Poisoning effects of H2S and HCl on the naphthalene steam reforming and water-gas shift activities of Ni and Fe catalysts

H2S and HCl are common impurities in raw syngas produced during gasification of biomass and municipal solid waste. The purpose of this study was to investigate the poisoning effect of H2S and HCl on synthesized and commercial catalysts during steam reforming of naphthalene. Four synthesized catalysts with different loadings of Ni and Fe on alumina support and two commercial catalysts were selected and evaluated in a fixed bed reactor at 790, 850 and 900 °C. The obtained results revealed that reforming and water-gas shift (WGS) activities of catalysts did not benefit from the Fe addition. The activities were influenced differently by H2S and HCl indicating that the reactions were catalyzed by different active sites on the nickel surface. In the presence of H2S and HCl, the poisoning of naphthalene reforming activity was caused by H2S and was not affected by HCl when both compounds were present in the gas. H2S chemisorbs on nickel surface forming NiS and decreasing the accessibility of active sites to hydrocarbons. The poisoning effect was only partially reversible. On the contrary, the poisoning of WGS activity could be caused by both H2S and HCl, and the activity could be completely restored when H2S and HCl were removed from the gas. Unlike naphthalene reforming activity, which was comparable for catalysts with similar Ni loadings, WGS activity depended on the catalyst structure and was less susceptible to poisoning by H2S and HCl in case of the catalyst with strong NiO-support interactions.

Besides nickel, other metals as well as bimetallic and polymetallic composites have been extensively investigated as reforming catalysts [15][16][17][18][19][20][21][22][23]. For instance, monometallic Fe and bimetallic Ni-Fe catalysts have shown satisfactory reforming activity and high catalytic stability during reforming of tar compounds under certain conditions [24][25][26][27][28]. Noichi et al. [24] found that higher Fe content in Fe-Al catalysts enhanced the catalytic steam reforming activity by increasing naphthalene conversion efficiency. In NiO-Fe 2 O 3 -Al 2 O 3 catalysts developed by Dong et al. [25] and Margossian et al. [26], syngas production and dry reforming activities of methane were influenced by Fe content. Furthermore, catalysts with optimized Fe content were reported to enhance thermal stability of the Ni-Fe catalysts by mitigating coke formation during tar reforming [25,26]. This superior effect was attributed to the formation of Ni-Fe alloys enriched with Fe-O species at the surface of nanoparticles that could catalyze coke oxidation [27,28].
It is well known that impurities present in syngas (e.g., particles, sulfur and chlorine species) can poison the catalysts during steam reforming process [29][30][31][32][33][34][35]. H 2 S is notorious poison for catalysts and only a few ppmv of H 2 S could rapidly deactivate a Ni-based reforming catalyst [29][30][31][32]. Upon contact with Ni-based catalyst, sulfur species (e.g., H 2 S) chemisorb on metal sites forming NiS according to the reaction (1) decreasing the accessibility of active sites to hydrocarbons [29]: Yung et al. [30] have attempted to regenerate the spent Ni catalyst which was contaminated during catalytic tar reforming at 850°C by 43 ppmv H 2 S in syngas produced from gasification of white oak. It was found that the Ni-S species in catalyst could not be completely removed during the steam/air regeneration procedure. As a result, the catalytic activity of Ni was only partially recovered and was lower than its initial activity levels [30]. The low melting point and high surface mobility of NiS can accelerate sintering [31], which may deteriorate the activity of catalyst. Furthermore, sulfur species can increase the carbon deposition on catalyst surface, which also decreases the catalytic activity [32].
The presence of HCl in syngas was reported to decrease the reforming and WGS activities of Ni catalysts [33][34][35][36][37]. Richardson et al. [33] found that the conversion of methane was extremely inhibited in the presence of HCl, due to the chemisorption of HCl by Ni. Coute et al. [36] demonstrated that HCl induced detrimental effect on WGS activity during steam reforming of chlorocarbons. Veksha et al. [37] investigated the mechanism for the activity loss of Ni catalysts during naphthalene reforming in the presence of 2000 ppmv HCl and demonstrated that naphthalene conversion is not influenced by HCl while WGS activity was poisoned due to the sintering of Ni. In the above mentioned studies, either H 2 S or HCl were present in gas streams during the reforming while in real syngas, these impurities are present simultaneously. To what extent the co-existence of both H 2 S and HCl in the gas can influence the catalytic activity during steam reforming has not yet been investigated.
The purpose of this work was to investigate the influence of H 2 S and HCl on the poisoning of synthesized and commercial catalysts during steam reforming of tar. It has been well known that Ni is an excellent metal for steam tar reforming. In this study, the addition of Fe is attempted, because Fe is a low cost material and Fe species has high redox activity [38]. Furthermore, the addition of Fe to Ni had beneficial effect on the performance of bimetallic catalysts under certain experimental conditions [24][25][26][27][28]. Four synthesized catalysts with different loadings of Ni and Fe on alumina support and two commercial catalysts were tested in a fixed bed reactor at different temperatures with varying contents of H 2 S and HCl in gas. Naphthalene was used as a model tar compound as it is one of the major tar species [39] which also has high stability during tar reforming [40][41][42]. In this study, 50 ppmv H 2 S and 300 ppmv HCl were used as they are in the range of typical concentrations of H 2 S and HCl present in syngas produced from biomass/ MSW [29][30][31][34][35][36][37]. The individual and combined effects of impurities on reforming and WGS activities of the catalysts and the reversibility of the catalyst poisoning are discussed.

Catalysts
Four catalysts with different Ni and Fe contents were synthesized using the method described elsewhere [37]. Briefly, the catalysts were prepared by impregnation of aluminum hydroxide (H 3 AlO 3 ·xH 2 O, Sigma-Aldrich) having particle sizes of 0.56-1.18 mm with known concentrations of Ni(NO 3 ) 2 ·6H 2 O (Sigma-Aldrich) and Fe(NO 3 ) 3 ·9H 2 O (Sigma-Aldrich) in aqueous solution. After evaporation of water in a rotary evaporator (Hei-Vap Precision, Heidolph Instruments, Germany), the materials were dried overnight in an oven at 105°C and calcined at 500°C for 2 h (heating rate 2°C min −1 ) in air, followed by sieving to obtain particle sizes between 0.56 and 1.18 mm. The synthesized catalysts are denoted as xNi -yFe, where x and y represent calculated molar contents of metals per 100 g of the resulted catalyst.
Two commercial catalysts from different manufacturers (6-holes monoliths from Xian Sunward Aeromat Co., China and 19 × 19 × 10 mm rings from Pingxiang Hualian Chemical Ceramic Co., China) were crushed and sieved to obtain 0.56-1.18 mm particles, and used as the reference materials. Fig. 1 shows the experimental setup for catalytic naphthalene reforming. A fixed bed reactor with a quartz frit (50-90 μm openings) was used. In a typical run, 0.5 mL of a catalyst was loaded into the reactor and heated at 15°C min −1 in reducing atmosphere containing 20 vol% H 2 -80 vol% N 2 (total gas flow 50 mL-STP min −1 ) to the reforming . In this study, 0.14 vol% naphthalene was used as it is within the range of typical concentrations of naphthalene produced from biomass/MSW gasification [43,44]. Naphthalene vapors were generated by purging N 2 gas through an evaporator containing heated naphthalene. H 2 , N 2 and H 2 S were supplied from gas cylinders using mass flow controllers. The steam and HCl vapor were generated from aqueous solution of HCl injected by a syringe pump into an evaporator. During experiment, all the gas lines were heated and kept above 150°C to avoid vapor condensation. After reforming reactor, the model gas passed through two water traps to capture HCl and one silica gel trap to remove moisture, and then was collected in Tedlar gas bags for analysis (collection time 5 min). Concentrations of CO, CO 2 and C 1 -C 5 hydrocarbons were measured by a calibrated gas chromatograph (Agilent 7890B, USA) equipped with one flame ionization and two thermal conductivity detectors. Steam reforming of naphthalene is analogous to steam reforming of other hydrocarbons [10,45]: (2)

Naphthalene reforming
Due to WGS activity, CO is partially converted to CO 2 over Ni catalysts [1,28,45]: Naphthalene conversion can be calculated by the following equation: where: n CO and n CO2 are the molar concentrations of CO and CO 2 generated during steam reforming of naphthalene, mol min −1 , and n naphthalene is the molar concentration of naphthalene in the feed, mol min −1 . All experiments were triplicated and the results are presented as averages of three experimental runs.

Characterization of catalysts
The catalysts were characterized by X-ray diffraction analysis with Cu-Kα radiation source (XRD, Bruker AXS D8 Advance), X-ray photoelectron spectroscopy with a dual anode monochromatic Kα excitation source (XPS, Kratos Axis Supra), X-ray fluorescence spectroscopy (XRF, PANalytical Axios mAx), transmission electron microscopy at 120 kV (TEM, JEOL JEM-2010) and N 2 adsorption at −196°C (Quantachrome Autosorb-1 Analyzer). Binding energies of elements in XPS spectra were corrected against an adventitious carbon C 1s core level at 284.8 eV. The processing of XPS peaks was carried out in the CASA XPS software. TEM images were used to measure the size of Ni nanoparticles in spent catalysts. The diameters were calculated using ImageJ software by analysing 150-200 Ni nanoparticles in each sample and assuming that nanoparticles have ideal spherical shape. Specific surface area of catalysts was calculated from N 2 adsorption isotherms using BET model. Total pore volumes were calculated from N 2 adsorption volume at P/ P 0 = 0.96. Temperature programmed reduction (TPR) was performed in a 5% H 2 /N 2 gas mixture at 30 mL min −1 flow rate with a temperature ramp of 10°C min −1 up to 900°C. Carbon content in the catalysts was measured by CHNS elemental analyser (Vario EL Cube).

Properties of catalysts
The properties of pristine Ni, Fe, Ni-Fe and commercial catalysts are presented in Table 1. Ni and Fe contents were determined from XRF analysis and used to calculate molar quantities of Fe and Ni. The molar Ni and Fe loadings per 100 g of catalysts were close to the corresponding theoretical values of x and y in xNi-yFe samples. The amount of Ni in 0Fe-0.4Ni and two commercial catalysts loaded into the reactor for steam reforming of naphthalene was nearly the same (approx. 1.90 mmol 0.5 mL −1 catalyst) due to the differences in bulk density, allowing comparison between the activities of Ni in the synthesized and commercial catalysts. The synthesized catalysts had higher BET specific surface areas and total pore volumes compared to the commercial catalysts. According to the high N 2 adsorption volumes at relative pressures P/P 0 > 0.1 and hysteresis loops between adsorption and desorption branches of isotherms ( Fig. S1), the synthesized materials were mesoporous. Among them, 0Fe-0.4Ni had the largest porosity (i.e. 213 m 2 g −1 and 0.31 mL g −1 , respectively) ( Table 1), which is one of the reasons for its better catalytic performance stated in the following study. The specific surface areas and total pore volumes of the synthesized catalysts decreased with increasing Ni + Fe contents, which can be attributed to the impregnation of the porous alumina with loaded metal species. X-ray diffraction (XRD) patterns of the synthesized catalysts in Fig. 2 consist of broad peaks with no distinct XRD peaks and also show no sharp XRD peaks indicating that alumina, nickel and iron oxides have non-crystalline and/or nanosized structures, so that alumina provides surface area for better dispersion of catalyst. On the other hand, in commercial catalysts, the XRD peaks of NiO (commercial 1 and 2) and α-Al 2 O 3 (commercial 1) can be clearly identified. Fig. 3 depicts the Ni 2p and Fe 2p core level spectra of the four pristine synthesized catalysts. Ni 2p spectra of 0Fe-0.4Ni, 0.1Fe-0.4Ni, 0.2Fe-0.4Ni and 0.5Fe-0Ni contain shake-up satellite peaks with binding energy (BE) of approx. 862 eV and peaks with BE of approx. 856 eV. In Ni-based catalysts, the binding energy of Ni 2+ typically increases with the strength of NiO-Al 2 O 3 interactions from approx. 854 eV for unsupported and weakly bound to the support NiO to approx. 856 eV for strongly bound to the support NiO [46][47][48][49]. At high NiO-Al 2 O 3 interaction levels, the binding energy of Ni 2+ in NiO of alumina supported catalysts becomes similar to that in NiAl 2 O 4 spinel (855.8 eV) [48,50]. Due to the shift in binding energy, it is uncertain whether the Ni 2+ state in the catalysts NiO or NiAl 2 O 4 solely based on Table 1 Characteristics of pristine catalysts.

Catalysts
Bulk density (g mL −1 ) BET surface area (m 2 g −1 ) Total pore volume a (mL g −1 )  [50]. TPR profiles provide useful information regarding reducibility of Ni and Fe oxides in the synthesized catalysts (Fig. 4). The reduction of catalysts occurred in a wide temperature range between 300 and 800°C. The catalysts 0.2Fe-0.4Ni and 0.5Fe-0Ni contained a distinct reduction peak at 475°C, which corresponds to the reduction of Fe 2 O 3 [51], i.e. the main Fe species in catalysts according to XPS. According to the TPR profile of 0Fe-0.4Ni, most of nickel was reduced at 500-700°C with the maximum reduction temperature at 590°C, which can assigned to highly dispersed NiO having strong metal-support interactions [48,52]. Small shoulder peaks at 350, 425 and 770°C were also observed. The reduction at 300-400°C is typically attributed to bulk and/or unsupported NiO, while the reduction > 700°C could be attributed to nickel aluminates formed due to sintering of NiO with Al 2 O 3 [48,53,54], indicating that minor quantities of these species could be also present in the synthesized catalysts. According to the similar positions of H 2 consumption peaks in the catalysts, the reducibility of Ni species was not influenced by the addition of Fe 2 O 3 and vice versa.
XPS and TPR data of the commercial catalysts are shown in Fig. S2. As it was reported elsewhere [37], in both catalysts, Ni 2+ was in the form of NiO. However, in Commercial 2, NiO was more strongly bonded to the support compared to Commercial 1. Considering the similar Ni loading per 0.5 mL catalyst bed for Commercial 1, Commercial 2 and 0Fe-0.4Ni, this allows investigation of the effects of H 2 S and HCl on the activity of catalysts with different strength of NiO-Al 2 O 3 interactions determined by the differences in porosity, crystalline structure, NiO dispersion etc. The addition of Fe to Ni-based catalyst provides further insight about the influence of H 2 S and HCl on the activity of catalysts with different metal composition.

Naphthalene reforming activity
Fig. 5 depicts naphthalene conversion using the six catalysts in the presence and absence of H 2 S and HCl at 850°C. CO and CO 2 were the only reaction products. No formation of C 1 -C 5 hydrocarbons was observed during the process. Naphthalene conversion over catalysts fluctuated during the first 30 min of experiment and was stabilized thereafter. In all catalysts containing Ni, the reforming activity was lower in the presence of H 2 S and HCl due to the poisoning effect (data in Fig. 5a against Fig. 5b). Furthermore, naphthalene conversion by 0.5Fe-0Ni was approx. 12% in the absence of H 2 S and HCl, and decreased to approx. 8% in the presence of H 2 S and HCl, suggesting the poisoning of Fe. Regardless of the presence of H 2 S and HCl, naphthalene conversion was stable during 5 h tests. The synthesized 0Fe-0.4Ni showed comparable conversion efficiency with commercial catalysts, which was likely due to the same amount of Ni loading per 0.5 mL bed in the three catalysts. These results suggest that there was similar poisoning effect on the naphthalene reforming activity for the catalysts with different strength of NiO-Al 2 O 3 interactions. Reforming activity of 0.1Fe-0.4Ni was similar to 0Fe-0.4Ni, while the higher content of Fe in 0.2Fe-0.4Ni resulted in the decreased naphthalene conversion. This could be attributed to the decreased porosity and specific BET surface area with the higher Fe content due to the occupation of surface sites (Table 1). Unlike Ni-based catalysts, 0.5Fe-0Ni merely achieved approx. 8% of naphthalene conversion, indicating that Ni is more active catalyst for naphthalene reforming compared to Fe. The lower catalytic toluene reforming activity due to Fe addition to Ni/zeolite catalyst was reported by Ahmed et al. [38], who found the depletion in basicity strength of  this Fe-Ni/zeolite catalyst leading to suppressed steam reforming. Elemental CHN analysis (Table S1) of the pristine and spent catalysts suggests that there was no significant increase in the amount of carbon after the reforming, indicating no coking happened in the presence of H 2 S and HCl. This can be attributed to the relatively high content of steam in the model gas (i.e. 26 vol%) that could assist in carbon gasification. Fig. 6 shows the TEM image of fresh 0.2Fe-0.4Ni after preheating in 20 vol% H 2 -80 vol% N 2 and spent 0.2Fe-0.4Ni after 5 h of reaction at 850°C in the presence of H 2 S and HCl. The comparison of the morphologies of fresh and spent catalyst indicates the absence of carbon deposition during naphthalene reforming, which is consistent with CHN analysis. After reforming, Ni was present in the form of discreet spherical nanoparticles. This is attributed to the sintering of Ni during the process [37]. On the contrary, Fe was evenly distributed over the catalyst surface (Fig. 6a). Fig. S3 shows that in other Fe-containing catalysts, Fe also remains in the dispersed state. The coverage of entire surface of the spent 0.2Fe-0.4Ni catalyst by S and Cl indicates that the chemisorption of these species occurred both on the Ni and Fe sites (Fig. 6b) [29,55,56], which explains the poisoning effect of HCl and H 2 S both on the reforming activity of Ni and Fe. Fig. 7 shows the XRD patterns of the spent catalysts after naphthalene reforming at 850°C in the presence of 50 ppmv H 2 S and 300 ppmv HCl. In the spent samples containing Ni element, the formation of metallic Ni phase was observed as suggested by the labelled NiO XRD peaks. As there were no XRD peaks of NiO in the fresh catalysts, these results indicate that upon reduction and reforming, Ni undergoes sintering into larger size crystalline nanoparticles, which is consistent with TEM data in Fig. 6a. Unlike NiO, the formation of crystalline FeO in 0.5Fe-0Ni was not observed as suggested by the absence of corresponding XRD peaks in this sample and even distribution of Fe in the TEM images of spent catalysts (Figs. 6a and S3). According to TPR data (Fig. 4), the reforming temperature was sufficient for the reduction of Fe 2 O 3 to FeO. Therefore, it is likely that in the spent catalysts iron was in metallic non-crystalline state. These observations are consistent with scanning TEM data in Fig. 6, showing the differences in Fe morphology compared to Ni.
There was no change in the position of NiO XRD peaks in 0Fe-0.4Ni, 0.1Fe-0.4Ni and 0.2Fe-0.4Ni with the addition of Fe (Fig. 7), which would have been observed with the formation of Ni-Fe alloys [27,57], indicating that there was no alloying between Ni and Fe in the spent catalysts. The amount of chemisorbed sulfur and chlorine species during reforming was typically low which explains the absence of XRD peaks corresponding to metal chlorides and sulphides in all catalysts.
The reaction temperature is one of the most important operating variables for steam reforming. 0Fe-0.4Ni, 0.1Fe-0.4Ni and Commercial 1 were further selected to investigate the effect of temperature on catalytic activity. Fig. 8 shows naphthalene conversion at 790, 850 and 900°C in the presence of H 2 S and HCl. Except for the decrease in conversion within the initial 30 min at 790°C, the activity of catalysts remained constant thereafter indicating that it is possible to maintain stable conversion efficiency of naphthalene in the presence of H 2 S and HCl within the studied period of time at each temperature. The catalytic activities of the three catalysts were similar in the presence of H 2 S and HCl at each temperature regardless of the strength of NiO-Al 2 O 3 interactions (0Fe-0.4Ni vs. Commercial 1) and the addition of Fe (0Fe-0.4Ni vs. 0.1Fe-0.4Ni). The reforming activities of all catalysts were greatly influenced by the reforming temperature, increasing from approx. 40% to approx. 100% efficiencies with the increase in reaction temperature from 790 to 900°C, respectively. These results can be attributed to the increased reaction rate of naphthalene with steam and the decreased H 2 S poisoning effect at higher temperature. It has been well known that H 2 S poisoning is caused by sulfur adsorbed on the nickel surface in the catalyst according to reaction (1). This reaction is reversible. With the increasing temperature desorption of H 2 S increases releasing surface active sites for the steam reforming reaction [58].
To determine the respective and relative roles of H 2 S and HCl in the catalyst poisoning effect observed in Figs. 5 and 8, the naphthalene reforming at four different conditions was compared: (1) 50 ppmv H 2 S and 300 ppmv HCl, (2) 0 ppmv H 2 S and 300 ppmv HCl, (3) 50 ppmv H 2 S and 0 ppmv HCl, and (4) 0 ppmv H 2 S and 0 ppmv HCl. The experiments were carried out at 790°C, as the poisoning was the most prominent at this temperature. According to Fig. 9, in the absence of H 2 S, naphthalene conversion was approx. 100% both at 0 and 300 ppmv HCl. In the presence of 50 ppmv of H 2 S, naphthalene conversion decreased to approx. 40% both at 0 and 300 ppmv HCl. These results suggest that the poisoning of naphthalene reforming was caused by H 2 S, while HCl had negligible effect on this reaction. Furthermore, since the naphthalene conversion in the presence of H 2 S was similar at 0 and 300 ppmv HCl, it can be concluded that H 2 S and HCl had no synergistic effect on the poisoning of reforming activity when both impurities were present in the stream.
Based on the obtained data, during the reforming of naphthalene from gas streams containing both H 2 S and HCl, the poisoning of catalysts is mainly caused by H 2 S and can be attributed to the decreased accessibility of surface active sites for hydrocarbons due to H 2 S chemisorption [29]. The poisoning effect on naphthalene reforming activity was similar for the catalysts with different strength of NiO-Al 2 O 3 interactions and Ni + Fe contents. Increasing reaction temperature could effectively improve catalytic activity of Ni and Ni-Fe based catalysts in the presence of H 2 S and HCl leading to approx. 100% naphthalene conversion (Fig. 8). Fig. 10 shows the ratios between CO and CO 2 in the gas during naphthalene reforming over the catalysts at 850°C in the presence of H 2 S and HCl. Steam reforming of hydrocarbons is typically presented as the combination of two reactions, namely, partial oxidation of hydrocarbon by steam into CO and H 2 (reaction 2) followed by WGS reaction   (3) [1,8,10,28,45]. Consequently, the lower CO/CO 2 ratio is probably attributed to the higher conversion of CO into CO 2 over catalysts via WGS reaction (3). Dashed black line shows the CO/CO 2 ratio at thermodynamic equilibrium (CO/CO 2 = 0.52 at 850°C). For all catalysts, the CO/CO 2 ratios were higher than 0.52 indicating that thermodynamic equilibrium was not attainable. This is due to the lower space velocity and longer residence time required for the equilibration of WGS reaction over the catalysts [37]. There were significant differences in the kinetics of WGS reaction as suggested by the different CO/CO 2 ratios for the catalysts. The CO/CO 2 ratios of synthesized Ni and Ni-Fe catalysts increased from 0.9 to 1.0 to 1.2-1.5 during the 5 h tests, depending on the sample. These changes were much lower compared to Commercial 1 and Commercial 2 catalysts (i.e. from 0.7 to 3.9 for Commercial 1 and from 1.0 to 2.2 for Commercial 2), suggesting higher stability of WGS activity of the synthesized catalysts. Based on the similar CO/CO 2 ratios for 0Fe-0.4Ni, 0.1Fe-0.4Ni and 0.2Fe-0.4Ni, the addition of Fe to catalysts did not alter the WGS activity of catalysts. Furthermore, the lower CO/CO 2 ratios over the synthesized Ni containing catalysts compared to 0.5Fe-0Ni indicate that the WGS activity over Ni was higher compared to Fe during naphthalene steam reforming. Although 0Fe-0.4Ni, Commercial 1 and Commercial 2 had similar NiO loading per catalyst bed volume, the strength of interactions between NiO and alumina support was different in the catalysts, eventually, leading to the different Ni-support interactions in the reduced catalysts. Specifically, the strength of interactions increased from Commercial 1 to Commercial 2 and, finally, to 0Fe-0.4Ni which is consistent with the increase in WGS activity in the same order (Fig. 10). One reason behind the observed phenomenon is the mechanism of WGS reaction over Ni based Al 2 O 3 catalysts. By combining density functional theory and microkinetic modelling, it was demonstrated that Ni-support interface provides catalytically active sites for WGS reaction, serving as a storage for oxygenated Ni 2+ species [59]. Therefore, the decrease in the strength of metal-support interactions in catalysts can result in the observed loss of WGS activity. In comparison, for the steam and dry reforming reactions of methane, the importance of metal-support interactions was found to be less important as the active sites for these reactions seem to be different. [59,60] Assuming that the mechanisms for reforming of hydrocarbons are similar [45], this could explain the negligible differences in naphthalene conversion over 0Fe-0.4Ni, Commercial 1 and Commercial 2 (as shown in Fig. 5).

WGS activity
For WGS reaction, oxygenated Ni 2+ sites are required [59], while higher reforming temperatures favor the reduction of NiO to metallic Ni. TPR profiles of synthesized and commercial catalysts (Figs. 4 and S2) show that the reduction temperature of Ni 2+ increased from Commercial 1 to Commercial 2 followed by the synthesized catalysts indicating that the synthesized catalysts can have the higher density of oxygenated Ni 2+ sites at the reforming conditions due to the stabilization of NiO by the support [48,52,61,62]. To confirm that, thermodynamic calculations using HSC Chemistry 9 software were carried out to calculate the content of oxygenated Ni 2+ in catalysts in the absence and presence of NiO-support interactions. For the simplicity of calculations, it was assumed that in the absence of interactions with the support, nickel can only be oxidized into NiO. In the presence of strong metal-support interactions, Ni can form stable oxidized species at the NiO-support interface. γand α-Al 2 O 3 were selected as the representative support materials, which allow for the formation of    Fig. 11a shows the standard Gibbs reaction energies (ΔG°) for the oxidation of Ni as the function of reforming temperature. It can be seen that ΔG°increases with temperature suggesting that higher temperature causes the formation of metallic Ni. However, at the same temperature, ΔG°is lower when γand α-Al 2 O 3 participate in the reaction, indicating that in the catalysts with strong NiO-support interactions, there is a higher content of oxygenated Ni 2+ . From the corresponding thermodynamic equilibrium constants, the content of Ni 2+ can be calculated at the experimental conditions. According to Fig. 11b, in the absence of NiO-support interactions, the content of Ni 2+ slightly increases with temperature and is 1.3%, 1.4% and 1.5% at 790, 850 and 900°C, respectively. In the presence of NiO-support interactions, the content of Ni 2+ is much higher at each reforming temperature (Fig. 11b). Notably, the γ-Al 2 O 3 favors the stabilization of Ni 2+ to a larger extent compared to α-Al 2 O 3 highlighting the importance of alumina material for the design of catalysts with tailored WGS activity. The provided thermodynamic calculations confirm that at the reforming temperatures, NiO-Al 2 O 3 interactions can indeed stabilize nickel in the oxidized form due to the participation of support in the reaction, which could be in turn responsible for the higher WGS activity on the synthesized catalysts. Since, the XRD patterns of the spent catalysts contain only metallic Ni phase (Fig. 2), it is likely that the oxygenated Ni 2+ species are mainly present at the NiO-Al 2 O 3 interface.
Previously, it was proposed that the exposure of catalysts to high concentration of HCl (2000 ppmv) during steam reforming of naphthalene causes the chemisorption of HCl on Ni followed by the sintering of Ni species into larger size nanoparticles. This process is irreversible and leads to a permanent loss of WGS activity [37]. The poisoning of WGS activity of catalysts by low concentrations of H 2 S and HCl (i.e. 50 and 300 ppmv, respectively) has not been investigated. Fig. 12 presents the CO/CO 2 ratios for 0Fe-0.4Ni, 0.1 Fe-0.4Ni and Commercial 1 at 790, 850 and 900°C. Among the tested catalysts, Commercial 1 showed the lowest WGS activity at each temperature. With the increase in temperature, CO/CO 2 ratios for Commercial 1 catalyst decreased, indicating that WGS activity of this catalyst could be improved by increasing the reforming temperature. Since the content of oxygenated Ni 2+ species is relatively high at all reforming temperatures, this could be attributed to the faster reaction rate that allows to approach closer to the thermodynamic equilibrium and/or enhanced desorption of S-and Cl-species at higher temperature. Nevertheless, the CO/CO 2 ratios for Commercial 1 remained high compared to those corresponding to thermodynamic equilibrium. The CO/CO 2 ratios for Ni and Ni-Fe catalysts were lower than that of Commercial 1 and closer to thermodynamic equilibrium at all temperatures, indicating higher WGS activity.
Since the poisoning of Commercial 1 was more pronounced at lower temperature, the individual and combined effects of H 2 S and HCl on WGS activities of two representative catalysts, namely, 0Fe-0.4Ni and Commercial 1, were compared at 790°C. Fig. 13a presents the CO/CO 2 ratios at four experimental conditions: (1) 50 ppmv H 2 S and 300 ppmv HCl, (2) 0 ppmv H 2 S and 300 ppmv HCl, (3) 50 ppmv H 2 S and 0 ppmv HCl, and (4) 0 ppmv H 2 S and 0 ppmv HCl. In the context with respect to WGS reaction, the presence of H 2 S and HCl had negligible effect on the poisoning of 0Fe-0.4Ni indicating high stability of the WGS activity to the action of both impurities. The deterioration of WGS activity of Commercial 1 was observed even in the absence of H 2 S and HCl. This could be attributed to the lower strength of NiO-Al 2 O 3 interactions in this catalyst compared to 0Fe-0.4Ni. As shown in Fig. 13b and c, the sizes of Ni nanoparticles were larger in the spent Commercial 1 compared to 0Fe-0.4Ni after using condition 4, which could result in the lower WGS activity [59].
For Commercial 1, CO/CO 2 ratios increased both under condition 2 (HCl only) and condition 3 (H 2 S only), indicating that both impurities contributed to the poisoning of WGS activity (Fig. 13a). The poisoning of WGS activity in the presence of H 2 S was faster compared to HCl as demonstrated by the rapid increase in CO/CO 2 ratio within the first 60 min of reaction (i.e. condition 2 against condition 3). The poisoning of catalyst was more pronounced in the presence of both H 2 S and HCl (condition 1), indicating the detrimental synergistic effect of impurities. According to Fig. 13b and c, at low concentrations of H 2 S and HCl, there was no change in the sizes of Ni nanoparticles of 0Fe-0.4Ni and Commercial 1. These data suggest that unlike at 2000 ppmv HCl in literature [37], low concentrations of H 2 S and HCl are unable to enhance Ni sintering, and the detrimental effect on WGS activity of Commercial 1 was most likely associated with the poisoning of catalyst surface solely via chemisorption. This could explain the increase in WGS activity of Commercial 1 with the increase in the reaction temperature from 790 to 900°C in Fig. 12 as higher temperature typically decreases chemisorption. If this hypothesis is correct and chemisorption is the main reason for the catalyst poisoning, after desorption of S and Cl species, the WGS activity of catalyst can be restored. On the other hand, if sintering causes the poisoning as observed for high concentrations of HCl (i.e. 2000 ppmv), the loss of WGS activity would be irreversible [37]. To test the hypothesis, the spent Commercial 1 and 0Fe-0.4Ni after 5 h of naphthalene reforming at 790°C in the presence of 50 ppmv H 2 S and 300 ppmv HCl (denoted as Exp. 1 in Fig. 14) were respectively used for the subsequent 5 h naphthalene reforming at 790°C in the absence of H 2 S and HCl (denoted as Exp. 2 in Fig. 14). According to Fig. 14a The structure of catalyst played an essential role in WGS reaction but not in reforming reaction (which was strongly influenced by temperature). The stronger NiO-Al 2 O 3 interactions provided beneficial effect to catalytic activity which could be probably attributed to the CO/CO 2 ratio (-)     formation of larger content of oxygenated Ni 2+ species that serve as active sites for WGS reaction [59]. The poisoning effect of HCl and H 2 S on WGS was more pronounced in a catalyst with weakly bonded NiO to the Al 2 O 3 support. At low H 2 S and HCl concentrations, the poisoning of WGS activity proceeds via chemisorption of S and Cl species and the loss of catalytic activity is reversible when H 2 S and HCl are removed from the gas stream.

Conclusions
The effects of H 2 S (50 ppmv) and HCl (300 ppmv) on catalytic steam reforming of naphthalene were investigated using Ni, Ni-Fe and Fe catalysts supported on alumina at 790, 850 and 900°C. Ni had higher reforming and WGS activities compared to Fe and the activities of Ni were not significantly influenced by the addition of Fe. H 2 S poisoned naphthalene reforming activity of the catalysts, while the addition of 300 ppmv to gas stream had no effect on this reaction at 0 and 50 ppmv H 2 S. On the contrary, both HCl and H 2 S could poison WGS activity of the catalysts and the poisoning effect was more pronounced when both impurities were present in the gas stream. The poisoning by H 2 S could be only partially restored by removing H 2 S from the gas stream indicating the strong chemisorption of H 2 S on Ni. However, H 2 S poisoning effect could be prevented by carrying out reforming of naphthalene at higher temperatures. Specifically, the increase in temperature from 790°C to 900°C increased naphthalene conversion from approx. 40% to approx. 100%. The poisoning of WGS activity during naphthalene reforming was significantly influenced to the structure of catalyst. The stronger NiO-Al 2 O 3 interactions provided beneficial effect minimizing the loss of WGS activity. This beneficial effect could be attributed to the formation NiO-support interfaces upon reaction serving as active sites for WGS reaction. At these concentrations of H 2 S and HCl (i.e. 50 and 300 ppmv, respectively), the loss of WGS activity was reversible when H 2 S and HCl were removed from the gas stream.