A Review on Combustion Characteristics of Ammonia as a Carbon-Free Fuel

A comprehensive review of combustion characteristics of ammonia (NH3) as a carbon free fuel is presented. NH3 is an attractive alternative fuel candidate to reduce the consumption of fossil fuel and the emission of CO2, soot, and hydrocarbon pollutants, due to its comparable combustion properties, productivities from renewable sources, and storage and transportation by current commercial infrastructure. However, the combustion properties of NH3 are quite different from conventional hydrocarbon fuels, which highlight the specific difficulties during the application of NH3. Therefore, this paper presents comparative experimental and numerical studies of the application of NH3 as a fuel during combustion process, including the combustion properties of laminar burning velocity, flame structures, pollutant emissions for the application of NH3 as a carbon free fuel. This paper presents the burning velocity and pollutant emissions of NH3 alone and mixtures with other fuels to improve the combustion properties. The aim of this paper is to review and describe the suitability of NH3 as a fuel, including the combustion and emission characteristics of NH3 during its combustion process.


INTRODUCTION
With the rapid economic development and population explosion of recent decades, the global primary energy consumption has increased dramatically. Coal, petroleum, and natural gas remain the major sources of energy across the world. Pollutants such as CO 2 , CO, SO 2 , NO x , volatile organic compounds, particulate matter et al., are released by the combustion of petroleum, coal, and natural gas fuels (IEA, 2021). CO 2 is the major contributor to global warming. In the recent years, the development of new technologies aiming to reduce problems related to energy consumption, for application in both developed and developing countries, has become urgent, because of the increasing strictness of governmental regulation of energy and CO 2 emissions. Such developments lead to the search for carbon-free fuels, and the increasing importance of renewable energy sources (Astbury, 2008).
Combustion process during various industry processes contributed huge amount of CO 2 emissions. In order to obtain the target of zero-carbon emissions, it is necessary to reduce CO 2 emissions with the application of carbon-free fuel in combustion system. NH 3 is an attractive hydrogen carrier with a high hydrogen density of 17.8% per unit weight, which can be considered as a carbon-free fuel with the advantage of 1) CO 2 , SO x , and soot emission free; 2) production from various sources, such as fossil fuels, renewable sources, and biomass; 3) transport and storage with existing facility, such as fuel tanks, ships, trucks, and pipelines, which makes NH 3 as a favorable alternative fuel candidate (Wang et al., 2007;Zamfirescu and Dincer, 2008;Zamfirescu and Dincer, 2009;Andersson and Lundgren, 2014;Habgood et al., 2015;Afif et al., 2016;Chai et al., 2021). As shown in Figure 1, NH 3 is synthesized from energy sources such as wind power, nuclear energy, and fossil fuels with CO 2 capture and storage (CCS), etc., which means that NH 3 is fully CO 2 free recyclable as a promising carbon-free fuel.
The potential of NH 3 as a carbon-free fuel owing to its suitable storage and transportation properties has been demonstrated, and applications for its combustion in vehicular engines and turbines are apparent. However, major challenges for NH 3 combustion are to provide its alternative fuels capabilities with outstanding performance, durability and reliability, optimize combustion properties, and lower pollutant emissions, which needs a strict determination of combustion properties during the application of NH 3 as a fuel. The lack of understanding of the NH 3 combustion characteristics, methods of combustion enhancement, and optimization of NO x formation in combustion have placed limits on the utilization of NH 3 as a fuel. It is necessary to have a comprehensive understanding of the combustion and emission characteristics of pure NH 3 and mixture with other fuels, which are expected to supply power for transport vehicles, electricity for power generation, and heating for industry with suitable combustion properties in future energy system.

PROPERTIES OF NH 3
As a carbon-free chemical compound, NH 3 is a colorless gas with a characteristic pungent smell. The chemical and physical properties of NH 3 are listed in Table 1. In 1909, German chemists Fritz Haber and Carl Bosch developed the NH 3 production process, now known as the Haber-Bosch process, which catalytically converts N 2 (from air) and H 2 (from industrial processes) to NH 3 using a metal catalyst under high-pressure (150-250 atm) and high-temperature (400-600°C) conditions. The production of NH 3 has dramatically increased in recent years. In 2020, the global NH 3 production was around 144 million metric ton (shown in Figure 2). The highest individual producers are China is (48.0 million ton: 33.3%), India (12.0 million ton: 8.3%), Russia (10.3 million ton: 7.2%), and the United States (9.2 million ton: 6.4%) (USGS, 2021). The majority of NH 3 is used in industrial and agricultural applications such as fertilizer, plastics, nitric acid, and explosives production, and refrigeration. NH 3 is a poisonous gas, and its safety hazard is a significant obstacle in its use. According to the US Department of Health and Human Services, the permissible exposure limit (PEL) for NH 3 is 35 ppm (27 mg m −3 ) as a short-term (15 min) exposure limit (CDC, 1994).  Some characteristics of ammonia are compared with conventional liquefied fuels such as diesel, gasoline, and natural gas, which are listed in Table 2 (Dimitriou and Javaid, 2020). The liquefied ammonia can be easily stored at 298 K and 10.2 atm. But the stored pressure of compressed liquid natural gas is 245.0 atm at 298 K, which is much higher than that of ammonia. Although the fuel density and energy density of ammonia are lower than those of gasoline and diesel, the fuel density of ammonia is three times as many as that of natural gas and the energy density is also much higher than that of natural gas. The octane rating of ammonia is 110 RON, which is higher than those of natural gas and gasoline. It indicates that the engine fueled by ammonia can be operated at higher pressure. What's more, ammonia exhibits much higher absolute minimum ignition energy than gasoline, which can be improved through some methods like co-firing with hydrogen. Ammonia has potential to be used as a fuel and the performance can be comparable to that of conventional fuel. For example, when directly using ammonia in an internal combustion engine, the driving range can reach 592 km, which is comparable to that of LPG and slightly lower than that of gasoline (Zamfirescu and Dincer, 2008). NH 3 has been widely used in the industry. It has been used in the manufacture of fertilizers, in organic and inorganic synthesis of chemical compounds as HNO, urea, plastics fibers and so on, in mining and metallurgy, in petroleum refining as neutralizing agent, and also NH 3 can be used as commercial refrigerant in food processing, as a substitute for calcium in bisulfite pulping of wood, and to purify municipal and industrial water supplies. Effective methods such as process enclosure, local exhaust ventilation, general dilution ventilation, and personal protective equipment, should be followed in controlling worker exposures or lead of this substance when used it as a fuel.
The use of NH 3 as a fuel by adding it to methane (Williams et al., 1991;Bell et al., 2002;Sullivan et al., 2002;Henshaw et al., 2005;Barbas et al., 2015), diesel (Reiter and Kong, 2011;Gill et al., 2012), DME (Gross and Kong, 2013), and H 2 (Skreiberg et al., 2004;Mendiara and Glarborg, 2009;Mørch et al., 2011;Joo and Kwon, 2012;Um et al., 2013;Choi et al., 2015) has been investigated. NH 3 has often been used as an agent for decreasing NO x emissions in industrial treatment processes through selective catalytic reactions (Xu et al., 2002;Grossale et al., 2009;Yun and Kim, 2013). For CH 4 combustion, both the CO and NO emissions decreased with increasing excess oxygen coefficient. A higher CO emission and lower NO emission would result from a decreased oxygen concentration in the oxidizer under the O 2 /CO 2 combustion atmosphere. Conversely, a higher NO x emission and a significantly lower CO emission are formed in oxy-fuel firing compared with those formed with air firing. CO and NO x formation and decomposition would be completely different when NH 3 is added to carbon-based fuels. In the absence of NH 3 seeding combustion, the most important NO formation process for CH 4 combustion is prompt-NO formation, followed by the NNH, thermal-NO, and N 2 O formation processes. When the combustion mixture is seeded with NH 3 at concentrations of several to several thousand ppm, fuel-NO formation becomes the dominant process; with higher NH 3 concentrations, a greater portion is converted to N 2 rather than to NO (Williams et al., 1991;Barbas et al., 2015). In the ignition engine, the addition of NH 3 leads to a longer ignition delay and limits the engine load because of the high autoignition temperature and low combustion rate (or low burning velocity). The combustion pressure and temperature would also be reduced with the addition of NH 3 , leading to higher CO and HC emissions. NO x formation would be enhanced because of NO x formation from fuel-nitrogen when NH 3 is added. However, soot formation would be extremely low (Gross and Kong, 2013). Furthermore, the substitution of NH 3 could be favored, not only because of the replacement of the carbonbased fuel, but also because it is an efficient method for reducing overall CO 2 emissions (Gill et al., 2012;Gross and Kong, 2013).  Laminar burning velocity is a significant important parameter during fuel combustion. The laminar burning velocity is the speed at which unburned gases move through the combustion wave in the direction normal to the wave surface, which can be calculated by the Metghalchi and Kech power-law relation as follows (Metghalchi and Keck, 1982): where S u is the laminar burning velocity (m/s), T u is the unburned temperature (K), T 0 298 K, p is pressure (atm), p 0 1 atm, S u0 , α, and β are constants. Many researches about laminar burning velocity of ammonia/ air flames have been reported, most of which were studied under ambient condition (Zakaznov et al., 1978;Ronney, 1988;Pfahl et al., 2000;Takizawa et al., 2008;Hayakawa et al., 2015a;Han et al., 2019, Mei et al., 2019. Zakaznov et al. measured the burning velocity by cylindrical-tube method at room temperature and atmospheric pressure (Zakaznov et al., 1978). Ronney et al. measured the flame speed by a constant-volume cylindrical combustion vessel at microgravity condition (Ronney, 1988). Pfahl et al. measured the behavior of ammonia/air flames by using the schlieren system in a constant volume combustion vessel (Pfahl et al., 2000). In order to minimize the uncertainty caused by the buoyancy effect, they also measured the horizontal component of the flame motion. Takizawa et al. measured the laminar burning velocity using the spherical-vessel method in a constant volume combustion vessel (Takizawa et al., 2008). Hayakawa et al. measured the flame speed using high speed schlieren photography in a constant volume combustion chamber (Hayakawa et al., 2015a). Recently, Han et al. measured the flame propagation of ammonia/air by using heat flux method (Han et al., 2019). And the laminar burning velocities of ammonia/air mixtures was also investigated using a high-pressure constant-volume cylindrical combustion vessel by Mei et al. (2019). Figure 3 shows the relationship between laminar burning velocities of NH 3 /air flames and equivalence ratio at 298 K and 1 atm. The range of the flame velocity is roughly from 1.4 to 8.23 cm/s at φ 0.7-1.3. It can be observed that the results measured by different groups reach their peaks around φ 1.1, while the maximum values vary from 6.3 to 8.2 cm/s. The relative differences are little at lean conditions, however, a discrepancy more than 2 cm/s appears under fuel-rich condition between the experimental results. The burning velocity of NH 3 /air flame is relatively low when compared with that of CH 4 /air flame whose maximum value is about 35 cm/s (Law, 2006). Ronney et al. measured the flame speed at p 0.066, 0.132, 0.329, 1, and 1.974 atm. They found that the change of pressure had a modest influence on the flame velocity (Ronney, 1988). Duynslaegher et al. investigated the ammonia/air flame speed by using Konnov mechanism at T 300K, P 1-40 atm and φ 1.0 (Duynslaegher et al., 2010). Results show an increase of pressure had a negative effect on the flame propagation, especially at the low pressure. With the pressure increased from 1 to 10 atm, the flame speed rapidly decreased from 12.67 cm/s to 8.15 cm/s. But with the pressure increased from 30 to 40 atm, the burning velocity decreased smoothly from 6.13 to 5.65 cm/s. Hayakawa et al. studied the laminar burning velocity of ammonia/air flames at elevated pressures by five detailed mechanism, which was validated by the measured experimental results (shown in Figure 4, Hayakawa et al., 2015a). Figure 4 shows the laminar burning velocities under different initial pressure P i and equivalence ratio φ, plotted along with simulation results by using Tian, Miller, Konnov, GRI 3.0, and Lindsted mechanism (Lindstedt et al., 1994;Miller et al., 1983;Frenklach et al., 2000;Konnov, 2009;Tian et al., 2009). The flame velocity decreases with the increase of initial pressure under the equivalence ratios of 0.9, 1.0, and 1.2. The tendency of the experimental and simulation results is the same as those of hydrocarbon. But it is different from the results of Ronney et al., which are indicated that pressure had a little influence on the flame speed (Ronney, 1988). The initial pressure has a larger effect on flame propagation speed at φ 1. For example, as pressure increased from 1 to 5 atm, the laminar burning velocity of measured stoichiometric NH 3 /air flame deceased from 6.9 to 4 cm/s (shown in Figure 4B), while the differences is about 1 cm/ s at φ 0.9 and 1.2 (shown in Figures 4A,C). As shown in Figure 4, most mechanism significantly over-predict the measured burning velocity. It is notable that the simulation results by Konnov mechanism is about two times larger than the measured values while the prediction results by GRI 3.0 mechanism closely agree with the experimental data. However, GRI 3.0 mechanism lacks some ammonia oxidation reactions which are very important for the numerical simulation of NO concentration in ammonia/air flame. Therefore, further improvements of those mechanisms are very necessary. Liu et al. experimentally and numerically studied the variations of laminar burning velocity with initial pressure for ammonia/air flame (Liu et al., 2019a). The flame speed has been measured at initial pressure from 0.3 to 1.6 atm, covering equivalence ratios from 0.5 to 1.3 and at 298 K. The measured laminar burning velocities of NH 3 flame firstly increased slightly and then decreased with the increasing pressure. The burning velocity of stoichiometric flame reaches the peak value at 0.5 atm. The maximum values exist at 0.7 atm when the equivalence ratio is 0.5 or 1.3. Generally, the laminar burning velocity decreases with an increase in pressure, which is agreeable with the observation obtained by Hayakawa et al. (2015a). The numerical simulation was calculated by using the GRI 3.0 mechanism under initial pressure ranged from 0.3 to 1.6 atm, and equivalence ratio of 1.0 and temperature of 298 K. The difference between the measurements and the calculations is around 12 cm/s. The influence of temperature is rarely investigated. Cohen measured the laminar burning velocity of ammonia/air flame at temperature ranged from 323 to 423 K by using a flat flame burner (Cohen, 1955). Recently, Lhuillier et al. measured the flame speed at 1 atm and temperature from 298 to 473 K by the spherical flame method (Lhuillier et al., 2020a). Han et al. measured the laminar burning velocity of NH 3 flame using the heat flux method at temperature from 298 to 448 K and 1 atm (shown in Figure 5, Han et al., 2020). Li et al. numerically investigated the effect of preheating temperature on the combustion and NO emission characteristics of ammonia flame (Li et al., 2016a;Li et al., 2016b). In all temperature ranges, the laminar burning velocities increase firstly before the equivalence ratio reaches 1.1, and then decrease when the equivalence ratio is large than 1.1. The maximum value of flame speed at different temperatures ranges from 6.98 to 14.87 cm/s. With the temperature increasing, the flame speed exponentially increases. The results measured by Han et al. at 423 K were compared with those from Cohen and Lhuillier (Cohen, 1955;FIGURE 4 | Laminar burning velocity of NH 3 flame with a function of initial mixture pressure at φ 0.9, 1 and 1.2, replotted from (Hayakawa et al., 2015a).
FIGURE 5 | Laminar burning velocity of NH 3 flame with a function of equivalence ratio at various temperature, replotted from (Han et al., 2020). Han et al., 2020;Lhuillier et al., 2020b). It was found that the experimental values obtained by Han et al. are lower than those from Cohen and Lhuillier. The difference between the measurements is around 2 cm/s, which is larger at fuel-lean conditions. The simulation results by using five mechanisms were also performed. Comparison of numerical and experimental data showed that the simulations results by using Okafor mechanism (Okafor et al., 2018) and Otomo mechanism (Otomo et al., 2018) performed best.
Oxygen enriched combustion can be considered as a promising approach to increase burning velocity. The experimental results measured by Takeishi are shown in Figure 6 (Takeishi et al., 2015). It can be observed that the velocities can be close to those of CH 4 /air flames when the O 2 concentration up to 0.35 and the maximum velocity improves to 36.1 cm/s, which is around 5 times the value measured under the air condition. Li et al. (2015) numerically investigated the effect of the oxygen content towards the burning velocity of NH 3 /air flame by using the Millar mechanism (Millar and Bowman, 1989) and the Reductive Konnov mechanisms (Duynslaegher et al., 2009;Duynslaegher et al., 2012). The maximum laminar burning velocity increases from 27.5 to 33.9 cm/s when the O 2 content increases from 0.27 to 0.30, and the values of laminar burning velocity at the O 2 content of 0.30 are always 6-7 cm/s higher than those at the O 2 content of 0.27. It is mainly due to an increase of the reaction rates of OH, H, O, and NH 2 radicals under a higher O 2 concentration. It is found that the laminar burning velocities calculated by Li et al. are over-predicted when compared with the results of Takeishi's experiment at the O 2 content of 0.30, especially at fuel rich conditions. For example, the difference between experimental and numerical results reaches 7.9 cm/s at the equivalence ratio of 1.1, while the discrepancy is around 1.5 cm/s at the equivalence ratio of 0.8. Recently, it was found that the burning velocities of ammonia are proper for the practical applications when the range of O 2 volume concentration in O 2 / N 2 mixture is roughly from 0.35 to 0.40 . Mei et al. measured the ammonia/air flame speed under the O 2 content ranged from 0.21 to 0.45 at 298 K, the pressure ranged from 1 to 5 atm and the equivalence ratios from 0.7 to 1.5 by using more complicated constant pressure spherical flame method (Mei et al., 2019). The flame speed increases with an increase of the O 2 concentration, but decrease with an increase of initial pressure. The experimental results showed that the flame propagation speeds are lower than those of CH 4 /air flames at the O 2 concentration of 0.35. It is different from the results of Takeishi et al., which indicated that the laminar burning velocities at the O 2 concentration of 0.35 are almost the same as those of CH 4 /air flames (Takeishi et al., 2015). In addition, they constructed a model of ammonia combustion with 38 species and 265 reactions, which can predict the results with a favorable satisfaction. Wang et al. measured the oxy-ammonia flame speed in a constant volume vessel at the temperatures ranged from 303 to 393 K, the O 2 content ranged from 0.6 to 1.0 and the equivalence ratios ranged from 0.6 to 1.4 . Results show that the peak value of laminar burning velocity can reach 125.1 cm/s at the O 2 content of 1.0. The equivalence ratio corresponding maximum value is around 0.9. Besides, the flame velocity increases with the increasing temperature, which is more sensitive to the change of temperature under lower O 2 content.
Due to low burning velocity, ammonia usually is burned with the addition possessing higher reactivity such as hydrogen. Ichikawa et al. measured the ammonia/hydrogen/air flame speed at pressure of 1, 3, and 5 atm by using a high-pressure constant volume chamber (shown in Figure 7, Ichikawa et al., 2015). The experimental results show an exponential variation of laminar burning velocity under different volumetric hydrogen fraction and the ammonia/hydrogen/air flame speed decreased with an increase in pressure under the same hydrogen concentration. It is obvious that the laminar burning velocity largely decreased when the pressure increased from 1 to 3 atm, while the values of burning velocities at 3 atm were almost the same as those of flame speeds at 5 atm. This means that the influence of initial pressure conspicuously appeared under low pressure, but that effects barely occurred under high pressure. Lee FIGURE 6 | Laminar burning velocity of NH 3 flame with a function of equivalence ratio at various oxygen enrichment level, replotted from (Hayakawa et al., 2015a, Takeishi et al., 2015Li et al., 2015;Okafor et al., 2018;Kobayashi et al., 2019).
October 2021 | Volume 9 | Article 760356 et al., Kumar et al., and Li et al. also investigated the flame propagation speed of ammonia/hydrogen/air mixture at 1 atm (Lee et al., 2010;Kumar and Meyer, 2013;Li et al., 2014;Li et al., 2017;Li et al., 2019a). The differences among experimental results were small and the tendencies were consistent with that measured by Ichikawa et al. (2015). It also can be observed that the value of flame speed was around 35.1 cm/s at the hydrogen concentration of 0.45 which was close to that of CH 4 /air mixture, indicating that the addition of hydrogen can effectively improve ammonia combustion. Han et al. measured the laminar burning velocities of NH 3 /H 2 /air by using the heat flux method (covering φ 0.7-1.6, χ H 2 0.15-0.4, P 1atm, K 298 K) (Han et al., 2019). Results show that the values of measurements reached their peaks at φ 1.05 when the hydrogen ratio ranged from 0 to 0.35. However, the peak velocity occurred at the equivalence ratio of 1.1 when the hydrogen ratio increased up to 0.4. This tendency has good agreement with the finding that the peak value of the hydrogen/air flame speed occurs at the equivalence ratio of 1.8 (Kwon and Faeth, 2001). They also found that mixing ammonia with the addition of H 2 is the most effective way to improve the flame propagation by comparing with the burning velocity of NH 3 /CH 4 mixture, NH 3 /H 2 mixture, and NH 3 /CO mixture. The flame speeds of NH 3 /H 2 flames (the hydrogen concentration ranged from 0.0 to 0.3) and pure ammonia at oxygen rich conditions (the oxygen concentration ranged from 0.21 to 0.30) were measured by Shrestha et al. at P 1-10 atm and T 298-473 K (Shrestha et al., 2021). They also developed a mechanism for the oxidation of ammonia/hydrogen mixtures and ammonia. Based on this model, they found that the N 2 H 2 formation path was favored in fuel rich condition, which is very important for numerical simulation. Different from adding hydrogen, hydrogen production can also be from ammonia decomposition, since ammonia has high hydrogen density (Valera-Medina et al., 2018;Kobayashi et al., 2019). It was found that ammonia started to crack when the temperature reached 473.15 K and the conversion efficiency can reach 98-99% when the temperature up to 698.15 K (Klerke et al., 2008;Lan et al., 2012). What's more, in contrast to co-firing hydrogen, using partially cracked ammonia is more economical and easier to implement (Yang et al., 2019). However, the deep insight into the combustion characteristics of NH 3 /H 2 /N 2 mixtures are limited (Mei et al., 2021;Wiseman et al., 2021). Very recently, Mei et al. experimentally investigated the laminar burning velocities of partially cracked NH 3 /air mixtures in a high-pressure constant volume cylindrical combustion vessel at the initial pressure of 1-10 atm, the equivalence ratios from 0.7 to 1.4 and the initial temperature of 298 K (Mei et al., 2021). Results showed that the laminar flame speed of NH 3 /H 2 /N 2 /air mixtures increased with an increase of the cracking ratio due to the increase of H 2 concentration. It was found that the flame speed could reach 38.1 cm/s at 1 atm and the cracking ratio of 40%, which is close to that of methane/air mixtures. They also found that the increase of N 2 concentration and the initial pressure had a negative effect on the flame speed. Then though the numerical simulation, they recognized that H + O 2 (+M) HO 2 (+M) plays an important in flame propagation.
Blending ammonia with methane is also an effective method to improve the combustion characteristics of NH 3 /air flames propagation. Konnov et al. experimentally investigated the laminar burning velocity of CH 4 /O 2 /N 2 flames doped with 0.5% NH 3 by using a heat flux method. Results show that ammonia had no significant influence on the flame speed (Konnov et al., 2006). Okafor et al. and Henshaw et al. measured the flame speeds of NH 3 /CH 4 /air flames as a function of the equivalence ratio at 298 K and 1 atm (Henshaw et al., 2005;Okafor et al., 2018). Zhou et al. experimentally and numerically investigated the laminar flames of NH 3 /syngas/air and NH 3 /bio-syngas/air at elevated temperature . Wang et al. experimentally and numerically investigated the laminar burning velocities of NH 3 mixing with CH 3 OH and C 2 H 5 OH in premixed flames . Figure 8 shows that the flame speeds decreased with the heat fraction of ammonia E NH3 increased and the flame speed were close to those of methane/air mixture when E NH3 reached 0.016. Obviously, the burning velocity decreased more sharply when E NH3 ranged from 0 to 0.1. The difference between the flame speed at E NH3 0.1 and E NH3 0.3 is about 8.8 cm/s, while the discrepancy is around 9.8 cm/s when the E NH3 increases from 0 to 0.1. Xiao et al. investigated the laminar burning velocity of NH 3 /CH 4 mixtures with the ammonia fraction ranged from 0 to 100% (Xiao et al., 2017b). They also found that ammonia had a negative impact on flame propagation. Shu et al. studied the fundamental combustion characteristics of NH 3 /CH 4 /air mixtures by using expending spherical flames (Shu et al., 2021). They found that the laminar burning velocity varied almost linearly with the methane volume fraction and the H/OH radicals were very important for the propagation of flames under fuel lean conditions. Furthermore, the laminar burning velocity had a strong linear relationship with the maximum mole fraction of (H + OH) radicals under lean conditions when the pressure ranged from 1 to 15 atm and the methane volume fraction ranged from 0.30 to 0.70. The laminar burning velocity of ammonia/ methane mixture under variable O 2 and CO 2 mole fractions of

NO X EMISSIONS
When ammonia is completely combusted, it only produces nitrogen and water without involving the production of NO x . However, NO x emission is relatively high in practical combustion, which is a main challenge of ammonia combustion. In order to reduce the production of NO x , some detailed mechanism of ammonia combustion is necessary. NO x is mainly composed of thermal NO x and fuel NO x . Thermal NO is usually produced by the oxidation of N 2 at temperature up to 1800 K. The extended Zeldovich mechanism is widely used to describe the formation of thermal NO. The reactions of thermal NO have three important pathways: N 2 + O NO + N, N + O 2 NO + O, and N + OH NO + H. The first reaction limits the reaction rate and usually take place when the temperature above 1,800 K. Therefore, controlling temperature is an effective way to reduce thermal NO production.
Fuel NO x emission is widely studied by many researchers. Miller et al. studied the reaction path of NH 3 oxidation and proposed a kinetics including 22 species and 98 elementary reactions (shown as Figure 9, Millar and Bowman, 1989). Firstly, ammonia is converted to NH 2 by the reaction with OH, which is the primary path under both fuel rich and lean conditions. It is also consumed by O to produce NH 2 under fuel lean condition and is consumed by H under rich condition. The reason of this phenomenon is that rich burning flame needs more oxygen or oxygen atoms, which results in the proportion of H in O/H radicals is high when compared to fuel lean conditions (Chai et al., 2021). It is obvious that the reactions involving NH i (i 0, 1, 2) have very significant impact on the formation and reduction of NO. And HNO intermediate make great contribution to NO formation via reacting with OH, NH 2 , and M. Lindstedt et al. also found that fuel NO is mainly produced by the reactions involving HNO (Lindstedt et al., 1994). For ammonia flames, HNO + M NO + H + M reaction plays a key role and HNO + OH NO + H 2 O reaction becomes important when the equivalence ratio less than 1. What's more, for the conversion of NO, the reaction between NO and NH 2 is the major path.
Dean et al. found that NNH dissociation plays an important role in the formation of N 2 under fuel rich conditions, which mainly produced by N 2 H 2 (Dean et al., 1984). The NH i radicals can be converted to N 2 without including NO, which has a potential to reduce the formation of NO. They developed the NNH mechanism (Dean and Bozzelli, 2000 (Miller and Glarborg, 1999;Klippenstein et al., 2011). They found that the rate constant of NNH + O N 2 + OH is over-predicted by Dean et al.

and the NNH + O reaction is very fast.
Some factors like equivalence ratio and pressure are also very important for NO x emissions. The main component of NO x is NO. It was found that with an increase of the equivalence ratio, the mole fraction of NO in NH 3 /air flame firstly increased before the equivalence ratio reaches 0.9 and then rapidly decreased when the equivalence ratio is large than 0.9 (Kobayashi et al., 2019). The production of NO is extremely low after the equivalence ratio up to 1.3. Therefore, it seems that burning under fuel rich condition is an effective way to reduce NO x emission. Hayakawa et al. experimentally and numerically studied the characteristics of  October 2021 | Volume 9 | Article 760356 NO x formation in ammonia/air flames at different pressures and equivalence ratios (covering p 1.0-3.0 atm and φ 0.7-1.1) (Hayakawa et al., 2015b). The experiments were performed in a nozzle-type burner and the numerical simulation used the Tian mechanism (Tian et al., 2009). Results show that the NO concentration decreased with an increasing equivalence ratio at pressure of 1 atm, which is mainly due to surplus production of NH i (i 0, 1, 2) under fuel rich conditions. Besides, with the increasing pressure, the concentration of NO decreased while the reaction rate of OH + H + M H 2 O + M increased. Therefore, they proposed that H and OH radicals mentioned above limited the formation of NO.
In order to use pure ammonia in practical applicants, some researches were performed to provide a deep insight into the characteristic of NO x emission. Lee et al. proposed a combustion strategy in order to use pure ammonia in the internal combustion engine (Lee and Song, 2018). In this strategy, they injected the lean ammonia/air mixture during the intake process which was auto-ignited in the next step. This method can raise the temperature and pressure of the cylinder and it has a positive effect on ammonia spray combustion. Then they studied the mechanism of NO formation and found that NO emission in this engine largely depend on the main SOI variation. For instance, the maximum value of NO formation up to 8,500 ppm, while the reduction of NO emission can reach 5,460 ppm by changing the SOI timing. Kurata studied the emission characteristics of nonpremixed NH 3 /air in gas turbine power generations (Kurata et al., 2017). It was found that the production of NO x and slip NH 3 were strongly influenced by the combustor inlet temperature. The existence of NO x and slip NH 3 at the mid-range of combustor inlet temperature implied that there is a possibility of NO reduction through selective non-catalytic reduction. They also developed a low-NO x NH 3 -air non-premixed combustor for actual gas-turbine operations and found that the NO production of this gas turbine power generation could be lowered to 337 ppm when O 2 content reached 16% . Okafor et al. designed a two-stage combustor for a micro gas turbine and found that NO x emission was around 42 ppm at initial pressure of 3 atm and initial temperature of 298K when the equivalence ratio of the primary stage reached 1.1 and the equivalence ratio of the secondary stage was 0.4 . Somarathne et al. numerically investigated the NO x and NH 3 emission of turbulent premixed ammonia/air flames in a gas turbine like combustor by using the large eddy simulation at elevated pressure up to 5 atm and various equivalence ratios (Somarathne et al., 2017). They found that NO production decreased when the initial pressure increased and the minimum value of NO and unburnt NH 3 emission could reach 200 ppm at initial pressure of 5 atm and the equivalence ratio of 1.2. What's more, the NO emission of ammonia/air flames with the secondary air injection could be lowered to 100 ppm at O 2 content of 16%. Rocha et al. numerically studied the NO x emission characteristics of ammonia/air mixtures under the typical conditions of commercial gas turbines by using three modern stationary gas turbine concepts (Rocha et al., 2020). They found that rich-burn, quick-quench and lean-burn (RQL) concept and moderate or intense low oxygen dilution (MILD) concept are able to reach low NO x emission while the NO x emission by using lean-burn drylow emissions (DLE) concept is high.
There are also some researches about the NO x emission of NH 3 /H 2 /air flames. Li et al. experimentally measured the NO x concentration of NH 3 /H 2 /air flames covering the NH 3 concentration from 0.440 to 0.544 at the equivalence ratios from 1.00 to 1.25 (Li et al., 2014). They found that NO x emission decreased when the equivalence ratio increased from 1.00 to 1.25 under different NH 3 contents. It can be observed that a more significant change occurs when the equivalence ratio increased from 1.00 to 1.10, while the change is small when the equivalence ratio increased from 1.10 to 1.25. As the NH 3 concentration increase the production of NO x decrease, which due to the low flame temperature. The peak value of NO x content is 1,450 ppm at NH 3 content of 0.500 and stoichiometric condition, while the maximum content is around 1,660 ppm when NH 3 content of NH 3 /H 2 mixture reaches 0.416 under stoichiometric condition. They also concluded that fuel NO x plays a significant role in NO x emission, whereas thermal NO x is negligible. Nozari et al. also studied the variation of the NO x emission with the equivalence ratio at elevated pressure of 17 atm and temperature of 673 K (Nozari, 2015). They found that the NO x concentration increased first and then decreased when the equivalence ratio ranged from 0.5 to 1.2 in all hydrogen content ranges. The maximum mole fraction of NO x occurred under fuel lean conditions, which is similar to that of ammonia/air flames. Wang et al. also found that the NO concentration in exhaust gas was non-monotonically changed with equivalence ratios and hydrogen mole ratio, because the competition between the effects by the reduction of N-atom and the enrichment of H/O radicals when hydrogen is added to ammonia flame . Lee et al. also found that both NO x and N 2 O emissions are low in fuel rich condition when compared to those under lean conditions (Lee et al., 2010).
The influences of pressure and hydrogen content on NO x emissions are also significant. Xiao et al. numerically investigated the influence of pressure in NH 3 /H 2 flames by using an improved mechanism based on the work performed by Mathieu et al. (Mathieu and Petersen, 2015;Xiao et al., 2017a). Results show that the concentration of NO x significantly decreased with an increasing pressure. NO x emission can be less than 5 ppm when the pressure reaches 10 atm and less than 1 ppm when the pressure up to 20 atm. Rocha et al. numerically investigated the variation of NO x emission with hydrogen addition ratio ranged from 0 to 1 by ten mechanisms for ammonia/ hydrogen/air flames at 1 atm and 298 K (Rocha et al., 2019). Results show that with an increase of the H 2 content, the NO x emission increased before the mole fraction of H 2 reached 0.80 and then rapidly decreased when the mole fraction of H 2 large than 0.80. They presumed that the presence of OH/O radicals at high temperature resulted in the increasing content of NO x before the H 2 concentration reached 0.80, while the decreasing concentration of ammonia leads to the decrease of NO x when the mole fraction of H 2 large than 0.80.
Lhuillier et al. experimentally investigated the NO x emission of NH 3 /H 2 /air mixtures in spark ignition engine at the intake Frontiers in Energy Research | www.frontiersin.org October 2021 | Volume 9 | Article 760356 temperature of 323 K and pressure of 1.2 atm (Lhuillier et al., 2020a). It was found that NO x production firstly increased and then decreased with an increased of the equivalence ratio. The peak values of NO x formation occurred at the equivalence ratio of 0.8-0.9 and the minimal NO x emission was found under rich conditions. What's more, as the H 2 concentration of NH 3 /H 2 /air mixtures increased, the emission increased. Very recently, Franco et al. also designed a laboratory combustor which had good performance in NO x emission (Franco et al., 2021). Valera-Medina et al. numerically and experimentally investigated the NO x emission of premixed 50%NH 3 /50%H 2 mixtures under fuel lean condition in a swirl combustor and found that the pollution emission was high when compared with that under fuel rich condition (Valera-Medina et al., 2017). Then they studied characteristics of 70%NH 3 /30%H 2 mixtures under gas turbine conditions through a numerical GT cycle model and found that NO x emission could as low as 50 ppm because the hot unburned ammonia was able to react with NO x . Hussein et al. found that injecting some NH 3 /H 2 blend downstream the primary zone could significantly reduce the NO x emission which was a promising method to minimize the pollution emission (Hussein et al., 2019).
Recently, many researches have studied the NO x emission of CH 4 /NH 3 flames. Among them, Tian mechanism is widely used to predict the NO x emission of CH 4 /NH 3 mixtures, which has a satisfactory prediction of the experimental results (Tian et al., 2009). Tian et al. investigated the intermediates and products for ammonia/methane/air/Ar flames at 0.04 atm and φ 1 by using tunable synchrotron vacuum ultraviolet photoionization and molecular-beam mass spectrometry (Tian et al., 2009). Results show that the formation of NO 2 decreased while those of NO and N 2 increased with the mole ratios of NH 3 /CH 4 increasing from 0 to 1. They also developed a detailed mechanism including 84 species and 703 elementary reactions, which can predict the major combustion species and intermediates well (shown as Figure 10). The process of CO formation is described below. Firstly, methane is converted to CH 3 by the reaction with OH and H. Then CH 3 mainly reacts with O addition to generate CH 2 O and reacts with OH to form singlet CH 2 which can transform to triplet CH 2 . CH 2 is converted to CH by reacting with H and then CH transforms to HCO by reacting with O 2 . Besides, HCO also can be produced by the reaction of CH 2 O with OH. Finally, HCO is converted to CO and H 2 by reacting with OH, O 2 , and H. However, NO is produced by the reaction of NH with O and the reaction of HNO with H, OH, and O. It seems that the addition of CH 4 has no significant influence on the major formation and reduction of NO. Though the numerical simulation, it also can be found that four reactions including H + O 2 O + OH, NH 2 + O HNO + H, NH 2 + NO N 2 + H 2 O, and NH + NO N 2 O + H are important for the conversion of NO and N 2 . Ramos et al. measured the NO x emissions from CH 4 /NH 3 flames at NH 3 content ranged from 0 to 0.7 and the equivalence ratio of 0.8, 0.9, and 1 by using a laboratory scale laminar flame burner (Filipe Ramos et al., 2019). They found that the production of NO x firstly increased and then slightly decreased with an increasing NH 3 content and the maximum NO x emission occurred when the concentration of NH 3 reached 0.50. Xiao et al. numerically studied the emission characteristics of CH 4 /NH 3 mixtures at the ammonia ratio from 0 to 1 by using an improved model based on the Konnov mechanism and found the same tendency (Konnov, 2009;Xiao et al., 2017b). It was found that the addition of ammonia had a significant positive effect on NO x emission under high methane proportion mixtures while it had a negative effect under high ammonia concentration. The probable reason of this phenomenon is that H + NO HNO inhibits the formation of NO at high ammonia concentration . The equivalence ratio and initial pressure also has a strong impact on NO x emission. Ramos et al. found that the peak values of the NO x concentration are around 4,300, 3,600, and 2,800 ppm at the equivalence ratio of 1, 0.9, and 0.8, respectively, (Filipe Ramos et al., 2019). It is obvious that NO x emissions decrease with a decrease of the equivalence ratio. Xiao et al. also studied the influence of the equivalence ratio and found that the maximum NO x emission is presented at the equivalence ratio of 0.9 when the ammonia mole fraction ranged from 0.20 to 0.80 (Xiao et al., 2017b). Then they numerically studied the reaction involving NO by using Tian mechanism (Xiao et al., 2020). It was found that NH 2 +O HNO + H reaction is the most promoting reaction at φ 0.8, while H + O 2 O + OH is the most promoting reaction at φ 1, 1.2. The most inhibiting reaction is NH 2 +NO N 2 +H 2 O, NH + NO N 2 O + H, NH 2 +NO NNH + OH at φ 0.8, 1 and 1.2, respectively. For NO x emission, the initial pressure plays an important role. The NO mole fraction decreased with the increasing pressure and the NO production was more sensitive to the change of the pressure than CO (Xiao et al., 2017b). What's more, it was found that the formation of NO is mainly through the NH + OH HNO + H reaction at high pressure (Valera-Medina et al., 2017). Zhang et al. investigated the emission characteristics of ammonia/air combustion in a model combustor with the addition of methane and hydrogen. They found NO and OH radicals showed a positive correlation, and the temperature had a secondary role on promoting NO x formation comparing with CH 4 /air flame (Zhang et al., 2021a). The thermal performance and NO x emission on a premixed methane/ammonia/flame at a micro-planar combustor were also investigated. Results showed that ammonia addition reduced CO 2 emission, but increased NO emission increased. While N 2 O emission increased first and then decreased with increasing ammonia ratio. (Cai et al., 2020;Cai et al., 2021;Han et al., 2021;Sun et al., 2021). An et al. found that the OH and NO were closely correlated in premixed CH 4 /NH 3 /air flames as they were both strongly related to flame temperature. While N 2 O had a correlation with NH and HNO components , which is similar with the results by the same combustion group. Zhang et al., 2021b;Wei et al., 2021).
Recently, CH 4 /NH 3 blends are also considered as a substitute fuel in practical applications. Liquefied natural gas is an alternative fuel and has been widely used in marine engines (Schinas and Butler, 2016). What's more, it also can improve the poor combustion characteristics of pure ammonia. Therefore, Oh et al. investigated the pollution emission of ammonia mixed with methane and found that NO x emissions of CH 4 /NH 3 blends in spark ignited engine rapidly increased before the split ratio of NH 3 reached 40% and then slightly decreased when the split ratio Frontiers in Energy Research | www.frontiersin.org October 2021 | Volume 9 | Article 760356 large than 40% (Oh et al., 2021). Besides, the maximum NO x emission of this study was lower than 40 g/kWh. Xiao et al. compared the simulation results under gas turbine conditions by using Tian, Konnov, Mendiara, GRI 3.0 and ÅA mechanism and found that Tian mechanism performed best in NO x prediction (Xiao et al., 2017c, Tian et al., 2009, Mendiara and Glarborg, 2009, GRI, 2000, Åbo Akademi University, 2005. What's more, they found that the oxidation of N and HNO played the most important role in the production of NO under gas turbine conditions. Li et al. also studied the NO x emission of CH 4 / NH 3 mixtures under gas turbine conditions and recognized that NO x formation increased when the NH 3 concentration increased, which is mainly due to the enhanced HNO pathway (Li et al., 2019b). Furthermore, they found that the availability of oxygen has significant influence on the production of NO x through HNO pathway and proposed a combustion system in which NO x production can as low as 30 ppm even when the NH 3 concentration up to 40%. Okafor et al. used rich-lean combustion to control NO x emission and proposed the optimum equivalence ratio of the primary zone ranged from 1.30 to 1.35 (Okafor et al., 2020). This method can significantly reduce the formation of NO x , for instance, NO x emission can be lowered to 49ppm. Okafor et al. investigated the characteristics of liquid NH 3 spray co-fired with CH 4 and found that the minimum NO x emission occurred when the equivalence ratio reached 1.06 (Okafor et al., 2021a). They also studied the NO x emission characteristics of this kind of fuel in two-stage micro gas turbine combustors and recognized that the increased inlet temperature and decreased wall heat loss led to the reduction of NO and N 2 O (Okafor et al., 2021b).

CONCLUSIONS AND PERSPECTIVES
This paper presents the burning velocity and pollutant emissions of NH 3 flames, NH 3 /H 2 flames and CH 4 /NH 3 flames. At 1 atm and 298 K, the laminar burning velocities increase before the equivalence ratio reaches 1.1, and then decrease when the equivalence ratio is large than 1.1. The maximum value of laminar burning velocity is less than 10 cm/s, which is relatively low when compared with that of CH 4 /air flame. Under different pressure conditions, it can be found that the flame velocity decreases with an increase of the initial pressure. The influence of temperature is also studied. With the initial temperature increasing, the flame speed exponentially increases. In order to increase the laminar burning velocity of NH 3 /air flames, oxygen enriched combustion is proposed as a promising approach. The flame speed of NH 3 /air flames under oxygen enrichment conditions increases with an increase of the O 2 concentration, but decrease with an increase of initial pressure under oxygen enrichment conditions. And the flame velocity increases with the increasing temperature, which is more sensitive to the change of temperature under lower O 2 content. Besides, ammonia is usually burned with the additions of hydrogen and methane to improve the combustion characteristics. The addition of H 2 or CH 4 can significantly increase the flame speed. NO x is composed of the thermal NO x and the fuel NO x , which is the major pollution of ammonia combustion. Thermal NO x is usually produced by the oxidation of N 2 when the temperature reaches 1,800 K. The generation of the thermal NO is satisfactorily described by the extended Zeldovich mechanism. It can be reduced by controlling temperature. Fuel NO x is mainly generated by the oxidation of NH 3 . HNO intermediate plays an important role in NO formation by reacting with OH and NH 2 . It was found that burning under rich condition and high pressure can effectively reduce NO x emission. The effects of equivalence ratio and pressure in NH 3 /H 2 flames are similar to those of NH 3 / air flames. The NO x production firstly increased with an increase of the H 2 content before the H 2 concentration reached 0.8, and then rapidly decreased after the mole fraction of H 2 large than 0.8. The reaction pathways in the combustion of NH 3 /CH 4 fuels are described by using Tian mechanism. It seems that the addition of CH 4 has no significant influence on the major formation and reduction of NO. The NO x production is high under high ammonia content, high temperature and stoichiometric condition. What's more, the concentration of CO 2 also influences the formation of NO x . The high CO 2 content has a positive impact on NO x production under fuel rich condition, while has a negative effect under stoichiometric and lean conditions.
Although the fundamental characteristics of ammonia-based flames has been widely studied, the practical application of this kind of fuel is still unsatisfying. Therefore, further studies are needed to overcome the problems, including low burning velocities and high NO x emission. For example, one of the main challenges in the investigation of ammonia-based fuel is that the numerical simulations by using mechanisms usually overpredict or underpredict when compared with those are obtained during the actual combustion processing. In order to fully understand the combustion characteristics of ammoniabased flames, developing an accurate mechanism is necessary. What's more, for emission characteristics of CH 4 /NH 3 mixtures, many researches focused on reducing NO x emission, and few studies on the carbon capture can be found. Therefore, it still need further studied.