A review on the effects of ethanol/gasoline fuel blends on NO X emissions in spark-ignition engines

Ethanol can be used as an alternative fuel for spark-ignition (SI) engines to increase the octane number and oxygen content of ethanol/gasoline blends, thereby reducing dependence on fossil fuels and the exhaust emissions of incomplete combustion products. Although it is widely agreed that ethanol can reduce CO and HC exhaust emissions, the literature on ethanol and NO X emissions is far from conclusive; hence there is a need for an in-depth, updated review of ethanol/gasoline blends in SI engines and the relative production of NO X emissions. In light of that, the present work aims to provide a comprehensive literature review on the current state of ethanol combustion in SI engines to shed definitive light on the potential changes in NO X emissions under various operating conditions. The first part of this paper discusses the feasibility of ethanol as an alternative transportation fuel, including world production and ethanol production processes. The physicochemical properties of ethanol and gasoline are then compared to analyze their effects on combustion efficiency and exhaust emissions. Then, the pathways of NO X formation inside the cylinder of SI engines are discussed in depth. Finally, we review and critically discuss the effects of ethanol concentration in blends and different engine parameters on NO X formation.

improved by the high oxygen content of the ethanol molecules (Costagliola et al., 2016).
Besides, as the flame propagation speed and heat of vaporization of ethanol are higher than gasoline, gasoline/ethanol-blended fuels can guarantee faster combustions and higher volumetric efficiencies. A further advantage is a reduction in GHG emissions through using ethanol in gasoline fuel blends. In better words, the lower carbon number of ethanol compared to gasoline could result in lower global dioxide emissions, while it should also be noted that the origin of carbon contained in ethanol is biogenic (Normann et al., 2009). However, the main drawback of ethanol/gasoline blends as a fuel substitute for SI engines concerns the lower vapor pressure of ethanol than gasoline, making a cold-start difficult (Chansauria and Mandloi, 2018).
Ethanol production and its use as an alternative fuel in SI engines have been extensively studied. It has been shown that gasoline/ethanol blends, as for other oxygenated fuels, can significantly reduce CO and HC exhaust emissions in SI engines compared to neat gasoline up to the 20% ethanol ratio limit in blends. However, there are substantial inconsistencies in accounting for NOX emissions, and the fundamental knowledge in this domain is still inadequate. As shown in Table 1, there are many comprehensive review articles on performance and exhaust emissions characteristics of SI engines using alcohols as alternative fuels. However, only a few provided detailed analyses concerning the effect of cold-start, compression ratio, and engine load on combustion characteristics and NOX emissions when using ethanol as a gasoline substitute. Besides, some of the review articles tabulated in Table 1 are outdated and do not focus solely on ethanol fuel.
To the best of the authors' knowledge, most of the research and review articles are focused on engine performance and exhaust emissions of warmed and steady running SI engines, monitoring the air/fuel ratio, but without providing specific analysis of the NOX emissions when the engine is fuelled with ethanol/gasoline blends under real operating conditions. In practice, during the warming-up process, SI engines require a fuel-rich injection to guarantee ignition and to overcome the low mixing of the inlet charge. In addition, the efficiency of sophisticated three-way catalytic converters is limited due to both their low temperature and rich air/fuel ratios. Hence, during the cold-start transient phase of last-generation SI engines, both the technological limits of three-way catalysts and the necessary rich air/fuel mixtures produce large amounts of exhaust emissions. Besides, NOX emissions and combustion characteristics under real operating conditions are rarely analyzed, which is noteworthy as exhaust emissions of SI engines depend on fuel combustion quality to a great extent.
Hence, given there are few review articles concerning the effect of coldstart transient, compression ratio, and engine load on NOX emissions in SI engines powered by ethanol/gasoline blends, the present review article aims to fill this gap. It should be highlighted that NOx emissions are classified by the Environmental Protection Agency (EPA) as one of the most harmful pollutants that can damage the respiratory system (Chong et al., 2010;Gonca, 2017). Therefore, NOX emissions from SI engines fuelled with gasoline/ethanol blends could represent a considerable obstacle to further market expansion of bioethanol.
In this review article, the research on NOX exhaust emissions related to the use of ethanol/gasoline fuel blends in last-generation SI engines under real operating conditions is comprehensively reviewed and critically discussed. The first part of this paper discusses the feasibility of using ethanol as an alternative transportation fuel, global ethanol production, potential sources, and ethanol production processes. In the next section, differences in ethanol and gasoline physicochemical properties are explained and compared to analyze their effects on combustion efficiency and NOX exhaust emissions. The pathways of NOX formation inside the cylinder of SI engines are also presented. Finally, the effects of both ethanol content in the blends and engine parameters on NOX formation are discussed in detail.

World production, sources, and production processes of ethanol
Global ethanol production has increased steadily during the past decade thanks to the efforts made by many countries to decrease their dependence on crude oil reserves, boost agricultural-based economies, and improve air quality (Mahmudul et al., 2017). The total production of ethanol in 2019 reached around 155 billion L, as shown in Figure 1. According to the statistics provided by the Renewable Fuels Association, the USA is the world leader in the production, use, and export of ethanol, representing around 54% of global ethanol production ( Fig. 2) (Ramli and Epplin, 2017). Ethanol is also widely produced and used in several other countries, including Brazil, the European Union, China, and India, where national programs for blending ethanol fuel into gasoline have been implemented (Masum et al., 2013;Shahir et al., 2015;Carneiro et al., 2017). In particular,   Brazil produced 8.62 billion gallons of ethanol in 2019, representing about 30% of world ethanol production. The Brazilian government has made the use of ethanol mandatory since the 1970s in pursuit of two main targets: obligatory use of gasoline-ethanol blends as a transportation fuel and gradual growth of the market of FFVs to the point that gasoline sold in Brazil now contains around 25% anhydrous ethanol and, at the same time, approximately 90% of new vehicles are equipped with flex-fuel engines (Flavin et al., 2006;Laurini, 2017). Based on the various feedstock sources used in industrial operations, bioethanol can be classified into two main categories (Varatharajan and Cheralathan, 2012): first-generation bioethanol and second-generation bioethanol. First-generation bioethanol is the ethanol produced from both sugars (sugar beet, sugar cane, and fruit) and starch (milo, rice, potatoes, maize, and cassava) (Kumar and Saravanan, 2016), with nearly 50 billion L produced annually. However, the first-generation bioethanol feedstock is unsustainable because the growing demand for bioethanol has led to both the shortage and rising prices of sucrose-containing and starchy feedstocks, which have food and feed applications (Yusoff et al., 2015).
Second-generation bioethanol is the ethanol produced from non-food feedstocks such as lignocellulosic biomass (Naik et al., 2010) and agro-residues (Szulczyk and McCarl, 2010). As benefits, such feedstocks can enhance the energy balance of ethanol due to the decreased use of fossil fuel to produce bioethanol. Also, being obtained from cheap, abundant, non-edible wastes and biomass, the produced ethanol is regarded as more environmentally friendly. Table 2 shows several potential biomass feedstocks to produce second-generation bioethanol and the related amounts of ethanol that can be potentially obtained per dry tonne of raw material (Naik et al., 2010). It should be highlighted that as reported by the United States Department of Energy's Centre for Transportation Research, the ethanol produced from lignocellulosic biomass reduces the life cycle of global GHG emissions further compared to its first-generation counterpart, e.g., maize-derived bioethanol (Fig. 3) (Yusoff et al., 2015). Table 2. Several biomass feedstocks used to produce second-generation bioethanol and their potential ethanol yields (Naik et al., 2010). Despite its advantages, the bioethanol produced from lignocellulosic biomass is still not economically viable. This is ascribed to the volatility of lignocellulosic feedstock prices and the high cost of pretreatments required, negatively affecting the overall production cost of lignocellulosic bioethanol (Hahn-Hagerdal et al., 2006). As a result, the main feedstocks for ethanol production are still coarse grains and sugar cane, with the second-generation bioethanol produced from lignocellulosic biomass contributing only 8% of the global ethanol production in 2019. Indeed, nearly 40% of the rise in world ethanol production over the last decade was due to increased sugar cane-based ethanol production mainly produced in Brazil.

Biomass
As for the ethanol production process, first-generation bioethanol, such as maize-derived bioethanol, is obtained from the alcoholic fermentation of sucrose-containing and starchy feedstocks (Sims et al., 2010). The starchy [n(C6H10O5)] feedstocks consist of long-chain glucose polymers and are first ground and mixed with water to break down into simpler sugars; subsequently, simple glucose (C6H12O6) is fermented by yeast into bioethanol. The ethanol production process through alcoholic fermentation is shown in Equations 1 and 2. Three different kinds of microorganisms can be used in the fermentation of glucose into bioethanol, namely: yeast (Saccharomyces sp.), mold (mycelium), and bacteria (Zymomonas sp.) ( ). In practice, 1000 kg of fermentable sugars produces around 583 L of bioethanol. Thus, about 42-50% of glucose can be efficiently converted into ethanol. Eq. 2 The second-generation bioethanol (lignocellulosic bioethanol) production involved additional steps (Yusoff et al., 2015): (i) pretreatment (solubilization of hemicellulose), (ii) enzymatic hydrolysis process (conversion of cellulose into sugars), (iii) fermentation (conversion of sugars into ethanol), and (iv) distillation of pure ethanol. In addition to that, the high cost of the enzymatic hydrolysis process causes the bioethanol production process from lignocellulosic biomass to be much costlier compared with the first-generation bioethanol (Lynd et al., 1996).
Ethanol can also be produced via catalytic hydration of ethylene (C2H4) with steam, as shown in Equation 3: The alcohol fermentation process contributes about 97% of the global ethanol production, whereas the catalytic hydration process contributes only 3% (Roozbehani et al., 2013). This is ascribed to the high cost of ethylene, making the synthetic ethanol production route less economically promising than alcoholic fermentation.

Physicochemical properties of ethanol and gasoline: a comparison
Ethanol contains about 34 wt% oxygen; it is transparent, neutral, colorless, flammable, volatile, and is easily mixable in non-polar solvents (Masum et al., 2013). The characteristics of ethanol and gasoline/ethanol blends as transportation fuels depend on several physicochemical properties such as density, viscosity, stoichiometric air/fuel ratio, heating value, latent heat of vaporization, RON, and oxygen content (Iodice et al., 2017;Awad et al., 2018a). Engine performance and exhaust emissions parameters, as well as the efficiency of the combustion process in last-generation SI engines, fuelled with gasoline/ethanol blends, rigorously depend on such physicochemical properties. The main chemical and physical characteristics of ethanol and gasoline and their differences are shown in Table 3.  Density. Generally, density affects combustion characteristics and fuel atomization efficiency. In effect, higher density can bring about greater fuel flow resistance which causes higher viscosity, thereby resulting in lower fuel injection (Mofijur et al., 2016). Since the density of ethanol (around 795 kg/m 3 ) is higher than that of gasoline (about 755 kg/m 3 ), the volumetric fuel economy is improved because the volumetric fuel pump injects less mass of ethanol/gasoline blend than pure gasoline.
Lower heating value. The energy content of ethanol (27 MJ/kg) is around 65% of gasoline (44 MJ/kg). Hence, ethanol and gasoline/ethanol blends are associated with higher fuel consumption as compared to gasoline. However, this does not lead to a significant decline in engine power with increasing ethanol proportion in blends. The heating value of the stoichiometric air/fuel mixtures (which is equal to the ratio between the lower heating value and stoichiometric air/fuel ratio) is around 3 MJ/kg for both air/ethanol and air/gasoline blends, and this value remains constant even when considering different ethanol/gasoline mixtures (Iodice and Senatore, 2016; Deng et al., 2018). In effect, since this factor indicates the energy content introduced in the engine with the unit mass of stoichiometric air/fuel mixture, the stoichiometric air/ethanol mixture holds the same amount of energy as the stoichiometric air/gasoline mixture (Benajes et al., 2018;Boulahlib et al., 2018).
Latent heat of vaporization. The heat of vaporization of ethanol (910 kJ/kg) is much higher than that of gasoline (390 kJ/kg). Thus, since ethanol requires more heat to evaporate than gasoline, the temperature of the intake manifold is lower when ethanol is used as fuel, thereby increasing the engine's volumetric efficiency (Amirante et al., 2017). However, higher heat of vaporization also results in lower combustion temperature and burning velocity, which lead to higher HC and CO exhaust emissions Oxygen content. Ethanol has 34.73% oxygen which results in higher combustion efficiency. In effect, because of the greater amount of oxygen in ethanol/gasoline blended fuels, a more regular and complete combustion phase and higher combustion temperatures can be achieved than neat gasoline (Fajri et al., 2017). Under such fuel-lean conditions, the exhaust emissions of CO and HC can also be abated (leaning effect) (Geng et al., 2017).

Reid vapor pressure.
The Reid vapor pressure of ethanol (17 kPa) is very low compared to neat gasoline (around 55 kPa). Hence, during the engine's warm-up phase, higher CO and HC cold emissions can occur due to the lower volatility of ethanol. As an advantage, the lower vapor pressure of ethanol can reduce evaporative emissions. However, the Reid vapor pressure of ethanol/gasoline blends does not depend linearly on the ethanol content in the blends.
Research octane number. The RON of ethanol (around 110) is higher than that of gasoline (approx. 95); as a result, to obtain more power from the engine, ethanol/gasoline blended fuels can withstand higher compression ratios before detonating in comparison with neat gasoline (Bhasker and Porpatham, 2017). Moreover, since the thermal efficiency depends on the volumetric compression ratio and ethanol fuel allows the use of high compression ratios without knocking, the use of ethanol can also enhance the thermal efficiency of SI engines (Genchi and Pipitone, 2014; Jamrozik et al., 2018). Table 3, combustion of ethanol clearly needs a lower quantity of air compared to neat gasoline (around 1.6 times less). Also, the airflow rate entering the cylinder keeps constant for fixed throttle valve opening and engine speed. Therefore, to hold the air/fuel equivalence ratio constant, the electronic control unit (ECU) has to increase the consumption of the gasoline/ethanol blend (Masum et al., 2013).

Stoichiometric air/fuel ratio. On comparing the stoichiometric air/fuel equivalence ratios shown in
Viscosity. Viscosity can generally affect combustion quality, spray characteristics, and fuel drop size (Mofijur et al., 2016). Since the viscosity of ethanol is higher than that of gasoline, larger droplets could be formed during injection, adversely affecting combustion quality, thus resulting in higher exhaust emissions (Hassan et al., 2015).

NOX formation pathways
The oxides of nitrogen represent some of the most troublesome emissions from SI engines. NOX in exhaust emissions comprises several chemical compounds: nitric oxide (NO), nitrogen dioxide (NO2), nitrous oxide (N2O), dinitrogen trioxide (N2O3), and dinitrogen tetroxide (N2O4). However, NO and NO2 are predominant, whereas the other compounds are present in negligible quantities (Normann et al., 2009). The ambient concentration of NO, which is an odorless and colorless gas, is typically far less than 0.5 ppm. Nitrogen dioxide, a reddish-brown, toxic, and corrosive gas, is relatively visible in high ambient concentrations. Oxidation of nitrogen inside the cylinder of SI engines at high temperatures leads to NOX formation (Chong et al., 2010). Hence, in effect, NOX formation from SI engines greatly depends on the concentration of available oxygen, the peak temperature inside the cylinder, the equivalence ratio and the residence time for the reaction to occur between atmospheric oxygen and nitrogen molecules (Masum et al., 2013;Shahir et al., 2015). Two mechanisms of NOX formation are discussed in this section: thermal and prompt NOX.
Thermal NOX formation is the predominant mechanism of total oxides of nitrogen. During combustion, nitrogen molecules react with the available oxygen inside the cylinder at temperatures higher than 1800 K through the following chemical reactions (Eqs. 4-7). Although the reactions leading to NOX formation are many more, such kinetic equations represent the basic steps of the Zeldovich mechanism that describes thermal NOX formation.
However, thermal NO formation advances more slowly than the oxidation of hydrocarbons since NO production needs high temperatures to advance owing to its high activation energy (314 kJ/mole) (Fajri et al., 2017). Equation 7 describes the thermal NO production rate in which T is temperature, and k1 and k2 represent reaction constants. As defined in Equation 4, there is good evidence of a strong dependence of thermal NO production rate on temperature. In effect, high residence times for the reaction to occur, high available oxygen concentrations, and high peak temperatures lead to high formation rates of thermal NOX (Varatharajan and Cheralathan, 2012).
The second mechanism leading to NOx formation in exhaust emissions is termed prompt NOX. Although the share of prompt NOX on total NOX emissions is lower than that of thermal NOX, the presence of such prompt NOX can be high in rich flames and the laminar premixed flame zone. Indeed, during the combustion of hydrocarbon fuels and before the production of thermal NOX, prompt NOX can be produced in a significant amount under specific operating conditions: fuel-rich conditions, short residence time, and low temperature. Prompt NOX formation entails some intermediate compounds in a complex sequence of chemical reactions summarised in the following equations (Eqs. Normally, under both fuel-rich conditions and combustion temperatures lower than 750 °C, nitrogen molecules react with CH and CH2 hydrocarbon fragments to produce amine species which are subsequently converted into prompt NO in exhaust emissions. Hence, as defined in the above equations, CH and CH2 hydrocarbon radicals are the major contributors to producing prompt NOX, which is much more sensitive to fuel chemistry than thermal NOX due to the high dependence of prompt NOX on hydrocarbon radicals (Eqs. 5 and 6). However, the concentration of nitrogen-containing fragments (hydrogen cyanide; HCN) rises with the increasing quantity of CH and CH2 hydrocarbon fragments, which in turn increases with growing equivalence ratios. Thus, prompt NOX production grows with rising equivalence ratios while it decreases when a lack of oxygen occurs (Masum et al., 2013).

Effect of gasoline/ethanol blends on NOX emissions
In the last three decades, SI engines have been extensively investigated to analyze the effect of varying ethanol concentrations in gasoline fuel blends on NOX emissions. In most such experiments, a decrease in NOX emissions was observed for ethanol-gasoline blends compared to neat gasoline. However, an opposite trend was also marked by other researchers (Song et al., 2007), while a few found no noteworthy differences in NOX emissions between gasoline and ethanol/gasoline blends (Jia et al., 2005). This section discusses the causes of NOX emissions for SI engines fuelled with ethanol/gasoline blends, focusing on different fuel compositions and engine parameters (i.e., compression ratio, equivalence ratio, engine load, and speed).

Effect of ethanol concentration in blends
The main parameters affecting NOX emissions in SI engines fuels by ethanol/gasoline blends are combined with combustion temperature, combustion duration, and ethanol concentration. The experimental investigations reviewed in this work included studies on SI engines powered by ethanol/gasoline blends containing various ethanol concentrations ranging from 5 vol% to as high as 100 % v/v (i.e., neat ethanol). Owing to the differences in physicochemical properties of ethanol and gasoline (Table 3), the properties of resulting ethanol/gasoline blends also differ (Liu et al., 2016). More specifically, with increasing ethanol content in ethanol/gasoline blends, the octane number, latent heat of vaporization, and density increase while the heating value of the blends decreases (Li et al., 2018).
Turner et al. (2011) measured NOX emissions of a direct injection SI engine fuelled with different ethanol/gasoline blends when the indicated mean effective pressure (IMEP) was set at 3.4 bar. They found that NOX exhaust emissions decreased by increasing ethanol concentration from 30% to 85% in the test fuels. This decrease could be ascribed to the reduction in flame temperature, which was confirmed by the decline recorded in the exhaust gas temperature. Further increments in ethanol concentration in the blends beyond 85% v/v led to slight increases in NOX emissions, possibly because of a more favorable combustion process resulting in higher maximum in-cylinder temperatures and pressures. Similar results were also reported by Bielaczyc et al. (2011), who used a vehicle with "Euro 4" European emissions standard (without engine modifications) and observed a linear decrease in NOX emissions in response to increasing ethanol concentration from10% to 85% v/v in the blends. These findings were in line with those of the study by Oh et al. (2010), in which a direct injection SI engine was fuelled with 25, 50, and 85% v/v ethanol blends. The authors attributed the decreases in NOX emissions with increasing ethanol concentration to the reduced peak in-cylinder pressure and temperature due to combustion retardation.
As described in the previous section, the formation of NOX emissions is strictly dependent on oxygen concentration, peak in-cylinder temperature, and residence time in the combustion chamber (Ozsezen and Canakci, 2011;Canakci et al., 2013;Masum et al., 2015). In an experimental study performed by Lin et al. (2010) on a small air-cooled engine, the decrease in NOX emissions as a result of increasing ethanol concentration in the test fuels was justified by the lower combustion temperature due to the high oxygen content of ethanol. They claimed substantial reductions in NOX emissions ranging from 77% to 86% when using gasoline/ethanol blends containing 6 to 9% v/v ethanol, attributing it to the higher heat of vaporization of ethanol in comparison with gasoline. In effect, with increasing ethanol concentration in the fuel blends, the temperature of ethanol/gasoline blends decreases at the end of the intake stroke, thus entailing a decrease in the maximum combustion temperature . Zervas et al. (2003) highlighted multiple reasons to explain the reductions observed in NOX emissions of the investigated SI engine fuelled with ethanolgasoline blends, including excess oxygen present in ethanol reducing the peak in-cylinder temperature, lower combustion temperatures owing to the lower heating value of the ethanol-containing blend, and higher latent heat of ethanol than gasoline. At an engine speed of 2000 rpm and in wide-open throttle conditions with different ethanol concentrations ranging from 5 to 80% v/v, Gravalos et al. (2011) found that NOX emissions decreased with increasing ethanol percentage in the gasoline/ethanol blends. They also cited the higher heat of vaporization of ethanol than gasoline leading to a lower combustion temperature of the blends under investigation as the main reason behind these observations. The maximum percentage decrease in NOX emissions was obtained with 80% v/v ethanol in ethanol/gasoline blends. Canakci et al. (2013), on a multi-port injection system engine, revealed that NOX exhaust emissions decreased as the ethanol concentration in the blended fuels increased. Reduced peak in-cylinder temperatures owing to the higher latent heat of vaporization and lower heating value of ethanol as compared to gasoline were indicated as the underlying reasons. For 10% v/v ethanol, the percentage decrease in NOX emissions was recorded at 15.5%.  The effect of low ethanol concentrations in blends on NOX emissions of SI engines was also investigated by Yang et al. (2012). They used nine four-stroke motorcycles equipped with carburetor systems tested on a chassis dynamometer using neat gasoline and gasoline blended with 3% v/v ethanol as test fuels but found no considerable differences between neat gasoline and the ethanol/gasoline blend. Jia et al. (2005) also found a slight reduction (6%) in NOX emissions using gasoline blended with 10% v/v ethanol compared with neat gasoline.
Using 20% v/v ethanol concentration in the fuel blend, the emission performance of an in-use fuel-injected vehicle (without engine modifications) was analyzed by Tibaquirá et al. (2018). The authors reported reductions in NOX emissions ranging between 3 to 17.5% when E20 was used compared with neat gasoline under steady-state conditions (Fig. 5). Some studies have reported both increases and decreases in NOX emissions depending on the ethanol concentration used in the fuel blend. For instance, Zhuang and Hong (2013), examining 0 to 60% v/v ethanol concentration, claimed that NOX emissions increased for up to 24% ethanol concentration in gasoline, attributing it to improved combustion process inside the cylinder, resulting in an increased peak in-cylinder temperature. However, NOX emissions decreased for ethanol concentrations higher than 24% v/v due to reduced peak in-cylinder temperature. The reduction observed could be ascribed to the higher latent heat of evaporation of the ethanol than gasoline's, lowering the peak in-cylinder temperature during vaporization of ethanol/gasoline blended fuels. Besides, for such high ethanol concentrations in the blends, the combustion temperature could also be reduced by more triatomic molecules in the combustion products of gasoline/ethanol-blended fuels than in those of neat gasoline (Schifter et  For the E10 test fuel and when the engine was operated at 1700 rpm, a 30% decrease in NOX emissions was measured compared to neat gasoline. However, further increases in ethanol concentration of blended fuels increased NOX emissions by 8.9%, 15.7%, and 76.9% for E15, E20, and E25, respectively.

Effect of transient cold-start
During the cold-start transient phase, fuel vaporization is slow due to the cylinder walls being cold. Consequently, the fuel flow increases with a resultant enrichment of the air/fuel mixture to overcome the poor mixing of the inlet charge, increasing unburned hydrocarbons. Furthermore, the gasoline forming a surface film inside the combustion chamber is weakly mixed and, therefore, weakly burned. Thus, during the warming-up process of the engine, until the cylinder walls warm up and the engine uses a rich mixture, CO and HC are increased. In addition, the efficiency of a three-way catalyst is also limited due to the rich values of the air/fuel ratio. Indeed, the best range of the air/fuel equivalence ratio for the optimum conversion efficiency of a catalytic converter is very narrow near the stoichiometric value. Hence, during the cold-start transient phase of last generation SI engines, both the technological limits of the three-way catalyst and the necessary rich values of the air/fuel mixture can produce higher CO and HC cold emission levels compared to hot operating conditions.
Iodice et al. (2018) compared NOX emissions for cold and hot operating conditions and in response to different ethanol concentrations in the gasoline/ethanol blends. As presented in Figure 6, both in cold and hot operating conditions, NOX emissions decreased with rising ethanol content in the test fuels as a result of the higher H/C atom ratio of ethanol compared to gasoline (See Table 3). NOX emissions are produced in the combustion chamber owing to high peak in-cylinder temperature when the nitrogen reacts with oxygen. In this regard, the chemical combustion equations of ethanol and gasoline ( Eqs. 13 and 14, respectively) show that ethanol/gasoline blends, under stoichiometric operating conditions, require a lower amount of air as compared to gasoline, with a subsequent decrease in NOX hot emission levels (Dogan et al., 2017). Likewise, ethanol/gasoline blends also require less air under rich operating conditions during the cold-start transient phase.  In addition to the justification presented above, NOX emissions decreased with the rise in ethanol content in the test fuels both in cold and hot operating conditions due to the higher latent heat of vaporization and lower heating value of gasoline/ethanol test fuels than those of commercial gasoline. Indeed, under such operating conditions, the temperature of the air/fuel mixture drops at the end of the intake stroke, thereby lowering the combustion temperature and reducing the thermal NOX.
To evaluate the effect of the transient cold-start on engine NOX emissions of turbocharged SI engines, Guo et al. (2020) investigated the exhaust emissions of a DI TG four-cylinder engine fuelled with 10% v/v ethanol (E10) under steady-state and transient cold-start conditions. They argued that the NOX transient cold-start emissions of the investigated DI TG engine fluctuated dramatically and strongly depended on the engine load. Besides, under steady-state operating conditions, the NOX exhaust emissions were increased with increasing coolant temperature from around 40 °C to 100 °C.

Effect of compression ratio
The thermal efficiency of SI engines depends on the compression ratio. A fuel with a high octane number can withstand high compression ratios without knocking before detonating, thus improving the knock tolerance (Yücesu et al., 2006). The octane number of neat ethanol (around 108) is higher than gasoline (91)(92)(93)(94)(95)(96)(97)(98)(99)(100). Hence, ethanol is particularly suitable for high compression ratios because it can improve the thermal efficiency of SI engines. For this reason, the fuel economy penalty resulting from the lower energy content of ethanol as compared to gasoline can be offset by an increase in the thermal efficiency of SI engines with high compression ratios when fuelled with gasoline/ethanol-blended fuels. However, NOX emissions should increase with increasing compression ratios, especially at high engine loads due to the resulting high combustion temperatures.
In a study performed by Al-Baghdadi (2008), NOX emissions were measured for an SI engine operated on ethanol/gasoline blends (ethanol concentration ranging between 0% to 30% v/v) at variable compression ratios (ranging between 8 to 9.25, proportional to the ethanol percentage in the blend). They confirmed decreases in NOX emissions with rising ethanol proportion in the blends but also showed that increases in compression ratio caused an opposite effect. The latter could be justified, on the one hand, by the higher peak temperature and pressure and, on the other hand, by the reduction in the time needed for NO dissociation into N2 and O2 (Al-Baghdadi, 2008). Celik (2008) analyzed different ethanol/gasoline blends (E25, E50, E75, and E100) in a four-stroke single-cylinder engine at constant load and speed, while the compression ratio varied from 6:1 to 10:1. On using gasoline blended with 50% v/v ethanol at a high compression ratio of 10:1, NOX decreased by 19% compared to neat gasoline under a compression ratio of 6:1. They attributed the reduced NOX emissions to a lower heating value of ethanol than gasoline. Koç et al. (2009) increased the compression ratio of a four-cylinder SI engine from 10:1 to 11:1 by using 0%, 50%, and 85% of ethanol in gasoline/ethanol blends. For the same ethanol percentage in the blends, NOX emissions were higher for higher compression ratios due to the resultant high combustion temperatures (Fig. 7). Besides, NOX emissions decreased with increasing ethanol concentration at both compression ratios. This phenomenon was due to the higher latent heat of vaporization of ethanol reducing the flame temperature, thereby resulting in lower NOX emissions. In effect, on using neat gasoline at a compression ratio of 10:1, NOX emissions were even higher than those of gasoline/ethanol blends at a compression ratio of 11:1.

Effect of equivalence ratio
The stoichiometric air/fuel ratio of ethanol is around 1.6 times lower than base gasoline. As is well known, for an SI engine, the airflow rate entering the cylinders remains the same for a given engine rotation speed and throttle valve opening. Hence, to obtain the same air/fuel equivalence ratio, the mass flow rate of the ethanol/gasoline blend is increased by the ECU as the ethanol content in the blended fuel increases. Following that, under fuelrich operating conditions, the leaner effect caused by the high oxygen content of ethanol can shift the fuel/air ratio from rich to stoichiometric values, thus reducing the flame temperature and, consequently, NOX emission levels. Najafi et al. (2009) investigated the effects of 5, 10, 15, and 20% v/v ethanol blends on the NOX emissions of a four-stroke SI engine and recorded higher NOX emissions when the ethanol percentage in the blend was increased. They attributed this finding to the higher oxygen content of the ethanol/gasoline blends than neat gasoline, increasing the air/fuel ratio in the fuel-rich regions of the engine. Hence, the air/fuel ratio moves from rich toward stoichiometric values as the ethanol fraction increases, thus bringing about a more complete and efficient combustion process (Thangavel et al., 2016). Finally, this efficient combustion process increases the in-cylinder temperature leading to an increase in thermal NOX emissions. Similar results were also found by Al-Farayedhi et al. (2000) and Hsieh et al. (2002).
A comparative study on emission and combustion characteristics of a multiport-fuel-injection SI engine powered by different fractions of butanol, methanol, and ethanol in gasoline-blended fuels was performed by Li et al. (2017). At equivalence ratios set at 1 and 3 bar BMEP, gasoline/ethanol blends (i.e., 10, 30, and 60% v/v) correspondingly decreased NOX emissions by around 3%, 5%, and 12% as compared to gasoline (Fig. 8). In the same study, the effects of gasoline blended with 30% v/v ethanol on NOX emissions were also analyzed under different equivalence ratios fluctuating from 0.83 to 1.25. The highest NOX emission occurred when the equivalence ratio ranged between 0.9 and 1.0, which could be explained by the resulting rapid and complete combustion process leading to increased combustion temperature. Then, as shown in Figure 9, the NOX emissions decreased as the equivalence ratio reached relatively leaner or richer values than the stoichiometric value due to the resultant re duced peak combustion temperatures.    Zervas et al. (2003) also explored the effects of different equivalence ratios on NOX emission of an SI engine fuelled with ethanol/gasoline blends. They argued that changing the equivalence ratio from 0.83 to 1.25 increased NOX emissions from 15% to 30%, respectively.

Effect of engine load and speed
Engine load and speed are the main parameters affecting NOX formation in SI engines: to increase the engine speed and load, a richer mixture is required, bringing about higher in-cylinder temperatures and hence higher NOX emissions. In this context, the faster flame speed of ethanol than gasoline can play an important role in complete combustion in rich-mixture operating conditions, which results in higher NOX emissions.
Melo et al. (2012) tested a flex-fuel SI engine fuelled with different ethanol percentages in the blends, from 5 to 100% v/v, at two different engine loads of 60 Nm and 105 Nm. They revealed that at the lower engine load, NOX emissions decreased with increasing ethanol concentration in the blends. In contrast, NOX emissions increased with the growing ethanol percentage in the blends at the same engine speed but at the higher engine load. The increase in NOX emissions at the higher engine load was attributed to the faster flame speed caused by ethanol-gasoline blends compared to gasoline because, under such operating conditions, the resulting complete combustion process led to an increased in-cylinder temperature. Citing the same justification, Li et al. (2017) reported that at the engine load of 5 bar BMEP, higher NOx emissions were generated compared to 3 bar BMEP (Fig. 9). However, compared to commercial gasoline, a reduced combustion temperature was caused by ethanol addition to the gasoline blends such that, under different equivalence ratios, the final NOX emissions were 0.4-10.4% lower for E30 (30% ethanol v/v), despite the higher oxygen content supplied.
Keskin and Guru (2011) examined the effects of ethanol blended with gasoline (0% to 20% v/v) on NOX emissions at different engine loads ranging between 800 and 2400 kW. NOX emissions increased with increasing ethanol percentage in the blends at high engine loads, while at low engine loads, NOX emissions were almost the same for all test fuels. Gomes et al. (2011) tested ethanol/gasoline blends (5% to 100% v/v) at engine loads of up to 30 bar IMEP. They concluded that at low engine loads, NOX emissions were lower using ethanol concentrations >25% v/v owing to the reduced peak temperature. In contrast, NO emissions were almost the same for all blended fuels at higher engine loads as the faster flame speed of ethanol than that of gasoline leads to similar peak in-cylinder temperatures (Pischinger, 2017). Pang et al. (2008) investigated 10% v/v ethanol in blended fuel at two different engine loads, 3 and 160 Nm torque, but found no significant variations in NOX exhaust emissions compared to neat gasoline.
NOX formation in SI engines is known to rise with increasing engine speed due to higher in-cylinder temperatures when more fuel is burnt in less time available for combustion than at low engine speeds (Ramasamy et al., 2017). In this context, ethanol's faster flame speed than gasoline can play a major role in completing the combustion in rich-mixture operating conditions caused by high engine speeds, thereby leading to higher NOX emissions than gasoline. Costa and Sodré (2010) investigated the effect of 22 and 100% v/v ethanol in blends on NOX emissions at low and high engine speeds. They claimed that at low engine speeds (2500-3000 rpm), there were no substantial variations in NOX emissions between the tested fuels. On the contrary, higher NOX emissions were observed at high engine speeds for neat ethanol than 22% v/v ethanol in the gasoline/ethanol blend due to the faster flame speed of ethanol than gasoline. Koç et al. (2009) also examined the effect of 0, 50, and 85% v/v ethanol in gasoline/ethanol blends on NOX emissions within a wide range of engine speeds ranging between 1500 and 5000 rpm. However, they observed opposite trends for gasoline-ethanol blends at high engine speeds compared to the previous reports. As shown in Figure 7, the increase in NOX formation with engine speed was lower for the 85% ethanol (around 11%) than the value recorded for neat gasoline (approx. 42%). The lower increase in NOX emissions as the ethanol content in the blended fuel increased under increasing engine speeds could be ascribed to the lower heating value and higher heat of vaporization of ethanol compared to gasoline.
In a recent study, Mohammed et al. (2021) assessed the effect of different ethanol inclusion rates (10, 20, 30, and 40%) in gasoline at various engine speeds ranging from 1500-2500 rpm. They found that with increasing engine speed, NOX emissions increased. Moreover, the lowest NOX emissions were recorded at the highest ethanol level (E40) by 21%, which could be ascribed to the delayed ignition timing; this improvement did not decrease thermal efficiency, though. Taghavifar et al. (2020) added 20% bioethanol into gasoline at an engine speed of 1400 rpm and different engine loads and measured NOX emissions under all conditions. Maximum NOX emission reduction was recorded by 60% at 75% engine load but offered no explanations for the observations made. In another study aimed at understanding the effects of ethanol addition and exhaust gas recirculation (EGR), Biswal et al. (2020) used E20 at different engine loads and an EGR of 10%. They found that although ethanol inclusion in gasoline (without EGR) increased NOX emission, using the EGR technique compensated for the ethanol effect, resulting in a 20% NOX emission reduction compared to neat gasoline without EGR. However, it is worth noting that E20+EGR resulted in less NOX emission reduction (17%) compared to E0+EGR (30%). Rosdi et al. (2020) combusted three different ethanol/gasoline blends (10, 20, and 30%) on a four-cylinder turbocharged SI engine (Multi-Point Injection). NOx emissions were proportionally decreased in response to increasing ethanol concentration in gasoline. For instance, NOx emission stood at 800 ppm for E30 while it was considerably higher for neat gasoline at 1700 ppm. They attributed the findings obtained to the high latent heat of vaporization, low heating value, and high oxygen content of ethanol. Agarwal et al. (2020) studied some of the toxic exhaust gases, including NO and NO2. They showed that the contribution of NO to NOX emissions was significantly higher compared to that of NO2. In their study, ethanol/gasoline blend (E10) showed various combustion behavior concerning NOX emission depending on engine operation conditions. More specifically, at the lower engine load tested (i.e., 50%), ethanol inclusion decreased both types of NOX emissions (NO and NO2) by 1%. While at a full load of engine operation, E10 increased NO and NO2 by 2%. They highlighted that at 50% load, the effect of ethanol's higher latent heat of vaporization was dominant over its oxygen, causing lower combustion temperature and NOX formation. On the contrary, at a full load of engine operation, the fuel-bound oxygen played a major role leading to more complete combustion and increased in-cylinder temperature and NOX emissions.

Effect of ignition timing
Little information is available in the published literature on the effect of ignition timing on the combustion behavior of ethanol/gasoline blends. In a very study, Liu et al. (2021) looked into the impact of ethanol port injection (25,50,75, and 100%) (ethanol injection using a separate injector) for different ethanol/gasoline blends (E0, E25, E50, E75, and E100) on engine performance and emission parameters at different ignition timings (10,15,20,25, and 30 CA BTDC). Their results showed that retarding the ignition strongly reduced NOX emissions. Minimum NOx emission was obtained for E25 at all ignition timings investigated. On the contrary, E75 and E100 caused higher NOX emissions than neat gasoline for all investigated ignition timings. In general, by increasing ignition timing from 10 to 30 CA BTDC, NOX emissions were reportedly increased for all fuel blends. These findings could be explained by the fact that advancing ignition timing to 30 CA BTDC must have decreased the time for reaching the peak pressure, subsequently leading to higher engine temperature and higher NOX emissions.
A summary of the main experimental results concerning the effect of ethanol addition in gasoline blends on NOX emissions under different ethanol concentrations and operating conditions is provided in Table 4.

Conclusions and future research directions
Of the alternative fuels available for the road transport sector, ethanol is widely accepted as a renewable fuel for SI engines when blended with neat gasoline due to its favorable physicochemical properties. However, there are substantial inconsistencies in current findings regarding NOX emissions from SI engines fuelled with gasoline/ethanol blends, which could represent a considerable obstacle to the market expansion of bioethanol. Therefore, we analyzed NOX emissions related to the use of ethanol/gasoline blends in SI engines, providing a comprehensive literature review on the current state of ethanol combustion in SI engines to explain the reported inconsistencies. Overall, it is envisioned that such wide-ranging and, at times, conflicting data should provide a platform for future studies in all areas of alcohols application as alternative fuels and related exhaust emissions to steer advancements in techniques used for reducing emissions and, in particular, NOX emissions.
The key conclusions which could be drawn from the present work can be summarised as follows: a. NOX emission levels of SI engines decrease with the rise in ethanol concentration in ethanol/gasoline blends due to the higher latent heat of vaporization and lower heating value of gasoline/ethanol blends than commercial gasoline. More specifically, upon using gasoline/ethanol blends, the temperature of the air/fuel mixture drops at the end of the intake stroke, thus leading to reductions in the combustion temperature and hence in thermal NOX emissions. Such decreases in NOX emissions also result from the lower combustion temperature due to the high oxygen content in the ethanol molecules. b. It can also be argued that NOX emissions decrease with increasing ethanol concentration in blended fuels due to the lower carbon number of ethanol than gasoline. In effect, the combustion process of ethanol/gasoline blends shows that, under stoichiometric operating conditions, a lower amount of air is required compared to neat gasoline. c. As the RON of ethanol is higher than that of gasoline, a higher compression ratio can be used in SI engines. Hence, the fuel economy penalty resulting from ethanol's lower energy content can be offset by an increase in the thermal efficiency of SI engines with high compression ratios when fuelled with gasoline/ethanol blends. However, NOX emissions should increase with increasing compression ratios, especially at high engine loads due to the resulting higher combustion temperatures. d. The faster flame speed of ethanol than gasoline plays an important role in achieving more complete combustion in rich-mixture operating conditions. However, at the same time, an increase in NOX emissions is observed at high engine loads. In effect, to increase the engine speed and load, a richer mixture is required. The resulting complete combustion process of ethanol/gasoline blends increases in-cylinder temperature and, hence, thermal NOX emissions. On the other hand, based on the experimental data available, under low engine speeds and loads, there are no considerable differences observed in NOX emissions of SI engines fuelled by different gasoline/ethanol blends.