AMMONIA AS A CONTAMINANT IN THE PERFORMANCE OF AN INTEGRATED SOFC REFORMER SYSTEM

As supply of natural gas (NG) is limited, more attention is being given to operating fuel cells on syngas derived from gasification of feedstocks such as coal and biomass. Ammonia (NH 3 ) is one of the problematic contaminants contained in syngas produced from these nitrogen containing feedstocks. NH 3 can be easily oxidized to nitric oxide (NO) in a combustion process and thus if present in the anode exhaust gas would be problematic. The potential effects of NH 3 (particularly at low levels) on fuel cell system performance have not been well studied. The former studies on NH 3 have been limited to either the reforming process alone or testing the fuel cell at the cell level with NH 3 containing gases. No studies have been accomplished on a fuel cell system level basis. Objectives of this work are to obtain a comprehensive understanding of fuel cell system performance on syngas containing NH 3 using an integrated SOFC reformer system. Detailed analysis is conducted within the three major reacting components – indirect internal reformer, SOFC stack and combustion zone. Various simulation tools (etc., CHEMKIN, ASPEN, APSAT) are utilized for analysis. Results show that NH 3 conversion (into N 2 and H 2 ) in the internal reformer is about 50% when temperature is 750°C. NH 3 conversion (into N 2 and H 2 ) in the SOFC stack can affect NO x emissions significantly. More than 50% NH 3 left from SOFC stack can convert into NO x in the combustion zone. Experimental study is also planned to validate the theoretical results.


INTRODUCTION
It is well recognized that solid oxide fuel cell has a significant advantage of fuel flexibility over the low temperature fuel cells [1]. Various fuels can be processed to produce a reformate (containing primarily H 2 and CO) for direct use in a solid oxide fuel cell. The main fuel sources used to produce this reformate include fossil fuels (e.g., natural gas, oil and coal), and renewable fuels (e.g., biomass and waste). The impacts of multi-fuel operation of integrated SOFC reformer systems on component performance, design point selection, thermal management and overall system efficiency were discussed [2]. However, the potential effects of one major fuel impurity -NH 3 , haven't been studied well.
Depending on the gasifier design, operating conditions and the fuel bound N 2 content, the concentration of NH 3 in the coal syngas can be as high as 0.5 mol% [3][4][5][6][7]. Water based NH 3 removal systems involve cooling the syngas to around 400K, and can remove a substantial amount of NH 3 , with residual levels of up to 400ppm as reported [4]. Although the sensible heat recovered in this cooling process may be effectively utilized elsewhere, the process results in reduced cycle efficiency and is not preferable in the advanced power generation design. High temperature clean up systems, which can provide higher efficiency and have less serious tar issues, however remove very little or no NH 3 at all [4,8].
In the state-of-the-art integrated coal gasification, high temperature fuel cell and gas turbine hybrid systems, NH 3 contained in the cleaned coal gas is first passed through fuel cell before entering gas turbine. Any remaining NH 3 after the fuel cell stack may be converted to nitrogen oxides in the gas turbine combustor. Understanding of NH 3 performance in the SOFC system is thus important for coal based SOFC hybrid systems.
In addition to being present in coal derived syngas, NH 3 is also present in gas derived from pyrolysis of nitrogen bearing feedstocks such as biomass [8,13].
Recently, a number of research groups [9][10][11][12] have been showing great interest in NH 3 as a H 2 carrier or a direct fuel for fuel cells, which are, however, beyond the current scope of this study.
The former studies on NH 3 conversion are limited to either reforming processes or the fuel cell [10,[12][13][14][15] but none on a system level basis. Very little litereature work can be found to have detailed study on NH 3 reaction at very low partial pressure (<2.6 torr) in the reformer. Objectives of this work are to theoretically study the performance of low content NH 3 within a pre-commercial integrated SOFC system. Detailed study will be conducted within the three major componentsreformer, SOFC stack and combustion zone. In the further step, strategies on NO x emissions control within integrated coal gasification, fuel cell and gas turbine combined cycle will be proposed. Experimental study is also planned based on the theoretical results.

SOFC SYSTEM DESCRIPTION
A typical integrated SOFC reformer system, the Siemens Westinghouse 25 kW SOFC system [2], is studied in this work. Figure 1 presents an overall system schematic and details of the SOFC stack design based on natural gas fuel. Compressed and desulfurized natural gas is fed to indirect internal reformers (or pre-reformers) with part of spent fuel from an anode off-gas recirculation plenum near the top of the fuel cell stack. In the nickel-based reformers, methane (CH 4 ) or any higher hydrocarbons react with the steam brought with spent fuel at a temperature as high as 750°C, and are converted to H 2 , CO, CO 2 , and remaining CH 4 . This reformed fuel mixture then enters a fuel manifold at the bottom of the stack where it flows upwards and is distributed to the outside surface of the tubular cells. Meanwhile, after preheating in a recuperator, air is fed by an injection tube and flows upwards along the inside surface of the tubular cells. With fuel on the outside and oxygen from the air on the inside electrochemical reactions take place along the length of the cells. The temperature inside the module and along the length of each cell varies somewhat but the maximum temperature is generally kept below 980 o C -1050 o C. Anode off-gas enters the recirculation plenum where a fraction of it is recirculated and the balance flows into the combustion plenum to mix with the depleted air. The small amount of remaining anode off-gas is combusted with the depleted air to preheat the air through the recuperator and provide heat to the reformers before it is exhausted. Notice that Figure 1 just presents the overall concept for stack design. More details about mixing phenomena for combustion zone will be described and discussed in the following modeling section.
The main interest of this work is to investigate the performance of NH 3 as an impurity within the integrated SOFC system, which contains three major components where NH 3 reactions can occur: the reformer, SOFC stack and combustor. The process flow diagram among those three reactors is shown in Figure 2. The detailed configuration of each component will be described in the following sections. Other components, such as recuperator and heater, are not presented in this diagram.

Reformer
Four identical reformers with nickel based catalyst are used within 25kW SOFC system, each per quadrant of SOFC stack. Operating temperature is maintained within the range of 720-750°C. The reformer annular geometry (see Figure 3) is: inside radius r in = 6.6 cm, outside radius r out = 8.4 cm, length L = 40 cm.

Figure 3: Reformer geometry and internal flow configuration
Commercial NH 3 cracking units are designed based on the above reaction mechanism to produce mixture of H 2 and N 2 from anhydrous NH 3 . When reaction takes place on nickel catalyst at a temperature of 850°C to 900°C, most of the NH 3 is cracked and residual NH 3 content is rather low, less than 100ppm without requiring additional purifier as reported [18,22].
Due to the lower reaction temperature and much lower NH 3 concentration compared to the commercial NH 3 cracking units, however, the internal reformer might not convert most of the NH 3 into H 2 and N 2 before it enters the SOFC stack. Quantitative assessment of residual NH 3 level in the reformer outlet requires a further understanding of NH 3 decomposition mechanism under the reaction conditions unique to the reformer.
In this work, NH 3 contained in the fuel is at impurity level of less than 0.5% by volume. At low NH 3 concentrations, reaction mechanism for NH 3 decomposition can be described using the following sequence [10,23]: Step 1: Step 3: where (ad) denotes that the species is adsorbed on the catalyst surface.
Depending on the reaction temperature, rate limiting step in catalytic NH 3 decomposition can be the NH 3 adsorption (1), the recombinative desorption of N 2 (step 6), or both.

a) Low Operating Temperature
When the temperature is low (<727°C based on nickel catalyst), the decomposition rate is independent of NH 3 partial pressure, where the rate-limiting step is the recombinative desorption of N 2 , with activation energies around 125-210 kJ/mol [19,21]. The NH 3 decomposition rate is closely approximated by [21]: where, is the nickel surface atom density and is given as ; is the rate constant for the N 2 recombination in step 6, and is given as [21]:

b) High Operating Temperature
When the temperature is high (>727°C based on nickel catalyst), the decomposition rate is first-order dependent on NH 3 partial pressure and the rate-limiting step is the NH 3 adsorption, with activation energies in the range of 16-42 kJ/mol [19,21]. When partial pressure of NH 3 is low, the decomposition rate is independent of N 2 and H 2 partial pressures [19,24].
Based on the above observations, a power law rate model was suggested [19,21] to express the NH 3 decomposition rate at high temperature and low NH 3 partial pressure: where, is rate constant for NH 3 adsorption in the step 1 of NH 3 decomposition, and is the partial pressure. is given as [21]: Pa s cm molecules k (11) At temperature higher than 520°C, rate constant was obtained to fit experimental data for NH 3 decomposition using a nickel based catalyst [19]:  (12) In this work, NH 3 partial pressure is less than 2.6 torr (0.05 psi). The reformer operates at a temperature range of 720°C-750°C, within the high temperature region as discussed before. Therefore, NH 3 conversion can be regarded as dominated by NH 3 adsorption (step 1).
Other than expressed as a series of elementary reactions, surface reactions are also described and studied using global reactions. To seek a simple global rate expression, kinetics of NH 3 decomposition on Ni/Al 2 O 3 catalysts are described using the Temkin-Pyzhev mechanism [25][26][27]: Kinetics of NH 3 decomposition on various catalysts, vanadium nitride (VN), palladium (Pd) and Iridium (Ir), have also been studied and expressed in Langmuir-Hinshelwood format [29][30][31].
However, none of the global rate expressions was found appropriate for reaction conditions of interest in this work, very low pressure and relatively high temperature (~750°C).

SOFC
All the NH 3 remained after the reformer will enter SOFC stack. Or, if a SOFC system is designed to operate directly on coal syngas without external or internal reformers, then the NH 3 contained in the coal syngas will be a potential concern for SOFC stack, either degrade the SOFC performance or generate nitrogen oxides (NO x ). NH 3 as a fuel contaminant in coal syngas or biogas has been tested on SOFCs and showed no strong association of cell degradation [13,32]. Instead, NH 3 has been considered as a direct fuel for fuel cell based on the reaction (18). In 1980, Farr at al. [12,15] constructed and tested a solid electrolyte fuel cell operating on NH 3 fuel to generate electric energy and nitric oxide (NO for the production of HNO 3 ). It was shown that the fuel cell, [NH 3 , NO, N 2 , Pt/ZrO 2 (8% Y 2 O 3 )/Pt, air], produced mainly NO when operating at temperature around 1100K, and showed that Pt based catalyst has high selectivity to convert NH 3 into NO via electrochemical oxidation. Instead of producing NO, more recent research efforts are to avoid the NO formation in SOFC. The more recent concept for using NH 3 as a direct fuel in SOFC is sending NH 3 directly to SOFC anode surface containing a catalyst, such as iron oxide, Fe 2 O 3 , or nickel-based compound. NH 3 is first cracked into N 2 and H 2 , and the generated H 2 is then utilized for the electrochemical generation of electricity. Based on the experimental results using silver anode and platinum anode with or without iron-based catalyst, Wojcik et al. [14] predicted that NH 3 could work very well in an SOFC system based on nickel anodes, although no actual experimental work has been conducted on nickel anode SOFC in their work. NH 3 performance in a SOFC with Ni/8YSZ anode was studied by Dekker et al. [10]. In their cell tests, the fuel cell outlet gas was measured and analyzed for NO x and NH 3 to determine the NH 3 conversion. It was concluded that at operating temperature of 800-1000°C, the conversion of NH 3 is higher than 99.996% due to the withdrawal of H 2 by the electrochemical reaction and is close to the thermodynamic equilibrium. Most of the NH 3 is cracked into H 2 and N 2 . The NO x outlet concentration of the fuel cell was measured to be below 0.5 ppm at temperature up to 950°C and around 4 ppm at 1000°C. Some researchers [10,13] argue that NH 3 as a fuel or fuel impurity can be completely converted into N 2 and H 2 over the SOFC nickel based anode when temperature is high (> 590°C as found in [13]). However, other research groups [19,21] found out that NH 3 conversion on nickel based catalyst can be high but never reach an equilibrium level even with nickel based catalyst and within high temperature range (>500°C). For example, NH 3 conversion on Ni-Pt/Al2O3 catalyst was measured higher than 80% but less than 99% in the temperature range of 520-690°C [19]. From theoretical understanding, high operating temperature of SOFC helps increase reaction rate, however, the overall limiting step for NH 3 conversion can be the mass transfer. Slow NH 3 diffusion especially at low concentration can limit the reach of NH 3 to catalyst surface and therefore leads to lower conversion. Moreover, H 2 existing at anode site, with much higher concentration and diffusivity compared to NH 3 , can compete with NH 3 to reach catalyst surface, and further limit NH 3 conversion. Therefore, an accurate prediction of NH 3 conversion within SOFC stack should cover the effects of various factors including catalyst, temperature, residence time and fuel composition, especially, H 2 concentration and NH 3 concentration.   3 and O 2-. It is noteworthy that nickel, an active catalyst for NH 3 cracking as introduced before, is contained in both anode and interconnection contact, as shown in Figure 4 [33].
In the typical SOFC system operating on natural gas, anode reactions are generalized as: Due to the existence of NH 3 and nickel catalyst, the following additional anode reactions are considered based on the suggestions by different studies [10,12,14,15]: a. NH 3 cracking: Selectivity to NO or N 2 for NH 3 reaction depends on: catalyst, temperature, O 2diffusivity and NH 3 diffusivity, residence time and NH 3 molar flow rate (content and total fuel flow rate). Within the temperature range of 700-1000°C, it was observed [34] that NO formation reaction (17) is selective with platinum based catalyst. NH 3 cracking reaction (1) is selective with nickel based catalyst.
Since none of the test results show that NH 3 would degrade SOFC performance, the only concern about NH 3 in SOFC stack is its possible causes for NO x generation. No report has been found showing that significant amount of NO x can be produced within fuel cell stack itself. Instead, NO x may be produced when the depleted fuel cell anode gas containing NH 3 is combusted in a combustor before entering turbine. Theoretical study and experimental work are needed to determine the significance of NH 3 content within anode depleted gas on NO x production or reduction (reaction (17) or (18)).

Combustion zone
Geometry and internal flow pattern of combustion zone are depicted in Figure 5 based on the schematics of the 25kW SOFC system provided by Siemens Westinghouse Power Corporation [33,35], though the exact configuration and dimensions were not well known. The zone outlets were assumed to be through the sides of the zone (as shown in Figure 1), based on knowledge of the unit. Combustion zone outlet temperature is observed about 860°C, where the anode depleted fuel meet with cathode depleted gas (~16% O 2 ) and combust to generate heat. NH 3 at that temperature range can react to generate various products (etc., N 2 , NO, N 2 O) as discovered in various studies [36,37].
No catalysts are utilized within the combustion zones. Therefore, the study focuses on non-catalytic reaction of NH 3 . Actually, NH 3 is commonly used as a reducing agent for NO in both selective catalytic reduction (SCR) and selective noncatalytic reduction (SNCR). It was found that when NH 3 is injected into a fuel-lean zone, which is the case in this work, NO is reduced by the reactions (18)

Figure 5: Schematic diagram of combustion zone
It was also reported that NH 3 gives effective reduction of NO emission within a narrow temperature window around 750-850°C. Average temperature (~860°C) of combustion zones studied in this work is slightly out of the temperature range. One negative effect is that if NH 3 is excess, NO

Overall strategy and tools
ASPEN PLUS, a widely used commercial simulation tool for process engineering, is applied to determine the reaction kinetics within the reformer. ASPEN PLUS enables the users to define detailed reaction mechanisms and reactor conditions to predict reaction conversions, and understand reaction behavior. Detailed reformer model will be described in the later sections.
Reactions within SOFC stack are studied in two steps using two analysis tools: A model comprised of a series of perfectly stirred reactors (PSR) and a plug flow reactor (PFR) from CHEMKIN 4.0.2 is built for simulating the combustion zone.

Reactor model setup
Geometry description and flow configuration of the annular steam reformer are shown in Figure 3. The detailed parameters and specifications are listed in Table 1. Reformer geometry was determined from the observed data [33,40,41], while the catalyst information was assumed based on the reference work [42]. The reformer is simulated as a plug flow reactor using RPlug model provided by APSEN PLUS (see Figure 6).   Reformer operating conditions and flow information based on both cases of natural gas and coal syngas are summarized in Table 2. Constant operating pressure and temperature are assumed to approximate the actual steady-state operating conditions. A typical NH 3 content of 0.2% was assumed based on the reference work [3][4][5][6][7]. Coal syngas fresh fuel flow rate is much higher than natural gas due to its lower fuel heat value. Anode recirculation ratio is 0.55 for natural gas case and 0.15 for coal syngas. Those design details can be seen in the former work [2].

Reaction kinetics
Reaction mechanisms and rate expressions for NH 3 decomposition within reformer were discussed in the former section. Global reaction (1) is incorporated into the ASPEN PLUS model, and different reaction rate expressions are considered and compared.
Due to the low content of NH 3 existing in the reformer, NH 3 decomposition is interesting, however, not a dominating reaction. A complete model of reformer need include all the other possible reactions. Major reactions over a nickel supported catalyst within a steam reformer were widely adopted as a three-step mechanism: 1. Endothermic steam reforming reaction, Langmuir-Hinshelwood mechanism, a mechanism for surface catalysis in which the reaction occurs between species that are adsorbed on the surface, is commonly applied to determine the reaction rates [26,[42][43][44]. Typical kinetic rate equations for steam reforming of methane are derived by Xu and Froment, whose work was carried out over a commercial catalyst (Ni/MgAl 2 O 4 , 15.2% nickel) [ where, p is partial pressure of each species in the reactor;  Kinetic factors , , and adsorption constants are defined in Arrhenius format: The constants used in the current model for each of the chemical expressions of interest are presented in Table 3, Table  4 and Table 5, which are derived from literatures [42][43][44]. It is noteworthy that the above rate equations for the major reactions are dealing with intrinsic kinetics of methane-steam reforming and water-gas shift on the nickel based catalyst, and do not account for diffusion limitations. A more accurate reformer model needs to combine both reaction kinetics and diffusion limitations to have a more accurate prediction. However, the focus of current study is on the NH 3 conversion within reformer. NH 3 rate expressions as given in equations (10) and (12) were derived from experimental data, which accounted for diffusion effects.

SOFC stack
Detailed kinetic modeling of NH 3 reaction within SOFC stack is not considered in the current work. Instead, conversion ratio of NH 3 based on reference work is investigated and summarized as reference for consideration. A complete model of NH 3 reaction including consideration of chemical kinetics, mass transfer, and heat transfer will be built in the future.

Combustion zone
Mixing phenomena and reaction modes are shown in Figure 7, where depleted fuel from recirculation zone is mostly reacted in the area between SOFC tubes. Gas from that area is mixed and reacted with more air in the area between air feed tubes. SOFC tube has an external diameter of 2.2cm, while distance between SOFC tubes is about 3.3cm.  A kinetic model for simulating the combustion zone was achieved using CHEMKIN 4.0.2 for both cases of natural gas and coal syngas (see Figure 8). Mass flow rates and temperatures for model inlets and outlet were predicted using APSAT simulation tool (See [2]). To approximate the mixing phenomena when anode depleted fuel enters the combustion zone, a series of Perfectly Stirred Reactors (PSRs) with stepwise addition of air are considered. Major reaction zone between SOFC tubes is equally divided into three zones and simulated using three PSR models. Total residence time for those three reactors is about 0.1 second. Area between air feed tubes is simulated with another PSR, with a residence time about 0.5 second. The phenomena including gas composition change due to the heat exchange and temperature drop after combustion products leaving the combustion zone are simulated using a Plug Flow Reactor (PFR). Reaction mechanisms (GRI 3.0) are provided in CHEMKIN, which can predict potential fuel NO x from NH 3 as well as thermal NO x . PSR model is applied due to its simplicity and also because it provides solutions to flame problems more quickly [45].

Preliminary equilibrium analysis
The equilibrium model from CHEMKIN 4.0.2, coupled with steady state analysis tool APSAT, was firstly used to estimate the gas composition after each of the three reacting components -reformer, SOFC stack and combustor. Typical coal syngas from Wabash coal gasification project is considered [1,46]. Major results are summarized in the Table 6, which suggest: a) Most of the NH 3 is cracked into N 2 and H 2 before entering combustion zone. In case of coal syngas containing 0.2% NH 3 , only 13ppm NH 3 remains in the reformate stream and enters SOFC stack, while NH 3 conversion within the reformer is 99.2%. More NH 3 conversion is found in SOFC stack assuming chemical equilibrium is reached, which leaves only 0.3ppm NH 3 entering the combustion zone.
b) When comparing NO x levels after the combustion zone with and without NH 3 contained in the fuel stream, it is found that NH 3 is not the major contributor to the significant amount of NO and NO 2 predicted at the combustor outlet. Due to the high conversion based on equilibrium analysis, very little NH 3 (0.3ppm) remains after reacting in the reformer and SOFC, and therefore generates negligible fuel NO x at combustion zone. NO x level after combustion zone is predicted to be lower in the coal gas case than natural gas case due to lower combustor temperature.
c) NO x level (less than 0.02ppm) in the system exhaust obtained from the equilibrium analysis is much underestimated when it is compared to the observed data. NO x emissions have never observed to be higher than 1ppm, but still higher than 0.1ppm in the exhaust of 25kW SOFC system based on natural gas fuel [35], which suggests that a more detailed analysis of combustion zone should be conducted. In this work, a kinetics model is set up for such a detailed study with consideration of mixing and reaction phenomena. d) CHEMKIN equilibrium analysis of NH 3 reaction in the reformer can provide reference values but inaccurate results for actual NH 3 conversion due to the low NH 3 concentration and short residence time (<0.5 second). More detailed kinetics study is necessary, and the results are shown in the following section.

Reformer: kinetics results
The performance of reformer in natural gas case with SOFC anode recirculation is first analyzed using the ASPEN reformer kinetics model (see Figure 9). Methane reformation and reverse water shift reaction reach equilibriums at a reformer length ~0.3 meter. Due to the high reformer operating temperature (750°C), a significant content of CO (~21%) remains when exiting the reformer.  Figure 10 shows the process stream profiles along the reactor obtained from the reformer kinetic model for coal syngas case. The reactions for major reactants (H 2 , CO, CO 2 , CH 4 and H 2 O) reach thermodynamic equilibrium at a distance of 0.15 meter from the reformer inlet when mass transfer limitations are ignored. Due to low CH 4 content and high CO 2 content contained in the coal syngas, steam reforming is insignificant and the species concentrations are not changed much, which suggests that a reformer may not be necessary for just operating on typical coal syngas, unless the SOFC system is designed for multi-fuel (e.g., natural gas, biogas and coal syngas) operation, or coal syngas contains significant amounts of CH 4 as in the case of syngas from Lurgi or British Gas Lurgi (BGL) gasifiers. Instead, since CO content is very high (>40%) and water gas shift reaction is exothermic, a shift reactor may be considered to couple with SOFC stack when operating on coal syngas for converting CO into H 2 and provide some flexibility to help relieve the challenge of thermal management within SOFC stack.
Residence times for natural gas and coal syngas are compared in Figure 11. Due to the lower inlet flow rate (see Table 2), reformer on coal syngas show a smaller residence time at the beginning compared to natural gas. At the very end of the reformer, coal syngas case shows a higher residence time, because coal syngas composition doesn't change much along the reformer, however, molar flow rate keeps increasing in natural gas case when steam reformation occurs. Coal syngas Natural gas Figure 12: NH 3 conversion along the reformer NH 3 reaction is considered and its conversion in the reformer is predicted based on the reaction mechanism and rate constant introduced in equations (10) and (12). For coal syngas case, as shown in Figure 12, the NH 3 reaction does not reach equilibrium within the reformer reactor, and the NH 3 conversion is limited to about 50% at the end of reformer (temperature = 750°C). For natural gas case, NH 3 reaction almost reaches equilibrium at the end of reformer due to its lower inlet NH 3 content, which is highly diluted by the recirculated anode depleted fuel.
It is noteworthy that NH 3 conversion has a nearly linear dependence on the reformer operating temperature. The increase of temperature enhances the conversion (see Figure  13), because NH 3 decomposition is endothermic. When temperature is lowered to 700°C from 750°C, the NH 3 conversion within the reformer will decrease to 16% from 50% for coal syngas case, and from 45% to 15% for natural gas case. Temkin-Pyzhev mechanism and rate constant based on equation (13) have also been incorporated into the reformer model, and the results show that NH 3 reaction reaches thermodynamic equilibrium very quickly (less than 0.0001 second). Considering equation (13) is derived at very high pressure, this fast reaction mechanism is not adapted in this work.

SOFC
As introduced before, an accurate prediction of NH 3 reactions and conversion in SOFC stack requires a comprehensive understanding of all different factors that could affect NH 3 reaction formats and rates, which could be very complicated and need a separate study. In this work, the following two assumptions are made for simplification based on the results shown in literature [10,13,19,21]: • All the reacted NH 3 is cracked into H 2 and N 2 .
• NH 3 conversion (with the formation of N 2 ) is in the range of 80-99.996%.

Combustion zone
NO x predictions from combustion zone model are summarized in Table 7, which are generated based on different cases with various fuel types, NH 3 contents and NH 3 conversions in SOFC stack. All NO x values are adjusted to a 15% O 2 dry bases.
When typical natural gas is used as the fuel (case NG 1), thermal NO x emissions are predicted to be about 0.3ppm which is close to the observed data of 0.35ppm [35]. When 0.2% NH 3 is added into natural gas (case NG 2), and 80% NH 3 conversion in the SOFC stack is assumed, then NO x level is about 3.5ppm. Combustion zone in Case NG 2 has almost the same operating conditions (e.g., temperature, residence time…) as Case NG 1 due to the negligible effects of low content of NH 3 . Therefore, the thermal NO x emissions from Case NG 2 can be considered the same as in Case NG 1 -about 0.3ppm. The increased NO x in Case NG 2, fuel NO x , is sourced from the adding of NH 3 . The conversion of NH 3 to NO x in the combustion ozone is about 53% (see Table 8), where the left part of NH 3 is converted into N 2 .
For coal syngas cases, thermal NO x , as shown in case CS 1, is about 0.16ppm, lower than NG Case 1 due to the lower temperature. Fuel NO x is found significantly depended on the NH 3 conversion in the reformer and SOFC stack. When NH 3 conversion in the reformer is about 50% as predicted by the model at a temperature of 750°C and the NH 3 conversion in SOFC is low (80%), 0.2% NH 3 contained in the coal syngas can cause 17ppm total NO x for a SOFC system integrated with a reformer, and 37ppm NO x without a reformer. Therefore, a reformer can help reduce NO x emissions when NH 3 conversion is not very high at SOFC stack.  When the conversion is high (99.996%), the NO x emissions are very low (less than 0.2ppm) for each case, and concern about NH 3 as a contaminant in the coal syngas is not necessary. It suggests that coal syngas from high temperature coal gasifier can be directly sent to SOFC system without special design for NH 3 removal. Also, it suggests that if NH 3 conversion is high in SOFC stack, a reformer is not necessary for SOFC system to crack NH 3 when operating on such type of coal syngas.
The mechanism routes for NH 3 to form NO x are plotted in Figure 14 for both cases of natural gas and coal syngas. NH 3 molecule is first broken down into NH 2 species. NH 2     It is also noteworthy that the NO x reduction mechanism as introduced in equation (19) is insignificant in this combustion zone due to the high local temperature in this study. For both cases of natural gas and coal syngas, the temperatures in different PSRs are higher than 860°C (higher ~1100°C in PSR 1 for both cases of natural gas and coal syngas as seen in Table  9), which is out of the temperature window (750-850°C) for the SNCR reaction to occur.
Quantitative analysis of NO formation in different routes is conducted and the results are summarized in Table 9. Because NO 2 comes from NO, the formation rate analysis of NO is equivalent to NO x . It is found that most of NO is produced in the first PSR reactor (PSR 1). The NO formations in the other PSR reactors are at least 2 orders of magnitude less than PSR 1. Thus, the analysis is focused on PSR 1. For both of natural gas and coal syngas cases, about 90% NO is formed from HNO and the left is mainly from NH. In PSR 1, the residence time is about 0.02 second. Sensitivity analysis shows that the change of residence time affects the net formation rate in each reaction, but doesn't change much the overall NO net formation rate.

CONCLUSIONS
The study of this work suggests the following major conclusions: 1. NH 3 as a contaminant in the SOFC integrated system will not degrade SOFC performance based on literature review, however, potentially causes NO x emissions from combustion zone according to the investigation in this work.

NH 3 conversion in the internal reformer is about 50%
when temperature is 750°C based on model results. 3. More accurate theoretical understanding and prediction of NH 3 conversion in the SOFC stack need a comprehensive study of mass transfer, heat transfer chemical reactions and electrochemical reactions, which should consider the following major factors: catalyst, temperature, O 2diffusivity and NH 3 diffusivity, residence time and NH 3 molar flow rate. 4. NH 3 conversion (into N 2 and H 2 ) in the SOFC stack can affect NO x emissions significantly. The current study shows that when the conversion is high (99.996%) as some research groups suggested, the concern about NO x emissions (less than 0.2 ppm) from NH 3 in the coal syngas is not necessary. It suggests that coal syngas may be directly sent to SOFC system without a reformer. 5. Lower NH 3 conversion in the SOFC stack may cause high fuel NO x . When the conversion is 80%, 0.2% NH 3 contained in the coal syngas can generate 17 ppm NO x with a reformer, and 37 ppm NO x without a reformer. 6. More than 50% NH 3 left from SOFC stack, either natural gas case or coal syngas case, can convert into NO x mainly (~90%) from the following formation route: