Formation and measurement of N2O in combustion systems

Direct N2O emissions from fossil fuel combustion have previously been reported to be equivalent to 25–40% of the NOx levels. At these levels, fossil fuels have been suggested to be a major anthropogenic source of N2O. Recent tests have shown these measurements to be in error, most of the N2O having been formed by reaction between NOx, SO2, and H2O in the sample containers. Time resolved measurements of gas samples stored in Tedlar bags, supported by chemical kinetic calculations, indicate that the majority of N2O forms over a time period of 6 hours. The conversion of NOx to N2O in the sample containers is shown to depend on the amount of SO2 present. This sampling artifact raises questions about the validity of the existing data base, collected by grab sampling methods. As a result, a continuous infrared analyzer, developed primarily for characterization of N2O emissions from full scale combustion sources, was used to perform on line N2O measurements at several full scale utility combustion systems. A variety of conventional and advanced utility combustion systems (firing pulverized coal, oil, and gas) were tested. The measurements from conventional systems (natural gas, oil, and pulverized coalfired) indicate that the direct N2O emission levels are generally less than 5 ppm and are not related to the NOx levels in the flue gas. However circulating fluidized bed units produced elevated N2O emissions. At one circulating fluidized bed combustor firing a bituminous coal, N2O levels ranged from 84 to 126 ppm as the load was varied from 100% to 55%, respectively. The N2O emissions from the circulating fluidized bed appeared to be inversely related to the bed temperature. However, temperature is not the only parameter affecting N2O emissions from fluidized beds; all three of the units studied operated at similar temperatures during full load operation, but the N2O emissions ranged between 25 and 84 ppm. N2O emissions were also elevated at a full-scale boiler using selective non-catalytic NOx reduction with urea; 11–13% of the reduced NOx was converted to N2O.


Background
The mean global concentration of N20 is approximately 300 ppb and has been increasing at a rate of 0.2-0.4% per year.l'z In the troposphere, N20 is a relatively strong absorber of infrared radiation, and, therefore implicated as a contributor to the "Greenhouse Effect."Being stable in the troposphere, N20 is transported to the strato-sphere.In the stratosphere N20 is the largest source of stratospheric NO; NO is the primary species responsible for establishing 3 the equilibrium stratospheric 03 concentration.
The increase of N20 in the atmosphere has been attributed to anthropogenic sources, although the dominant source of N20 is still uncertain.Weiss, 2 Hao et al., 4 and Tirpak 1 suggest that combustion of fossil fuels results in emissions of N20 that can ac-245 NO, KINETICS count for the observed increase in N20.Previous N20 measurements from combustion sources indicate that systems fired with natural gas do not produce significant concentrations of N20.However, substantial levels of N20 were reported from systems fired with residual oil or coals (e.g., fuels containing nitrogen and sulfur).In fact, the measurements of Hao et al. 4 and Castaldini et al. 5 suggest that the N20 emissions are related to the emissions of NO~ (NO + NO2); the N20 level being equal to 25-40% of the NO~ concentration.
The N20 measurements from combustion sources referenced above were made by gas chromatographic analysis of gas samples collected in glass or stainless steel containers.Recent measurements have shown that these grab samples can undergo chemical reaction in the containers creating N20 concentrations substantially higher than those originally formed in the combustion process.6'7 This raises questions regarding the existing data base on N20 emissions from combustion sources and the role of combustion in contributing to the increases in atmospheric N20 concentration.This paper will address two issues: 1) the formation of N20 in sample containers; and 2) N20 emissions from combustion systems using an on-line continuous infrared analyzer.

Formation of NzO in Sample Containers
The artifact in measuring N20 with grab sampiing techniques was initially identified in the course of combustion experiments in a bench-scale combustor.6 '7 The tests showed that the NzO was being formed in the sample containers through a reaction mechanism involving SO2, NO, and 1"I20.6'7 The SO2 concentration was found to influence the amount of N20 formed.To better quantify the progress of the reactions occurring in the sample containers, tests were conducted during this study to determine the time history of N20, NO, NO2, and SO2 in the sample containers.Gas samples from a natural gas fired combustor were collected in 270 liter Tedlar bags and subsequently analyzed for NO, NO2, N20 and SO2.The 270 liter Tedlar bags provided sufficient sample to allow samples to be withdrawn and analyzed as a function of time using continuous gas analyzers.N20 was analyzed with an infrared continuous analyzer described in Ref. 8. NO and NO, were measured with a TECO chemiluminescent analyzer and SOz with a Western Research ultraviolet analyzer.
The primary variable for the series of tests was SO2 concentration, which was varied from 700 ppm to 2500 ppm.Initial SOz and NOx concentrations were varied by adding SO2 to the combustion air and NH3 to the natural gas.The time histories of N20, NOt, and SOz are shown in Fig. 1.As can  -c, at the high SO2 condition, virtually all of the NOx initially present in the Tedlar bag is transformed to N20 in the final product.At the lowest SO2 concentration (716 ppm), N20 continues to form until virtually all of the SOz is removed from the gas phase, after which the N20 stabilizes while the NOx continues to decrease.At the intermediate SOz level (1210 ppm), the N20 formation rate is initially rapid becoming more gradual after 2 hours, as the NOx and SOz are consumed.For this intermediate case, there appears to be sufficient SO2 to allow the reaction to proceed longer than at the low SOz condition, but not sufficient SO2 to convert all of the initial NO~ to NzO, as was observed at the higher SO2 level.
The gas phase and aqueous reactions between SO2, NOx, and H20 leading to the formation of NzO have been documented in the literature.Martin et al. 9 discuss the reaction mechanism; the general chemistry has also been discussed in the wet FGD scrubbing literature.1~ Lyon and Cole 12 proposed a detailed mechanism for N20 formation in the sample containers.In the sample flasks, NO is oxidized to NO2.NO2 and SO2 in the gas phase are taken to be in equilibrium with the aqueous phase.In the aqueous phase, the NO2 and SO2 ultimately form N20 and H2SO4 through a reaction sequence involving HNO. 12 The detailed mechanism of Lyon and Cole was integrated for the same conditions as the experiments conducted in the Tedlar bags (Fig. 1).The results of these calculations are included with the experimental results in Figure la-c.In general, the model calculations are in qualitative agreement with the experimental results.The characteristic time for N20 formation is basically the same between experiment and model, with the experiment exhibiting somewhat faster NzO formation than predicted by the model.However, there are noteworthy differences between the experiments and the model calculations.First, for the medium and high SO2 cases, the kinetics underprediet the amount of NzO formed.At the high SO2 condition, the experiments exhibited virtually 100% conversion of the NO~ to NzO, whereas the model predicts 65% conversion.Second, in the Tedlar bags the rate of disappearance of NO~ decreased as the SO2 level decreased, whereas the kinetic calculations indicate little effect of SO2 level upon the NOx time histories.Another difference between the experimental observations and calculations is the effect of the chemical reactions on the pH of the aqueous phase.Experimentally, the pH in sample containers containing 600 ppm NO and 1500 ppm of SOz have been reported to be 1.8.~ However, the calculations indicate an aqueous phase pH of only 4.5-4.8after 24 hours for initial SO2 concentrations ranging from 2470 ppm to 100 ppm.This is expected to influence the calculated levels of N20 produced since previous work suggests that N20 increases as the pH decreases.1~ At present, the model does not include a reaction allowing the direct oxidation of sulfite ion to sulfate ion, although this step is known to occur.By including this reaction in the model, most of the deviations noted in Fig. 1 could be accounted for.For example, the sulfite ion concentration would be depleted, reducing the resistance of the liquid phase to SO2 dissolution.This would increase the predicted rate of SO2 decay in the gas phase, improving agreement between model and experiment (Fig. 1).Also, the depleted sulfite ion in the liquid phase would slow the liquid phase conversion of nitrogen, thereby increasing the resistance of NOz entering the liquid phase.Finally, the missing reaction would accelerate the conversion of a weak acid into a strong acid, reducing the predicted pH toward the experimentally observed values of 1.7-1.9.

N20 Emissions from Combustion Sources
Reliable measurements of N20 emissions from combustion systems are needed in order to assess the role of combustion as an anthropogenic source of atmospheric NzO.With grab sampling procedures in question, a continuous N20 analyzer, utilizing a non-dispersive infrared principle was used to characterize N20 emissions from a variety of utility scale combustion systems.The continuous N20 .... 8 analyzer system is described m detail elsewhere.The basic analyzer is a non-dispersive infrared analyzer manufactured by Horiba Ltd. based on the work by Montgomery, et al. 8 The analyzer utilizes a 500 mm sample cell and uses the 7.8-8.5 micron region of the infrared for the N20 measurement.
To eliminate interferences, NOz and SOz are removed from the sample gas using 1 M solutions of sodium sulfite and sodium carbonate.8 The continuous NzO analyzer was used to characterize the N20 emissions from the following systems: Natural Gas-Fired Utility Boilers The units firing natural gas, oil, and pulverized coal comprised a variety of designs including single wallfired, opposed wall-fired and tangentially fired, and ranged in size from 50 to 750 MW (electric).The majority of the testing was performed at full load.For selected sites, the testing was performed over a range of loads.A summary of the results is pre-sented in Table I.At four of the boilers, on-line gas chromatographic measurements sponsored by the U.S. EPA were made in parallel and showed similar results.13 Natural Gas Firing: The N20 emission from three natural gas-fired utility boilers ranging in size from 215 to 750 MW were consistent with the previously reported results. 2 The N20 emissions from these natural gasfired utility boilers were low, less than 2 ppm.Also, as seen in Table I, there was no correlation with the NO~ emissions.

Residual Oil Firing:
The N20 emissions from the two residual oil-fired utility boilers (110 and 215 MW) were again low, 1 ppm, and exhibited no correlation with the NOx emissions.

Pulverized Coal Firing:
Seven pulverized coal-fired boilers were tested ranging in size from 180 MW to 790 MW.As can be seen in Table I, except for the 620 MW opposed wall-fired boiler, the N20 emissions were less than 6 ppm.The 12 ppm N20 concentration measured from the 620 MW opposed-fired unit may represent an upper limit on the emissions from this unit.Because of the high SO2 levels and NOx levels in this flue gas, a small amount of N20 may have formed in the unheated sample lines during transport to the analyzer.

Circulating Fluidized Bed Combustion:
N20 emissions were also characterized at three circulating fluidized bed combustors.Compared to pulverized coal-fired boilers, the circulating bed units exhibit elevated levels of NzO emissions.At full load operation of these units, NzO levels ranged from 84 to 26 ppm (dry at 3% 03).14 The corresponding NOx levels varied from 81 to 283 ppm (dry at 3% 03).13 At one of the units, tests were conducted over a range of operating loads, from 55 to 100% of rated load.A summary of this unit's emissions are presented in Table II.The N20 and NOx levels are plotted as a function of load in Figure 2a.For this unit, the N20 levels decrease and NOx increases as the load increases.Further, for this unit the N20 and NOx emissions appear to correlate with bed temperature (see Fig. 2b).The N20 levels decrease and NOx increases as the bed temperature increases.The trends in N20 concentrations with bed temperature are consistent with the results obtained from pilot-scale fluidized bed combustors reported by Amand and Anderson 15 and the mechanism suggested by Kramlich et al. 16 Kramlich et al. 16 suggest that NzO formation can occur from HCN through a mechanism of the form HCN + OH---> HNCO + H HNCO + OH ~ NCO + H20 NCO + NO--~ N20 + CO Peak conversion of HCN to N20 via the above  mechanism occurs at a temperature of about 1250 K. Below 1250 K, N20 concentrations are controlled by the formation of NCO and the conversion of NCO to N20.Above 1250 K, removal of N20 through reactions with H and OH control the level of N20.At the circulating fluidized bed tested over a range of loads, bed temperatures ranged between 1116 and 1140 K, suggesting that the N20 levels from this unit are controlled by formation processes rather than destruction reactions.Further investigations are needed to assess what other operational variables yield higher or lower N20 levels in CFBC's.While N20 levels correlated with bed temperature for the unit where tests were conducted over a range of loads, all three of the units tested operated at similar bed temperatures at full load.Yet, the N20 levels varied from 26 to 84 ppm.Other parameters that might influence the level besides temperature include the fuel type (char N evolution), char and/or solids loading, residence time, calcium species, sulfur species, and NOx level.Certainly the field data suggest there are operational conditions that minimize NzO in CFBC's, offering possible keys to its control.Once the primary N20 formation mechanisms are understood, potential control mechanisms can be developed.

Selective Non-Catalytic NOx Reduction: Urea Injection
A number of selective non-catalytic NOx reduction processes (SNCR) are currently under development for utilization in industry.These include injection of ammonia (NH3), urea (NHzCONH2), or cyanuric acid ([HNCO]3) at temperatures nominally between 1200 and 1500 K where the decomposition products react with NOx.Caton and Siebers 17 reported bench-scale results showing N20 as a product of the reaction between cyanuric acid and NOx, presumably via reactions 3 and 4. Further, they suggested that urea injected into high temperature combustion products would decompose to NH3 and HNCO.N20 measurements were also made at a unit using urea injection to reduce NOx.The boiler is rated at 110 MW and was firing residual oil.A summary of the full load data is presented in Table III.The urea/NOx molar injection ratio was varied between 0.0 (baseline) and 0.6.Without urea injection, the baseline NOx and N~O emissions were 291 and 1 ppm, respectively.As the amount of urea injection increased, the NOx level decreased, and the N20 level increased.Over the range of urea injection rates tested (urea/NOx molar ratio of 0.2 to 0.6), between 11 and 13% of the NOx removed was converted to N20.Further work is needed to 1) verify these first measurements, 2) thoroughly characterize N20 formation from SNCR processes, and 3) develop a better understanding of the basic NzO formation chemistry.

FIc. 1 .
Time histories of N~O, NO,, and SO2 during reaction in 270 liter Tedlar bags: experimental versus model results: a. N20, b.NO,, c. SO2 (.experimental, ---model).be seen in Fig. la, the majority of the N20 formation occurs in the first six hours, irrespective of the initial SOz level.The accompanying time histories of NO~ and SOz are shown in Figs.lb and lc.As seen in Figs.la

,
FIG. 2. N20 and NOx emissions versus load from a circulating fluidized bed eombustor.(a. as a function of load; b. as a function of bed temperature).