Environ. Sci. Technol. 1991, 25, 255-260
Measurement of Ambient Nitrous Acid and a Reliable Calibration Source for Gaseous Nitrous Acid Zbyn6k VeEe?at and Purnendu K. Dasgupta"
Department of Chemistry and Biochemistry, Texas Tech University, Lubbock, Texas 79409-1061 A diffusion scrubber based automated instrument measures HONO utilizing ion chromatography and UV detection with time resolutions of 7-15 min and a detection limit of ca. 20 pptv. The thermodynamic equilibrium of NH3(g) and HONO(g) over NH4N02(s)can be described by In (pHONO, atm) = -8018 f 84/T 16.81 f 0.30 and provides a convenient and reliable source for HONO(g) free from major contamination with NO,. Ambient measurement data for Lubbock, TX, are presented. Persistent daytime concentrations of HONO in the 100-500 pptv range, weekday morning peak concentrations up to 1ppbv following peak traffic hours, and typical nighttime peak concentrations between 1 and 2 ppbv are observed.
+
Nitrous acid plays an uniquely important role in atmospheric chemistry. It is efficiently photolyzed by near-UV radiation during daytime with a near-unity quantum yield at 368 nm to produce NO and 'OH ( I ) and is unequivocally a major daytime source of 'OH. The exact extent to which nitrous acid and nitrites account for acidic deposition is unknown; however, there are some reports that in areas with high traffic density [HONO(g)] can far exceed total nitrate (2). Formation mechanisms of HONO are not well-characterized at this time; nighttime formation schemes invoke the reaction of NO, 'OH, and a third body. Other sources include direct vehicular emission and heterogeneous hydrolytic reactions involving NO and NO2 (ref 3 and citations therein). Despite the importance of HONO, relatively few reported measurements exist. Since its unequivocal detection is ambient air a decade ago by long-path differential optical absorption spectroscopy (DOAS; see, e.g., ref 4), this technique has been responsible for much of the extant HONO data (refs 5-7 and citations therein). Rodgers and Davis (8)described the limitations of the DOAS approach and proposed a sensitive method involving the photofragmentation of HONO and laser-induced fluorescence of the excited NO thus formed. Routine utility of this technique is deterred by the sizable instrument cost. Alkaline adsorbent coated diffusion denuders provide an affordable alternative for HONO measurement; this was pioneered by Ferm and Sjodin (9) and since utilized by others, typically in the annular denuder format (see, e.g., refs 10-12). Ambient data on HONO concentrations obtained by this technique have been reported (12-14), and at least in once case (12),the nighttime results have been shown to agree with DOAS measurements. These measurements can, however, be subject to error from artifact production of NO2- from other atmospheric N compounds mediated by an alkaline denuder surface (14,15) or from the 0,-induced oxidative loss of the collected NOf (16). Also, the technique is labor-intensive, is difficultly automated, and provides relatively poor temporal resolution. Although one automated approach based on serial tungstic acid and potassium iron oxide coated denuders has been reported for the determination of HN02, HNO,, NO, and 'Permanent address: Institute of Analytical Chemistry, Czechoslovak Academy of Sciences, Leninova 82,611 42 Brno, Czechoslovakia. 0013-936X/91/0925-0255$02.50/0
NOz (17), the ability of tungstic acid denuders to make selective measurements may be questionable (18). Integral to the measurement of HONO is the problem of calibration. An early approach involves reacting NaN02 and dilute H2S04in solution and taking the headspace vapor, a mixture of NO, NOz, and HONO (1). Although the limitations are obvious, variations of this method have been used for calibration (8,9). A continuous flow/carrier gas sparging approach using the same chemistry has been reported very recently (19) to result in a stable HONO source in which NO and NO2 concentration are substantially reduced. However, a number of parameters, e.g., gas and liquid flow rates, acid and nitrite concentrations, etc., must be carefully controlled. Previously, Braman and de la Cantera (20)described a simple sublimation source for HONO in which oxalic acid, sublimed onto NaN02(s) produces HONO(g) by displacement. The relative humidity (RH) must be maintained within a controlled range: at high humidities, the deliquescence of oxalic acid is a problem; at low humidities, too much of the acid sublimes and generates NO by a redox reaction. We have failed to obtain a stable HONO(g) output from this source in our laboratories over prolonged periods. In this paper, we report a diffusion scrubber based (21, 22) inexpensive automated instrument for the measurement of HONO(g) and a calibration source based on NH4N02(s)at thermodynamic equilibrium with NH3(g) and HONO(g). Experimental Section Reagents. Ammonium nitrite was purchased as a 18% w/v aqueous solution from Pfaltz and Bauer (Stamford, CT). The solution (ca. 80 mL), put in a flask equipped for vacuum aspiration, was prefrozen by immersion in a dry icelacetone bath and the solvent was then removed by freeze-drying. The solid NHINOz thus obtained conas determined by suppressed tained -5 w t % ",NO3 ion chromatography (IC) with tandem UV and conductivity detection (vide infra) and was used without further purification. The composition of the solid reagent (in regard to the percentage of nitrate) did not change during storage in a refrigerator for at least 3 months. Sodium hydroxide used as diluent was prepared from 50% NaOH solution (J.T. Baker) and protected from C02 intrusion by soda lime guard tubes during use. Freshly prepared solutions of NaNO, were used for the liquidphase calibration of the IC system. Ndion cation-exchange resin was supplied by Du Pont as a gift in K+ form, this was converted to H+ form before use. Water used as solvent was distilled and deionized and met or exceeded all specifications of ASTM type I water. All other chemicals used were of reagent grade. HONO Generation System. The generation system is shown schematically in Figure la. Cylinder nitrogen gas, controlled by a precision pressure regulator, passed through a critical orifice and a rotameter A through a Mg(C1O4),-fi1leddrying tube B into a thermostated (-10 to +16 "C) bath filled with 25:75 ethanol/water. At the very low Nz flow rates (1-30 mL/min) used in this study, the entrance length of the tube in the bath was sufficient for thermal equilibration. The nitrogen gas then passed
0 1991 American Chemical Society
Environ. Sci. Technol., Vol. 25, No. 2, 1991 255
Figure 1. (a) Generation system. A, rotameter; B, drying tube; C, ",NO,-packed tube in thermostated enclosure; D, dilution chamber; E, mass flow meter; (b) Diffusion scrubber coupled ion chromatography system. DS, diffusion scrubber: M, mass flow controller; P, air pump: N, pressurized water reservoir; V, injection valve; S, peristaltic pump: Q, HPLC pump; K, column; H, suppressor; U, UV detector; J, conductivity detector.
through a 0.3 X 20 cm bed of solid ",NO2 packed in a polytetrafluoroethylene tube C, which is externally shielded by a water-impervious tube to prevent intrusion of water vapor. The passage of the carrier gas through NH4N02(s)generates NH,(g) and HONO(g) in the exit stream. The exact flow rate was measured at the exit point with a digital soap bubble meter. The exit flow proceeded into a 4-L mixing chamber D where it was diluted by pure dry dilution air (filtered through silica gel, molecular sieves, and activated carbon columns) a t flow rates up to 20 L/ min metered by mass flowmeter E. The diluted stream was sampled by a bubbler or the diffusion scrubber (vide infra) with the balance being vented outside the laboratory. Source Calibration and Purity Determination. The diluted source output was calibrated by sampling with a fritted glass midget bubbler containing 15 mL of water or 0.03-0.10 M NaOH as absorber. After being sampled (typically 1 h at 30 mL/min), the solution was diluted to a known volume and NO2- and NO3- were measured by IC. A t the concentration levels and sampling periods used, no difference in absorber efficiency between water and NaOH was noted. Additionally, by serial bubbler experiments, the capture efficiency of the first bubbler was shown to be essentially quantitative. The purity of the generated gas was assessed also by determining the NH4+collected in the water absorber by the reaction of the sample with o-phthaldialdehyde and sodium sulfite and measuring the resulting fluorescence (23). Further, spectroscopic measurements were made for HONO(g) in equilibrium over NH,NO,(s) using a 5-cm path-length cell and a PerkinElmer Model 559A spectrophotometer equipped with a double monochromator. Diffusion Scrubber Coupled Ion Chromatograph. The diffusion scrubber coupled ion chromatography system (DS/IC) is shown schematically in Figure lb. This is largely similar to the configuration used by Lindgren and Dasgupta (24), except as noted below. Ambient air, calibration gas, or zero air was drawn through the DS inlet by a sampling pump P and a mass flow controller M at a rate of 0.5-2.1 L/min, 1.5-2.1 L/min being used for ambient measurements. The DS was composed of a glass256
Environ. Sci. Technol., Vol. 25, No. 2, 1991
jacketed 6-mm-i.d. fluorinated ethylene propylene copolymer tubing 60 cm in length inside which 1.0 m of a microporous membrane tubing (Celgard X-20,400-pm i.d., 25-pm wall thickness, 40% surface porosity, 0.02-pm mean pore size; Hoechst-Celanese, Charlotte, NC) was suspended concentrically as a 3 X 400 mm coil. A Nylon monofilament (6-lb strength fishing line, 275-pm diameter) was inserted within the membrane tube and the assembly coiled and thermoset as described for Nafion membrane tubes (25). Initial testing established that at the low aqueous concentrations (ppb levels) and small residence times involved, no loss of HNOz or HNO, occurs on monofilament Nylon surfaces. The scrubber liquid was water and the flow was maintained countercurrent to the sample air flow. Components in the sample gas diffuse through the membrane into the scrubber liquid flowing through the DS (22, 24). The scrubber liquid was contained in a pneumatically pressurized (-2 psi) bottle N and was aspirated through the sample loop (120 pL) of the chromatographic injection valve V by peristaltic pump S at a flow rate of 16 pL/min. The valve V is shown in its normal, loading configuration; the valve switches to the inject position for 1 min in an automatically repeated cycle period ranging from 7 to 15 min. Chromatography was conducted on a Dionex AS-4A column K with 1 2 mM NaOH being pumped through it at 0.6 mL/min by a Beckman Model llOA pump Q. The column effluent flowed through an 0.4 X 1000 mm filament-filled Nafion membrane helical suppressor H externally packed with cation-exchange resin, with 10 mM H2S04at 1mL/min being used as regenerant. The suppressor effluent flowed through an absorbance detector U (Model 757, Kratos/ Schoeffel, Applied Biosystems, Ramsey, NJ) of 8-mm path length set at 220 nm and then through a conductivity detector (ion chromatography module, Wescan Instruments, Santa Clara, CA). The heat exchanger associated with the UV detector was removed to avoid excessive back-pressure to the suppressor. UV detection is significantly more sensitive for the determination of NO,-; we find, however, that the presence of a suppressor significantly reduces background noise and baseline absorbance, thus improving detection limits. Conductivity detection provides a good adjunct to UV detection when HONO levels are sufficiently large to obtain acceptable signal to noise (S/N). Under these conditions, the UV/conductivity response ratio provides a further confirmation of the NO,peak identification beyond retention indexes. Unless otherwise stated, the reported data are based on the UV detector output. The DS membrane was periodically cleaned with methanol as previously described (22).
-
Results and Discussion Purity and Stability of the Source Output. HONO decomposes according to 2HONO F+ NO + NO2 + HZO (1) The equilibrium constant for reaction 1, based on the free energy data available (27), ranges from 2.3 to 25 atm at 25 "C for trans- and cis-HONO, respectively. Even for a system saturated with water vapor (0.031 atm at 25 "C), and assuming the more favorable case of trans-HONO, a total nitrogen species concentration of 1 ppmv will result in only 8% of the total N being present as HONO at equilibrium, The fraction present as HONO will decrease further at lower concentrations. Initially we attempted to use partially humidified (25-40% RH) N2 as the carrier gas to limit the decomposition of HONO according to reaction 1. Such attempts led to unsatisfactory results-the source output gradually
I
300
I
350
400
Wavelength. nm
Figure 2. Near-UV absorptlon spectrum of HONO over NH,NO,(s).
decreased, and even at the relatively low RH of the carrier gas used, ",NOz deliquescence resulted in a visibly moist bed, with occasional droplets. Thoroughly dry carrier gas was used for all further experiments. Fortunately, reaction 1is slow, albeit it may be subject to heterogeneous surface catalysis (27). The extent to which HONO decomposition occurred in our system was studied and is reported below. In bubbling mixtures of sub-ppmv levels of NO and NO2 through a water absorber, we observed very poor conversion to HONO, as determined by measuring NOz- in the resulting solution by the Griess-Saltzman (G-S) method (28). Based on published kinetic data (29), it may also be expected that a water bubbler will capture an insignificant fraction of NOz. The following gaseous species may exist in the source output: ",NOz, NH,, HONO, NO, NOz, and HzO, the last three originating from the decomposition of HONO. The source output was sampled through a NazC03-impregnatedfilter (29), which removes the bulk of the HONO(g), followed by two sequential bubblers, the first containing water to remove the remaining HONO(g) and the second filled with G-S reagent which measures NOz. The filter extract, as well as the bubbler contents, was measured by the same spectrophotometric procedure, after postsampling addition of the G S reagent to the filter extract and the contents of the first bubbler. The relative amounts of NOz- found in the filter and the two bubblers were 42:2:1, suggesting only small amounts of decomposition products in the HONO generated. A second experiment involved sampling the source output with a water absorber and determining both NOzand NH4+in it. The excess of NH4+over NOz- measures the extent of HONO decomposition. Within the combined precision of the two analytical methods (f3%), there was no difference between the NH4+ and the NOz- concentrations. The absorption spectrum in the 300-400-nm region of the headspace over solid ",NOz is shown in Figure 2. There was no significant structured absorption in the 400-500-nm region, indicating the absence of measurable quantities of NOz. Further, rapidly thermally cycling the cell caused the absorption bands to increase and decrease reversibly with temperature, indicating that reaction 1 did not significantly proceed within the limited observation period of this experiment (several minutes). Figure 2 represents a UV absorption spectra of HONO(g) in the demonstrated absence of NO and NOz and for which, therefore, no correction is necessary, unlike most previous
studies. The exact positions of the four principal absorption bands are 342.2, 353.9, 367.9 and 385.4 nm, respectively, as calibrated in reference to a Ho203filter. During calibration of the source output with an aqueous bubbler and IC analysis of the absorber, some amount of NO, (up to 5% of the NOz-, depending on sampling time) was observed. However, we believe this to be an artifact due to the oxidation of the collected NOz- to NO3-. The available data for the NH4N03(s)* NH3(g) + HNOJg) equilibrium (30) show that the NH,NO,(s) present in the NH4NOZ(s)cannot possibly result in this amount of HN03(g). Further, with the DS/IC instrument, which utilizes a much shorter liquid residence time than bubbler sampling, no NO3- peak was found. If present at concentrations indicated by bubbler sampling, the H N 0 3 would have been easily detectable even after accounting for the potentially lower collection efficiency of the DS for HNO, vs that for HONO. The question of whether NH,NOz(g) exists in equilibrium with NH3(g) and HONO(g) is also relevant in view of the work of de Kruif (31),in which the gas-phase dissociation constants for various NH,X species (X- = F,Cl-, Br-, I-, CNS-, NO3-, HC03-) were measured. For the HX(g), the equilibrium process NH4X(g) + NH,(g) constants are reported to range from 4.84 X lo-' atm for NH41 to 5.97 X atm for NH,F. In our experiments, attempts to avoid the presence of undissociated gaseous ",NOz by removing the NH, with a bed of acid-form Nafion (a perfluorinated cation-exchange resin in H+form) immediately following the ",NOz bed were unsuccessful in that the concentration of HONO, as measured by the DS/IC system, actually decreased and was unstable over long periods. This avenue was not further pursued. While we cannot offer unequivocal evidence regarding the significant absence of NH,NO,(g), for the "high concentration level of 20 ppbv HONO(g) used for calibrating the DS/IC system, even the lowest dissociation constant observed by de Kruif will result in -4% of the total gaseous nitrite being present as NH4NOZ(g). If the dissociation constant is closer to that for ",NO3 (3 X lo4 atm), 10.7% of the total should be present in the undissociated form. The long-term stability of the source was determined over a 15-day period at 5 "C; the source emitted 17.5 f 0.8 ng of HONO/mL.min Nz ( n = 29). The output concentration was independent of carrier gas flow rate in the range of 5-30 mL/min, indicating that complete equilibration occurred. A large portion of the long-term variability of the output (rsd 4.6%) can be ascribed to thermal variations. Measurement of the bath temperature with a recording thermometer indicated a temperature stability of f0.2 "C over this period. On the basis of the measured temperature dependence of the output (vide infra), at 5 "C this can be responsible for an output variation of f4%. Short-term stability is substantially better, the output of the complete DS/IC system over 3-h-long periods at sampled concentrations 1 10 ppbv show a typical relative standard deviation better than 1%.The performance of the system a t low calibrant HONO concentrations along with low daytime ambient levels is shown in Figure 3. Thermodynamic Characterization. The concentration of HONO(g) over NH4NOZ(s)is shown in Table I as a function of temperature. A Clausius-Clapeyron plot of these data, weighted on the basis of the number of measurements, yields the best fit linear equation ( r = 0.996, n = 45): In (pHONO, atm) = -(go18 f 84)/T + 16.81 f 0.30 (2) The mean AH for the process NH,NO2(s) + NH,(g)
+
+
Environ. Sci. Technol., Vol. 25, No. 2, 1991
257
I I
;\ E
20 min
1260
kL II
850
20 min
440
Flgure 3. Typical system output for calibrant and low ambient levels. Triplicate or quadruplicate repeated measurements are presented. The peak with a dot over It is NOz-; the peak immediately to the left of it is due to COP.
Table I. Vapor Pressure of Gaseous Nitrous Acid overSolid Ammonium Nitrite as a Function of Temperature
temp, O C -7.4 -3.9 1.5 2.2 2.3
PHONO, ppmv
(SD)
temp, "C
1.50 (0.05) 2.28 (0.14) 4.15 (0.28) 4.52 (0.15) 4.77 (0.11)
5.0 10.4 15.4 15.8
PHONO, PPmv
(iSD)
6.24 (0.28) 10.65 (0.11) 14.88 (0.00) 16.86 (0.26)
HONO(g) within the temperature range studied is computed from eq 2 to be 133.2 f 1.4 kJ/mol, a t a weighted mean temperature of 277 K in comparison with the AHo value of 131.8-133.9 kJ/mol at 298 K that can be obtained from data in standard compilations (26),depending on whether the HONO formed is assumed to be in the cis or the trans form. Neglecting changes in heat capacities, we compute PHONO over NH4N02a t 25 OC to be 4.18 X atm. In comparison to pm0 over NH4N03(31),this value is 4 orders of magnitude higher. There is little likelihood that the NH4N02(s)+ NH3(g) + HONO(g) equilibrium plays any role in the chemistry of the ambient atmosphere. The AGf" value for NH4N02(s)is not available in the literature; a value of 109.8 f 2.89 kJ/mol can be estimated from our data, assuming an equilibrium distribution of cisand trans-HONO. Instrument Performance. The coiled membrane geometry of the DS was developed for the present application to permit the placement of a greater length of the membrane in the DS to improve the collection efficiency without a proportionate increase in the actual device length. This is particularly attractive in the present case, because significant concentrations of particulate NOz- are unlikely to exist. Collection efficiency (f) of the present DS was determined (from a knowledge of the source output and direct calibration of the chromatographic system with aqueous NOz-) to be 0.842, 0.544, 0.384, and 0.303 at respective sampling rates (Q)of 0.52, 1.01, 1.53, and 2.10 L/min. The device exhibits Gormley-Kennedy behavior (32)in that a plot of In (1 - f ) vs 1 / Q shows good linearity (r = 0.997). For a straight membrane device of the same overall length, typically a 40-cm length of membrane can 258
Environ. Sci. Technol., Vol. 25, No. 2, 1991
be utilized. As a comparison with the present DS, the collection efficiency for the linear membrane device for H,02, a gas with a Graham's law diffusion coefficient 1.37 times that of HONO, was reported to be 0.78,0.47,0.37, and 0.28 at respective sampling rates of 0.50, 1.00, 1.50, and 2.00 L/min (22). The increase in collection efficiency as a result of the increased membrane length is evident. The DS/IC system response was calibrated and checked on a routine basis, the response was linear (r = 0.999) up to a sampled concentration of at least 55 ppbv with a calibration slope reproducibility of better than f5% from day to day. The lowest calibrant concentration studied was 200 pptv; dilutions required to obtain lower concentrations were impractically large for the present experimental system. The Y intercept of the calibration plot at low concentrations varied day to day from negative to positive values, both relatively small. This seemed largely to be a function of whether calibration was performed in ascending or descending concentration order and suggested a small memory effect in the generation system for rapid changeover of concentrations. For interpretation of ambient data below the lowest calibrant concentration, a linear response behavior between this concentration and zero concentration (zero response observed with zero air) was assumed. Based on a S / N criterion of 3, the liquid-phase limit of detection (LOD) was 1.0 pg of NOz-/L, corresponding to an observed noise level of 8 X absorbance units (AU). With zero air flowing through the DS, the mean noise level of the chromatographic baseline was essentially the same. This corresponds, under the condition of the scrubber liquid flow rate and the injection volume stated in the Experimental Section, to calculated gas-phase LODs of 24, 19, 17, and 16 pptv for a 7-min cycle (6-min sample load time) a t sampling rates of 0.5, 1.0, 1.5, and 2.1 L/min, respectively. However, since an actual study of the response behavior at these concentrations was not possible, actually attainable LODs may be higher. Nevertheless, the signal for even the lowest ambient concentration reported herein was much higher than the system noise level, as is evident in Figure 3; we did not pursue further refinements. The memory characteristics of the DS/IC system was measured by applying a step function (calibrant to zero gas or zero gas to calibrant) at the gas-phase input end as the chromatographic valve switched to the inject mode. In the step-up mode, 86%, 98%, and 100% of the steady-state calibrant response was reached by the first, second, and third sampling cycles; the corresponding values for the step-down cycle were 59%, 12%, and 5% (and -0% for the fourth cycle) of the calibrant response, respectively. These data also show that, quite unlike HNO,, HONO is not especially susceptible to slow adsorption/ desorption on the inlet manifold surfaces. Potential interference from NO2 and NO was checked by sampling these gases generated from a permeation tube and a certified cylinder blend (balance N2), respectively. Any HONO present in the generated NO, was removed by a NazC03-impregnatedfilter, the exact NO, concentration sampled was verified both by the weight loss of the permeation source and by the G-S method (28). Over a concentration range of 0-350 ppbv NO,, the slope of a plot of the NO2- peak height vs [N02(g)]was 0.022% of the corresponding calibration slope for HONO. With the addition of a constant amount of 330 ppbv NO to varying concentrations of NO,, this slope increased to 0.027% of the value for HONO; the difference, however, is barely significant relative to experimental uncertainties. Overall,
'ooolIi B
m
800
>
4
Q Q
o-600 Z
0 I
40 1y, ,1,y ;.
200
7
,,,,, , , , , ,
115
116
,,,,,
y,,;,y,,81,7:,
117
Julian Day, 1990
118
Flgure 4. Ambient data for HONO over a 4-day period with 12-min resolution. M and N indicate midnight and noon, respectively.
no major interference from NOz and NO is expected. Regarding effects of sample humidity, although the calibration source output is very dry, the evaporation of water through the membrane pores makes the humidity of the actual stream sampled quite significant. In one set of experiments, partially humidified rather than dry air was used for diluting the source output; no differences were evident. Ambient Measurements. Several multiday measurements have been made with the DS/IC instrument, with periodic calibration during each experiment. The results for one 4-day measurement period are shown in Figure 4 as an example. The sampling point was located ca. 15 m aboveground with two major parking lots within 150 m. The data shown in Figure 4 were obtained with 12-min resolution. Few detailed HONO data with reasonably short temporal resolution are available for comparison. The first notable aspect of the data in Figure 4 is that although there is a clear diurnal pattern, the daytime HONO concentrations are not zero. While this has also been observed by other investigators (12-14), a viable, sufficiently fast mechanism for the daytime formation of HONO that allows its persistent presence at low levels is still to emerge. One feature we observe is a weekday morning peak of HONO shortly after the peak morning traffic hour. The concentration decays rapidly and minimum daytime concentrations can be as low as -100 pptv. While low, the persistent daytime concentrations are likely not measurement artifacts-the ambient HONO signal is removed by a NazC03-impregnated filter with the same efficiency with which it acts on the HONO calibrant. The morning peak feature, albeit less pronounced, is also apparent in the DOAS data obtained at Glendora, CA (33). Our results indicate that at the present location, the occurrence of the morning peak is a function of other meteorological variables; e.g., the peak is much less pronounced on day 117, a windy day (and even less so for day 118, a Saturday). We have not observed HONO corresponding to the evening rush hour. In the absence of washout by hydrometeors, evening peaks (as high as 1300 pptv) are reached 2-3 h after sundown and then fall again or fluctuate; a second peak during the night hour (see days 115/ 116) is not uncommon. The particularly intriguing aspect of the nighttime measurements is that the HONO
levels frequently start decreasing before sunrise. The DOAS data from Glendora in Urban Los Angeles (33) represents the impact of far more numerous sources, both mobile and stationary. The peak HONO concentrations range up to 4.5 ppbv at night and during the day rapidly drop to below the LOD (800 pptv) following the previously mentioned morning peak. It is not possible to determine from these data (33) if there is any clear pattern showing a decrease in the HONO concentration before sunrise. When Lubbock, TX, is compared with Glendora, CA, it is useful to note that the difference in the NO, concentration is actually far higher than the difference in the peak HONO levels. It is logical to assume that some significant nonphotolytic loss mechanism must be operative. While it is obviously impossible to deduce what major pathways are involved in the daytime formation and nighttime production and loss of HONO solely from ambient measurements at isolated locations, we make the following observations. Despite the thermodynamic improbability, the presence of HONO in automotive exhaust has been unequivocally demonstraed (34) and the correlation of morning HONO peaks with morning traffic is also clear. Approximate calculations suggest, however, that if the concentration levels of primary HONO emissions in vehicular exhaust reported (34) are taken to be typical, they fall short of accounting for the peak morning levels. Taking into account the absence of corresponding evening peaks, we propose that significant accounts of HONO are also formed by secondary reactions involving vehicular exhaust that are photomediated. Although laboratory experiments suggest that hydrolytic disproportionation reactions of NOz with liquid water to produce HONO and HNO, are relatively slow (29),a variety of potential hydrogen donors can dramatically improve the rate of production of HONO-this is indeed the basis of the G-S method for the determination of NOz (28). Significant conversion rates of NOz to HONO, for example, have been observed in simulated indoor environments (3). Further, although the heterogeneous reaction of NO, with aqueous S(1V) has been studied ( 3 5 ) ,its implications toward the formation of HONO have not been considered. This reaction produces both H+ and NOz-, and the latter can easily outgas from solution as HONO as the pH of the droplet falls. Mechanistically, this reaction is not fully understood and metastable intermediates may be involved. HONO itself can react with S(1V) in a sequence of reactions to produce a variety of products (36). In the presence of sufficient liquid water, as may occur in early morning/late night hours due to increased relative humidity, this loss mechanism may become important. Acknowledgments
We thank H. W. Biermann, California Department of Food and Agriculture, for a preview of the 1989 DOAS HONO measurement data. Literature Cited (1) Cox, R. A.
J . Photochem. 1974, 3, 291-304.
(2) Sjodin, A,; Ferm, M. Atmos. Environ. 1986, 20, 409-411. (3) Atkinson, R.; Carter, W. P. L.; Pitts, J. N., Jr.; Winer, A. M. Atmos. Enoiron. 1986, 20, 408-409. (4) Perner, D.; Platt, U. Geophys. Res. Lett. 1979,6,917-920. (5) Harris, G. W.; Carter, W. P. L.; Winer, A. M.; Pitts, J. N., Jr.; Platt, U.; Perner, D. Environ. Sci. Technol. 1982, 16, 4 14-4 19.
( 6 ) Kessler, C.; Platt, U. Proc. 3rd Eur. Symp. Physico-Chem.
Behavior Atmos. Pollut. 1984, 412-422. (7) Pitts, J. N., Jr.; Biermann, H. W.; Atkinson, R.; Winer, A. M. Geophys. Res. Lett. 1984, 11, 557-560. Environ. Sci. Technol., Vol. 25, No. 2, 1991 259
Environ. Sci. Technol. 1991, 25, 260-267
Rodgers, M. 0.;Davis, D. D. Enuiron. Sci. Technol. 1989, 23, 1106-1112. Ferm, M.; Sjodin, A. Atmos. Enuiron. 1985, 19, 979-983. Allegrini, I.; De Santis, F.; Di Palo, V.; Febo, A,; Perrino, C.; Possanzini, M. Sci. Total Enuiron. 1987, 67, 1-16. Brauer, M.; Koutrakis, P.; Spengler, J. D. Enuiron. Sci. Technol. 1989,23, 1408-1412. Appel, B. R.; Winer, A. M.; Tokiwa, Y.; Biermann, H. W. Atmos. Enuiron. 1990, 24A, 611-616. Sjodin, A; Ferm, M. Atmos. Enuiron. 1985, 19, 985-992. Sjodin, A. Enuiron. Sci. Technol. 1988, 22, 1086-1089. Penkett, S. A.; Sandalls, F. J.; Jones, B. M. R. VDI-Ber. 1977, NO.270, 47-54. Sickles, J. E., 11; Hodson, L. L. Atmos. Enuiron. 1989,23, 2321-2324. Braman, R. S.; de La Cantera, M. A.; Han, Q. X. Anal. Chem. 1986,58, 1537-1541. Appel, B. R.; Tokiwa, Y.; Kothny, E. L.; Wu, R.; Povard, V. Atmos. Enuiron. 1988, 22, 1565-1573. Taira, M.; Kanda, Y. Anal. Chem. 1990,62, 630-633. Braman, R. S.;de La Cantera, M. A. Anal. Chem. 1986,58, 1533-1537. Dasgupta, P. K. Atmos. Enuiron. 1984, 18, 1593-1599. Dasgupta, P. K.; Dong, S.; Hwang, H.; Yang, H.-C.; Genfa, 2. Atmos. Enuiron. 1988, 22, 949-964. Genfa, Z.; Dasgupta, P. K. Anal. Chem. 1989,61,408-412. Lindgren, P. F.; Dasgupta, P. K. Anal. Chem. 1989, 61, 19-24. Dasgupta, P. K. Anal. Chem. 1984,56, 96-103.
Dean, J. A., Ed. Lunge’s Handbook of Chemistry, 12th ed., McGraw-Hill: New York, 1979. Kaiser, E. W.; Wu, C. H. J . Phys. Chem. 1977, 81, 1701-1706. Lodge, J. P., Jr., Ed. Methods for Air Sampling and Analysis; Lewis: Chelsea, MI, 1989; pp 389-393. Lee, Y.-N.; Schwartz, S. E. J. Phys. Chem. 1981,85,840-848. Stelson, A. W.; Friedlander, S. K.; Seinfeld, J. H. Atmos. Environ. 1979, 13, 369-371. de Kruif, C. G. J. Chem. Phys. 1982, 77,6247-6250. Gormley, P. G.; Kennedy, M. Proc. Ir. R. Acad. Sci. Sect. A 1949, 52, 163-169. Atkinson, R.; Winer, A. M. Measurements of NOz, HONO, HCHO, PAH, Nitroarenes and Particulate Mutagenic Activities During the Carbonaceous Species Methods Comparison Study. Final Report, California Air Resources Board, Sacramento, CA, February 1988. Pitts, J. N., Jr.; Biermann, H. W.; Winer, A. M.; Tuazon, E. C. Atmos. Enuiron. 1984, 18, 847-854. Clifton, C. L.; Alstein, N.; Huie, R. E. Enuiron. Sci. Technol. 1988, 22, 586-589. Oblath, S. B.; Markowitz, S. S.; Novakov, T.; Chang, S. G. J . Phys. Chem. 1982,86, 4853-4858.
Received for review May 18,1990. Revised manuscript received August 30,1990. Accepted September 18,1990. This work was supported by the Electric Power Research Institute through Grant RP1630-55 and by the State of Texas Advanced Research Program.
Measuring Chlorinated Hydrocarbons in Combustion by Use of Fourier Transform Infrared Spectroscopy Matthew J. Hall,+ Donald Lucas,t and Catherine P. Koshland*g* Departments of Mechanical Engineering and Biomedical and Environmental Health Sciences, University of California, Berkeley, and Applied Science Division, Lawrence Berkeley Laboratory, Berkeley, California 94720
The quantitative measurement of products of incineration of chlorinated hydrocarbons using Fourier transform infrared (FTIR) spectrometery was evaluated during a study of the fundamental processes of toxic waste incineration. Combustion products were sampled from a turbulent flow reactor into which chlorinated hydrocarbons were injected. Chlorinated species were detected mainly in the C-C1 stretching region, away from H 2 0 and C 0 2 interferences. Absorbances of species having rotational lines narrower than the instrument resolution, such as CO, C02,and HCI, were nonlinear under most conditions. The measured absorbance of these species was very sensitive to total pressure, increasing with an increase in total pressure; at very low optical densities, however, absorbance became independent of total pressure. Thus, while FTIR spectroscopy has many features that make it attractive for measuring combustion species, care must be taken to ensure the accuracy of quantitative measurements. Introduction
Infrared absorption spectroscopy has long been used for molecular identification and quantification in a variety of chemical systems. Most molecules, with the exception of homonuclear diatomics such as N2,02,and CI,, absorb light t Department of Mechanical Engineering, University of California, Berkeley. Lawrence Berkeley Laboratory. I Department of Biomedical and Environmental Health Sciences, University of California, Berkeley. f
260
Environ. Sci. Technol., Vol. 25, No. 2, 1991
in the mid-infrared region (500-4000 cm-l). The development of Fourier transform infrared (FTIR) spectrometry in the 1970s greatly increased the speed, resolution, and sensitivity of measurements ( I ) . Coupled to long path length cells, FTIR spectroscopy has provided a wealth of information about species at low concentrations, especially in the study of atmospheric phenomena (2-4). For many species the limit of detection is in the sub-ppm range ( 5 ) . FTIR spectroscopy has not been used extensively in combustion. One major problem is the presence of large quantities of water and carbon dioxide, molecules with numerous infrared absorption features. There have been some studies using FTIR spectroscopy for gas analysis (3, 6-11) and several for temperature measurements (8,9,12, 13). In addition, there has been research on the use of tunable diode laser spectroscopy for combustion products (14), and nondispersive infrared analyzers (NDIR) for single-species detection have been used extensively. In studying the combustion of chlorinated hydrocarbons, we are using FTIR spectroscopy as our principal diagnostic technique. At the present time there are no simple methods for detecting chlorinated hydrocarbons during their incineration; several methods are currently being investigated (15,16). Infrared analysis is attractive because the C-C1 stretching region is from 500 to 750 cm-’ (17), a region relatively free from water and carbon dioxide interferences, and the spectra of HCI and other simple chlorine-containing species are well-known (18). The spectrometer has a large dynamic range, allowing quantification from the sub-ppm range to the percent range,
0013-936X/91/0925-0260$02.50/0
0 1991 American Chemical Society