Indoor, outdoor, and regional summer and winter concentrations of PM10, PM2.5, SO4(2)-, H+, NH4+, NO3-, NH3, and nitrous acid in homes with and without kerosene space heaters.

Twenty-four-hour samples of PM10 (mass of particles with aerodynamic diameter < or = 10 microm), PM2.5, (mass of particles with aerodynamic diameter < or = 2.5 microm), particle strong acidity (H+), sulfate (SO42-), nitrate (NO3-), ammonia (NH3), nitrous acid (HONO), and sulfur dioxide were collected inside and outside of 281 homes during winter and summer periods. Measurements were also conducted during summer periods at a regional site. A total of 58 homes of nonsmokers were sampled during the summer periods and 223 homes were sampled during the winter periods. Seventy-four of the homes sampled during the winter reported the use of a kerosene heater. All homes sampled in the summer were located in southwest Virginia. All but 20 homes sampled in the winter were also located in southwest Virginia; the remainder of the homes were located in Connecticut. For homes without tobacco combustion, the regional air monitoring site (Vinton, VA) appeared to provide a reasonable estimate of concentrations of PM2.5 and SO42- during summer months outside and inside homes within the region, even when a substantial number of the homes used air conditioning. Average indoor/outdoor ratios for PM2.5 and SO42- during the summer period were 1.03 +/- 0.71 and 0.74 +/- 0.53, respectively. The indoor/outdoor mean ratio for sulfate suggests that on average approximately 75% of the fine aerosol indoors during the summer is associated with outdoor sources. Kerosene heater use during the winter months, in the absence of tobacco combustion, results in substantial increases in indoor concentrations of PM2.5, SO42-, and possibly H+, as compared to homes without kerosene heaters. During their use, we estimated that kerosene heaters added, on average, approximately 40 microg/m3 of PM2.5 and 15 microg/m3 of SO42- to background residential levels of 18 and 2 microg/m3, respectively. Results from using sulfuric acid-doped Teflon (E.I. Du Pont de Nemours & Co., Wilmington, DE) filters in homes with kerosene heaters suggest that acid particle concentrations may be substantially higher than those measured because of acid neutralization by ammonia. During the summer and winter periods indoor concentrations of ammonia are an order of magnitude higher indoors than outdoors and appear to result in lower indoor acid particle concentrations. Nitrous acid levels are higher indoors than outdoors during both winter and summer and are substantially higher in homes with unvented combustion sources.

hbttp://ehpnetl. niehs. niggov/docs/I999/107p223-231leadeer /abstracthtmI There is an increasing body of epidemiologic evidence which suggests that exposures to short-term ambient levels of suspended particles are associated with adverse health effects. The effects range from changes in respiratory function and symptoms and exacerbation of respiratory disease to excesses in daily mortality (1). Several studies have suggested that particles less than 10 pm in diameter (PM10), particles less than 2.5 pm in diameter (PM2 5), and the sulfate or strong acid aerosol component of the ambient aerosol are implicated in the observed particle/effect associations (1).
Exposures to particulate matter occur in a variety of microenvironments (outdoors, residences, public buildings, etc.). Because outdoor concentrations can vary considerably in time and space and indoor aerosol concentrations are associated with both indoor and outdoor sources, particle mass concentrations must be measured for different microenvironments. Size and chemical composition of the aerosols are also important. Altogether, this exposure assessment information serves the needs of epidemiologic studies, risk assessment evaluations, and the development ofmitigation strategies.
As part of a prospective epidemiologic investigation of the nature of an association between particulate exposures and daily reported (over a 1-year period) respiratory symptoms in 918 infants and their mothers, we conducted an extensive exposure assessment study (2). The study protocol employed a nested design that utilized questionnaires and passive and active pollutant monitors. Active monitoring consisted of measuring particle and gaseous species both indoors and outdoors at residences as a function of season and indoor sources. Outdoor central site daily monitoring was also conducted during the summer months. In this paper we report on the following measurements: PM10; PM2.5; particle sulfate (S042-), nitrate (NO;3), ammonium (NH4t), and strong acidity (H+); and gaseous S02, nitrous acid (HONO), and NH3. Measurements were made indoors and outdoors at residences in Connecticut and southwest Virginia and at a central outdoor regional site in southwest Virginia. Indoor/outdoor/central site comparisons by indoor source and season for particle size and chemical composition are presented and discussed.

Methods
Sites and residence selection. Twenty-fourhour particle sampling was conducted at 20 residences in Connecticut between August 1994 and June 1995 and at 261 homes in southwest and central Virginia between July 1995 and January 1998. Sampling was conducted as part of a prospective epidemiologic study of the respiratory effects on infants and their mothers from indoor exposures to vapor and particle phase acids associated with kerosene heater use (2). Sampling was conducted in 58 residences during the summer seasons and in 223 residences during the winter seasons. Air-conditioning use during the summer period was reported in 49 of the residences, with 21 reporting the presence of a gas cooking stove. Kerosene heaters were used in 74 residences during the winter period, whereas 61 reported use of a gas cooking stove. The epidemiologic study design excluded all homes where tobacco combustion occurred, so tobacco smoke was not a source in any of the homes actively sampled.
Central site ambient sampling was conducted in Vinton, Virginia. This site, located approximately 6 km east of Roanoke, Virginia, was selected to represent regional air quality. Twenty-four-hour partide samples were collected during the period from 15 May through 15 September for 1995 and 1996 for comparison with daily summer respiratory symptoms in the infants and their mothers (3) and for comparison with twice-daily peak flow measurements recorded by the mothers over a 2-week period during the summer (4). In this paper 50 days of data from the Vinton site are used, corresponding to days for which indoor or outdoor partide sampling was conducted at residences in the region. Distances from the Vinton site to residences monitored varied from 1 to >175 km, with an average distance of 96 km.
Measurements. Samples for PMIO and PM2.5 were collected using inertial impactor samplers. These impactors collect particles with aerodynamic diameters <10 and .2.5 pm at flows of 4 and 10 liters per minute, respectively. The filters used for partide sampling were equilibrated for 48 hr at a temperature of 23 ± 3°C and relative humidity of 40 ± 5% before determining pre-and postsampling weights. Partide sulfate, nitrate, strong acidity, and ammonium, and gaseous species nitrous acid, nitric acid, ammonia, and sulfur dioxide were measured using the Harvard glass honeycomb denuder/filter pack sampler (HDS; Ogawa & Co., USA, Pompano Beach, FL). The HDS sampler (5,6) consists of an impactor to remove coarse particles (>2.1 pm in diameter) from the air samples, two glass honeycomb denuders, and a three-stage filter pack to collect fine partides. The denuder system draws air at a sampling rate of 10 l/min. Air travels first through the inlet section of the sampler, where an acceleration jet directs the air stream onto a sintered stainless steel impactor plate coated with mineral oil, which removes partides >2.1 pm. The air then passes through a transition section that provides a uniform flow through the honeycomb denuders. The first honeycomb denuder is coated with sodium carbonate/glycerol to collect gaseous nitric acid, nitrous acid, and sulfur dioxide. The second honeycomb denuder is coated with citric acid/glycerol to collect ammonia. Fine particles are collected from the air stream leaving the denuders on a Teflon (E.I. Du Pont de Nemours & Co., Wilmington, DE) filter in the front of the filter pack. Sodium carbonate-coated and citric acid-coated glass fiber filters were used downstream of the Teflon filter to collect acidic gases and ammonia volatilized from the collected fine partides.
The concentration of aerosol acidity was determined from pH analysis of the Teflon filter extract. The denuder and filters were extracted and analyzed by ion chromatography to determine both gaseous (sulfur dioxide, nitrous acid, nitric acid, and ammonia) and particle (sulfate, nitrate, nitrite, and ammonium) components.
The sampling preparation, chemical analysis, and quality assurance procedures used in this study are described in detail elsewhere (5)(6)(7)(8). Limits of detection (LOD) for PM2 5 and PM10 samples, which were estimated to equal three times the root mean square error of the blank filter measurements, were 3.6 and 3.3 pg/m3, respectively. These values are similar to those found in a Philadelphia-based study that used the same sampling methods as Suh et al. (7), where the LOD for PM2.5 and PM10 were 3.4 and 2.8 pg/m3, respectively. Coarse particles (2.5<d < 10 pm) mass concentrations were calculated as the difference between measured PM1O and PM2.5 concentrations. Because PM2.5 concentrations cannot by definition exceed PM1O, negative coarse values were set to 0. LOD for chemical determinations from the HDS system were equal to those previously estimated for 24-hr HDS samples, which for so42-, H, and NH3 are 6.0 ilmol/m3, 4.0 nmol/m3, and 0.3 ppb (7).
There was concern that the complex nature and amount of particle and gas phase contaminant emissions from kerosene heaters might introduce interferences in the HDS system, resulting in lower collection efficiencies for gases. It is possible, for example, that semivolatiles including organic acid emissions from the kerosene heaters could deposit on the denuder surfaces, thus blocking the intended gases (i.e., nitrous acid) from reaching the sodium carbonate coating. Also, the citric acid coated denuder could become similarly overloaded or masked, such that ammonia could have passed through the denuder and into the filter pack. In the first case the denuders would have underestimated nitrous acid levels, and in the second case the ammonia that passes through the denuders could neutralize acid aerosol collected on the first filter (Teflon) of the filter pack. In an effort to address these potential interferences our protocol used four honeycomb denuders for indoor sampling (two sodium carbonate coated and two citric acid coated) to minimize the potential for saturation by higher indoor levels of ammonia and other gaseous contaminants. Two honeycomb denuders (one sodium carbonate coated and one citric acid coated) were used for all outdoor sampling at the Vinton site and four denuders were used outside homes. In addition, parallel HDS systems were used with sulfuric acid treated (doped) Teflon filters during winter sampling in 15 homes where kerosene heaters were used and in 20 homes where kerosene heaters were not used. By comparing acid loss on the aciddoped filters collected in the homes with and without the use of kerosene heaters, a qualitative evaluation of the potential for kerosene heater generated acid aerosol could be made.
Samplers in residences were located in the main living area of the home, typically the family room or living room. Outdoor samplers were located within 8 m of the residence and away from any potential sources. Indoor and outdoor samples at residences were collected at a distance of approximately 1 m above the ground or floor. Partide and denuder samplers at the central Vinton site were 1.5 m off the ground. Sampling times for the denuder and particle mass measurements inside and outside residences as well as at the central site were 24-hr samples and were collected over the same time periods. Denuder systems and partide mass samplers were colocated at all sites. Available resources prevented the simultaneous measurement of all particle variables inside and outside at all residences monitored. All parameters were, however, measured daily at the central outdoor site during the summer months. Results Summer concentrations. Mean summertime concentrations of PM10, PM2 5, coarse mass (PM10-PM2.5), 042 H+, NH4 X NO3-, NH3, HONO, and SO2 by location (inside and outside of residences or at the central outdoor site) and by use of air conditioning in the homes are summarized in concentrations of coarse mass were higher than outdoors. S042and H+ were lower indoors, whereas NH3 and HONO concentrations were markedly higher indoors. Concentrations of sulfur dioxide were low at all sites, with outdoor concentrations higher than indoor concentrations. Nitric acid levels were typically at or below the LOD (0.2 ppb). Table 2 shows the results of the statistical analyses of the differences between concentrations measured at different sites for selected particle contaminants shown in Table 1. The analysis is for paired measurements among sites (paired t-test). Because paired samples were obtained for only five homes reporting no air conditioners, these homes were combined with homes reporting the presence of an air conditioner. The correlation coefficients for paired site measurements for those contaminants are also shown in Table 2. Overall, the correlations are low, indicating considerable scatter. PMto concentrations measured at the regional site were not significantly different from those measured either inside or outside of homes, nor were PMIO concentrations measured outside homes different from those measured inside homes. A weak statistically insignificant correlation was seen between outdoor and indoor PM1O concentrations with even weaker correlations for home versus regional site measurements. Although no statistically significant differences were observed for PM2.5 concentrations for any of the three comparisons, significant moderate correlations were observed between PM2 measured at the regional site and outside homes and between PM2.5 concentrations measured inside and outside of homes. Figures 1 and  2 show the regression equation and scatter plot for the correlated PM2,5 comparisons (regional site vs outside homes and inside vs outside homes, with and without outliers). The explained variation in comparisons improved with the elimination of outliers, particularly for the comparison for the comparison of PM2 5 inside and outside homes. No significant correlations among sites were observed for coarse mass, although significant concentration differences were observed for the comparisons of regional site with inside homes and for outside homes with inside homes.
Significant differences (p<0.05) were found in the concentration values for so42and H+ for the regional site versus inside homes and outside versus inside homes, but not for the sulfate comparison between the regional site and outside homes. Significant correlations by site for so42and H+ were found for regional site versus outside homes and for inside versus Abbreviations: SD, standard deviation; PM10, particle mass <10 pm in diameter; PM25, particle mass <2.5 pm in diameter; coarse, particle mass between 10 and 2.5 pm in diameter (PM10-PM2.5); AC, air conditioned.
,This is not a paired comparison, thus there is not always a PM10 for every PM2.5 and vice versa. -0.02 0.32** Abbreviations: SD, standard deviation; PM10, particle mass <10 pm in diameter; PM25, particle mass <2.5 pm in diameter; coarse, particle mass between 10 and 2.5 pm in diameter (PM10-PM25). PM2.5 at regional Site (gig/rn3) sulfate measured at the regional site and outside of homes was 0.58; the correlation between sulfate measured outside and inside of homes was 0.51. Correlations between the various measured particle parameters by the site of measurement for the summer data are shown in Table 3. PM10, PM2.5, So42, H+, and NH4 concentrations were significantly and strongly correlated with each other at the regional site and significantly, but somewhat less strongly, correlated outside and inside residences. Coarse mass was correlated with all aerosol parameters at the regional site, but was not correlated with these parameters outside or inside homes.
A positive and significant (p<O.O5) correlation was found between coarse mass and PM2.5 for inside air-conditioned homes,  Table 4. A comparison of 0 mean ratio of 0.85 ± 0.62); PM1O, PM2.5, and coarse mass by site and unted for 43% of the PM2.5 by indoor source use is shown in Figure 3. 5 Table 3. Pearson correlation coefficients between selected summer pollutants at regional site, outside, and inside homes PM10 PM25 C SO2-H+ Regional site (n Abbreviations: PM10, particle mass .10 pm in diameter; PM25, particle mass <2.5 pm in diameter; C, particle mass between 10 and 2.5 pm in diameter (PM1,-PM25); AC, air conditioned. *p<0.05.
'This is not a paired comparison, thus there is not always a PM10 for every PM2.5 and vice versa. without a kerosene heater (21.6 ± 3.37 nmol/m3) and lower than levels in homes with a kerosene heater (82.8 ± 76.6 nmol/m3) ( Table 4). Strong acidity was low at all sites and did not vary by location. NH4+ concentrations followed the pattern of sulfate, with the highest concentrations observed in homes where kerosene heaters were used; the lowest concentrations were in homes without kerosene heaters. Ammonia levels followed a pattern similar to that observed during the summer, with indoor concentrations an order of magnitude or more higher than those outdoors. No significant differences in indoor ammonia concentrations were observed between homes with or without a kerosene heater. Measured indoor concentrations of ammonia for homes with kerosene heaters may be underestimated because of potential masking/collection inefficiencies of citric-acidcoated denuders. Indoor sulfur dioxide concentrations were higher in homes with kerosene heaters. Outdoor SO2 levels were higher than in homes without kerosene heaters. Nitrous acid levels were considerably lower outdoors (0.81 ± 1.32 ppb) than indoors (3.50 ± 3.61 ppb). For homes without kerosene heaters, indoor concentrations were higher in homes with gas stoves (5.46 ± 3.75 ppb) than in those without (2.43 ± 0.14 ppb). Homes with kerosene heaters and no gas stove had an average HONO concentration of 6.74 ± 6.4 ppb (n = 65)-levels comparable to homes with gas stoves only (data not shown). As with the ammonia measurements, possible inefficiencies of the carbonate-coated denuders may have resulted in the underestimation of HONO levels in homes with kerosene heaters.
Correlation coefficients for selected pollutants by location and use of a kerosene heater are shown in Table 5. Unlike the summer data, most correlations are not significant (p>0.05) for both outside and inside homes. Correlations among indoor sulfur dioxide, sulfate, PM2.5, strong acid concentrations, and hours of kerosene heater use indicate the importance of the contribution of kerosene heater emissions. In homes with kerosene heater use, the strong correlation between NH4+ and SO42and the high indoor levels of NH4+ suggests that the major form of the sulfate associated with kerosene heater kerosene heaters averaged 29 ± 7.5 nmol/m3. The decrease in strong acidity on the doped filters in the homes with the use of kerosene heaters suggests that neutralization of particle strong acidity may be occurring by ammonia not adequately collected by the citric-acidcoated denuders. This suggests that the acid aerosol levels measured in homes with G13 ,E40

Discussion
Numerous studies have been conducted to characterize the physical and chemical nature, spatial and temporal coiicentration distribution, and sources of ambient aerosol in the northeastern quarter of the United States, particularly during the summer season when fine particle and sulfate concentrations are high and regional in nature. Among the most recent studies was the EPA-sponsored Metropolitan Acid Aerosol Characterization Study (7). Relatively few studies, however, have sought to characterize the physical and chemical relationship of outdoor to indoor particles and the nature of the relationship of summer regional ambient particle concentrations to those measured inside and outside of homes in a region. This study investigated the relationship between indoor and outdoor particle concentrations in summer and winter for a sample of homes drawn primarily from southwest Virginia. Figures 4 and 5 contrast measured summer and winter PM2,5 and SO42~concentrations by location of measurement. Measured values of outdoor summer PM1o, PM2.5, So42 , and H+ con centrations are similar to those measured in a more densely populated portion of the same region (Washington, DC) during an intensive aerosol characterization conducted during 1994 (7). These measured values are also similar to PM1O, PM2.5, and coarse particle levels measured in 1992 and 1993 summer studies conducted within the Philadelphia metropolitan area (8).
In our summer study, homes were located as far as 175 km from the regional sampling site, yet no significant differences in mean concentrations of PM10o PM2.5? or sulfate were observed between concentrations outside the homes and the regional site. However, for all of these parameters, although most had statistically significant correlations, these correlations were all relatively low (a lot of scatter); for coarse mass, correlations were actually slightly negative (r = -0.20). The PM2*5/PM10 and SO42-/PM 2.5 ratios were similar between the regional site and outside homes, and PM2.5 and SO42concentrations at the central site and outside homes were correlated. These findings suggest a strong regional nature to the summer aerosol and that during the summer in our study area an ambient regional sampling site is a reasonable predictor of fine particle concentrations measured outside homes. This finding is consistent with the identified regional nature of aerosol in both the Washington, DC (X, and Philadelphia (8) studies, although these studies used results Volume 107, Number 3, March 999 . Environmental Health Perspectives emissions is ammonium sulfate or bisulfate. Strong acidity concentrations measured in 20 homes without kerosene heaters and using the sulfuric acid-treated (doped) Teflon filters in the denuder samplers averaged 273 + 171 nmol/m3. Strong acidity concentrations measured using the doped filter denuder systems in

Summer
Winter Figure 4. Comparison of 24-hr particle mass <2.5 pm in diameter (PM25) measurements made inside and outside homes and at the regional Vinton, Virginia, site for homes with and without kerosene heaters during winter and summer sampling periods for homes in Connecticut and Virginia. from sampling stations whose location was intended to represent geographical areas not immediately outside homes. It is also consistent with the findings of the particle total exposure (PTEAM) study (9), the Nashville, Tennessee, study (10) and the Uniontown, Pennsylvania, study (11). The PTEAM study was conducted in Riverside, California, in the fall of 1990 and the Nashville study was conducted in the summer of 1995. Concentrations in the PTEAM study were generally 2-3 times those observed in our study. In the PTEAM study, PM10 and PM2.5 levels measured at a central monitoring site, although significantly different than those measured outside of homes, were good predictors of levels outside of 178 homes in the region. In the PTEAM study, however, homes monitored were within 5 m of the central site. The Uniontown study (11) investigated the relation among indoor, outdoor, personal, and centrally measured acid aerosol concentrations monitored for 27 days during the summer of 1990 for 24 children. In the Uniontown study, concentrations of S042, NH4 , and H+ at the central monitoring site were found to be strong predictors of, and not significantly different from, concentrations of these same species measured outside homes, suggesting a strong regional nature to the sulfate aerosol consistent with the findings of this study. Strong acidity levels indoors in our study, however, were markedly lower than those measured outdoors or at the regional site, indicating that during the summer particle acidity levels indoors are low. Higher indoor levels of ammonia may result in the neutralization of acid aerosol. Correlations were found, however, between H+ measured at the regional site and outside of homes and between outside and inside of homes.
Winter concentrations of PM2,5 exhibited a pattern different from that of the summer. Concentrations outside of homes during the winter were approximately 57% of the concentrations measured during the summer. Indoor levels during the winter in homes without a kerosene heater were approximately 39% higher than outdoor concentrations and similar to indoor summer levels. Sulfate levels in these homes in the winter were approximately 70% of the outdoor level, suggesting a substantial contribution of outdoor PM2.5 to indoor levels (roughly 9 pg/m3 on average). Indoor winter sources of PM2.5 in homes without kerosene heaters contribute about as much as outdoors. Correlations between PM2, S042, NH4 , and H+ measured outsie and inside of homes without kerosene heaters were much poorer than those measured during the summer. Outdoor winter PM2.5 concentrations and indoor/outdoor PM25 ratios for homes without kerosene heaters found in this study are similar to those found in the winter New York State Energy Research Development Authority (12)(13)(14) study for nonsmoking, nonkerosene-heater homes in their sample of more than 400 homes drawn from Onondaga and Suffolk Counties in New York State.
Coarse mass concentrations measured during the summer months at the regional site were not correlated with coarse mass concentrations measured inside or outside of homes and the concentration differences for both were statistically significant. Concentrations inside homes were higher than concentrations outside and those outside of homes tended to be higher, though not significantly higher, than concentrations measured at the regional site. This suggests that the larger particle size (2.5-10 pm) aerosol is not as regionally well distributed as the fine aerosol. Regional ambient coarse aerosol mass measurements may not adequately represent levels outside or inside of homes. The results which indicate that indoor coarse levels during the summer are on average 33% higher than levels outdoors reflect both the expected relatively low penetration of outdoor coarse partide to indoors as well as the presence of significant indoor sources. Coarse particle concentrations were higher both outdoors and indoors during the winter as compared to the summer. Indoor coarse mass concentrations during the winter were not significantly different from outdoors. Higher outdoor levels during the winter may be related to higher wind speeds and street salting. Additional factors contributing to higher indoor winter concentrations are related to greater amounts of time spent indoors by residents, possibly greater occupant activity, and indoor winter sources (i.e., wood-burning fireplaces or stoves). These higher winter indoor concentrations could also be explained by typically lower air exchange rates in the winter as compared to rates in the summer. Given the same amount of indoor emissions, decreased air exchange results in higher indoor concentrations. The interaction of factors contributing to indoor concentrations of pollutants, under equilibrium conditions for a single compartment with complete mixing and no air deaning, can be expressed as: surfaces or chemical transformations (equivalent air changes per hour), S = indoor source strength (micrograms/hour), and V= volume of the indoor space (cubic meter).
Differences between summer indoor and outdoor concentrations of PM2.5 were not significant and the indoor and outdoor values were well correlated. Indoor SO42concentrations were significantly different from those measured outside of homes, but were significantly correlated. The corresponding indoor/outdoor ratios for PM2.5 and s042for homes were 1.03 ± 0.74 and 0.74 + 0.53, respectively, reflecting the strong dependence of indoor concentrations on outdoor levels. Because there are no known indoor summer sources of sulfate and because sulfate particles are generally < 1 pm, sulfate can serve as a marker for the contribution of outdoor PM2.5 (9).
Thus the sulfate ratio suggests that, on average, approximately 75% of the indoor fine aerosol during the summer is contributed by outside aerosol and 25% may be generated by indoor sources or activities. In this sample of homes 85% reported using an air conditioner and 15% reported no air-conditioner use. Doors and windows in air-conditioned homes are closed, resulting in lower air exchange rates and longer particle residence times, with greater potential for particles to deposit on interior surfaces. Inline filters, typically found in air conditioners, and deposition to the interior of air-conditioning systems also contribute to particle removal. The associations between indoor and outdoor particles would presumably be stronger for homes without air conditioning than for those with air conditioning, as it is likely that homes without air conditioning would be more open with higher air exchange rates. Our small sample size of homes without air conditioning does not allow for a statistical distinction to be made between indoor/outdoor ratios for air-conditioned homes versus homes without air conditioning. A comparison of the indoor/outdoor ratio for s042for air-conditioned homes (0.71) versus homes without air conditioning (0.86), however, indicated a trend toward outdoor aerosol contributing a higher portion of the fine mass in homes without air conditioning. Other studies have investigated the indoor/outdoor relationship for particle mass (7,9,11), but these studies have typically included smokers, have been conducted over only summer periods, or have not monitored a comparable set of variables (i.e., SO42-).
Indoor levels of ammonia and nitrous acid (Tables 1 and 4) were significantly higher than outdoor levels measured either at the regional site or outside homes.
Indoor summer levels of ammonia measured in this study were approximately 40% higher than those observed in the Uniontown (10) and Nashville (11) studies. Winter levels were approximately 40% higher than summer and were probably related to occupants and their indoor activities. Higher indoor levels in the winter (approximately 40 ppb vs 30 ppb in the summer) in homes with and without the use of kerosene heaters may be explained by the likelihood that occupants spend more time indoors during the winter, and also by lower air exchange rates during the winter. These results are the first reported indoor winter ammonia levels. High indoor ammonia levels have been proposed to be responsible for lower indoor acid aerosol levels because of ammonia's ability to neutralize strong acidity (10).
This study represents the most extensive database to date on indoor levels of nitrous acid. Nitrous acid levels during winter and summer were higher indoors than outdoors. Indoor levels were higher in homes with gas stoves, especially during the winter season. Indoor levels in homes with gas stoves and kerosene heaters were several times higher than homes without gas stoves. Winter indoor levels of HONO in homes without NOX sources are three times the levels of those homes in the summer. While the air exchange rates are lower in the winter, the outdoor NOX concentrations are higher. With these conditions, in the winter there is more time for the accumulation of nitrous acid formed through heterogeneous reactions indoors. Twentyfour-hour average HONO concentrations as high as 36 ppb were recorded. Nitrous acid concentrations indoors represent an important gas phase acid exposure. The heterogeneous reaction of nitrogen dioxide originating from outdoors with water vapor on indoor surfaces is thought to be the mechanism responsible for indoor HONO levels in homes without nitrogen dioxide sources (15). HONO levels in homes with nitrogen dioxide sources (i.e., gas stoves) may result from direct emissions as well as the heterogeneous reaction of gas stovegenerated nitrogen dioxide on internal surfaces. This study found that indoor exposures to HONO were appreciable in both winter and summer, particularly in homes with unvented combustion sources. The strong correlations observed between NH4+ and So42and the levels of NH4+ observed at all sampling locations and seasons suggests that a major form of sulfate indoors and outdoors during both winter and summer is ammonium sulfate or bisulfate. Ammonia concentrations both outside and inside of homes probably were responsible for the lower values of particle strong acidity measured at these sites. Sulfate appears to make up a major portion of the PM2.5 aerosol indoors and outdoors during both seasons and at the regional site during the summer. If ammonium sulfate is assumed to be the form of the sulfate for winter and summer outdoor samples, then There was concern that the complex nature and amount of particle and gas phase contaminant emissions from kerosene heaters might introduce interferences in the HDS system, resulting in inefficient collection of gases, and thus affecting measurements of both particle and gas phase acids. The doped filter sampling protocol used in this study suggests that such interferences were encountered, and may have resulted in an underestimation of both H' and HONO in homes with kerosene heaters.
Chamber and field studies have identified unvented kerosene heaters as an important source of both gas and particle phase air contaminants indoors (16,17). One chamber study measured emission rates of PM2.5 and SO42and determined the chemical composition of the sulfate emissions for a variety of kerosene heaters operated under a range of burner conditions (17). That chamber study estimated that under typical use conditions, kerosene heaters could add approximately 20 pg/mi3 or more to residential concentrations of PM2.5 and 7-15 pg/m3 of So42. In the field study reported here, kerosene heaters added approximately 12.5 pg/m3 of PM25 and 6 pg/m3 of So42 to residences during an average use period of 6.9 hr over the 24hr sampling period. This compares to 15.8 pg/m3 observed in the New York State Energy Research and Development Authority study for kerosene heaters in Suffolk County, New York (12). A simple regression model of hours of heater use against fine particle mass and sulfate indicates that PM2.5 concentrations during heater use, on average, added approximately 40 pg/m3 of PM25 and 15 pg/m3 of 5024 to background residential levels of 18 and 2 pg/m3, respectively. The present study did not measure the elevated residential H+ concentrations associated with kerosene heater use that were predicted by the chamber studies. A comparison Volume 107, Number 3, March 1999 * Environmental Health Perspectives of indoor winter samples using acid-doped Teflon filters and nondoped Teflon filters in kerosene-heater and nonkerosene-heater homes suggested that substantial amounts of collected strong acidity in homes with kerosene heater use may be neutralized on the Teflon filter in the denuder system used to collect particle acid. The mechanism for this possible neutralization is suspected to be denuder breakthrough of ammonia. In the present study, kerosene heater use resulted in elevated indoor concentrations of PM25 and so42-, with the potential for a substantial portion of the sulfate to be in the form of acid particles. Occupants in homes using kerosene heaters are likely to experience peak exposures (several hours at a time) to PM2 and So42and possibly H+ in excess ofievels typically experienced outdoors during the summer months. Frequent users of kerosene heaters are likely to experience longer term exposures (weeks or months) to PM25 and SO42and possibly H+ during the winter months that are in excess of summer exposures and substantially in excess of winter levels in nonkeroseneheater homes. Only tobacco combustion indoors is likely to result in higher indoor fine particle exposures.

Conclusions
Our study results indicate that a regional air monitoring site may provide a reasonable estimate of concentrations of PM2.5 and So42during summer months outside and inside of homes (in the absence of tobacco combustion) within a region, even when a substantial number of the homes use air conditioning. Kerosene heater use during the winter months in homes of nonsmokers results in a substantial increase in indoor concentrations of PM2.5, S042, and possibly H+. During the summer and winter periods, indoor concentrations of ammonia are an order of magnitude higher indoors than outdoors and appear to result in lower indoor acid particle concentrations. Nitrous acid levels are higher indoors than outdoors during both winter and summer and are substantially higher in homes with unvented combustion sources.