Anal. Chem. 2001, 73, 3624-3631
Purge-and-Membrane Mass Spectrometry, A Screening Method for Analysis of VOCs from Soil Samples Marja Ojala,† Ismo Mattila,† Virpi Tarkiainen,† Timo Sa 1 rme,† Raimo A. Ketola,† Anniina Ma 1a 1 tta 1 nen,‡ §,⊥ ,† Risto Kostiainen, and Tapio Kotiaho*
VTT Chemical Technology, P.O. Box 1401, FIN-02044 VTT, Finland, Golder Associates, Ruosilankuja 3E, FIN-00390 Helsinki, Finland, Department of Pharmacy, Division of Pharmaceutical Chemistry, P.O. Box 56, FIN-00014 University of Helsinki, Finland, and Viikki Drug Discovery Technology Center, Department of Pharmacy, P.O. Box 56, FIN-00014 University of Helsinki, Finland
Purge-and-membrane mass spectrometry (PAM-MS) is a combination of dynamic headspace sampling and membrane extraction. A new and simple purge-and-membrane sampler is introduced and its basic testing results for the analysis of VOCs in soil samples are reported. Soil moisture had no effect on desorption times in the case of sand, but the desorption times increased when the content of organic matter in the soil sample (garden soil) increased. The longest desorption times were measured with dry garden soil samples. For both types of samples, minor differences in desorption peak areas were observed between 10 and 20% moisture. Detection limits of the VOCs varied in the range 2-150 µg/kg, depending on the soil type. Good linearity (correlation coefficient > 0.990) was observed in the range 0.5-50 mg/kg. Aging of the spiked soil samples had only a slight effect on desorption peak areas for samples stored at 5 °C up to two weeks, but after six months of storing, differences were observed between dry sand and moistened garden soil. In both cases, peak areas were diminished. On average, 46% of compounds could be desorbed from the aged sand and 86% from the aged garden soil. The modified vapor fortification method was used in preparing standard soil samples, which were analyzed by static headspace gas chromatography (HSGC) and PAM-MS. Some authentic soil samples were also analyzed using both of these techniques. Many of the vapor fortification samples and the authentic samples were also analyzed in another laboratory by HSGC. The agreement between the methods and the laboratories was generally good. Volatile organic compounds (VOCs) are the most common contaminants of soil. The main emission sources are industry, traffic, and energy production. Because of the high volatility of VOCs, their analysis in soil samples is very difficult and special * Corresponding author. Tel: +358-9-4565277. Fax: +358-9-4567026. E-mail:
[email protected]. † VTT Chemical Technology. ‡ Golder Associates. § Department of Pharmacy. ⊥ Viikki Drug Discovery Technology Center.
3624 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
care must be taken in the sampling and storing of the samples before analysis. The pretreatment of the samples should be as simple as possible to avoid loss of volatile analytes. Various analytical methods1 have been used to analyze VOCs in contaminated soils. The most frequently used methods are headspace and purge-and-trap technique connected either to a gas chromatograph2-5 or a mass spectrometer.6 Multiple headspace extraction7 and headspace solid-phase microextraction8,9 have also been used. To avoid transportation of samples, various portable on-site analysis systems10,11 have been developed. In headspace analysis, special care must be taken because quantitative results are strongly dependent on the partitioning of the VOCs between gas, liquid, and solid phases.2 Hewitt et al.12,13 compared laboratory purge-and-trap gas chromatography/mass spectrometry and onsite aqueous extraction headspace/gas chromatography in the analysis of VOCs in soils. Usually the agreement was good, but significant differences in the results were also observed. On-site and on-line analytical methods14 have also been applied for VOCs in water and soil samples. Among these methods, membrane inlet mass spectrometry (MIMS) is one of the most suitable techniques for VOCs in air15 and water samples;16,17 (1) Clement, R. E.; Yang, P. W.; Koester, C. J. Anal. Chem. 1997, 69, 251R287R. (2) Pavlostathis, S. G.; Mathavan, G. N. Environ. Technol. 1992, 13, 23-33. (3) Volce, T. C.; Kolb, B. Environ. Sci. Technol. 1993, 27, 709-713. (4) Kolb, B.; Bichler, C.; Auer, M.; Voice, T. C. J. High Resolut. Chromatogr. 1994, 17, 299-302. (5) Markelov, M.; Guzowski, J. P., Jr. Anal.Chim. Acta 1993, 276, 235. (6) Schumacher, B. A.; Ward, S. E. Environ. Sci. Technol. 1997, 31, 22872291. (7) Maggio, A.; Milana, M. R.; Denaro, M.; Feliciani, R.; Gramiccioni, L. J. High Resolut. Chromatogr. 1991, 14, 618-620. (8) Llompart, M.; Li, K.; Fingas, M. Talanta 1999, 48, 451-459. (9) James, J. J.; Stack, M. J. High Resolut. Chromatogr. 1996, 19, 515-519. (10) Denley, E.; Hopper, D. Field Analytical Methods for Hazardous Wastes and Toxic Chemicals, Las Vegas, NV, January 29-31,1997; pp 752-770. (11) Eckenrode, B. A. Field Anal. Chem. Technol. 1998, 2, 3-20. (12) Hewitt, A. D.; Miyares, P. H.; Leggett, D. C.; Jenkins, T. F. Environ. Sci. Technol. 1992, 26, 1932-1938. (13) Hewitt, A. D. J. AOAC Int. 1994, 77, 458-463. (14) Kotiaho, T. J. Mass Spectrom. 1996, 31, 1-15. (15) Ketola, R. A.; Ojala, M.; Sorsa, H.; Kotiaho, T.; Kostiainen, R. K. Anal. Chim. Acta 1997, 349, 359-365. (16) Lauritsen, F. R.; Kotiaho, T. Rev. Anal. Chem. 1996, 15, 237-264. (17) Johnson, R. C.; Cooks, R. G.; Allen, T.; Cisper, M. E.; Hemberger, P. H. Mass Specrom. Rev. 2000, 19, 1-37. 10.1021/ac001504i CCC: $20.00
© 2001 American Chemical Society Published on Web 06/26/2001
however, MIMS is not suitable for solid samples. Recently, static headspace-MIMS has been introduced for VOC analysis in soil samples,18 and it has also been used in water analysis.19 Purgeand-membrane mass spectrometry (PAM-MS), which combines dynamic headspace sampling and membrane inlet mass spectrometry,20-22 has proved to be very promising for the analysis of VOCs in soil samples. The method has also been applied in the analysis of VOCs in pharmaceuticals.23 In this technique, VOCs are purged from the samples using inert gas, and the gas stream is directed through a membrane into the ion source of a mass spectrometer for mass spectrometric measurement. This work describes a detailed evaluation of the suitability of the PAM-MS method for analysis of VOCs from soil samples and compares the method with HSGC. Furthermore, a new smaller and simpler-to-use second generation PAM sampler and its testing results are presented. Testing results of the new PAM sampler are also compared with the corresponding results that were obtained with the first sampler22 in order to demonstrate that the new device functions as well as or even better than the first one. Literature data indicates that the soil type, moisture content, and aging of the samples are the basic parameters to study in characterization of a new method for analysis of VOCs in soil samples.24-34 The comparison of various analytical methods is also essential for this task.12,35 In this study, the effect of soil type and moisture content was studied using sand, garden soil, and their mixtures, and moisture contents of 0, 10 and 20%. The effect of aging was studied with two different soil types, namely dry sand and moistened garden soil. In addition, numerous standard samples prepared by the vapor fortification method and authentic soil samples were analyzed using HSGC and PAM-MS techniques. EXPERIMENTAL SECTION Reagents. 1,2-Dichloroethene [540-59-0], trichloroethene [7901-6], 1,1,1-trichloroethane [71-55-6], tetrachloroethene [127-184], chloroform [67-66-3], toluene [108-88-3], benzene [71-43-2], (18) Mendes, M. A.; Sparrapan, R.; Eberlin, M. N. Anal. Chem. 2000, 72, 21662170. (19) Wenhu, D.; Kuagnan, C.; Jianli, L.; Zhenying, D. Mass Spectr. 1987, 35, 122-132. (20) Kostiainen, R.; Kotiaho, T.; Ketola, R.; Virkki, V. Chromatographia 1995, 41, 34-36. (21) Kostiainen, R.; Kotiaho, T.; Mattila, I.; Mansikka, T.; Ojala, M.; Ketola, R. A. Anal. Chem. 1998, 70, 3028-3032. (22) Ojala, M.; Mattila, I.; Sa¨rme,T.; Ketola, R. A.; Kotiaho, T. Analyst 1999, 124, 1421-1424. (23) Ojala, M.; Poutanen, M.; Mattila, I.; Ketola, R. A.; Kotiaho, T.; Kostiainen, R. Rapid Commun. Mass Spectrom. 2000, 14, 994-998. (24) Ong, S. K.; Lion, L. W. Soil Sci. Soc. Am. J. 1991, 55, 1559-1568. (25) Petersen, L. W.; El-Farhan, Y. H.; Moldrup, P.; Rolston, D. E.; Yamaguchi, T, J. Environ. Qual. 1996, 25, 1054-1063. (26) Raihala, T. S.; Wang,Y.; Jackman, A. P. J. Hazard. Mater. 1999, B65, 247265. (27) Unger, D. R.; Lam, T. T.; Schaefer, C. E.; Kosson, D. S. Environ. Sci. Technol. 1996, 30, 1081-1091. (28) Piatt, J. J.; Brusseau, M. L. Environ. Sci. Technol. 1998, 32, 1604-1608. (29) Ruiz, J.; Bilbao, R.; Murillo, M. Environ. Sci. Technol. 1998, 32, 10791084. (30) Ruiz, J.; Bilbao, R.; Murillo, M. Environ. Sci. Technol. 1999, 33, 37743780. (31) Cabbar, H. C. J. Hazard. Mater. B 1999, 68, 217-226. (32) Costanza, M. S.; Brusseau, M. L. Environ. Sci. Technol. 2000, 34, 1-11. (33) Hewitt, A. D. J. AOAC Int. 1994, 77, 735-737. (34) Minnich, M. M.; Zimmerman, J. H.; Schumacher, B. A. J. Environ. Qual. 1997, 108-114. (35) Askari, M. D. F.; Maskarinec, M. P.; Smith, S. M.; Beam, P. M.; Travis, C. C. Anal. Chem. 1996, 68, 3431-3433.
o-xylene [95-47-6], 1,3,5-trimethylbenzene [108-67-8], 1,2,4,5-tetramethylbenzene [95-93-2], heptane [142-82-5], 2-methoxy-2-methylpropane (MTBE) [1634-04-4], methyl-2-methyl-2-butyl ether (TAME) [994-05-8], and chlorobenzene [108-90-7] were obtained from Mallinckrodt, Fluka, and Sigma. The stock solutions of test compounds were prepared by dissolving 1 g of a compound in 100 mL of methanol (Mallinckrodt Specialty Chemicals Co.; Paris, KY; nanograde purity). Further dilution was made with methanol. The soils used in testing were sand (Seesand, purum, Fluka; Switzerland) and commercial garden soil (Kekkila¨ Oy, Parkano; Finland). The water and organic content of the soils were determined by drying samples at 102 °C and thereafter burning off the organic compounds at 550 °C. The following results were obtained: garden soil had a water content of 17% and 9% organics, and sand had a water content of 0% and 0% organics. The authentic soil samples used in testing were obtained from customers of VTT Chemical Technology. Vapor Fortification Standards. Fluka sand or natural sand samples (5 g) were weighed in 20 mL of headspace and PAM vials (volume, ∼20 mL) in order to prepare standard samples for both methods using the vapor fortification method.36-38 The vials containing soil samples and empty vials (for background analysis) were dried in a desiccator over sodium sulfate for 2 days. After this, an evaporating dish containing 50 mL of a solution of selected compounds in methanol was placed on the bottom of the desiccator and left there for about 10 days. Three different methanol solutions were used for spiking, which contained (1) trichloroethene, toluene, benzene, tetramethylbenzene, heptane, and MTBE, ∼0.5 g of each, (2) a simulator of gasoline, and (3) gasoline. When the desiccator was opened, the vials were left open for 5 min to allow the vapor from the headspace to evaporate, and after that, caps were put on the vials but not crimped. The PAM vials were crimped and analyzed; no pretreatment was needed. To headspace vials, 10 mL of methanol was added before the caps were crimped, and they were sonicated for 5 min. For the HSGC analysis, an aliquot of methanol (100-500 µL) was taken and placed into another 10-mL headspace vial that contained water (9.9-9.5 mL) and 1 g of NaCl. The caps of the vials were crimped, and the samples were analyzed by HSGC. Preparation of Soil Samples for Aging. Two soils, namely dry Fluka sand and moistened garden soil (moisture 17%), were weighed into the PAM vials (5 g) and spiked with the selected compounds in methanol (Table 3) (2 mg/kg). The vials were closed and stored at 5 °C for periods ranging from 1 day to 6 months before analysis. Six different storing times were used, and three replicates were made for each measurement. m-Fluorotoluene was used as an internal standard, and it was added through the septum of the sample vials before analysis. To mimic authentic soil samples, spiking using aqueous standards would be preferable,39 but in the case of volatile and not very water-soluble compounds, the preparing of homogeneous spiking standard is a very difficult task. Preparation of Spiked Soil Samples. Three soils, namely dry Fluka sand, moistened Fluka sand (moisture 17%) and (36) Hewitt, (37) Hewitt, (38) Hewitt, 28. (39) Hewitt,
A. D.; Grant, C. L. Environ. Sci. Technol. 1995, 29, 769-774. A. D. Environ. Sci. Technol. 1998, 32, 143-149. A. D.; Jenkins, T. F.; Grant, C. L. Am. Environ. Lab. 1995, 7, 25A. D. Environ. Sci. Technol. 1997, 31, 67-70.
Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3625
Table 1. Desorption Times, Defined as the Time between 0 and 90% of the Desorption Curve, of Selected Compounds from Different Soil Types and Moisture Levelsa soil type/desorption time, s compound
sand/ garden moisture, % sand garden soil, 1:1 soil
benzene o-xylene 1,3,5-trimethylbenzene tetrachloroethene MTBE
a
0 10 20 0 10 20 0 10 20 0 10 20 0 10 20
20 17 19 23 22 19 21 20 18 22 21 17 24 25 27
36 24 25 70 57 68 83 68 70 44 39 34 42 26 24
43 27 26 84 77 67 98 93 83 59 49 44 44 27 23
The errors in desorption times are estimated to be (15%.
Table 2. Detection Limitsa of Selected Compounds Measured from Dry Sand, Moistened Sandb and Moistened Commercial Garden Soilb,c ion
soil type/ detection limit, µg/kg
compound
m/z
dry sand
moist sand
moist garden soil
benzene toluene o-xylene 1,3,5-trimethylbenzene 1,2,4,5-tetramethylbenzene 1,2-dichloroethene trichloroethene 1,1,1-trichloroethane tetrachloroethene MTBE TAME
78 92 106 105 119 61 97 130 166 73 73
5 5 5 2 2 10 20 50 10 50 40
10 10 8 4 4 50 70 100 40 100 80
15 30 50 40 40 50 70 150 80 100 90
a S/N ) 3 b Moisture content, 17%. c Detection limits were measured using the SIM technique and estimated using the desorption peak heights.
M100, Specialty Silicone Products Inc.; Ballston Spa, NY) with dimensions: thickness, 25 µm; contact area, 28 mm2. PAM Device. A schematic diagram of the new semiautomatic purge-and-membrane mass spectrometric apparatus is presented in Figure 1. The main parts of the device are a membrane inlet mass spectrometer, an oven with adjustable temperature up to 200 °C (Meyer-Vastus; Monninkyla¨, Finland), a four-port valve (E4C2UWT, Valco Co; Schenkon, Switzerland), and a mass flow controller (Aalborg GFC-17, Aalborg Instruments & Controls; Orangeburg, NY). All gas lines in contact with the gas stream containing VOCs were 1/16′′ in. (1.59-mm o.d.) EFNi tubing, and they were heated to 150 °C to minimize contamination. The sample vessels were made by cutting off the bottoms of two commercial 10-mL headspace vials and connecting the truncated vials together from the cut ends. A glass sinter for supporting the sample was mounted on the bottom of the vessel during the manufacturing process. The sample was purged with synthetic air at 100 mL min-1. The air flow was directed to the bottom of the vial through the septum by a stainless steel needle, and the purged analytes were passed over the custom-built membrane inlet to the ion source of a mass spectrometer. Gas Chromatography. The gas chromatograph (GC) used for headspace GC analysis in lab 1 was an HP 5890 series II, equipped with two FIDs and an HP 7694 headspace sampler (Hewlett-Packard; Palo Alto, CA). Two separate columns were used to improve the identification capability of the headspace GC method, namely a DP-1 and a DP-1701 (J&W Scientific; Folsom, CA). Both of the columns were 30 m × 0.32 mm i.d. with a stationary-phase thickness of 1.0 µm. The temperature program 10°C/min
used in the separation was 45 °C, 5 min 98 210 °C, 10 min. The temperatures of injector and detectors were 220 °C and 250 °C, respectively. The gas chromatograph used in lab 2 was an HP 6890, equipped with an FID and a PID and with an HP 7694 headspace sampler (Hewlett-Packard; Palo Alto, CA). Two columns were used, namely an HP-5 for FID and an HP-1701, for PID. Both columns were 30 m × 0.32 mm i.d. with a stationary-phase thickness of 1 µm. The temperatures for FID and PID were 300 °C and 280 °C, respectively. The temperature of the injector was 5°C/min
220 °C. The temperature program was 40 °C, 5 min 98 80 15°C/min
moistened garden soil (moisture 17%) were weighed in PAM vials (5 g) and spiked with selected compounds (Table 2) (1-200 µg/ kg). The vials were closed, and the samples were used immediately for detection-limit measurements. For the linearity measurements, the natural soil was weighed in PAM vials (5 g) and spiked with the same compounds at six concentrations from 0.5 to 50 mg/kg. Samples were stored usually from 1 to 2 days before analysis. Mass Spectrometry. The mass spectrometer that was used was a Balzers QMG 421C quadrupole mass spectrometer (mass range, 1-500 amu) equipped with an open crossbeam electron impact (70 eV) ion source (Balzers Aktiengesellschaft, Balzers, Liechtenstein). A custom-built membrane inlet utilizing a sheet membrane, built at VTT Chemical Technology,15 was used. The temperature of the membrane inlet was typically 130 °C. The material of the sheet membrane was dimethylpolysiloxane (SSP3626 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
°C 98 275 °C, 4 min. The portable gas chromatograph used in the field tests was an HNU 311 (HNU System; Newton Highlands, MA) GC equipped with a PID (10.2 eV source). Manual injection of headspace air was used, and the temperature of the injector and detector was 100 °C. Isothermal analysis was used, with an oven temperature of 70 °C. The column used was a Wcot Ultimetal CP sil 5 CB (Chrompack International BV; EA Middelburg, The Netherlands) steel column with dimensions 25 m and 0.53-mm i.d. and with a stationary-phase thickness of 5.0 µm. RESULTS AND DISCUSSION The advanced semiautomatic PAM-MS instrument is presented in Figure 1 in the sampling mode. This PAM-MS instrument was developed on the basis of the first PAM device.22 Oven size was optimized, and all of the lines were made as short as possible. In the first PAM version, the sampling occurred through a six-port valve. To minimize contamination in the new instrument, the six-
Table 3. Results of Soil Samples Prepared by the Vapor Fortification Method PAM-MS compound
content mg/kg
benzene toluene MTBE heptane trichloroethene 1,2,3,5-tetramethylbenzene TVOCa
44 47 63 42 45 103 343
benzene toluene xylenes ethylbenzene tetramethylbenzene trimethylbenzene aliphatics MTBE TAME TVOCa
23 150 180d 2.7 128 102 225g 812
HSGC lab 1
portable HSGC
HSGC lab 2
RSD %
content mg/kg
RSD %
content mg/kg
RSD %
content mg/kg
RSD %
12 10 12 16 12 9 8
16 29 17 12 25 45 144b
12 9 12 9 9 9 9
15 36 15 nm nm nm 137c
12 8 10 nm nm nm 8
15 29 15 nm nm nm nm
16 13 13 nm nm nm nm
nm nm nm nm nm nm nm nm nm nm
nm nm nm nm nm nm nm nm nm nm
nm nm nm nm nm nm nm nm nm nm
nm nm nm nm nm nm nm nm nm nm
Gasoline-Contaminated Soil Sample 4 19 14 5 174 12 9 177 12 40 12 22 4.2e 6 2 49f 9 6 6 119 16 39 14 5 1032b 13
a TVOC (total volatile organic compounds) is the sum of identified compounds and unknown compounds. The amount of unknown compounds was estimated using toluene (lab 1) or xylene (lab 2). b Estimated with toluene. c Estimated with xylene. d The sum of xylenes and ethylbenzene. e Only 1,2,3,5-tetramethylbenzene. f Only 1,2,4-trimethylbenzene. g The sum of MTBE and TAME. nm, not measured.
Figure 1. Schematic picture of the PAM sampler showing the sampling mode of operation. Arrows indicate the direction of gas flow.
port valve was replaced by a four-port valve, and sampling did not occur through the valve, as in the first version.22 All of the lines of the new PAM device were made of electro-formed nickel (EFNi) tubing, also to minimize possible memory effects. The sample holder had places for five samples in order to allow preheating of the following samples during analysis of the previous one. Compared with the first PAM device, the new instrument is smaller, easily portable, and simpler to build. The measurement procedure with the PAM device starts by preheating the sample in the oven (80 °C) in the sample holder. After a selected preheating time (10 min), the four-port valve is switched to the sampling position, and finally, the sampling lines with needles are manually punctured through the sample vessel
septa. This last step directs the flow of purge gas through the sample. Desorption of VOCs occurs, and ion chromatograms or mass spectra for VOCs can be measured with the mass spectrometer. In the stand-by mode, a backflush of purge gas flows through all of the gas lines in contact with VOCs and the membrane inlet in order to prevent contamination and to provide a constant background for the mass spectrometer. Effects of Soil Type and Moisture Content of the Sample on Sorption. The effects of soil type and moisture on desorption peak areas and desorption times were studied using garden soil, sand, and a mixture of sand and garden soil (50/50 w/w). Three replicates of each analysis were made. The amounts of water in the samples were adjusted to 10 and 20% before the analysis. In Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3627
Figure 2. Effect of moisture on peak areas for (a) sand and (b) commercial garden soil.
addition, dry soils were also used as samples. A selected set of test compounds was used, i.e., benzene, toluene, o-xylene, 1,3,5trimethylbenzene, 1,2,4,5-tetramethylbenzene, chlorobenzene, 1,2dichloroethene, trichloroethene, 1,1,1-trichloroethane, tetrachloroethene, heptane, MTBE and TAME. Figure 2 shows the desorption peak areas from two different soil types at three different moisture contents. The highest peak areas were obtained from dry sand, especially for trimethyl-, tetramethyl-, and chlorobenzenes. In the case of the garden soil sample (Figure 2b), the highest peaks were obtained from the sample with 10% moisture. For both soil types, only minor differences in desorption peak areas were observed between 10 and 20% moisture. The greatest difference in peak areas was between dry sand and dry garden soil. When analyzing typical authentic samples, the results of the PAM method can be considered to be independent of soil type for practical purposes, because the authentic samples are never dry. The average moisture content of 54 authentic soil samples sent to our laboratory for analysis, was 20%, with variation from 54 to 4%, and more than 80% of the samples had a moisture content higher than 10%. As an example of the effect of moisture on the desorption times, typical results are presented in Table 1. The moisture content of a sample had no significant effect on desorption times 3628
Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
in sand samples, but in the case of garden soil, the longest desorption times were recorded from dry samples. Desorption times of 1,2,4,5-tetramethylbenzene and trichloroethene from dry garden soil were so long that they could not be measured. A general trend was that desorption times increased when the garden soil content, i.e., the content of organic matter in the sample, increased. In addition, desorption times normally decreased when the water content of the sample increased. Interestingly, desorption times of benzene, o-xylene and MTBE for sand/ garden soil mixtures and garden soil at 20% moisture content were the same. These results differ from those obtained with the first PAM sampler,22 but they are in good agreement with some other studies.26,30 Raihala et al.26 reported that the addition of water caused the previously adsorbed toluene to desorb immediately after the water addition. Ruiz et al.30 showed that increasing the air humidity accelerated the removal of VOCs from sand and that desorption rate increased more for nonpolar compounds than for polar compounds; however, Yeo et al.40 reported that wet desorption is slower for chlorobenzene than dry desorption, and that water had only a slight effect on the desorption of trichloroethene. The reason for the difference between our earlier results and the results presented here are not clearly understood, but the (40) Yeo, S-D.; Tuncer, E.; Akgerman, A. Sep. Sci. Technol. 1997, 32, 24972512.
Table 4. Analytical Results of Two Authentic Soil Samples Measured by PAM-MS and HSGC sample/content, mg/kg sample 1
sample 2
compound
PAM-MS
HSGC lab 1
portable HSGC
PAM-MS
benzene toluene xylenes ethylbenzene tetramethylbenzene trimethylbenzene aliphatics MTBE TVOCf
6.3 140 680a
5.0 110 610 130 27b 280c
17 >140 >350 110
6.8 23a
100 1000 400 18 2400
32 2900d
9.2 52 26 37 >490e
120
HSGC lab 1
portable HSGC
0.1 7.0 43 5.8 6.1b 33c
1.0 19 70 16
240d
300e
a The sum of xylenes and ethylbenzene. b Only 1,2,3,5-tetramethylbenzene. c Only 1,2,4-trimethylbenzene. d Estimated with toluene. e Estimated with xylene. f TVOC (total volatile organic compounds) is the sum of identified compounds and unknown compounds. The amount of unknown compounds was estimated using toluene (lab 1) or xylene (lab 2).
difference might be due to the different spiking procedure. In the earlier study, the volume of the spiking methanol solution was 100 µL instead of the 10 µL volume used in this study. Several studies concerning the effects of soil type and moisture content on the VOC sorption can be found in the literature.26,28,29,31,40 The organic matter content has been reported to have an effect on desorption times, and even the type of organic matter (humic soil vs fulvic soil) at low concentrations (less than 1%) can have an effect on adsorption and desorption. For example, the adsorption/desorption times for naphthalene from humic soil are approximately double those from fulvic soil. Yeo et al.40 reported that the initial soil loadings, compounds, and moisture content have an effect on desorption. According to their results, the effect of moisture on desorption efficiency was significant for chlorobenzene. Raihala et al.26 and Ruiz et al.30 reported that moisture decreases desorption times (see above). Our observations are similar to the results presented in the literature. On the basis of the results presented, partitioning of VOCs in organic matter is most effective in dry soil samples, because the smallest desorption peaks were observed for samples having high organic content, and desorption times were observed to increase with the amount of organic matter; however, for the samples containing a relatively high moisture content, the results show that the prevailing effect is dissolution of VOCs into the water phase. Support for this interperation comes from the facts that the desorption peak areas measured for moist sand and garden soil are almost the same, and that desorption times decreased when the moisture concentration increased. Linearity, Detection Limits and Repeatability. The linearity of the method was tested using 11 compounds (Table 2). Six concentrations of the compounds (0.5-50 mg/kg) in methanol were spiked in natural sand samples (moisture 17%) and analyzed after 2 days of storing. Good linearity, as in our earlier studies,22 was obtained, with correlation coefficients varying from 0.990 for 1,2,4,5-tetramethylbenzene to 1.000 for 1,3,5-trimethylbenzene. Detection limits (S/N ) 3) were measured for the same compounds (1-200 µg/kg) from three different matrixes, namely dry and wet Fluka sand (moisture 17%) and wet garden soil (moisture 17%) (Table 2). From Table 2, it can clearly be seen that larger amounts of moisture and organic matter in the samples increase the detection limits, possibly as a result of the increased
chemical background level observed when the organic matter or water content in samples increased. Measured detection limits were of the same order of magnitude as they were in our earlier observations.22 Repeatability of the PAM method was tested using spiked garden soil samples. Four compounds, namely MTBE, 1,1,1trichloroethane, o-xylene, and tetrachloroethene were used in spiking. Ten replicates were measured, and the relative standard deviation varied from 4.2% for o-xylene to 7.7% for 1,1,1-trichloroethane. The relative standard deviations calculated using mfluorotoluene as an internal standard varied from 2.5% for tetrachloroethene to 3.4% for 1,1,1-trichloroethane. According to these results, the repeatability of the PAM method is very good. These results, like those obtained with the first PAM equipment,22 are better than our earlier results obtained using the manual PAM method.22 This is simply due to the fact that the constructed PAM apparatus makes working simpler and more reliable. Aging of the Soil Samples. To define the effect of aging on the desorption of the compounds from different soils, spiked samples were stored for different times before analysis. Another aim of this study was to define the length of time for which the soil samples can be stored before analysis. Two types of samples, namely dry Fluka sand and moistened garden soil (moisture 17%) were used in these experiments. The same compounds as in the linearity tests (Table 2) were used in spiking. The storing times before analysis were 1 day; 2 weeks; and 1, 2, 4, and 6 months. Both peak areas and the desorption times for six samples of each storing time were measured. m-Fluorotoluene was used as an internal standard and was added prior to analysis. For the data treatment, the analyte peak areas/internal standard peak area of the samples stored for 1 day were given a value of one, and all of the other desorption peak areas were normalized relative to these values. Figure 3 summarizes the results, showing the mean of the normalized peak area value of each of the compounds and the corresponding mean standard deviation values. The storage time of up to 2 weeks had no significant effect on peak areas in garden soil, but in the case of dry sand, the peak areas decreased by 14% on average (Figure 3). Over 6 months of storage in the case of garden soil, only a rather small decrease of peak areas was observed (on average 14%), but in sand samples, the final area was on average only 46% of the original. Especially Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3629
Figure 3. Effect of aging on peak areas using the dry sand and moistened garden soil. The averages and standard deviations of all compounds are presented.
for tri- and tetrachloroethene, the decrease was significant, because the desorption peak areas decreased from 1 to 0.07 and 0.17, respectively. According to our studies, storage had no effect on the desorption times in sand samples, but in garden soil samples after six months of storage, the desorption times for substituted benzenes were about twice as long as in the beginning. The results obtained show that soil samples can be stored at 5 °C in closed bottles for two weeks before analysis without very significant loss of even the most volatile compounds. The difference in results between soil type may originate from either soil type or moisture. The garden soil contains more organic matter, and therefore, it can possibly adsorb more volatile components. On the other hand, it has been reported that it can be difficult to desorb volatile compounds from dry soil samples.4 According to the results of Kolb et al.,4 the time of adsorption has an effect on the recovery of desorption. The recovery of trichloroethene after 3 h of adsorption was 91%, whereas after 2 days of adsorption, the recovery was only 27%. In addition, it has been reported that biodegradation can cause loss of VOCs during storing.39 Vapor Fortification Tests. The preparing of homogeneous and reliable soil standards for VOC analysis is very difficult, because of the high volatility of compounds to be analyzed. It has been shown that this can be accomplished by the vapor fortification method;36,37 therefore, the vapor fortification method was used to make identical standard soils for comparison of different analytical methods. Two examples of the results of these tests are presented in Table 3. One sample was contaminated with a mixture of compounds presented in Table 3 (upper part), and the other was contaminated with gasoline. The PAM-MS analysis was performed using selective ion monitoring (SIM) technique, and the following ions were used in quantitation: m/z 78, 92, 119, 106, 73, 97, 56, and 105. The results are means of five replicate samples, and two HSGC (lab 1) analyses were performed from each bottle. The samples analyzed by lab 2 with the portable HSGC and HSGC were taken from the same bottles. All of the GC results are very close to each other, but the PAM results were about 3- or 4-fold for the first sample. It was also noticed earlier12,33 that differences of this magnitude in the analysis of soil samples can occur when samples are analyzed using different techniques; however, the analytical results of PAM-MS and HSGC of soil samples contaminated with gasoline by vapor fortification methods agree well with each other. It is worthy of note that TAME and MTBE have similar mass spectra, so they cannot be analyzed 3630 Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
Figure 4. Mass spectrum of (a) a gasoline-spiked soil sample, (b) a diesel fuel-spiked soil sample, and (c) an authentic soil sample contaminated with both gasoline and diesel fuel.
separately with PAM-MS. The result for xylenes also includes all xylene isomers and ethylbenzene, and the results for tri- and tetramethylbenzene are the sums of all isomers in the PAM-MS analysis. For HSGC analysis, only single isomers of tri- and tetramethylbenzenes were chosen, namely 1,2,4-trimethylbenzene and 1,2,3,5-tetramethylbenzene. Analysis of Authentic Soil Samples. Many authentic soil samples were also analyzed using different techniques, namely the PAM-MS method with selected ion monitoring (SIM) and two headspace gas chromatographic (HSGC) methods. The samples that were analyzed were typically contaminated by gasoline or diesel oil. As an example, the mass spectra of a soil sample spiked with gasoline (4a), a soil sample spiked with diesel fuel (4b), and an authentic soil sample (4c) are presented in Figure 4. It can be observed that the authentic sample is probably contaminated with
Figure 5. Correlation of the results obtained by HSGC and PAMMS. Both vapor fortification and authentic samples are included.
both gasoline and diesel fuel, because the charasteristic ions m/z 91, 105, and 120 of the aromatics in gasoline and the typical ions m/z 57, 71, and 83 of diesel components are present. Figure 4 demonstrates that the mass spectra of unknown samples can be used to aid identification of the source of contamination. As an example, quantitation results of two authentic samples are presented in Table 4. The results obtained with the three different methods are in relatively good agreement. As a summary of the vapor fortification samples (79) and authentic samples (35) that were analyzed, Figure 5 shows the correlation between the results obtained by the PAM-MS method and the HSGC method of lab 1. The compounds that were analyzed are mainly those presented in Table 3. The slope of the line is 1.034, and the
coefficient of regression (r) is 0.953. Figure 5 shows that results obtained with the two methods are in reasonable agreement. The Student t-test was also used to compare results obtained with the PAM-MS and the HSGC methods. The t-value obtained for samples was lower than the theoretical value (0.718 and 2.00, respectively, 2-sided test), indicating that the results from both of the methods have no significant differences at 5% confidence level. It is worthy of note that an additional source for the deviation between the HSGC and the PAM-MS results may be the small amount (5 g) of sample used, because authentic soil samples are not necessarily very homogeneous, and because of the volatile compounds to be analyzed, they cannot be homogenized very well. In conclusion, the results show that the PAM-MS method is suitable for the quantitative determination of volatile organic compounds in soils, especially for rapid screening of the samples. The PAM method can also be used for analysis of other types of solid samples (e.g. building materials) and it can easily be modified for the analysis of volatiles in water samples. In the PAM-MS method, the results are rather independent of the soil type, especially if the moisture is higher than 10%, as is normally the case in authentic samples. Other advantages of the method are short analysis times, the nonrequirement for pretreatment of samples, and for environmental and health risk reasons, the fact that solvents are not used. In addition, the PAM-MS method is easily applied for on-site analysis. ACKNOWLEDGMENT The authors acknowledge financial support from the Technology Development Centre (Tekes) and Fortum Oil and Gas. Received for review December 20, 2000. Accepted May 10, 2001. AC001504I
Analytical Chemistry, Vol. 73, No. 15, August 1, 2001
3631
