisorption analyser (Micromeritics, Norcross, GA, USA). Samples of 0.1 g of each catalyst were loaded in a U-type quartz sample tube. The analysis gas used was 10% NH 3 /He. First, samples were pre-treated at 500 • C for 1 h under a He stream (30 mL/min). This was followed by saturation with NH 3 at 120 • C. A thermal conductivity detector provided quantitative measurements of NH 3 desorption in the temperature range of 100-900 • C, and a ramp rate of 10 • C/min.
Scanning electron microscopy (SEM) imaging was carried out using a FEI Quanta 250 field emission gun scanning electron microscope (ThermoFisher Scientific, Waltham, MA, USA), operating at an accelerating voltage of 15 kV. Energy-dispersive X-ray spectroscopy (EDS) line analysis was carried out using the Oxford Inca software. Line point samples were taken at an equidistance of 40 µm. All samples were coated with carbon using a magnetron sputtering, to allow conduction between the sample and a beam of electrons from the SEM-EDS instrumentation. The samples were then embedded in resin (conditions: 121 • C, 290 bar, 20 min), using a PR-10 Mounting Press (Leco, MI, USA). After mounting, the samples were cut to expose the pellets' cross-sectional area, as indicated in Figure 13. Powder X-ray diffraction (XRD) analysis was performed using a D8-Advance allpurpose X-ray diffractometer (Bruker, Karlsruhe, Germany), operated in a locked coupled mode, with Cu-Kα radiation (λ = 1.5406 Å) at 40 kV tube voltage and 40 mA tube current. The measurements were within the range 2θ (0-90 • ) and a position-sensitive Lynx-Eye detector was used. Diffraction data were recorded at a speed of 0.5 s/step, which is equivalent to 92 s/step for a scintillation counter.

Analysis of LOHC Molecule
Analysis of the LOHC molecule was performed using gas chromatography coupled to a single quadrupole mass spectrometer system (GC-SQ-MS) (Scion, Goes, The Netherlands). The system was equipped with a CP-8400 autosampler, a GS-Tek GsBP-50 + MS column (length 30 m, internal diameter 0.25 mm, film thickness 0.25 µm) and a single quadrupole mass spectrometer. Helium of 99.999% purity was used as a carrier gas. To perform liquid analyses, samples of the reaction mixture were drawn from the reactor and diluted with hexane (1:100 wt./wt.), and 1 µL aliquots of each sample were injected, at a split ratio of 1:100.

Dehydrogenation Activity Tests
Dehydrogenation of H18-DBT was performed using a system comprising of a horizontally oriented plug flow reactor made of quartz glass, see Figure 14. Prior to dehydrogenation experiments, the system is purged with nitrogen to remove air. Hydrogenrich molecule (H18-DBT, >99% doh) is pumped into the reactor using a peristaltic pump (Medorex TL/150). The reactor is divided into three zones: (i) pre-heating zone packed with inert particles, (ii) reaction zone packed catalyst pellets and (iii) cooling zone packed with inert particles. An electrical heating jacket with three heating zones was used to supply heat and a thermocouple with the corresponding measurement points was centred horizontally through the reactor. During dehydrogenation reaction, the spent LOHC molecule from the reactor is cooled before reaching the product tank. Furthermore, the spent LOHC required for analysis is sampled before the product tank. Hydrogen gas passes through a cooling condenser and a carbon filter prior flow measurement by a mass flow meter (EL-Flow F11B-500-AGD-22V, Bronkhorst; flow range 0.16 to 3.5 nL/min; accuracy ± 0.5% Reading. This dehydrogenation system is controlled by Siemens PLC and for all experiments, a constant feed flow (1.3 mL/min) and reaction temperature (300 • C) were used. The catalyst productivity (P), H 2 -yield, conversion (X) and selectivity (S) are calculated using the following equations: P = m H2 /m pt .t H 2 -yield = [VH 2 released /VH 2 max ] × 100 (4) X H18-DBT = [mol H18-DBT (in) − mol H18-DBT (out) )/(mol H18-DBT (in) )] × 100 (5) S H0-DBT = [(mol H0-DBT )/(mol H0-DBT + mol H12-DBT + mol H6-DBT + mol by-products )] × 100 (6) where m H2 is the mass of released hydrogen, m pt is mass of platinum, t is time in minutes, VH 2 released is the volume of hydrogen released and VH 2 max is the theoretical maximum amount of hydrogen stored in H18-DBT.

Conclusions
The addition of Mg and Zn dopants strengthened metal-support interaction, decreased Pt particle size and dispersion and changed the acidity of supports. Pt/Mg-Al 2 O 3 showed improved catalytic performance when compared to Pt/Zn-Al 2 O 3 and Pt/Al 2 O 3 catalysts, and the higher the reaction activity, the higher the amounts of by-products. This is because the higher the catalytic activity, the more aromatic molecules are produced, and these aromatic molecules are susceptible to the acid sites of the support, which provides the cracking function. Furthermore, the formation of by-products could be facilitated by the hydrogenolytic effect of Pt/Mg-Al 2 O 3 . The improved activity of Pt/Mg-Al 2 O 3 can be attributed to the fact that Mg improves reduction of Pt-oxide and this has led to higher metallic Pt concentration (active sites) on Pt/Mg-Al 2 O 3 than that on Pt/Al 2 O 3 and Pt/Zn-Al 2 O 3 . The Al 2 O 3 grain size could be densified by Mg addition, and the larger the grain size, the larger the pores. This has an effect on minimizing the pore diffusion limitation and increasing the catalytic activity.