Document jmXXRRO6aR368ZEa72q5V7YM9

Ozone Reactive Chemistry on Interior Latex Paint RICHARD REISS, * ' P. BARRY RYAN, PETROS KOUTRAKIS, AND SARAH I. TIBBETTS Harvard University, School of Public Health, 66S Huntington Avenue, Boston, Massachusetts 02115 The heterogeneous chemistry of ozone on interior latex paint was investigated in a tube flow reactor. The emissions of several polarvolatile organic compounds (VOCs) including organic acids and carbonyls (aldehydes and ketones) were measured while a glass tube coated with latex paintwas exposed to clean air and ozone. Four different commercial brands of latex paint were tested. Formic and acetic acids were not found to be generated via ozone reactions; however, both were found to off-gas from the latex paints, and the off-gasing increased with increasing relative humidity. The off-gasing rates are large enough, particularly for acetic acid, to impact residential concentrations significantly. Formaldehyde was found to be produced by reactions related to the ozone concentration. There was some evidence that acetal dehyde and acetone may also be produced by processes related to the ozone concentration. A steadystate model is presented that is used to extrapolate the chamber results to a representative indoor environment. The model is based on an experimentally derived parameter termed the VOC formation factor, which is defined as the number of VOC molecules of a particular species formed via an ozone reaction divided by the total number of ozone molecules sticking to the surface. Using this model, it was found that formaldehyde production via ozone reactions is significant enough to impact indoor concentrations of formaldehyde. Introduction Indoor volatile organic compounds (VOCs) are ofincreasing concern because of their potential as irritants and car cinogens and their relation to sick-building syndrome, (1). For example, Molhave et al. (2) have found that the indoor totalVOC concentration correlates with occupantirritation. Many of the VOCs that have been detected in indoor environments are present at higher concentrations than in the ambient environment, indicating that there are indoor sources forVOCs (I). Recent research efforts have focused * E-mail address: rick@sonomatech.com. f Present address: Sonoma Technology, Inc., 5510 Skylane Blvd., Suite 101, Santa Rosa. CA 95403. on the specific sources of the VOCs. It has been shown that numerous VOC species originate from off-gasing by household products and materials (3). However, a new avenue of research is emerging that studies the secondary formation of VOCs via the indoor reactions of other pollutants (e.g., refs 4-6). Weschier et al. (7) have found, in a laboratory chamber study, that ozone reacts with carpet to form formaldehyde, acetaldehyde, and other C5-C10 aldehydes. The postulated pathway for these products is the reaction between ozone and olefins (5, 6). The ozone--olefin reaction yields a carbonyl and a Criegee biradical (fl). The Criegee biradical may isomerize to an organic acid, with the isomerization rate increasing with the presence of water vapor. In a residential field study in Boston, MA, Reiss et al. (5) found a correlation between the removal of ozone in indoor residences and the formation of secondary polar VOCs including carbonyls and organic acids. Environmental variables including temperature and relative humidity are confounders in this relationship, but it still appears that ozone reactions may account for a measurable portion of the concentration of polar VOCs found in residential environments. In a similar study in New Jersey, Zhang et al. (6) found thatthe indoorozone correlates with the indoor concentrations of acetaldehyde, n-valeraldehyde, vaieraldehyde (isovaleraldehyde plus n-valeraldehyde), and formic acid. These carbonyl and organic acid compounds con stitute about 15% of the total VOC concentration (5). Latex paint is a common indoor surface, and the resins used in the latex paints in our study were all variations of vinyl polymers. Vinyl compounds have the following functional form: CH2=CHR. The double bonds in the vinyl open as the polymer is formed; however, there are normally unreacted monomer resin-containing free double bonds. These unreacted monomers are susceptible to attack by oxidizing agents such as ozone. There are also some VOCs in latex paints such as toluene and xylene that can react with ozone (3), but the rates of these reactions are very slow. In this paper, we studied the reaction of ozone with several brands of interior latex paint. Several polar VOCs, including carbonyls (aldehydes and ketones) and organic acids, were measured duringexposure oflatexpaint surfaces with ozone and clean air. From these measurements, the emission rates of the polar VOCs were calculated for each of the exposures. The emission rates for the zero-air exposure corresponds to a natural off-gasing rate of the material, while the emission rate of the ozone exposure corresponds to natural off-gasing plus formation from processes related to the ozone concentration. This process is likely to be the formation ofVOCs as a result ofthe reaction of ozone with some constituent of the paint. Another possibility is that ozone may break down the polymer structure ofthe paintandmaycauseVOCsthat were trapped in pores below the paint surface to be released. Thus, from a comparison of emission rates of the zero-air and ozone exposure experiments, any VOC products related to the ozone concentration will be quantified. Finally, a steadystate model will be presented to extrapolate the results of this chamber study to an actual residential environment. 1906 a ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL 29. NO. 8. 1995 0013-936X/95/0929-1906S09.00/0 1995 American Chemical Society FIGUR Mate Desc react sche? repre ozon (i.e., glass airsy tube coate paint to th< place to dr of po meas the ir in or The : relati typic expo ozon 3h. 75 pi 1002 was W. colie were diffu togn tion orga ofth the t lates )wn 5 by lew lary lier ber rde, .ted Dne sa ical ion na rnd oor )Cs itai )) are hat i of tiai ; et oor :alnic )n- ins ; of ing nyl illy ds. by Cs act sry ith 3s, lie :es he ch air he ire >m !SS on ier ter ed >m ne he ety FIGURE 1. Schemstic diagram of tube flow apparatus. Materials and Methods Description ofApparatus. PolarVOC formation via ozone reactions was measured in a laminar tube flow reactor. A schematic of the apparatus is shown in Figure 1. This represents a modification of an apparatus used to study ozone deposition onto latex paint (see ref 9). A zero-air (i.e., pure air) system is used to expose a latex paint-coated glass tube sequentially to pure air with ozone. The zeroair system is described in detail in Reiss et al. (9). The glass tube (inside diameter of 2.1 cm and length of 30 cm) was coated by standing the tube vertically and pouring the latex paint through it and allowing the excess paint to drip down to the bottom onto a towel. After about 2 h, the tube was placed horizontally on a shelfin the laboratory and allowed to dry 5 days prior to being exposed. The concentrations of polar VOCs including organic acids and carbonyls were measured before and after the latex paint test section. Also, the inlet and outlet concentrations ofozone were measured in order to determine the ozone deposition to the surface. The measurement methods for ozone, temperature, and relative humidity were described in Reiss et al. (9). For a typical experiment, a latex paint-coated tube was first exposed to zero-air for 3 h. The tube was then exposed to ozone for 3 h and then to a higher ozone concentration for 3 h. The first ozone exposure was typically between 50 and 75 ppb ozone, and the second ozone exposure was between 100 and 150 ppb ozone. The flowrate for these experiments was typically 2.5 L/min. Modification of our apparatus was made to afford collection of organic acids and carbonyls. Organic adds were collected using potassium hydroxide (KOH) coated diffusion denuders and were analyzed using ion chroma tography. Lawrence and Koutrakis (10) provide a descrip tion ofthis measurementmethod. In order to measure the organic acid production, one denuder was placed upstream ofthe test section, and another was placed downstream of the test section. For a typical organic acid experiment, a latex paint-coated tube was exposed to zero-air for about TABLE 1 Summary of Limits of Detection for Carbonyl Compounds compound molecular formula LOD (ppbl aldehydes formaldehyde acetaldehyde acrolein propionaldehyde crotonaldehyde butyraldehyde benzaldehyde isovaleraldehyde n-valeraldehyde n-hexaldehyde ketones acetone butanone HCHO CH3CH0 CHz-CHCHO CH3CH2CH0 c h 3c h -c h c h o CH3(CH2}2CHO CeHsCHO (CH3)2CHCH2CHO CH3{CH2)3CHO CH3(CH2l*CHO CH3COCH3 CH3COCH2CH3 1.28 0.70 0.39 0.44 0.89 0.64 3.20 0.71 1.21 2.28 0.47 0.98 18 h, and then the next day the same tube was exposed to ozone (60-110 ppb) for 18 h. The limit of detection (LOD) for formic acid was 6.6 ppb, and for acetic acid it was 1.6 ppb. Our apparatus does not allow collocated sampling. Therefore, these LODs were determined by multiplying three times the standard deviation ofthe laboratoryblanks and using a representative flow rate and duration of experiments. For carbonyl measurements, two ports, before and after the test section, draw off 1.0 L/min air. This air is pumped through a2,4-dinitrophenylhydrazine- (DNPH) silica SepPak cartridge (Millipore Corporation). A detailed descrip tion of this measurement method is given by Tejada et al. (11). Also, Amts andTejada (12) note that ozone interferes with the analysis. To eliminate these interferences, copper tubing coated with potassium iodide was placed upstream of the cartridge to act as an ozone scrubber. The limits of detection (LODs) are shown in Table 1. Only acetone had consistently measurable blank values. Thus, these LODs were determined by multiplying three times the standard VOL. 29. NO. 8, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1907 deviation of the lowest HPLC standard and assuming a typical flow rate and experimental duration. Homogeneous versus Heterogeneous Chemistry. It is not known whether the VOC formation in residences is effected via homogeneous or heterogeneous chemistry or both. However, Reiss et al. (5) have shown, via a math ematical model, that ozone can react homogeneouslywith some indoor unsaturated hydrocarbons to account for about 20% of the ozone removal. Ozone also reacts homogeneously with nitric oxide, especially when gas appliances are present (5), but this reaction does not produce VOCs. However, it is known that ozone deposits on surfaces at significant rates (see refs 9 and 13). Our apparatus is designed to test only the heterogeneous reactions as the residence time in die test section is too short for homogeneous reactions to be significant. This can be shown by reaction kinetic principles given rate constants for ozone reactions ofcompounds known to off gas from latex paint such as xylene and toluene (3). Thus, all values reported in this paper represent a lower limit for ozone-generated pollutants as homogeneous production and production from heterogeneous reactions on other surfaces are not taken into account. Reaction Kinetic Calculations. We are interested in determining the rate of VOC formation as a function of ozone deposition in order to extrapolate our laboratory results to actual residential environments. Thus, we first need to quantify the ozone deposition. This can be done using the concept of the mass accommodation coefficient, which is defined as the number of "sticks" of a molecule . colliding with a surface divided by the total number of collisions. Reiss et al. (9) present a method for determining the mass accommodation coefficient, a, of latex paint in a tube flow reactor. The mass accommodation coefficient in combination with knowledge of indoor air flows can be used to extrapolate chamber deposition results to predict ozone deposition in indoor residences (see refs 9 and 14). Since not all ozone "sticks" to the surface result in the formation of a single VOC constituent, it is necessary to define an additional parameter, k , which we will refer to as the VOC formation factor. This parameter is defined as the number of molecules of a specific VOC constituent formed on the surface divided by the total number of "sticks" of ozone to the surface. The VOC formation factor is specific to each ofthe VOC molecules that is formed through ozone reactions. It is conceivable that this factor will be greater than unity if a single ozone molecule were to initiate a chain reaction resulting in the formation of multiple VOC molecules. TheVOCformation factor was determined from our experimental data by dividing the total number ofozone molecules that deposited in the chamber by the total number of molecules of a particular VOC that is attributed to ozone formation. The emission rate attributable to ozone reactions can be determined from the emission rate of an ozone exposure minus the emission rate of the corre sponding zero-air exposure. The emission rate may decrease with time, so this procedure provides a lower bound to the amount of VOC formed that is attributable to ozone reactions. At steady state, the ozone flux across the boundarylayer to the surface must be equal to the VOC flux across the boundary layer to the core region (defined as the area away from the room away from the surface boundary layer), modified by k , which reflects the fraction of the ozone flux converted to VOC molecules, as follows: This condition will be true even if some of the VOC molecules that are formed axe adsorbed back onto the surface provided that there is an adsorption--desorption steady state. The ozone flux to the surface can be determined from the deposition velocity as follows: *d,=^J[03] where kj, is the deposition velocity corresponding to a particular surface. Cano-Ruiz et al. (14) developed a mathematical model for determining deposition velocities as a function of the nature of the air flow in the residence. Theyconsideredfourpossibleairflowscenarios: (a)forced laminar convecdon parallel to a flat plate, (b) laminar natural convection flow along an isothermal vertical plate, (c) laminar natural convection flow in an enclosure, and (d) homogeneous turbulence in an enclosure. We will use the first and fourth scenarios as they are most likely to be present during the summer when the ozone is highest (9). We can also use some of these concepts to model the mass emission rate of the VOCs off-gasing from the latex paints during steady-state conditions. Clausen et al. (IS) have shown that VOC emission from latex paint is boundary layer limited as opposed to being dependent on diffusion of the VOCs through the condensed phase of the latex paint to the paint-air interface. Nevertheless, at steady state, we do not need to be concerned about the characteristics of the boundary layer of the residence because the flux of VOC off the surface will need to equal the flux of VOC at the edge ofthe boundary layer releasinginto the core region. It will only be necessary to adjust for the different surface areas between our chamber apparatus and the surface area of latex paint in an actual indoor environment. It should be mentioned that some researchers have developed empirical models for the change over time ofVOC emission rates (e.g., refs 16 and 17). However, we dldnothaveenough data to use any of these models. Results and Discussion Organic Acid Results. The organic add results are shown inTable 2. There was no evidence oforganic acid formation via ozone reactions. However, significant quantities of acetic acid and smaller quantities of formic acid off-gased from all of the paint surfaces. Also, the emission rates increased dramatically with increasing relative humidity. The relative humidityeffect is best observed for latex paint brand A. Without ozone, the emission rate for experiment A4 (a low relative humidity experiment) was 1.1 /fg/s, while for experiment A2 (a high relative humidity experiment), it was 60.6 pgls. Reiss et al. (5) have observed, in a residential field study in the greater Boston area, that the acetic acid emission rate was correlated with the indoor relative humidity. To examine the effect of prior ozone exposure on the off-gasing, we can compare experiments A2 (new) and A5 (10 months),whichwere bothhigh relative humiditybrand Aexposures. The acetic acid emission rate was significantly lower for experiment A5 while the formic acid emission fate increased a small amount In all but one of the experiments (experiment A5), the emission rate for acetic acid was higher for the zero-air exposure compared to the ozone exposure. One explana tion for the loweremission rates ofacetic acid during ozone exposure is that the mass emission rate is lowered as the 1908 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 29. NO. 8. 1995 TABI Sumi exp Al A2 A3 A4 A5 A6 A7 A8 . A9 * Tl could tube durir thez hypo expo expo cons resul inter expo C; mar> resul bran (see throi hum C9ai form few i high) off-g pain: ingre not a It is i prod we a acetc resul the e alsot resul the a Thei may' suggi occu In tion react voc ) the Dtion ci be to a ed a :ities mce. need rural (0 itd) ethe isent 1 the latex U5) dary sion aint tate, sties ix of C at ;ion. face area juld >ped sion ugh own tion s of tsed ates iity. aint tent hile ait). na the .oor one >nts tive tate mic the -air naone the TABLE 2 Summary of Organic Acid Results i formic acid acetic acid exp surface age (months) RH {%) ozone(ppb) in'(ppb) outb (ppb) ER=(pgM in' (ppb) out* (ppb) ER'tM/h) A1 latex brand A new 50-55 0 1.6 2.7 0.3 2.7 86.5 28.6 45-50 65 4.1 6.3 0.6 1.4 55.5 18.4 A2 latex brand A new 75-80 0 2.9 13.0 2.6 0.9 178.8 60.6 i 80-85 78 2.2 7.2 1.3 1.0 132.8 44.9 A3 latex brand A new 80 0 1.7 1.6 N/A<* 11.9 146.0 45.7 75-80 73 1.8 4.9 0.8 0.9 81.4 27.4 A4 latex brand A 10 5-23 0 1.7 2.2 0.1 2.1 5.4 1.1 4-5 75 4.8 5.0 0.05 0.7 1.8 0.4 A5 latex brand A 10 76-84 0 3.0 17.8 3.9 2.8 63.0 20.5 78-80 89 3.1 16.5 3.5 1.2 63.9 21.4 A6 latex brand B new ' 4-5 0 0.5 3.2 0.7 0.3 3.3 1.0 4 110 3.0 0.5 N/Ad 0.4 2.4 0.7 A7 latex brand B new 48-54 0 1.0 2.2 0.3 3.4 56.8 18.2 53 88 4.6 3.3 N/A* 3.3 35.4 11.0 A8 latex brand B new 46-56 0 1.7 7.2 1.4 0.0 43.9 15.0 45-46 75 3.9 7.9 1.0 3.2 26.9 8.1 A9 latex (red) brand C new 47-54 0 0.8 4.5 1.0 2.4 80.9 26.7 48-56 99 5.3 8.9 1.0 1.6 33.7 11.0 The flow rate to the inlet denuder was 2.85 L/min. `'The flow rate to the outlet denuder was 1.85 L/min. c R, emission rate, tfThe emission rate could not be calculated because the outlet measurement was lower than the inlet measurement. tube is exposed, and thus, the mass emission rate is lower TABLE 3 during the ozone exposure because it was always done after the zero-air exposure. Future experiments should test this hypothesis by alternatingthe order ofthe zero-airand ozone exposures. However, it should be noted that the ozone Summary of Carbonyl Compound Experiments age RH ozone deposition exp surface (months) i%) (ppb) i%) exposure was always conducted second in order to err Cl latex paint brand A new 24-26 0 conservatively in our estimate of the VOC production as a result of ozone exposure. Thus, one must be cautious in C2 latex paint brand A 18-20 55 31-35 105 13 40-43 0 75 77 interpreting results of an experiment where the ozone 48-52 71 4 exposure preceded the zero-air exposure. Carbonyl Compound Results. Table 3 shows a sum C3 latex paint brand B 46-50 151 new 28-30 0 26-28 49 2 44 mary of the carbonyl compound experiments. The raw 20-24 113 48 results and mass emission rates are shown inTable 4. Several C4 latex paint brand B 14 10-16 0 i brands of paint were tested including a non-white paint (see experiment C8). The relative humidity was varied C5 latex paint brand C 12-16 49 6-8 147 new 52-54 0 26 30 throughoutthe experiments because it is knownthat relative 50-53 61 74 humidity affects ozone deposition (28). Also, experiments C9 and CIO were run as replicates. This table only includes C6 latex paint brand C new 53-55 0 36-39 56 35-37 112 80 77 formaldehyde, acetaldehyde, and acetone because only a C7 latex paint brand C 1 35-40 0 few samples showed detectable quantities of any of the 38-40 61 45 higher weight compounds. Formaldehyde does appear to off-gas from latex paint. This is particularly true for latex C8 latex paint (red) brand C new 32-35 110 21-25 0 22-28 62 42 50 paint brand C. Formaldehyde is not listed as a paint 22-26 119 52 ingredient (although due to proprietary concerns we were C9 latex paint new 54-58 0 I not able to obtain listings of all compounds that are used). It is possible that air oxidation of the paint polymer could CIO latex paint brand D 51-53 65 52-54 112 new 50-52 0 66 56 produce formaldehyde. For several of the experiments, 48-52 63 80 we also observed small quantities of acetaldehyde and acetone off-gasing from the latex paints. However, the C11 latex paint brand O 52 112 73 2 20 0 27 52 59 results for these compounds were not consistent. Some of 18-20 101 48 the experiments did not show any off-gasing. This was also truefor the replicates (C9 and CIO). The formaldehyde of these compounds when the latex paint tubes were results were consistent between the two experiments, but exposed to ozone. For experiments C4-C11, excess form the acetaldehyde and acetone results were not consistent. aldehyde was produced from the introduction of ozone to The inconsistency in the acetaldehyde and acetone results the air stream. For a few of the experiments, the emission maybe due to the small quantities that were detected. This rate is linear with ozone (although there were not enough suggests a conclusion that off-gasing of these species, if it data points to conduct an adequate statistical analysis). occurs, is likely to be small. Figure 2 shows this linear relationship for experiment C8. In several of the experiments, we observed the produc The ozone flux to the latex paint surface in the chamber tion of a secondary pollutant from an ozone--latex paint is greater than a typical ozone surface flux in an actual reaction, as evidenced by the increase in the emission rates residential environment, and the nonlinearityin deposition VOL. 29. NO. a, 1995 I ENVIRONMENTAL SCIENCE & TECHNOLOGY 1909 I t f| ;^!j V. ' ti ..JLj :d IS i -5;| ji 3 il .4* 1-f 'Sl fet s -ii'l ` ! &} jsi* ' 4 I TABLE 4 Summary of Raw Results from Carbonyl Experiments" formaldehyde exp ozone (ppb) in (ppb) out (ppb) ERMtil Cl C2 C3 C4 C5 06 C7 C8 ' C9 CIO C11 0 55 105 0 71 151 0 49 113 0 49 147 0 61 0 56 112 0 .61 110 0 62 119 0 65 112 0 63 112 0 52 101 0.8 1.9 0.7 1.7 0.9 1.9 1.1 1.8 1.3 1.9 1.3 1.9 2.0 7.3 3.1 6.2 2.8 7.8 0.5 1.5 0.6 4.4 1.1 5.7 1.8 14.5 3.7 28.5 1.0 20.5 1.3 26.4 2.1 35.5 0.7 4.3 1.5 9.7 5.0 19.3 2.5 3.6 2.3 11.1 2.9 20.0 0.5 0.9 0.8 2.0 0.8 3.1 1.9 2.1 1.9 3.4 3.2 6.0 0.3 2.5 1.1 4.0 1.0 5.0 0.18 0.19 0.18 0.14 0.13 0.13 0.95 0.58 0.96 0.16 0.66 0.83 2.42 4.96 3.45 4.29 6.29 0.68 1.64 2.86 0.21 1.71 3.32 0.07 0.26 0.46 0.05 0.30 0.56 0.38 0.58 0.77 -ER, emission rate. N/A, not applicable. in{ppb) 0.5 0.4 0.7 0.03 0.5 0.3 0.5 0.8 0.9 0.0 0.6 0.6 0.6 0.4 0.6 0.6 1.0 0.1 0.7 2.7 0.5 1.0 0.8 0.0 0.0 0.4 0.4 1.5 2.3 0.0 0.8 0.8 acetaldehyde out (ppb) EH Ws/b) 3.4 0.69 2.4 0.53 2.8 0.55 0.2 0.04 1,0 0.16 0.5 0.07 0.4 N/A 0.8 N/A 1.6 0.20 0.1 0.02 1.2 0.14 2.2 0.42 0.5 N/A 1.4 0.30 1.4 0.22 2.9 0.58 4.5 0.97 0.3 0.06 0.6 N/A 2.4 N/A 0.4 N/A 1.3 0.09 1.0 0.04 1.1 0.34 0.9 0.27 2.6 0.66 2.0 0.43 1.4 N/A 1.7 N/A 0.1 0.03 1.4 0.15 1.3 0.14 in (ppb) 1.5 2.1 2.2 0.8 1.6 0.6 0.0 3.8 1.6 1.7 1.5 2.4 1.2 3.6 0.9 1.1 1.5 0.3 2.9 7.0 0.2 3.9 3.5 0.3 0.0 1.9 1.6 3.3 4.6 0.1 2.0 3.0 acetone out(ppb) 2.0 2.0 2.4 0.3 2.0 2.3 0.1 1.8 0.8 1.7 2.4 5.2 0.0 1.6 0.8 1.1 1.6 2.1 6,4 2.3 0.3 1.1 1.1 0.5 0.4 1.3 1.8 2.6 1.6 0.4 0.5 2.2 ER (jug/h) 0.16 N/A 0.06 N/A 0.18 0.69 0.02 N/A N/A 0.01 0.32 0.97 N/A N/A N/A 0.01 0.68 1.35 N/A 0.03 N/A N/A 0.05 0.31 N/A 0.07 N/A N/A 0.09 N/A N/A Outlet minus Inlet Ozone Concentration (ppb) FIGURE 2. Outlet minus inlet ozone concentration versus formal dehyde emission rate for experiment 08. usually begins at higher ozone fluxes. This suggests that the formaldehyde production via ozone--latex paint reac tions ina residence may also be linear with respect to ozone concentration. Latex paint brand C showed the most formaldehyde production. The red brand C paint showed similar formaldehyde production to the white brand C latex paint, indicating that the pigment may not be a factor in the reactivity. Latexpaint brand D also showedsome excess production but less than brand C. Brand B may have some reactivity, but the results are inconsistent, Brand A did not show any reactivity and also showed verylittle formaldehyde off-gasing. Thus, it is clear that there are significant differenes amongthe different brands ofpaint. Thesepaints are all based on vinyl polymers, but details about the formulations were not available because this information is proprietary. The cause of these differences in reactions is important for future research where more sophisticated techniques to study chemistiy may be available. Several of the experiments (C2, C4, CS, and C6) showed production of acetaldehyde via the ozone--latex paint reaction. Figure 3 shows a plot of ozone outlet minus inlet concentration versus the emission rate of acetaldehyde for experiment C4. Two experiments (C2 and C4) showed acetone production. The fact that fewer experiments showed acetaldehyde and acetone production as compared to formaldehyde production may be a result of the low concentrations that were being measured, and thus, any changes may be difficult to distinguish from experimental error. Changes in these smaller values may not be measurable with our apparatus. Nonetheless, if ozone reacting on latex paint truly forms acetaldehyde and acetone, it does so in small quantities. The effect of aging may be manifested as a decrease in the ozone deposition. Reiss et aL CS) observed that the ozone mass accommodation coefficient decreased by about an order of magnitude for several tubes that were about 1 1910 ENVIRONMENTAL SCIENCE & TECHNOLOGY t VOL. 29. NO. 8, 1995 . 3 < FIGl defy yea in i dep fact rese tiga that inti COIT fom I mei trap indc fom cart com whe cent com to t sign data duri aver indc font emis 26.3 resic also mod whic (b) ti mas.' (ref $ m_ 1 sam{ nr1, ahig TABLE 5 Summary of Reaction Kinetic Parameters VOC formation factor /g/h) exp ozone (ppb) HCH0 CHsCHO CHiC0CH3 1 16 Cl 55 0.0 N/A8 N/A A 105 0.0 N/A N/A )6 C2 71 0.0 0.13 0.34 A 151 0.0 0.03 0.77 18 C3 49 0.0 0.0 N/A 19 113 0.0 0.01 N/A (2 C4 49 0.23 0.04 0.07 \\ 147 0.09 0.04 0.06 C5 61 0.28 0.02 N/A 1 C6 56 0.11 0.03 0.0 2 112 0.17 0.03 0.0 7 C7 61 0.17 N/A 0.06 X 110 0.24 N/A N/A X C8 62 0.25 0.01 N/A X 119 0.26 0.0 N/A 0 1 Outlet minus Inlet C9 65 0.02 N/A 0.02 112 0.03 0.02 N/A 3 Ozone Concentration (ppb) CIO 63 0.03 N/A N/A FIGURE 3. Outlet minus inlet ozone concentration versus acetal C11 112 52 0.03 0.03 N/A 0.01 dehyde emission rate for experiment C4. 101 0.04 0.01 N/A N/A N/A i 4 IS ':) "i a ii^ lii .'if 1 ,':i| year old. The particular deposition mechanism that results in the VOC formation will determine how the decrease in depositionaffects VOC production. We do not have enough a N/A denotes that the VOC formation factor was negative, which is a nonphysical result that usually occurred when low concentrations were being measured. si data to determine the rate at which the VOC formation rate of formaldehyde is 0.25 /(g/s for laminar flow and 0.35 factor decreases with time. This is an area for future tigIs for turbulent flow, which is 10.9% and 15.2%, respec research. Another potential factor that should be inves tively, ofthe formaldehyde emission rate measured by Reiss tigated is the paint film thickness. Clausen etaL (15) showed et al. (51. Fora low-endvalue ofk (0.03), the source emission that the film thickness ofthe paint, whichwas not controlled rate of formaldehyde is 0.029 /(g/s for laminar flow and in this study, influences the VOC off-gasing rates ofseveral 0.041 /(g/s for turbulent flow, which is 1.3% and 1.8%, compounds. The effect of paint film thickness on VOC respectively, ofthe measured formaldehyde emission rate. in formation via ozone reactions should be investigated. We can use the modelin amore generalsense bycalculating 2SS Extrapolation of Model Results to Indoor Environ the source emission rate of VOCs produced by an ozone ne ments. Using the model developed above, we can ex reaction as a function of k . A plot of the VOC formation lot trapolate the results from this laboratory study to actual factor versus the formaldehyde emission rate for laminar de indoor air environments. The key variable is the VOC and turbulent flow is shown in Figure 4, given the int formation factor (at ). The VOC formation factors for the assumptions listed above. its carbonyl experiments are shown in Table 5. Reiss et al. (51 We can also examine the effects ofthe naturaloff-gassing he conducted a residential field study in the Boston, MA, area of the VOCs from the latex paint surface. In the above d ii where 24-h average indoor and outdoor carbonyl con model for the ozone reactions, we were able to adjust for ns centrations and emission rates were measured. We will the aging effect byuse ofamass accommodation coefficient 2d compare the emission rate ofcarbonyls from the latexpaint foi an aged surface. However, for the off-gasing, wecannot to the total carbonyl emission rate to determine the make adjustments for the age of the paint Nonetheless, ;d nt et or id ts d *v significance ofthe ozone--latexpaint reaction. The summer data from this studywill be used as ozone is most prevalent during the summer. These researchers found the following averages among the nine homes that were sampled: (a) indoor formaldehyde concentration, 16.1 ppb; (b) outdoor formaldehyde concentration, 2.4 ppb; (c) formaldehyde emission rate, 2.3/^g/m3; (d) outdoor ozone concentration, 26.3 ppb; (e) air exchange rate, 2.6 h_1; and (f) average residencevolume, 350 m3. [Severalother assumptionswere our estimates will be valid for new paints. The largest offgasing was observedfor acetic acid. Fora typicalnewpaint, we observed an off-gasing of 0.011 /igls from our chamber, which after adjusting for differences in the surface area of the chamber and a typical residence (see above) translates into an emission rate of 168 /igls in a residence. This is a very high emission rate compared to the average of 10.4 Ag/s observed in a summer field study by Reiss et al. (5). This indicates that freshly coated paints will result in avery y u e e also necessary to apply the Cano-Ruiz et al. deposition model. These include (a) indoor air flow, 10 cm/s (ref 14). which is used to determine the boundary layer thickness; (b) turbulence intensity, 1 s"` for m -- 2 (ref 14); (c) typical large indoor acetic acid concentration in an actual resi dence. For the two aged experiments, we observed an emission rate of about 0.0057 /(g/s for the high relative humidity experiment and 3.1 x lO'Vg/s fordie lowrelative mass accommodation coefficient for latex paint, 5 x 10"6 humidity experiment. This translates into 86 and 4.6 /tg/s, (ref 9).] We will use a surface to volume ratio (A/V) of 3.3 respectively, for an indoor residence. The high relative m"* (20). By examining the descriptions of the residences humidity emission rate gives an unrealisticallyhigh indoor sampled in this study, we estimate the Ap/ Vto be about 1.0 emission rate, about eight times what was observed in the [t rrr*, where Ap is the surface area ofonly the latex paint. For field study. It may be that latex paints age faster in actual a high endvalue of k (0.25), the calculated source emission residences. This could occur because there is more air VOL. 23, NO. 8, 1995 / ENVIRONMENTAL SCIENCE & TECHNOLOGY . 1911 O Laminar Turbulent Acknowledgments This studywas funded by the Center for IndoorAir Research under Contract number 90-31. Support for R.R. was provided by the National Institute ofHealth Training Grant ES07155. Bob Weker provided advice and technical assistance for the analytical work in this study. We would also like to thank Dr. Haluk Ozkaynak and Dr. Joseph Harrington for providing a critique ofthe paper and making helpful suggestions. FIGURE 4. VOC formation factorversus formaldehyde mass emission rate for laminar and turbulent flow given assumptions listed in the text. flow across the surface of latex paints in residences than our tubes in the laboratory. For formaldehyde, an average new paint emission rate was 0.0014 pgls, which gives an indoor emission rate of 21 figls. The emission rate for an aged paint or a new paint for some brands was 2.8 x 10-4 fig/s, which gives an indoor emission rate of 4.2 pg/s. The average emission rate for formaldehyde observed by Reiss et al. (5) was about 2.3 ag/s. Thus, the formaldehyde concentrations will also be higher after application of an interior latex paint. Conclusions Formaldehyde was observed to be formed when latex paint surfaces were exposed to ozone. Therewas some evidence that acetaldehyde and acetone were also formed during ozone exposures. The formaldehyde production is suf ficient enough to impact indoor formaldehyde production measurably. We also observed significant quantities of several of the carbonyls and organic acids off-gasing from the latex paints. The mass emission rates of the off-gasing from new latex paints were significantly higher than the totalmass emission rate observed in a residential field study. A mechanism involving the ozonation of impurities in the vinyl resin of the paint is proposed for the production of the polar VOCs. Future research in this area should concentrate on determining the effect of aging, film thickness, and brand oflatexpaint on these processes. Also, given that the reactions studied in this paper only account for up to about 30% of the ozone deposition, future work should aim toward explaining what happens to the rest of the ozone. Glossary A surface area of the residence Ap latex paint surface area of the residence J?1 ozone flux to surface Hac VOC flux from surface Kd deposition velocity V volume of residence Greek Symbols k VOC formation factor Literature Cited U) Wallace L. In Indoor Air Pollution: A Health Perspective; Johns Hopkins University Press: Baltimore, MD, 1991; pp 253-272. (2) Molhave, L. Indoor Air 1991, I, 357-376. (3) Sack, T. M.; Steele, D. H.; Hammerstrom, K.; Remmers, J. Atmos. Environ. 1992, 26A, 1063-1070. (4) Ryan, P. B.; Koutrakis, P. In The 5th International Conference on Indoor Air Quality and Climate; International Conference of Indoor Air Quality and Climate: Toronto, Canada, 1990; Vol. 2; pp 489-494. (5) Reiss, R.; Ryan, P. B.; Koutrakis, P. Air Waste, in press. (6) Zhang, J.; Uoy, P. J. Indoor Air 1994, 4, 95-105. (7) Weschler.C. J.; Hodgson.A.T.;Wooley, J. D. Environ.Sci. Techno! 1992, 26, 2371-2377. {8) Finiayson-Pitts, B. J.: Pitts, J. N. Atmospheric Chemistry: Fun damentals and Experimental Techniques; John Wiley & Sons: New York, 1986: pp 441-459. (9) Reiss, R.; Ryan, P. B.; Koutrakis, P. Environ. Set. Technol. 1994, 28, 504-513. CIO) Lawrence. J. E.; Koutrakis, P. Environ. Set. Technol. 1994, 28, 957-964. (11) Tejada, S. B. Int. J. Environ. Anal. Chem. 1986, 26, 167-105. (12) Arms, R. R.; Tejada, S. B. Environ. Sci. Technol. 1909,23,14281430. (13) Weschler, C. J.; Shields, H. C.; Naik, D. V. Tropospheric Ozone . and the Environment II: Effects, Modeling and Control; Air and Waste Management Association: Pittsburgh, PA, 1991. (14) Cano-Ruiz, J. A.: Kong, D.; Balas, R. B.; Nazaroff, W. W. Atmos. Environ. 1993, 27A, 2039-2050. (15) Clausen, P. A. Indoor Air 1993, 4, 269-275. (16) Dunn, J. E. Atmos. Environ. 1987, 21, 425-430. (17) Colombo. A.; De Bortoli, M. Indoor Air 1992, 2, 49-57. (18) Sabersky, R. H.; Sinema, D. A.; Shair, F. H. Environ. Set Technol. 1973, 8, 347-353. Received for review August 15, 1994. Revised manuscript received January 26, 1995. Accepted May 9, 2995. ES940520S Abstract published in Advance ACS Abstracts, June 15, 1995. 1912 * ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 29. NO. 8. 1995