Effect of H2S and HCl on solid oxide fuel cells fed with simulated biosyngas containing primary tar

Integrated biomass gasifier solid oxide fuel cell systems are an alternative to fossil‐fuel‐based combined heat and power generators. However, biosyngas contaminants represent a bottleneck for small‐scale systems. In this work, we present the results of experiments on the effects of H2S, HCl, and acetic acid as model primary tar on Ni‐GDC SOFC. First, the effects of 17‐128 g/Nm3 dry basis acetic acid were studied. On a second cell, 0.8 and 1.3 ppm(v) H2S were added to the simulated biosyngas anode flow. After a full recovery, the cell was exposed to 42 g/Nm3 acetic acid and 0.8 ppm(v) H2S. On a third cell, 3.4, 20, and 50 ppm(v) HCl were tested and, after a recovery period, 42 g/Nm3 acetic acid and HCl were added. Even 0.8 ppm(v) H2S caused an immediate voltage drop. H2S affected CH4 reforming and water‐gas shift reaction. Differently, even 50 ppm(v) HCl appeared not to significantly affect these reactions. Acetic acid increased the cell voltage but caused carbon deposition at the cell inlet. The voltage increase seemed not to be affected by H2S or HCl, and no acetic acid was measured at the cell outlet, indicating that these contaminants do not affect the primary tar conversion.


| INTRODUCTION
Electricity and heat production accounts for the 25% of global greenhouse gas emissions. 1 Integrated biomass gasifier solid oxide fuel cell systems might represent an alternative to fossil-fuel-based combined heat and power (CHP) generators, thus reducing carbon dioxide emissions in the atmosphere. The low energy density and scattered distribution of biomass make it is necessary to develop small-scale systems to exploit the biomass potential sustainably.
Among the various gasification technologies, updraft gasifiers are suitable for this application since they can handle different types of biomass in terms of moisture and size, emit low amounts of particulate matter, and have high cold gas efficiency. 2 Therefore, they can be fed with both energy crops, that is, plantations grown specifically for energy production, and residual waste streams, including by-products from agriculture, forestry, farm, and agro-industry. However, updraft gasifier biosyngas contains large amounts of light condensable compounds, volatile organic compounds, and tar (around 150 g/ Nm 3 dry basis 2 ). These compounds might create several issues as plugging filters pores, gas lines and heat exchangers, deactivation of catalysts, and condensation in cold spots and even at high temperature if polymerization occurs. However, these | 2457 compounds might act as fuel in the SOFC, and due to the endothermicity of reforming reactions, they could reduce the excess cathode air required to maintain constant the SOFC temperature, thus increasing the system efficiency. Moreover, avoiding or simplifying the tar cleaning step reduces system complexity and costs, thus facilitating the technology entry in the market.
The SOFC operating temperature, the presence of Ni catalyst, steam, and carbon dioxide in the inlet gas stream as well as generated at the triple phase boundary make direct internal reforming, that is, the reforming of tar and hydrocarbons directly in the anode chamber, theoretically possible. Various authors have investigated with experimental work the feasibility of direct internal reforming. Usually, model tar compounds are used in these studies. Toluene is probably the most used model tar compound. [3][4][5][6][7][8][9] However, also naphthalene [10][11][12] and benzene 13,14 are used. Only few works can be found with primary tar (eg, methanol, ethanol, acetic acid, and hydroxyacetone), and they are mostly focused on direct utilization and not on primary tar as biosyngas contaminants. [15][16][17] These compounds are called tar despite they generally have a molecular weight lower than that of benzene. They are the first compounds released during biomass pyrolysis and gasification, and they further evolve to form secondary and tertiary tar, such as toluene, naphthalene, and coronene. 18 At present, there is not yet general agreement on the fate of tar in SOFC. Papurello et al reported an irreversible increase in the low frequency range impedance spectra of a Ni-YSZ cell fed with simulated biosyngas and an amount of toluene as low as 0.1 g/Nm 3 . 19 Pieratti et al coupled a cocurrent fixed bed biomass steam gasifier with two SOFC stack. A filter with calcined dolomite and manganese oxide supported on zirconium silicate was used for tar and H 2 S removal. After a short period of time, the stack performances decreased, may be due to the insufficient dimension of the filter. 20 On the other hand, Baldinelli et al indicated that 10 g/Nm 3 toluene in a simulated biosyngas mixture on Ni-YSZ did not harm the cell and no carbon was observed in postmortem analysis. 21 Some groups have carried out short-term tests with biosyngas from an autothermal downdraft gasifier and from an allothermal bubbling fluidized bed gasifier with real tar up to 3 g/Nm 3 . Despite the presence of tar, a Ni-GDC anode cell fed with the gas showed continuous stable performance. However, in a test where biosyngas particulate reached the cell, postmortem SEM analysis identified carbon-based tubular structures in the anode functional layer. 22 Hofmann et al integrated a circulating fluidized bed biomass gasifier with a single cell test station. They measured a stable performance when feeding a Ni-GDC cell with biosyngas containing 10 g/Nm 3 tar. 23 Long-term tests with single cells and stack tests using real tar are however still required.
Biosyngas contains other minor species, among which H 2 S and HCl, that can harm the SOFC and negatively affect the catalytic reactions occurring in the anode chamber. The effects of H 2 S have been widely studied with different fuels, anode materials, and operating conditions. With Ni-YSZ anode cells, when H 2 is the fuel, H 2 S is adsorbed on Ni active sites partially inhibiting electrochemical reactions. 24 Ni coverage causes a sudden drop in the cell operating voltage. 25 This is followed by a quasi-steady state voltage, 26 or by a slow degradation. 27 The higher the H 2 S concentration, the more severe the drop. Rasmussen and Hagen indicated that for an anode-supported Ni-YSZ cell under load at 850°C, saturation coverage is reached around 40 ppm(v) H 2 S. 28 Norheim et al indicated 80 ppm(v) H 2 S at 800°C. 29 The effect was reversible and decreased with increasing operating temperature. 30 The presence of steam seems to help the desorption of adsorbed sulfur. 31 Different authors reported contradicting results on the effects of current density, with some suggesting a beneficial effect 32 and some a worsening impact. 33 Analysis of Ni-YSZ anode cell impedance indicated that H 2 S causes an increase in the cell polarization at high frequency, which was ascribed to charge transfer process. However, at low hydrogen partial pressure, also the low frequency part of the cell polarization resistance was affected. 34 Ni-GDC anodes seem more tolerant to sulfur poisoning, due to the large surface area of ceria remaining active for electrochemical reactions. 12 For H 2 S concentrations higher than 0.1% in volume, sulfur can react with the anode material. 35 However, Dong et al detected Ni 3 S 2 using Raman Spectroscopy for the analysis of a cell contaminated with only 100 ppm(v) H 2 S. 36 The formation of sulfur compounds is considered responsible for irreversible cell performance degradation. 37 When carbon-containing fuels are used, H 2 S might affect also the catalysis of reforming and water-gas shift reactions. 38 Rasmussen indicated that in a Ni-YSZ anode-supported cell fed with a mixture of CH 4 /H 2 /H 2 O and operated at 850°C, 4 ppm(v) H 2 S has a larger effect on reforming than on electrochemical activity, with more than 60% of the reforming activity suppressed. 39 In fact, as reported by Rostrup-Nielsen, the sites for reforming and electrochemical reactions are different. 40 Similarly, Hagen observed an increase in both the high and low frequency range of a Ni-YSZ impedance spectra fed with H 2 /H 2 O/CO and 2-9 ppm(v) H 2 S. The increased H 2 diffusion resistance was ascribed to poisoning of the watergas shift reaction. 41 Cells with Ni-ScSZ anodes are reported to have better sulfur tolerability, 42 that is, outrun by Ni-GDC anodes, which remain active for H 2 and CO electrochemical oxidation 43 and are reported to be active toward catalytic CH 4 reforming. 44 In addition to methane reforming, Sasaki et al studied the effect of H 2 S on ethane, propane, butane, and iso-octane with Ni-ScSZ cells. They found that 3 ppm(v) H 2 S accelerates carbon deposition due to the suppression of reforming. 24 Seemingly contradictory results were obtained by Boldrin et al when investigating toluene reforming on Ni-GDC composites. The presence of H 2 S above 11 ppm(v) resulted in a decrease in the carbon formation. 45 Also, the effect of HCl contamination on SOFC anodes has been investigated in the past years. 12,[46][47][48][49][50] According to equilibrium calculations, HCl reacts with Ni forming NiCl x . These compounds are gaseous at SOFC operating conditions, and their concentration increases fast with low HCl amount and is almost stable at high HCl amount. 51 The poisoning effect is believed to be due to the formation and subsequent sublimation of NiCl 2 , as observed by Sasaki et al using Cl 2 as chlorine precursor. 24 Experimental results showed that poisoning by low HCl concentration is reversible. 52 At low concentrations in hydrogen feed, Aravind et al found that 9 ppm of HCl does not affect the performance of cells with Ni-GDC anodes. Similarly, Li et al and Błesznowski et al reported that a Ni-YSZ anode-supported cell can tolerate up to 10 ppm of chlorine on short-term operation at 750°C. 12,53 At concentrations higher than 100 ppm, Haga et al observed an only partially recoverable degradation that was ascribed to adsorption on active sites and sublimation of NiCl 2 . 49 Kuramoto et al found that Ni-YSZ cells operated at 900°C with a current density of 150 or 200 mA/cm 2 were not affected by the presence of 10 ppm(v) HCl when fed with simulated post-CCS syngas (H 2 /N 2 /H 2 O = 70.9/23.6/5.5 vol%). 26 A different and faster degradation mechanism takes place when a carbon-containing fuel is used compared to hydrogen. Marina et al reported that the presence of 50-800 ppm HCl in coal syngas causes immediate and reversible minor power losses in SOFC. The authors attributed the losses to the contaminant adsorption on Ni, thus blocking its electrocatalytic activity. 46 Trembly et al tested Ni-YSZ button cells at 800°C with simulated coal syngas and observed performance losses right after the injection of the contaminant even at concentrations as low as 20 ppm; they explained the poisoning effect of HCl with adsorption and blocking of the active sites. 50 Conversely, Bao et al observed no significant degradation when feeding an electrolyte-supported Ni-YSZ cell with 40 ppm HCl. 54 Xu et al observed with SEM a morphological change in the surface of Ni particles that were more scabrous after exposing the cell to syngas with 100 ppm HCl. Despite no significant performance loss, the presence of HCl might cause long-term degradation and also cause wear and corrosion of other parts made of stainless steel. 47 Recent studies carried out by Reeping et al with Ni-YSZ cells operated at 700°C with methane as fuel and 100-300 ppm HCl resulted in rapid cell performance degradation. Adsorbed HCl blocked at least part of the nickel active sites for CH 4 adsorption while hydrogen continued to dissociate and diffuse to the TPB. At higher temperatures, the cell degraded less but the degradation was irreversible due to the sublimation of NiCl x . 55 Conversely, Madi et al reported more sever degradation when the cell was fed with H 2 rather than with syngas. They explained the observation by the decrease in HCl adsorption as due to competitive CO adsorption and oxidation. 56 Papurello et al showed that below 20 ppm(v) a cell fed with biogas reformate does not suffer degradation, and below 40 ppm(v) there is only a slight influence on the electrochemical processes occurring at the anode. Above this concentration, the loss in performance was considered to be probably caused by HCl adsorption on Ni. 19 Despite the presence of publications on the topic, there is no unanimity on the tolerance level indicates the necessity to carry out more studies.
Direct internal tar reforming might cause carbon deposition and failure due to thermal and mechanical stress. 57 Despite the presence of some literature on direct internal tar reforming, not sufficient research is available on the impact of biosyngas primary tar on SOFC. Moreover, excluding the studies of Sasaki et al on copoisoning of H 2 S and hydrocarbons, 24 and Papurello et al 19 and Boldrin et al 45 on cross-influence of sulfur and toluene, the effect of H 2 S on tar reforming and in particular on primary tar has not been investigated. The same holds for the cross-influence of HCl and tar, with exception for a previous work where toluene was used as tar. 58 Furthermore, there is no agreement on the effect of HCl on catalytic reactions involving methane and carbon monoxide. The aim of this study was therefore to investigate the cross-influence of acetic acid, H 2 S, and HCl on the anode of solid oxide fuel cells. A commercial electrolyte-supported Ni-GDC SOFC was fed with simulated biosyngas and known concentrations of acetic acid, H 2 S, and HCl separately. Successively, the cross-influence of the tar with one of the other two contaminants was studied. SOFC tolerance limits to tar, H 2 S, and HCl are not yet well defined and are based on single contaminant effects. The results are expected to help the design of integrated biomass gasifier SOFC systems and therefore facilitate the development of this technology.

| METHODOLOGY
In the present study, we used simulated biosyngas to evaluate the effect of H 2 S and HCl on primary tar reforming, as well as on CH 4 reforming and water-gas shift (WGS) reactions in an operating Ni-GDC SOFC. As reported in literature, tar and biosyngas components might compete for anode active sites, and therefore, the effect of tar is more thoroughly discerned under syngas. 45 A representative updraft biomass gasifier composition was taken from literature. 59 The H 2 S concentrations tested were 0.8 and 1.3 ppm(v) dry basis, while for HCl 3.4, 20, and 50 ppm(v) dry basis were tested. The H 2 S and HCl concentrations were selected based on SOFC tolerance limits found in literature. 60 Acetic acid was selected as primary tar, since it is one of the most abundant light condensable species generated from updraft gasifiers. 18 The concentration of acetic acid was varied from 17 to 128 g/Nm 3 dry basis, which is roughly the expected tar amount from updraft gasification. 2  . Thermodynamic equilibrium calculations were performed using the software FactSage version 5.4.1 (Thermfact/CRCT and GTT-Technologies) to calculate the required amount of water to avoid carbon deposition at 400°C, that is, the assumed operating temperature of a gas cleaning unit being developed by TU Delft and HyGear in FlexiFuel-SOFC project 61 for removing H 2 S around the ppm(v) level. The software takes as inputs the mass of the reactants, process temperature, and pressure and gives as outputs the products and their amounts based on Gibbs free energy minimization. 62

| Setup and equipment
A scheme of the test station used in this study is illustrated in Figure 1. A ceramic housing with platinum gauzes as current collectors on both anode and cathode sides was positioned in a temperature controlled furnace. A weight of 10 kg on top of the housing was used to assure contact between electrodes and current collector, and to improve the anode chamber sealing that was achieved using gaskets of Thermiculite™ 866. Electrolyte-supported cells (HC Starck) with 100 μm thick electrolyte, 40 μm anode, and 44 μm cathode were used. The anode was made of Ni-GDC and the cathode LSM mixed with 8YSZ; both electrodes had an area of 81 cm 2 . The electrolyte was 8YSZ and had an area of 100 cm 2 . The flow rates were regulated using Mass Flow Controllers (MFC) Bronkhorst EL-FLOW F201C (Bronkhorst). Steam was added to the fuel gas stream using a Controlled Evaporation Mixing (CEM) unit W202A (Bronkhorst). Acetic acid of 99.7% purity (Sigma Aldrich) was added using a peristaltic pump BT100-2J (Longer Precision Pump Co.). The liquid acetic acid was evaporated by trace heating the pipe where the contaminant was injected. The stainless steel pipes after the CEM and the acetic acid injection were trace heated to 150°C. Hydrogen sulfide was added using a gas bottle containing 500 ppm(v) of the contaminant in N 2 while for HCl a bottle containing 300 ppm(v) in N 2 (Linde). A polytetrafluoroethylene pipe was used for connecting the gas bottles with the MFC and this with the anode inlet as close as possible to the furnace to avoid any interaction with the stainless steel piping. The polarization (i-V) curves were recorded using an external load PLZ603W (Kikusui Electronics Corp.) and a DC power supply SM120-25D (Delta Elektronika BV). The open circuit voltage (OCV) measured was compared with the Nernst voltages calculated using Equation 1 below where R is the universal gas constant, T is the cell operating temperature, F is the Faraday constant, and P O 2 the equilibrium oxygen partial pressure at cathode and anode sides calculated using FactSage. A microGC Agilent 490 with a CP-Molsieve 5Å capillary for measuring CO, H 2 , N 2 , and CH 4 and a PoraPlot U capillary for measuring CO 2 (Agilent) was used to measure the outlet gas composition. Before reaching the microGC, the gas was passed through a condenser and a desiccator containing silica gel to remove the moisture contained in the gas. The anode outlet flow rate was back-calculated from the inlet N 2 flow rate and the N 2 outlet concentration measured with the microGC. This was then used to calculate the flow rates of H 2 , CO, CO 2 , and CH 4 . The results are used for a qualitative analysis of the trends observed at different operating conditions. The isopropanol used for tar sampling was analyzed with a Varian 430 GC-FID (Agilent).

| Testing procedure
After having reduced the cell at 950°C by stepwise replacing with H 2 the total anode flow rate of 1500 NmL/min N 2 used during heating up, the furnace temperature was lowered to obtain a cell temperature of 800°C. An anode flow rate of 1000 NmL/min composed of 15% H 2 , 20% CO, 15% CO 2 , 2% CH 4 , and 48% N 2 and a cathode flow rate of 3000 NmL/ min simulated air were used. Each contaminant concentration was maintained for 24 hours. In the H 2 S test, after the first 24 hours of exposure to the contaminant, the cell was operated for 24 hours under load with clean biosyngas to verify eventual recovery. In all the tests, the cells were operated at 800°C and nearly atmospheric pressure with a current of 68 mA/cm 2 . The outlet gas composition was measured continuously during the tests; i-V curves were recorded with clean biosyngas, and at the beginning and at the end of the exposure to each contaminant concentration. When measuring i-V curves, the current was varied only between 0 A (Open Circuit) and the operating current kept during the contaminant exposure. The amount of acetic acid at the cell outlet was measured by bubbling the gas in two impinger bottles in series containing isopropanol at room temperature and at 0°C, respectively. The sampling was carried out at the beginning and at the end of the injection of acetic acid, as illustrated in Figure 2.
After having performed the test with H 2 S, the cell temperature was increased to 950°C and the anode gas composition was changed to pure H 2 . The cell was kept at open circuit for 12 hours before decreasing the temperature to 800°C and changing the anode gas to biosyngas. The same cell was used to test the cross-influence of acetic acid and H 2 S. The cell was first exposed to 42 g/Nm 3 acetic acid; then also, 0.8 ppm(v) H 2 S was added. Acetic acid was then removed, and the cell was kept with H 2 S only for five hours. Also after testing HCl alone, the cell was kept at OCV at 950°C with pure H 2 for 12 hours. The temperature was then decreased to 800°C and the anode flow changed to simulated biosyngas. Successively, 42 g/Nm 3 acetic acid was added for 24 hours before adding an increasing amount of HCl, with each concentration kept for 24 hours. The flow of acetic acid was stopped 90 minutes  4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

| RESULTS AND DISCUSSION
Equilibrium calculations indicated that to avoid carbon deposition at 400°C, taking as input the selected gas composition and 150 g/Nm 3 of acetic acid, it is necessary to add 0.47 g of water per normal liter of dry gas, corresponding to a volume flow rate of 584 NmL/min. Figure 3 shows the cells performances at 950°C with dry H 2 , and at 800°C with dry and wet H 2 as well as with biosyngas. Table 2 shows the cell anode flow rates set at the cell inlet, measured at the cell outlet, and the equilibrium composition calculated using FactSage. The gas composition was measured during the operation of the cell with 68 mA/ cm 2 ; therefore, the additional oxygen flow reaching the anode was considered for calculating the equilibrium composition. Methane was not completely reformed in the cell anode. Therefore, also H 2 and CO flow rates were lower than the expected equilibrium values. The incomplete reforming might explain the difference between OCV measured and calculated using Nernst equation. As shown in Table 3, the OCV calculated replacing 8 NmL/min of methane with an inert gas (ie, considering that 8 NmL/min of methane do not react in the anode chamber) is very close to the measured value. A lower OCV could be due also to gas leakages. However, comparing the calculated and measured OCV with humidified hydrogen, the sealing of the cell appeared as decent. Figure 4 shows the OCV measured and calculated with the different concentrations of acetic acid tested. The presence of acetic acid resulted in an increase in the cell OCV, thus indicating that the primary tar was converted into fuel. However, the measured OCV was lower than the calculated values. Therefore, not all the acetic acid injected was converted into usable fuel.

| Test with acetic acid: cell 1 test 1.1-1.4
The presence of acetic acid resulted in an increase in H 2 , CO, CO 2 , and also CH 4 outlet flow rates. An increase in methane might indicate that acetic acid somewhat hinders methane reforming, or that it is converted at least partly via thermal decomposition rather than catalytic reforming. Interestingly, the cell temperature decreased from 793°C to 790°C when 128 g/Nm 3 acetic acid was present. The endothermic reforming of acetic acid might be responsible for the temperature decrease, but also the increased anode flow rate might have caused this effect. While the amount of CO 2 measured was equal to the expected equilibrium amount, the measured flow rates of CO and H 2 were lower than the equilibrium ones. This is in agreement with the lower OCV measured as compared to the calculated one. No acetic acid was measured at the cell outlet by tar sampling. Only with the highest concentration of acetic acid tested, that is, 128 g/ Nm 3 , traces (below 1 g/Nm 3 dry basis) of hydroxyacetone were measured. Therefore, acetic acid was completely converted. At the end of the test, a significant amount of carbon was present in the cell inlet ceramic pipe, at the inlet of the ceramic housing, and in a minor amount on the anode surface close to the inlet. The carbon deposited on the cell might have been formed in the inlet pipes and entrained by the gas or might be the result of incomplete catalytic reforming.
Even the highest concentration of acetic acid seemed to be completely converted. However, this compound caused severe carbon deposition. The formation mechanism is not completely clear. In a biomass gasifier SOFC system, severe carbon deposits could cause plugging of pipes and reactors. Large amount of acetic acid should therefore be considered as a contaminant in systems operating at these conditions, unless appropriate conditions for preventing the cracking of this hydrocarbon before it reaches the anode are adopted. Nonetheless, this compound is a valuable fuel and it would be preferable to convert it into H 2 and CO rather than removing it from the gas stream.

| Tests with H 2 S: cell 2 test 2.1-2.4
In the test with hydrogen sulfide, the presence of the contaminant caused an immediate and significant voltage drop, as visible from the open circuit voltage values presented in Table 4, in agreement with literature. 28 The 24-hour recovery between the tests with 1.3 ppm(v) and 0.8 ppm(v) H 2 S allowed an almost complete recovery of the cell OCV. Sulfur poisoning is considered reversible for concentrations lower than 100 ppm(v) and temperature above 600°C. 63 However, the 7-hour period after the addition of 0.8 ppm(v) H 2 S resulted in an incomplete recovery. This might have been due to an insufficient recovery time rather than an irreversible poisoning, since the concentration tested was even lower than the previous one. The analysis of the gas composition at the outlet helps understanding the causes of the decreasing OCV. The flow rates of CH 4 and CO increased while those of H 2 and CO 2 decreased, as visible from the values presented in Table 5. In fact, the presence of H 2 S prevented methane steam reforming, as reported in, 42 and partially hindered WGS reaction, as suggested in Refs. 41,64,65 After the 12-hour recovery period at 950°C with pure H 2 , the cell was again fed with biosyngas, and after 24 hours of operation, the test on the cross-influence of acetic acid and H 2 S started. The anode outlet flow rates are presented in Figure 5. As noticed during the previous test, adding acetic acid resulted in an increase in H 2 , CO, CO 2 , and CH 4 outlet flow rates, in addition to the OCV increase. Moreover, when also 0.8 ppm(v) H 2 S was added at the inlet flow, methane reforming was completely prevented and WGS partially hindered. Interestingly, when both contaminants were present, the amount of methane at the cell outlet was higher than the inlet set value. Moreover, when the flow of acetic acid was stopped, there was a clear decrease in the outlet flow rates of H 2 , CO, and CO 2 . This shows that part of the acetic acid undergoes thermal decomposition T A B L E 2 Inlet, outlet, and equilibrium anode flow rates rather than being fully catalytically reformed, according to the reactions presented in Ref. 66 : The OCV values presented in Table 6 also show that H 2 S does not significantly affect acetic acid conversion. The removal of acetic acid caused a voltage drop close to the voltage gain measured when acetic acid was initially added. Tar sampling at the cell outlet did not show the presence of acetic acid or other compounds, thus confirming that the primary tar was fully converted.
In summary, the presence of around 1 ppm(v) H 2 S caused a voltage drop due to the inhibition of methane and water-gas shift reaction. However, the effect was reversible. H 2 S does not significantly affect acetic acid conversion. In a biomass gasifier SOFC system, hindering WGS is not necessarily a drawback. Before reaching the SOFC, the biosyngas passes through a gas cleaning unit. If ZnO-CuO are used for sulfur removal at 400°C, the WGS reaction is catalyzed by the sorbents and its equilibrium is highly shifted toward the product. Since SOFC operates at temperatures higher than 400°C and the WGS is an exothermic reaction, inside the SOFC reverse WGS would happen, thus decreasing the availability of fuel. Differently, the negative effect H 2 S has on methane reforming might result in fuel starvation in systems operating without an external reformer. Therefore, despite literature indicates that oxidation reactions are less affected than WGS and CH 4 reforming, 67 the effect H 2 S has on these two reactions should be taken into account when considering a SOFC heat and fuel management.

| Tests with HCl: cell 3 test 3.1-3.7
The presence of hydrogen chloride did not appear to cause a significant effect on the anode catalytic reactions, as visible from the open circuit voltage values presented in Table 7, which decrease by only 4 mV when HCl was present.
After the 12-hour recovery period at 950°C with pure H 2 , the anode gas feed was again changed to biosyngas to test cross-influence of acetic acid and HCl. Table 8 shows the average outlet flow rates measured during the test.
As noticed during the test with cells 1 and 2, acetic acid caused an increase in H 2 , CO, and CO 2 outlet flow rates. Also CH 4 outlet flow rate increased, despite only marginally.
(2)  46,50,68 Nonetheless, the authors did not measure the outlet gas composition and there is therefore no direct observation of the phenomenon. Upon checking the gas composition with thermodynamic equilibrium calculations, it can be noticed that in both cases, the anode inlet composition was not at equilibrium at the cell operating temperature and HCl might have hindered WGS reaction. Since the syngas composition used did not contain any CH 4 , no information on the effect of HCl on direct methane reforming can be obtained. Reeping et al indicated that in a cell operating at 700°C with CH 4 , HCl inhibits CH 4 reforming on the Ni catalyst. 55 Also in this case, the gas outlet composition was not measured but the observation was derived from operando in situ and ex situ optical methods. The lower operating temperature and the higher chlorine concentration in Reeping et al experiments might explain the difference with the present study.
The results presented in this manuscript are however in agreement with other studies available in literature. In a study by Madi et al, anode-supported Ni-YSZ cells operating at 750°C with simulated syngas (CH 4 0.3%, CO 2 10%, CO 19.3%, H 2 50.6% and H 2 O 19.9%) showed no voltage drop when 10 to 90 ppm(v) HCl were added. However, when calculating the expected thermodynamic equilibrium gas composition, it is noticed that the inlet gas composition used was equal to the equilibrium composition at cell operating temperature, and HCl hindering CH 4 reforming would have caused a decrease in voltage of only 1 mV. By comparing the degradation of cells fed with HCl in H 2 or syngas, and observing no degradation in a stack fed with syngas and up to 500 ppm(v), the authors of the study suggested that CO might compete with HCl for Ni active sites thus acting as a protection for chlorine poisoning. 56 With a gas composition of 50.8% H 2 , 19.5% CO, 9.1% CO 2 , 0.7% CH 4 , and 19.9% H 2 O, Papurello et al measured the cell impedance spectra and observed no change in the charge transfer resistance of an anode-supported cell with up to 20 ppm(v) HCl. At higher contaminant concentrations, the effect of the contaminant was mostly visible on the high frequency resistance associated with charge transfer process. However, also in this study, the inlet gas composition used was already at equilibrium at the cell operating temperature of 750°C. 19 Also, Xu et al observed no voltage drop when 100 ppm HCl was added to simulated syngas with 30% H 2 , 26% H 2 O, 23% CO, and 21% CO 2 . Similar to the previous studies, the gas inlet composition was very close to the equilibrium composition expected at the cell operating temperature. Therefore, no information on the contaminant effect on Ni catalytic activity can be obtained. Nonetheless, there was no significant performance loss visible from the impedance spectra analysis. 47 54 In this case, the gas inlet composition was not at equilibrium at the cell operating temperature. Therefore, it appears that water-gas shift reaction was not hindered by the presence of chlorine.
In our previous tests reported in, 58 we studied the cross-influence of toluene and HCl on a Ni-GDC SOFC fed with hydrogen, and HCl appeared to affect the reforming of toluene. The different behavior observed with toluene and acetic acid can be due to the higher reactivity of acetic acid. This compound in fact might undergo at least partially thermal decomposition rather than catalytic reforming; thus, HCl does not necessarily affect its conversion. In the same study, we observed the formation of methane when toluene was present in the gas feed, and an increase in the outlet CH 4 flow rate when  55,58 In summary, a concentration of HCl up to 50 ppm(v) seems not to largely affect the catalytic reactions occurring in the anode of a Ni-GDC SOFC operating at 800°C. The contaminant does not affect the conversion of acetic acid. Based on these results, it might appear that HCl does not always require a significant cleaning effort and it is therefore possible to simplify the gas cleaning section of a biomass gasifier SOFC system. However, cross-influence studies of HCl and other tar species are required to determine HCl tolerance limits in systems with direct internal tar reforming. Furthermore, the long-term effect of the contaminant on the anode catalytic and electrochemical reactions should be further investigated. Large quantities of HCl might however cause corrosion in other downstream equipment, especially in the heat recovery section of a micro-CHP system. Therefore, the implications of having a simplified gas cleaning unit should be analyzed not only at cell but also at system level.

| CONCLUSIONS
Biosyngas contaminants can be a techno-economic bottleneck for the development of small-scale integrated biomass gasifier solid oxide fuel cell systems. Tar compounds are usually removed or reformed externally since there is not yet complete understanding of their fate in solid oxide fuel cell anode. The concentration of other biosyngas contaminants, such as H 2 S and HCl, is decreased in the ppm(v) range. However, this range is based on single contaminant effects. This might result in incorrect understanding of solid oxide fuel cell tolerance limits and improper design of the system gas cleaning unit. Therefore, the aim of this study was to investigate the influence of acetic acid, H 2 S, and HCl on the catalytic reactions occurring in the anode of Ni-GDC SOFCs.
The results presented clearly show that acetic acid is at least partially converted into useful fuel at SOFC operating conditions. Both thermal decomposition and catalytic reforming seem possible conversion mechanism. Severe deposition of carbon was observed at the inlet of the ceramic housing. The presence of H 2 S in biosyngas appears to completely stop direct CH 4 internal reforming, even at concentrations as low as 0.8 ppm(v). This results in lower amount of fuel available for the cell electrochemical reactions. Moreover, H 2 S partially hinders WGS reaction. Interestingly, the presence of H 2 S seems not to affect the conversion of acetic acid. This, together with the carbon deposited before the anode chamber, indicates that at least part of acetic acid might be thermally decomposed even before reaching the SOFC anode. The presence of HCl in concentrations up to 50 ppm(v) seems not to heavily affect the anode catalytic reactions. This manuscript provides additional insights on the fate of acetic acid in SOFC anode and on the cross-influence of H 2 S, HCl, and acetic acid. The results are expected to help the integration of biomass gasifiers and SOFC systems.