Separation and Isolation of Volatile Organic Compounds Using Vacuum Distillation with GC/MS Determination
Michael H. Hiatt
U.S. Environmental Protection Agency, National Exposure Research Laboratory
Environmental Sciences Division. P.O. Box 93478, Las Vegas, Nevada 89193-3478
Phone: 702 798 2381. Fax: 702 798 2142.
E-mail: hiatt.mike@epa.gov.
and
David R. Youngman and Joseph R. Donnelly
Lockheed Environmental Systems & Technologies Co.
980 Kelly Johnson Dr., Las Vegas, NV 89119
[Note: minor content and formatting differences exist between
this web
version and the published version]
Abstract
Vacuum distillation of water, soil, oil and fish samples
is presented as an alternative technique for determining volatile organic
components (VOCs). Analyses of samples containing VOCs and non-VOCs were
performed to evaluate method limitations. Analyte recoveries were found
to relate closely with boiling point unless a compound's water solubility
exceeded 5 g/L. Recovery, precision, and method detection limits for VOCs
demonstrate this technology is appropriate for environmental samples.
Introduction
Determining volatile organic compounds (VOC) concentrations in environmental
matrices is one of the most important and routine analysis. VOCs are addressed
as a major group of analytes for the Comprehensive Environmental Response
Compensation and Liability Act (CERCLA), Resource Conversation and Recovery
Act (RCRA), and the Clean Water Act (CWA). The widespread occurrence of
VOCs in the environment and the potential of using VOCs as indicator parameters
for contamination plumes from a point source make their accurate and routine
measurement an important issue. These considerations make any improvements
in VOC determinations worthwhile to the EPA and relevant to the analytical
community.
The most widespread technologies used to determine VOCs is purge and trap, developed by the EPA to determine VOCs in water.1,2 Purge and trap is incorporated in EPA methods for water and soil.3-5 This technique is optimal for drinking water, but has difficulties when purging is hindered by elevated organic content. The trapping material also introduces difficulties that have been summarized.6
Vacuum distillation was developed as an alternative technique to determine VOCs in nonwater environmental matrices. The vacuum distillation of sediments and fish tissue provided greater VOC recoveries compared to purge and trap.7 This approach was utilized by other investigators for analyzing larger water samples8 and algae.9
A further refinement of vacuum distillation was incorporated to eliminate an adsorbent trap and directly interface the apparatus to a gas chromatograph/mass spectrometer (GC/MS) equipped with a fused silica capillary column.10 The removal of the adsorbent trapping material eliminated a major source of problems that plague volatile compound determinations.6 This new apparatus, however, was cumbersome and required continuous operator activity. The vacuum distillation apparatus described in this study provides further improvements to streamline operation. The utilization of a single condenser coil facilitated temperature control and also eliminated a series of temperature control baths. This makes vacuum distillation conducive to automation and an attractive approach to determine volatile compounds.
This work is a result of EPA's investigation of vacuum distillation using the new apparatus to develop a method for determining volatile organic compounds in environmental samples and hazardous waste. A secondary goal was a single distillation procedure addressing multiple environmental matrices. Such a procedure eliminates costs and confusion associated with specializing apparatus, and calibration by matrix. The vacuum distillation procedure used for this study is in the approval process for an EPA test method.4
Vacuum distillation apparatus
The vacuum distillation apparatus (see Figure 1) consists of a sample
chamber connected to a condenser which is attached to a heated six-port
sampling valve (V4). This sampling valve is connected to a condenser,
vacuum pump, cryotrap, and gas chromatograph (GC). This study used a mass
spectrometer, interfaced to the GC, for detection and quantitation.
The circulating system which controls the condenser coils temperature includes a cryogenic cooler with a cold reservoir (Neslab ULT80DD) and an elevated (45 °C) temperature bath. The fluid for both baths is isopropyl alcohol and the routing of isopropyl alcohol through the system is controlled by the bath fluid valve (V2). The cold isopropyl alcohol (-5 °C) is pumped through the condenser by the cryogenic cooler. The warm isopropyl alcohol is pumped through the condenser using a peristaltic pump (Cole Parmer 6-600 rpm).
The apparatus valves, transfer lines and condenser exterior are heated to 80 °C using thermal strips. This temperature is sufficient to prevent condensation of analytes onto condenser walls, valves, and connections. The temperature of the transfer line from the sampling valve to the gas chromatograph was maintained at 150 °C.
Pirani gauges (Edwards Model 1001 with gauge head Model PRH10K) were installed at the sample chamber, condenser and vacuum pump (Alcatel Model ZM2012A) for pressure monitoring.
Some key dimensions of the apparatus are: (a) cyrotrap, 8" x 1/8" stainless steel tubing; (b) condenser, 12" x 2" with ground glass ends and Buna-N O-ring seals; (c) tubing between six port valve and GC inlet, 1/16" fused silica lined stainless steel; (d) a 1/8" fittings six port Valco sampling valves with (V4) port diameter (0.040") stainless steel with Teflon internal parts; (e) 0.187" I.D. manual valves (V1, V3); Whitey SS-43XS4, stainless steel with Teflon internal parts; and (f) Valve V2 is a simplification of Multiple 3-way Whitey B-43x54 valves.
Reagents
Stock standard solutions were prepared in methanol using assayed liquids
or gases, and were stored with minimal headspace at -10 to -20 °C and
protected from light. Fresh gas standards (chloromethane, bromomethane,
vinyl chloride, and chloroethane) were prepared weekly. Stock solutions
in methanol were prepared from pure standard materials or purchased as
certified solutions (Supelco, Bellefonte PA). These solutions were used
for calibration and spiking sample matrices.
Reagent water was generated by passing tap water through a carbon filter bed containing about 450 g of activated carbon (Calgon Corp., Filtrasorb-300), or by using a water purification system (Millipore Super-Q). Purge-and-trap grade methanol (Burdick and Jackson, Muskegon, Michigan) was utilized for the study.
Pharmaceutical grade Osco brand cod liver oil, and food-grade Starkist brand canned tuna in water were used for sample matrices.
Spiked samples
Samples utilized in this study were spiked with 250 ng of each analyte
dissolved in methanol. Water, soil, and tissue were spiked directly using
a gas-tight syringe after the sample aliquot was transferred to the sample
chamber.
The methanol spike solutions, however, were not miscible with the oil matrix studied and, therefore, an intermediate spiking step was required. A 200 µL solution consisting of water saturated with lecithin was added to the oil in the sample chamber. The spike was then injected into the aqueous phase and the sample swirled until there was an even emulsion. When combinations of oil and other matrices were to be investigated, the oil component was first spiked and then the additional matrix was added.
Vacuum distillation procedure
The sample chamber, containing the prepared spike sample, was attached
to the apparatus (Figure 1) and sampling valve (V4) is switched to the
distillation position allowing the cryotrap and condenser to be evacuated.
The sample chamber valve (V1) is opened allowing sample vapors to pass
over the condenser coil, (chilled to below -5 °C), resulting in the condensation
of water vapor on the condenser coil. The vapors not condensed on the
condenser coils were collected cryogenically in a section of 1/8 inch
stainless steel tubing, chilled with liquid nitrogen (-196 °C). After
ten minutes of vacuum distillation, sampling valve (V4) is switched to
the desorb position, connecting the cryotrap to the gas chromatograph.
The cryotrap condensate is then thermally desorbed and transferred to
the gas chromatograph using helium carrier gas.
During the cryotrap desorption, a ten minute decontamination cycle is performed which requires heating the condenser coil with 45 °C isopropyl alcohol and evacuating resultant vapors through the pump valve (V3).
GC/MS Procedure
A fused silica capillary GC column (30 m x 0.53 mm ID, 3 µm film thickness,
DB-624 from J&W Scientific, Folsom, CA) was used for this study. The
GC column was temperature programmed from 10 °C (3 min) to 230 °C at 5
°C/min. The Hewlett-Packard 5970B mass spectrometer was operated in EI
mode at 70 eV, scanning 35 to 350 daltons at a rate of 0.82 second per
scan. A heated jet separator was employed to interface the GC column to
the mass spectrometer.
Results and Discussion
An analyte's presence in the vacuum distillate was found to be related
to its boiling point. Samples were spiked with VOCs and semivolatile compounds
representing a wide range of boiling points (-24 to 285 °C). The recovery
of analytes compared to analyte's boiling points are presented in Figure
2. Only compounds with water solubility less than 5 g/L were used to generate
Figure 2. Recoveries were calculated as comparison to analyte standards
transferred through the apparatus with the condenser coil at 45 °C.
Volatile compounds that are very water soluble, such as alcohols, ethers and amines are not usually considered as VOCs due to their low volatility in the presence of water. To evaluate solubility effects when using vacuum distillation, water samples spiked with analytes of varied solubility (less than 1 g/L to infinitely miscible) were analyzed. Compounds with water solubilities greater that 5g/L demonstrated lower recoveries than expected by their boiling points estimated using Figure 2. Compounds that were very soluble in water demonstrated the greatest drop in recovery and made their detection difficult. We did not attempt to minimize analyte losses due to solubility, however, sample pre-treatment, such as adding salt, would likely improve recoveries of miscible analytes.
Spiked samples, representing the range of environmental matrices, were analyzed to determine the VOCs recoveries and precision. Recovery data for the VOCs resulting from five replicate analyses of water, soil, and tissue samples are presented in Table 1.
The effect of organic content on recovery of VOCs from water and soil was also evaluated. When the organic content (emulsified cod liver oil) of water reached 20% the VOC recoveries were similar to those listed for oil in Table 1. However, when high organic water samples are analyzed using a purge and trap procedure, an uncontained froth is produced. The purge pushes the foam and contaminates the adsorption trap destroying the analysis. These samples were successfully analyzed and were easily contained within the sample chamber using the vacuum distillation method.
Oil samples proved to be the most difficult matrix from which to recover VOCs and, therefore, only 1g aliquots were analyzed. Heating and sonication of the sample and extending distillation times were evaluated as means to increase recoveries. Of these only heating the sample during the analyses improved recoveries measurably. Similar recovery improvements were also obtained by creating an emulsion of water and oil. It was reasoned that heating the sample was not a practical improvement as heating could introduce artifacts. Heating the sample could also diffuse undesirable compounds throughout the apparatus that would be difficult to remove during the decontamination cycle. The recoveries of VOCs from the oil and water emulsion are those data contained in Table 1.
Fish tissue behaved similar to cod liver oil in its impact on VOC distillation yields. As with oil, the addition of water did not substantially improve VOC recovery but there were improvements in sample handling, precision and minimization of the amount of polar compounds in the distillate that interfered with VOC GC/MS quantitation.
Method detection limits for water, soil, oil and tissue were determined in accordance to RCRA guidelines.4 These data are provided in Table 2.
Conclusions
It was found that VOCs that are not miscible in water were minimally impacted
by matrix and losses were closely related to boiling points. Method variations
to address different matrices are not necessary. Vacuum distillation is
an attractive technique for determining VOCs in both low and high organic
content samples.
Notice
Although the research described in this article has been funded by the
United States Environmental Protection Agency, through Contract Number
68-CO-0049 to the Lockheed Environmental Systems and Technologies Company,
it has not been subjected to Agency review. Therefore, it does not necessarily
reflect the views of the Agency, and no official endorsement should be
inferred. Mention of trade names or commercial products does not constitute
an endorsement for use.
References
1. Bellar, et al.; J.American Water Works Assoc. 1974, 66,739.
2. Bellar, T.A.; Lichtenberg, J.J.; Kroner, R.C. J. Amer. Water
Works Assoc. 1974, 66, 703.
3. U.S. Environmental Protection Agency. 1990. Contract Laboratory
Program, Statement of Work for Organic Analysis, Multi-media, Multi-concentration.
Document Number OLM01.0, Including Revisions, OLM01.1 - OLM01.8, Dec.
1990-Sept. 1991. U.S. Environmental Protection Agency, Cincinnati, Ohio.
4. U.S. Environmental Protection Agency. 1992. Test Methods for
Determining Solid Waste, SW-846.
5. U.S. Environmental Protection Agency, 1988, Methods for the
Determination of Organic Compounds in Drinking Water.
6. Shirey, R.E. Cole, S.B. Supelco Reporter, 1993, 12,21.X3.
7. Hiatt, M.H.; Anal. Chem. 1981, 53,
8. Kozloski, R.P. J. Chromatogr. 1985, 346,
408.
9. Newman, K.A.; Gschwend, P.M. Limnol. Oceanogr. 1987,
32, 702.
10. Hiatt, M.H.; Anal. Chem. 1983, 55,
506.