Stability of Trace-Level Volatile Organic Compounds Stored in Canisters and Tedlar bags

Fifty-six volatile organic compounds (VOCs), known to be ozone precursors, were stored in three media (SUMMA and Silocan canisters and Tedlar bags) to evaluate their stability in these storage media. An analysis of samples of air followed the procedures described in the US EPA Method TO-15, and was performed using gas chromatograph (GC) equipped with a mass spectrometer (MS). The first-order decay model matched 87 % of the observations. These 56 VOCs were classified into four groups-alkanes, alkenes, aromatics and biogenics. Alkenes and biogenics exhibited lower recovery rates than those of alkanes and aromatics. After a seven-day (7-d) storage period, 87 % of alkenes could be recovered from canister storage and 82 % were recovered from Tedlar bag storage. Isoprene, a major component in biogenic VOCs, exhibited a recovery rate of only 75±8 % after storage for seven days in canisters and Tedlar bags. Storage conditions (humidity and temperature) affected the degradation constant of each VOC. The calculated average half-lifes of 56 VOCs for SUMMA canisters, Silcocan canisters, and Tedlar bags were 45±6, 52±6, and 37±4 days, respectively. The VOCs stored in Tedlar bags had a lower recovery than those stored in canisters.


Introduction
Many diverse media are used for collecting gaseous or whole air samples, including for example, stainless steel canisters and Tedlar bags.Fax: +886-5-531-2069 E-mail address: hsiehcc@yuntech.edu.twTO-15 (U.S. EPA, 1997) is used to measure 97 VOCs that represents 189 hazardous air pollutants (HAPs) listed in the Clean Air Act Amendment of 1990.Method TO-14 has been used to measure non-polar VOCs, while TO-15 is used for both non-polar and polar VOCs.The stability of several compounds to which Method TO-15 applies, has been evaluated under different storage conditions.
The loss of VOCs may be due to the physical adsorption of VOCs on canister walls, the dissolution of VOCs in water condensed in the canisters, chemical reaction, hydrolysis and biological degradation (Coutant, 1993).
To support calibration and performance audits, the stability of 39 and 34 VOCs in compressed gas cylinders was determined from 1 ppbv to 10 ppmv (parts per million by volume) and 5 to 700 ppmv, respectively (Jayanty et al;1992).The results indicate that the concentrations of several VOCs varied by under 10 % over several months, but that several compounds were found to be unstable at the ppbv or even the ppmv level.The unstable compounds at the ppbv level were ethylene oxide, propylene oxide and trans 1,4-dichloro-2-butene.
Six other compounds were found to be unstable in compressed gas cylinders at the ppm level (aniline, cyclohexane, p-dichlorobenzene, ethylamine, 1,2-dibromoethylene, and formaldehyde).A review of the literature on the stability of 52 HAPs stored in stainless steel canisters showed that canister stability data were widely variable for a dozen compounds (Kelly and Holdren, 1995).The stability of 194 compounds stored in SUMMA canisters at the ppbv level was examined (Brymer et al, 1996).Although the concentrations of 168 of the 194 compounds studied did not change, a few of the compounds studied including methyl mercaptan, ethyl mercaptan, butyl mercaptan, dimethyl acetal and bis ether) exhibited high variability, low recovery or poor storage stability.
Seven aldehydes and four terpenes stored in SUMMA canisters were recovered at concentrations of 3-5 ppbv (Batterman et al., 1998).
The concentrations of all terpenes and most aldehydes decreased markedly during the first hour.
On the 16th day, the recovery of most compounds was reduced to 50% of the initial concentration.
An alternative method for collecting air samples involves Tedlar bags.Tedlar bags are used according to EPA Method 18 for measuring the emission of VOCs from high-concentration process streams (U.S. EPA, 1993).The bags are made of a chlorinated substance and are resistant to corrosion by most solvents and chemicals.However, some VOCs have been reported to degrade significantly.
(For example, 70 %of methanol degrades after being stored in 10L Tedlar bags for 6 hrs and ethylbenzene and o-xylene degrades to 13% and 11% concentrations, respectively after 2 days) (Lipari, 1990).In contrast, methanol and formaldehyde stored in 60L Tedlar bags tend to degrade slowly (Andino and Butler, 1991), apparently because they have a smaller surfaceto-volume ratio than other substances.

Materials and Methods
The research included the preparation and storage of these 56 VOCs.These 56 VOCs are classified into four categories-alkanes, alkenes, aromatics and biogenics.Good results were achieved.Table 1 shows MDL and linearity coefficient (R2) for each VOC.
Triplicate samples were prepared and analyzed, so the reported concentrations are the average measured concentrations of the compounds.

7-day Recovery Estimates
A storage period of seven days was used for evaluating stability since this is the period typically required to complete collection, transportation and analysis of such samples.Recoveries are obtained from the initial analysis of each sample, and are averaged across the three samples.The decay kinetic constants in Table 3 were used to estimate the recovery of the VOCs with R 2 > 0.7.The results shown in Fig. 3 show that the recovery of each VOC varies widely among the three media (SUMMA canister, Silcocan canisters and Tedlar bag).Of the four categories, the highest recovery was obtained for alkanes (middle in Fig. 3) and lowest was of biogenics.Isoprene, as one of the most importantly reactive hydrocarbons in atmospheric chemistry, with a boiling point of 34 ℃, had the lowest recovery (75 ± 8 percent) rate of any of the nine biogenics evaluated in this study.

Half-life Estimates
The VOC decay kinetic constants can be further used to estimate the compound's half-lives, which are defined as the time for 50 % of the initial concentration to disappear.Figure 4 shows the canisters, Silcocan canisters, and Tedlar bags were 45±6, 52±6 and 37±4 days, respectively.The half-lives could be used to calculate the maximum storage periods.However, such estimates may be too low because losses at the beginning of the storage and the effects of old or damaged canisters are neglected; rather, the estimates apply under laboratory conditions.Thus, a "safety factor" should be included, by, say, halving the estimated maximum storage time (Batterman et al., 1998).

Inter-canister Reproducibility
Reproducibility was calculated as the average coefficient of variation (COV) of concentrations measured among the three types of canisters (or bags) for each set of samples.Adsorption was considered as the chemical interaction of vapors on "active sites" at the canister walls (Freeman et al., 1994).The results imply that Silcocan canisters had fewer active sites and thus adsorbed less.Tedlar bags gave more active sites than canisters, and thus absorb more.
A second mechanism that affects recovery is the dissolution of soluble compounds in condensed water present in the canister.As the pressure in a canister drops because of withdrawal of a sample, the quantity of condensed water falls and concentrations of dissolved compounds may thus T: Temperature (℃), RH: Relative Humidity increase (Coutant, 1993).This effect depends on the compound's solubility and the presence of Different analytical methods may require different sampling media.The Compendium Method TO-14 targets 41 VOCs (U.S. EPA, 1988).These compounds have been successfully stored in canisters over periods of several days to months at ppbv (parts per billion by volume) levels.Method *Corresponding author: Tel: +886-5-534-2601 ext.4413 improve sample-wetting characteristics.The TO-14 canister cleaning procedure consisted of repeated cycles of pressurization with humidified zero air, followed by evacuation to 0.1 torr for 1h, with heating to 120℃.Each canister was first filled with humidified zero air and analyzed by the procedure described below to determine its cleanliness.Samples prepared according to the procedures ofBatterman et al. (1998).Clean evaluated canisters were humidified by injecting 69 µl and 207 µl of HPLC-grade water and then filled with dry zero air to near atmospheric pressure, to make up 30 and 90% RH at 25 ℃, respectively.Volumes of 108 µl and 323 µl of HPLC-grade water were injected into clean evaluated canisters to yield 30 and 90 % RH at 35 , respectively.The samples thus prepared ℃ were stored in an oven to maintain the target temperature.Organic compounds were prepared using a gas standard (Environ-Mat Ozone Precursor Mixture, Matheson Gas Products, Georgia, USA) and biogenic compounds were prepared using a liquid standard (Supelco, PA, USA).The gas sample was diluted with dry zero air in Tedlar bags and then transferred into stainless steel canisters.Liquid samples were diluted in methanol.Then, the mixture was injected into Tedlar bags and canisters, with proper dilution.All samples thus prepared were heated using an infrared lamp to keep VOC in the vapor phase.Most samples were prepared at concentrations (5 to 30 ppbv) similar to ambient levels.All Tedlar bags were kept in black bags to prevent photochemical reactions.The procedures described in Method TO-15 were followed to analyze the samples.Samples of air were concentrated cryogenically using a preconcentrator (7100, Entech Instrument, Inc., CA) and transferred by a heated transfer line and a cryofocus unit to a gas chromatograph (GC) (5890 SeriesⅡplus, Hewlett Packard, Palo Alto, CA).The flow of a 400ml sample of air was metered by a mass-flow-controlled meter (Sierra Instrument Inc., Monterey, CA).The GC used a HP-1 60m capillary column (0.25-mm diameter x 1.8-µm film thickness, Hewlett Packard) and nitrogen as the carrier gas.VOCs were detected by massspectrometry (MS) (Model 5972, Hewlett Packard).The MS scanned a wide range of mass to charge ratios (40-200).A PC workstation (G1034C Chemstation, Hewlett Packard) was used to acquire and manipulate data.The starting temperature in the GC oven was 10 ℃ and was maintained for 3 min.The temperature was then increased at 8℃ per min to 130℃ , at which it was maintained for 5 min.Subsequently, the temperature was increased at 5 ℃ per min to 180 ℃ , at which it was maintained for 5 minutes.Quality control and quality assurance requirements involved establishing GC retention times, calibration curves and MDLs, as well as studying the reproducibility of the results for all compounds analyzed.Furthermore, an internal standard (toluene-d8) was used for each analysis to verify the stability of the MS.Additionally, cyclohexanal-d12 was analyzed daily as a reference standard for determining the performance of the instruments used.A response deviation of more than 15% resulted in full instrument recalibration.

Figure 2 .
Figure 2. Decay rate constants for four categories of VOCs under various storage conditions.

Figure 3 .
Figure 3. Recovery rates of 56 VOCs under variousstorage conditions after seven days

Figure 4 .
Figure 4. Half-life of 56 VOCs under various storage conditions.
condensed water in the canister.A third mechanism involves a gas-phase reaction and the subsequent transformation and loss of objective compounds.Such a reaction may be promoted by the presence of ozone or other reactive species.Information available in the literature and in this study is insufficient to identify the details of the loss mechanisms.A general mechanism that affects all tested compounds is expected since compounds stored at a lower temperature are more stable and their are more reproducible, probably because the reaction rates are lower.Moreover, adding water or increasing humidity may also reduce adsorption losses of several VOCs.The low recoveries in dry canisters verify the occurrence of physical adsorption.Those storage conditions applied in this study were limited to typical environmental conditions in the Taiwan area.Since the potential for adsorbing VOCs and water in a canister changes with temperature, recovery must be further investigated over a range of temperatures (such as 0-40 o C) and humidities, to separate out sorption losses.

Table 1 .
Properties of targeted VOCs Table 1 lists 56 VOCsalong with their CAS numbers, boiling points, method detection limits (MDL), and whether they are listed in EPA Methods TO-14 and TO-15.Inc., Eighty Four, PA).According to the manufacturer's information, the interior of each S UMM A cani st er was subj ect ed t o an electropolishing treatment to reduce the surface adsorption of VOCs.Each silcocan canister had been subjected to siloxane to impart resistance and MW: molecule weight (g); BP: boiling point (℃); TMP: trimethylpentane; TMB: trimethylbenzene

Table 2 .
Correlation coefficients (R 2 >0.7) for 1 st order decay under four storage conditions.

Table 3
VOCs have k values close to the average k value of the 56 compounds.Since the k values of compounds stored in SUMMA and Silcocan canisters do not statistically significantly different,Table 4 presents the mean k values for the two kinds of canisters was tabulated, and compared

Table 3 .
Degradation constant (k, 100 d -l) of 56 VOCs in three media for R 2 ＞0.7

Table 4 .
Degradation constant (k, 100 d -1 ) of four Table 5 shows the mean COV for each of the four groups under different storage conditions.Reproducibility was in all cases between 9 and 30%.The reproducibility

Table 5 .
Average reproducibility of four groups under for storage conditions expressed as COV in