Environ. Sci. Technol. 2005, 39, 8214-8219
Prediction of Initial Emission Rates of 2-Butoxyethanol from Consumer Products Using Equilibrium Headspace Concentrations: An Application of the Vapor Pressure and Boundary Layer (VB) Model J I P I N G Z H U , * ,† H E N R I K L I , ‡ MARK KORCHINSKI,§ AND PHIL FELLIN‡ Chemistry Research Division, Health Canada, Tunney’s Pasture, Ottawa, Ontario, K1A 0L2 Canada, AirZOne One, Inc., 222 Matheson Boulevard East, Mississauga, Ontario, L4Z 1X1 Canada, and Management of Toxic Substances Division, Health Canada, Tunney’s Pasture, Ottawa, Ontario K1A 0K9
The initial emission rate of volatile organic compounds (VOCs) from consumer products is important for assessing potential human exposure to VOCs in products. The vapor pressure and boundary layer (VB) model developed in the past was used to predict the emission rates of VOCs in the fast decaying phase from petroleum-based wet materials. This study has extended the model to largely water-based products. Study results have shown a good agreement (ratio ) 1.01, r2 ) 0.89) between model-predicted initial emission rates (ER0) of 2-butoxyethanol (2-BE) based on its equilibrium headspace concentration and experimentally measured ER0 in a small dynamic environmental chamber for 20 consumer products. These water-based products included wood surface treating stains, general cleaning agents, and degreasers with 2-BE concentrations over a wide range. The results also demonstrated a dependency between the headspace concentrations of the target analytes and the water content in the liquid. But dependency on water content had no effect on the use of headspace concentration to predict the ER0. The ER0 of 2-BE in the products ranged from 100 to 3000 mg m-2 h-1. In the majority of cases, the 2-BE concentration range in individual products indicated in the Material Safety Data Sheet agreed with the measured data.
Introduction Emission characteristics of volatile organic compounds (VOCs) from building materials and consumer products have an impact on the indoor air quality in buildings (1-3). In the past two decades there have been several major studies in both North America and Europe to generate VOC emissions data for products (4-6). The emissions from wet materials such as wood stains, paints, and cleaning liquids are often characterized by the double-exponential model with a high * Corresponding author phone: 613-946-0305; e-mail:
[email protected]. † Chemistry Research Division, Health Canada. ‡ AirZOne One, Inc. § Management of Toxic Substances Division, Health Canada. 8214
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
and fast decay emission rate in the first phase and relatively low and slow decay rate in the second one (7). The initial emission rate (ER0), representing the emission rate at the beginning of an emission process, is one of the two parameters (the other one being the decay constant k1) used to describe the emissions in the first phase (7). The vapor pressure and boundary layer (VB) model has been developed to describe the VOC emissions in the first phase, mainly from petroleum-based surface-coating materials such as wood stain, polyurethane wood finish, and floor wax (8, 9). In the VB model, the emission rate (ER) is described as a function of the concentration difference at the emission surface of a product (Cs) and in the surrounding air (Ci) as shown in eq 1
ER (mg m-2 h-1) ) km (m h-1) × (Cs - Ci) (mg m-3) (1) where km is the gas-phase mass transfer coefficient. At the beginning of an emission process, the equilibrium vapor concentration in the headspace (Cv) can be used to replace Cs, while Ci can be considered to be zero. So for the initial emission rate (ER0) eq 1 can be rewritten as eq 2.
ER0 (mg m-2 h-1) ) km (m h-1) × Cv (mg m-3) (2) km is dependent on both the chemical and physical conditions of an emission process, which can either be directly measured or estimated from the diffusivity of a VOC and the boundary layer of the emission surface (10). Cv meanwhile can be determined directly through headspace analysis, or if the saturated vapor pressure (P0) of a given VOC is available, calculated from its mole fraction (x) in the liquid mixture according to Raoult’s law (eq 3).
Cv (mg m-3) ) P0 (mg m-3) × x
(3)
The VB model has been validated so far largely for nonpolar VOCs such as alkanes and alkyl benzenes (9, 10) in petroleum-based products. It would be interesting and valuable to extend the model to polar or semi-polar chemicals in water-based products. Unlike the nonpolar chemicals in petroleum-based products, the hydroxyl group in the polar or semi-polar compounds has stronger interactions between molecules and between the compounds and water because of hydrogen bonding. In the present study, the VB model was applied to predict the ER0 of a water-soluble VOC (2butoxyethanol) in selected consumer products, mostly waterbased products, using its equilibrium vapor concentration determined through headspace analysis. The study provided a direct comparison between model-predicted and measured ER0. The study focused on 2-butoxyethanol (2-BE) because this compound has recently been assessed under the Canadian Environmental Protection Act (CEPA), and is currently in the risk management phase (12). The aquatic toxicity of 2-BE has been reviewed along with that of other ethylene glycol ethers (13). Moreover, hematological effects of 2-BE have been reported (14) in both experimental animals and humans, likely through its metabolite, butoxyacetic acid. Although the presence of 2-BE in a limited number of consumer products in Canada has been reported (15), additional information was needed to facilitate the risk management process. As a result, the current uses of 2-BE in Canada have been surveyed (16) recently; results showed that 2-BE is present mainly in cleaning products, and in waterbased paints and coatings. On the basis of the survey data, 10.1021/es051080f CCC: $30.25
2005 American Chemical Society Published on Web 09/24/2005
TABLE 1. Information on the Properties of Target Glycol Ethers and the Composition of the Solvent Mixtures chemical name
abbr.
CAS
formula
M. W. (amu)
2-methoxyethanol 1-methoxy 2-propanol 2-ethoxyethanol 1-ethoxy 2-propanol 2-butoxyethanol 1-butoxy 2-propanol 2-hexyloxyethanol water
2-ME 1-MP 2-EE 1-EP 2-BE 1-BP 2-HE
109-86-4 107-98-2 110-80-5 1569-02-4 111-76-2 5131-66-8 112-25-4
C3H8O2 C4H10O2 C4H10O2 C5H12O2 C6H14O2 C7H16O2 C8H18O2 H2O
76.1 90.1 90.1 104.2 118.2 132.2 146.2 18.0
Da (m2 h-1)
vapor pressure (20 °C) (mmHg)
0.0337 0.0298 0.0298 0.0271 0.0249 0.0232 0.0218
6.2 8.6 3.8 5.3 0.76 0.4 0.051
amount in mixture (% weight) M-1
M-2
M-3
M-4
M-5
4.2 4.2 4.2 4.2 5.0 4.2 4.2 70
0.8 0.8 0.8 0.8 25 0.8 0.8 70
7.5 7.5 7.5 7.5 25 7.5 7.5 30
2.5 2.5 2.5 2.5 55 2.5 2.5 30
0.0 0.0 0.0 0.0 100 0.0 0.0 0
a The diffusivity coefficient (D) values were calculated using the diffusion coefficient estimation method from U.S. EPA available at http://www.epa.gov/athens/lean2model/part-two/onsite/estdiffusion-ext.htm.
FIGURE 1. Dynamic air generation system and exposure chamber. 30 consumer products were selected for determination of the concentrations of 2-BE and related compounds in the products and in headspace samples. Due to constraints in funding, only 20 of the 30 products, covering a wide range of 2-BE concentrations, were tested in an environmental chamber to determine ER0. The objective of the chamber tests was to validate the model-predicted initial emission rates in products.
Materials and Methods Testing Materials and Standards. All 30 products were acquired from retail sources in standard containers as sold to consumers. As such, they are representative of formulations available to typical consumers or users. The basis for the selection of the 30 products included commercial availability, indoor use, and representation of a range of different product types, such as paints and coatings, cleaners, degreasers, etc. Five solvent mixtures (M-1 to M-5, Table 1), each made up of seven glycol ethers in water, were prepared in the laboratory. 2-Methoxyethanol (99.9+%) and 1-methoxy 2-propanol (99.5+%) were purchased from Sigma-Aldrich (Sigma-Aldrich Canada Ltd., Oakville, ON), and2-ethoxyethanol (99.0%), 1-ethoxy 2-propanol (90%), 2-butoxyethanol (99.0%), 1-butoxy 2-propanol (96.0%), and 2-hexyloxyethanol (98.0%) were purchased from TCI America (Portland, OR). Methylene chloride and methanol were from Caledon Laboratories Inc. (Georgetown, ON). Toluene-d8 was obtained from Restek Corp. (Bellefonte, PA). Headspace Analysis. Aliquots of liquid sample (4.0 mL) were transferred into 6-mL amber glass vials fitted with
Teflon-coated septum caps. After 16 h standing at a temperature of 22 ( 0.5 °C, 0.5 mL of the headspace was slowly withdrawn using a 1.0-mL gastight syringe. During the collection of headspace samples, a separate syringe needle was inserted through the septum into the headspace to avoid creating a vacuum in the vial. The headspace sample was injected directly into a gas chromatograph/mass spectrometer (GC/MS) for analysis. Measurement of Chemical Concentrations in Materials. A 5.0-µL aliquot of each product was transferred to a 4.0-mL amber glass vial containing 4.0 mL of solvent (methylene chloride/methanol, 95:5 v/v, spiked with internal standard toluene-d8 at 2.5 ng µL-1). After the solution was shaken for 30 min it was left at room temperature for 16 h, then transferred to 2.0-mL auto-sampler vials for GC/MS analysis. Chamber Tests. The homogeneity, stability, and suitability of the background air supply (2.5 L min-1) of the test chamber (0.3 m3 made of 304-stainless steel with a glass door) were evaluated in a previous study (17). Figure 1 shows a diagram of the air supply system and test chamber. The temperature and relative humidity were measured by a digital thermohygrometer (thermometer NIST traceable, ISO 17025 certified). The tests were run at 22 ( 1 °C, 50 ( 5% relative humidity (RH), and 0.5 h-1 air exchange rate (ACH). The air in the emissions generation zone was gently mixed with Teflon paddles rotated at 2 rpm giving an air velocity of less than 0.2 m per second to ensure representative collection of chamber air. The chamber was cleaned with methanol and flushed with clean air. Four dishes with total surface area of 0.14 m2 were placed in the chamber at the start of each test. VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8215
The dishes contained accurately weighed amounts of product totaling 120 ( 0.1 g. The weight loss was determined at the end of the test period. Air from the chamber was drawn through ORBO-32 charcoal absorbent tubes (front/back sections 100/50 mg), obtained from Supelco (Bellefonte, PA), at a nominal flow rate of 0.12 L min-1. Samples collected from chamber tests were extracted with 4.0 mL (front section of charcoal tube) and 1.0 mL (back section of charcoal tube) of 95:5 (v/v) methylene chloride/ methanol solvent, which had been spiked with internal standard toluene-d8 at 2.5 ng µL-1. The glass wool plug was extracted with the back-up adsorbent. After adding solvent, sorbents were extracted on an automatic shaker for a least 60 min and left for an additional 60 min before GC/MS analysis. GC/MS Analysis. All samples were analyzed on an HewlettPackard (HP) 5890 GC Series II coupled with HP 5971A MSD (Agilent Technologies, Palo Alto, CA), equipped with a DB624 capillary column (60 m × 0.32 mm × 1.8 µm film). The GC oven temperature was set at 40 °C for 2 min, raised to 250 °C at 10 °C min-1, and held for 2 min. The injector and detector temperatures were 220 and 260 °C, respectively. The column head pressure was 6.0 psi at 40 °C of UHP helium gas. A 1.0 µL portion of solution was injected in splitless mode. The mass spectrometer was operated in scan mode for headspace and product analyses, and in SIM mode for analysis of samples from chamber tests. Target compounds were identified by retention time and comparison of spectra with standard solutions. Other compounds were identified by comparison with the NIST library of mass spectra and quantified as toluene-d8 equivalents. QA/QC. The instrument was calibrated with multi-level standard solutions of the target analytes in concentrations ranging from 0.4 to 100 ng µL-1. Toluene-d8 in the extraction solvent was used to correct variations in injection volume and GC/MS sensitivity. In SIM mode, 3 ions, one target ion (T) and two qualifier ions (Q1 and Q2), were selected for each target compound. A peak was accepted if its retention time was within (0.3 min of the retention time of the standard, and the ion ratios of Q1/T and Q2/T were within (15% of the corresponding ion ratios. The method detection limits (MDL) were developed according to the CFR 40 (1984) procedure (18). The instrument detection limit (IDL) was defined as 3 times the baseline noise level. Blank levels in solvents and on absorbent media were very low and reproducible. Average recoveries at spiking levels of 0.5, 1.5, 2.5, 5, and 10 ng µL-1, were in the range of 76-92% for the 7 target compounds. The average recovery was used for quantitative calculations. Precision of recovery levels was good with RSD values less than 8% for all spike levels except the lowest one (0.5 ng/µL) which was less than 16%. Correlation coefficients for calibrations exceeded 0.999 for all compounds. Minimum detection levels were about 0.30.4 ng µL-1 and 0.04-0.07 ng µL-1 in Scan and SIM modes, respectively. No breakthrough was observed in charcoal tube samples for all compounds. The emission rate (ER) was calculated according to eq 4 as described in ASTM Standard 5116-97 (19). Because of the large quantity (120 g) of product introduced into the chamber, the emission rates during the first hours were relatively constant. The average of these ER values was deemed to be the initial emission rate (ER0)
ERi ) (∆Ci/∆ti + NCi)/L
(4)
where ERi is the emission rate at time ti, mg m-2 h-1; Ci is the chamber concentration at time ti, mg m-3; ∆Ci/∆ti is the slope of the time-concentration curve at time ti; ti is the test elapse time, h; N is the air exchange rate, h-1; and L ) A/V, 8216
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
FIGURE 2. Ratios of headspace concentration to mole fraction in the solvent mixtures. Water content in the solvent mixture: M-1 and M-2, 70%; M-3 and M-4, 30%; M-5, 0%. in which A () 0.14 m2) was the sample emitting surface, and V () 0.3 m3) was the volume of the test chamber.
Results and Discussion Effects of Water Content on Headspace Concentrations of Solvent Mixtures. So far, the VB model has been validated using petroleum-based products (7-10). These products, such as wood stains and polyurethanes, do not contain noticeable amounts of water and the compounds in the emissions are largely nonpolar volatiles. As glycol ethers are mostly used in water-based products, the potential impact of the water content on the headspace concentrations of glycol ethers was evaluated. Five mixtures (M-1 to M-5, Table 1), with different combinations of seven glycol ethers and water, were examined to determine the relationship between mole fraction of a compound in the liquid phase and its resulting equilibrium concentration in headspace samples. The ratio of concentrations of a compound in the headspace to its mole fraction in the liquid mixture represents the saturated vapor pressure of the compound in its pure form according to Raoult’s law (eq 3). However, in the case of these solvent mixtures, the ratio appeared to depend on the water content in the mixture (Figure 2), as the ratios were higher in M-1 and M-2 samples (70% water) than in M-3 and M-4 (30% water). In the case of 2-BE, such water content dependency of headspace concentration to mole fraction was clearly evident; the ratio decreased from M-1 and M-2 samples (around 40 000), to M-3 and M-4 samples (9500), and further to M-5 (about 3700) where the water content was zero. Such a dependence on the water content indicates that the equilibrium headspace concentration (Cv) of a compound from a nonideal liquid mixture cannot simply be predicted with Raoult’s law (eq 3). Moreover, the degree of the reduction, as the water content decreased, was different among the target glycol ethers. The impact of water content on the ratio values progressed as the vapor pressure of the compounds decreased, with the greatest reduction observed for compounds 1-BP (about 80%) and 2-HE (about 90%). Estimation of Emission Rates. The next step was to examine whether the dependence on water affected the correlation between the initial emission rate (ER0) and the headspace concentration (Cv). For this, the same five mixtures (M-1 to M-5) were subjected to dynamic chamber tests to determine the ER0 of all seven glycol ethers under defined environmental conditions (19). The ER0 was calculated from the time-dependent concentrations in the chamber according to eq 4. Linear relationships between measured ER0 and headspace concentration (Cv) were observed for all seven glycol ethers. Figure 3 illustrates such relationship for
FIGURE 3. Relationship between headspace concentrations and measured initial emission rates for 2-butoxyethanol in five solvent mixtures. The slope (0.5627 m h-1) represents the mass transfer coefficient.
TABLE 2. Correlation between Initial Emission Rate (ER0) and Headspace Concentration (Cv) or Mole Fraction in the Mixture (x) ER0 ) km × Cv
2-methoxyethanol 1-methoxy 2-propanol 2-ethoxyethanol 1-ethoxy 2-propanol 2-butoxyethanol 1-butoxy 2-propanol 2-hexyloxyethanol
ER0 ) km × P0 x
km
r2
km P0
r2
0.2216 0.2311 0.1805 0.2330 0.5627 0.0980 0.0046
0.99 0.99 0.98 0.99 0.99 0.99 0.99
17719 32954 14083 40454 1991.7 9064 147.14
0.99 0.97 0.99 0.89 0.15 0.10 0.23
2-butoxyethanol as an example. The slope (equal to km) and coefficient of determination (r2) of the correlations for all seven glycol ethers are summarized in Table 2. For comparison, the correlation of ER0 to mole fraction in the liquid mixture is presented in Table 2 as well. It is evident that the correlation of ER0 to the headspace concentration (ER0 ) kmi × Chs) is much better than that to mole fractions in the mixtures (ER0 ) kmi × P0 × x). In the first case, high r2 values (0.98-1.00) were achieved for all glycol ethers in the mixture. But in the latter case, the correlations for the three glycol ethers with higher boiling points (and lower vapor pressures) were very poor. Results in Table 2 indicated that although the water content in the mixtures had an impact on the ratio of headspace concentration to its mole fraction in the liquid, it did not affect the correlation between ER0 and Cv. The different slopes for the seven glycol ethers confirmed that the mass transfer coefficient (km) is not only a function of the physical environment in which the emissions occurred, but also depends on the properties of the compound (8). The impact of VOC diffusion coefficients (D) and surface air velocity on emissions has been discussed by others (10, 11). It was mentioned that km is a function of D2/3 when the Schmidt number (Sc) > 1 (in the case of majority of VOCs) or D1/2 when Sc < 1. In addition to the relationship between km and D in these water-based solvent mixtures, the correlation between km and the vapor pressure of VOCs, which determines the partition of the chemical in liquid and gas phases, was also examined. Both diffusion coefficient (raised to the 2/3 power) and vapor pressure correlate well with km values (Figure 4). Excluding the value of 2-BE, the correlation value, r2, was greater than 0.98 in both cases. A vapor pressure dependent km of alkanes (nonane, 1.16 m h-1; decane, 1.10; undecane 1.02; dodecane, 0.94) in petroleum-based products showed a similar trend (4). The km (0.5627) for 2-BE was not in line with the other glycol ethers tested. If the curve in Figure 4 applies to all glycol ethers, the km value of 2-BE
FIGURE 4. Relationship of mass transfer coefficient and diffusivity (left scale) or vapor pressure (right scale) of glycol ethers. 2-BE value was not included from the correlation equation. should be around 0.15. The reason for the higher km value for 2-BE is not clear. However, as described below, the km value of 0.5627 for 2-BE resulted in a calculated ER0 that was close to that measured. 2-BE and Other VOCs in Consumer Products. Before conducting dynamic chamber tests, concentrations of 2-BE and other VOCs in the samples as well as in their headspace concentrations were determined in 30 consumer products, which were selected based on a recent survey on the current uses of 2-BE and 2-ME in Canada (16). Measured 2-BE levels in the products (expressed as percent weight) and in headspace (mg m-3) are summarized in Table 3, along with the concentration range of 2-BE in products stated in the Material Safety Data Sheet (MSDS) by the product suppliers. In most cases, the measured 2-BE concentrations were within the range stated by the product supplier. However, there were a few exceptions, notably product 16 (lacquer thinner) where the measured concentration was twice the stated value, and other products (1, 2, 3, and 20). Besides 2-BE, three other target glycol ethers (1-methoxy 2-propanol in products 12 and 19, 1-butoxy 2-propanol in 19, and 2-hexyloxyethanol in 13) were also detected in the products as well as in their respective headspace samples (Table 4). This indicated that because of their similar properties, these glycol ethers are also being used in consumer products. Glycol ethers other than 2-BE were also reported in another study of seven consumer products and their headspace concentrations (15). In addition to the target analytes, levels of other major VOCs in both products and their headspace were also measured. The list in Appendices A1 and A2 is available as Supporting Information and provides screening results on the levels of VOCs in the products. Predicted and Measured ER0 of 2-BE from Products. Dynamic chamber tests are a time-consuming and costly undertaking. To verify the observations on the five solvent mixtures, 20 out of 30 products covering a wide range of 2-BE concentrations (1.0-58.1%) were tested to determine the ER0s of 2-BE under the same environmental chamber test conditions as the solvent mixtures. In this way, the mass transfer coefficient (km) of 2-BE (0.5627) generated from the solvent mixtures (Table 2) could be applied to the prediction of ER0 of 2-BE from consumer products. Table 3 lists both measured and predicted ER0 for 2-BE. When measured 2-BE initial emission rates were plotted against predicted ones, the slope and the correlation r2 value were 1.01 and 0.89, respectively (Figure 5), indicating good agreement between the two. The results presented in this paper further demonstrated that the VB model can be applied to water-based products as well. Although the ratio of headspace concentration to mole fraction in the liquid mixture varied depending on the VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8217
TABLE 3. Concentration Values of 2-BE and Its Initial Emission Rates from Products
product
product type
stated concn range (%)a
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
polyurethane interior stain water-based interior stain latex interior stain semi transparent stain alkyd interior stain aerosol paint interior wood stain general paint water-based glass and mirror cleaner water-based cleaner aqueous acidic cleaner instant spot remover glass and surface cleaner heavy duty degreaser paint and varnish remover lacquer thinner liquid carpet stain remover tire and wheel cleaner glass cleaner water-based cleaner water-based cleaner and brightener water-based odor neutralizer general cleaner, water-based water-based cleaner general cleaner, degreaser water-based cleaner, degreaser foaming cleaner all purpose cleaner, water-based glass cleaner, water-based general cleaner, water-based
3-7 1-5 0.5-1.5 1-5 10-30 5-10 5-10 1-5 >1 15-25 1-4 7 0.5-1.5 3-7
