Sampling of airborne polycyclic aromatic hydrocarbons - Analytical

Determination of Averaged Long-term Air Concentrations of Semivolatile Polycyclic Aromatic Hydrocarbons: First Results. Gerd Kloster , Reinhard Niehau...
0 downloads 0 Views 698KB Size
Anal. Chem. 1987, 59, 1701-1705

1701

Sampling of Airborne Polycyclic Aromatic Hydrocarbons Rein Otson* Environmental Health Directorate, Health a n d Welfare Canada, Room E-19,E.H.C., Tunney's Pasture, Ottawa, Ontario, Canada K1A OL2 John M. Leach and Lambert T. K. Chung

B.C. Research, 3650 Wesbrook Mall, Vancouver, British Columbia, Canada V6S 2L2

Llmltations of NIOSH sampling method P and CAM 183 were defined for airborne standard mixtures of polycycllc aromatk hydrocarbons (PAH) generated as vapors In a flow-through apparatus. The PAH fell Into three categorles: those that were too volatlle to be collected by the NIOSH flltratlon method at normal ambient temperatures and were best sampled with Tenax or XAD-2 sorbent (Le., Indane, naphthalene, blphenyl, acenaphthene, fluorene, 9,10-dlhydrophenanthrene, phenanthrene, and anthracene); those that were quantltatlvely collected by fllters, even afier a brief alrborne resldence t h e (Le., benz[a]anthracene, chrysene, benro[a]pyrene, dlbenr[a ,h]anthracene, and benzo[ghi]perylene); and those that partltloned between fllter and sorbent (Le., fluoranthene and pyrene). A comblnatlon glass fiberhiher membrane fllter backed by two sorbent tubes in series gave overall mean recoveries of 94-96% for the 15 PAH studied at total concentrations of, nominally, 0.2 and 0.02 mg/m3. Indlvldual PAH concentrations were 0.03-0.05 and 0.003-0.005 mg/ma, respectively.

A recent study of emissions to the workplace from benchscale coal liquefaction units ( I ) indicated that concentrations of total polycyclic aromatic hydrocarbons (PAH) were usually less than 0.1 mg/m3 and levels of individual PAH rarely exceeded 0.001 mg/m3. However, there was concern that use of NIOSH sampling procedure P & CAM 183 (2), which employs glass fiber and silver membrane filters, may have resulted in low recoveries of PAH. Several studies have shown that recovery of airborne PAH containing two to four rings is inefficient when conventional filter capture techniques are used. Collection of the more volatile compounds is incomplete, and fractions of those that are initially retained by the filters subsequently vaporize. Ratios of particulate and vapor-phase PAH in automobile emissions (3-5) and urban air (6-8) have been recorded, but none of these studies included method validation with known concentrations of standard compounds. NIOSH procedures for determining recovery efficiencies of organic compounds involve injection of a solution containing the contaminant of interest onto the collection medium. A more realistic method is to use dynamic, flow-through calibration with atmosheres containing known concentrations of the compounds of interest. In this way overall collection and recovery efficiencies can be determined taking into account factors such as humidity, temperature, sampling time and air flow rate, contact with aerial oxygen, and exposure to light, which are believed to affect recoveries of PAH. One study (9) has used a flow-through vapor generation system to determine the collection and recovery efficiency of four vaporphase PAH on Amberlite XAD-2 resin. However, there was no attempt to generate a known, constant concentration of vapor, and filters were not used to determine vap0r:particulate phase ratios for standard mixtures of PAH. Recently, an apparatus was designed and used to determine the effectiveness of filters and sorbents for collection of air-

borne aromatic amines (10). This apparatus was also used to measure the effectiveness of NIOSH filter method P & CAM 183 for collecting 15 airborne PAH and to determine recoveries on back-up tubes containing polymeric sorbents (Tenax or XAD-2) of PAH that were not retained by the NIOSH glass fiber/silver membrane filter combination. The concentration of total PAH in air was adjusted as closely as practical to the American Conference of Governmental Industrial Hygienists (11)threshold limit value (TLV) of 0.2 mg/m3 (1 x TLV) for coal tar pitch volatiles and to 0.02 mg/m3 (0.1 X TLV). Airborne particulates affect the partitioning of PAH between vapor and solid phases by providing nuclei for condensation and surfaces for adsorption. In some experiments a chamber was used to provide a known contact time for vapors with particulate-containing laboratory air. EXPERIMENTAL SECTION Materials. PAH were obtained from the following sources (specified purity in parentheses): naphthalene (95%1, biphenyl (95%),fluorene (95%),phenanthrene (95%),anthracene (95%), fluoranthene (90%),pyrene (95%),and chrysene (95%)were from MicroTek, Baton Rouge, LA; indane (95%),benz[a]anthracene (99% ), benzo[a]pyrene (99% ), dibenz[a,h]anthracene (99% ), and benzo[ghi]perylene (99%)were supplied by Ultra Scientific, Hope, RI; acenaphthene (purity not specified) was from K & K Laboratories, Inc., Plainview, NY; and 9,lO-dihydrophenanthrene (97%) was from Aldrich Chemicals, Milwaukee, WI. PAH were tested in three groups with similar volatilities (Table I). Standard solutions of the PAH groups were prepared by dissolving the compounds in diethyl ether (glass distilled, BDH Limited) ta give concentrations of 0.64 and 0.064 mg/mL of each group 1PAH, 1.0 and 0.1 mg/mL of each group 2 PAH, and 0.4 and 0.04 mg/mL of each group PAH. Inadvertently, the concentration of acenaphthene was twice that for each of the other group 1 PAH. Calibration solutions (group 1, 16 and 1.6 hg/mL; group 2,25 and 2.5 pg/mL; group 3,20 and 2 pg/mL) containing n-tridecane internal standard (30 ,ug/mL for 1 X TLV and 4 pg/mL for 0.1 X TLV tests) were prepared from the appropriate standard solutions. Solutions were stored in the dark at -10 " C and warmed to room temperature before use. Glass tubes containing 100 mg front and 50 mg backup sections of Tenax or 150 and 75 mg sections of XAD-2 resin were obtained from SKC, Inc. Glass fiber and silver membrane filters (37 mm diameter) from SKC, Inc., and filter support pads (37 mm) from Millipore Corp. were Soxhlet extracted with cyclohexane (10). Gas Chromatography. Gas chromatography (GC) analyses were generally done as previously reported (10). The GC oven temperature was adjusted to suit each PAH group and concentration. Typically, the oven temperature was held at 100 "C during injection and was then immediately raised at a rate of 8 "C/min for at least 14 min. The injector split ratios were 1:50 and 1 : l O for experiments at 1 X TLV and 0.1 X TLV, respectively. Vaporization and Sampling Tests. Vaporization and collection of PAH were generally done as previously described for aromatic amines (10). Laboratory air was normally 22 " C and at 4560% relative humidity (RH). Standard solutions of group 1 or group 2 PAH (25-50 WL)were injected into the quartz capillary tube of a Hewlett-Packard Model 18580A pyroprobe. The pyroprobe temperature was programmed t o vaporize PAH into a stream of high-purity nitrogen (120 mL/min) at a steady rate over 3 to 4 h. Nitrogen flow rate and temperatures required were

0003-2700/87/0359-17Ol$OlSO/OPublished 1987 by the American Chemical Society

1702

ANALYTICAL CHEMISTRY, VOL. 59, NO. 13, JULY 1, 1987

Table I. Polvcvclic Aromatic Hydrocarbons and Some " _ Properties bp, "C (760 mmHg)"

compound

compound indane

Group 1 IND indane NAP naphthalene BIP biphenyl ACE acenaphthene FLU fluorene DHP 9,10-dihydrophenanthrene

178 218 255 279 294

0.40 0.71 5.4 x 105 1.24 1.54 1.8 X lo4 1.82 5.4 x 103

biphenyl

338 340 383 393

acenaphthene 2.64 1.2 x 103 2.67 6.0 X 10' 3.53 1.0 x 102 3.69 4.9 X 10 7.4 x 10e

Group 3 BaA benz[a]anthracene CHR chrysene BaP benzo[a]pyrene DBA dibenz[a,h]anthracene BPE benzo[ghi]perylene

naphthalene

2.11

Group 2 PHE phenanthrene ANT anthracene FLA fluoranthene PYR pyrene

Table 11. PAH Extraction Recoverieso ( % ) with Cyclohexane

435 441 496 270d 278d

4.61 2.6, l.le 4.64 5.60 0.85 X 6.37 6.50 0.16 X

Handbook of Polycyclic Aromatic Hydrocarbons; Bjorseth, A., Ed.; Marcel Dekker: New York, 1983; p 709. bRelative to n-tridecane. Equilibrium vapor concentration at 25 "C calculated from vartor uressures in ref 18. dMelting uoint. 'EVC from ref 17. a

fluorene 9,lO-dihydrophenanthrene phenanthrene anthracene fluoranthene pyrene benz[a]anthracene

determined in prior tests involving analyses of PAH residues in the pyroprobe tube after 1, 2, 3, and 4 h. The pyroprobe outlet was connected by a Swagelok adapter (3.2 to 6.4 mm) to one of two types of mixing units. In one mixing unit, which gave an airborne time of ca. 0.5 s between generation and collection, PAH vapors and air were mixed in a T- or Y-shaped unit made from glass tubing (10 cm X 6.4 mm id.) encased in an electrical heating mantle to minimize condensation of PAH. A second apparatus (10)consisted of a glass vortex mixer and 2-L residence chamber wrapped with heating tape and gave a mean airborne time of ca. 37 s, based on smoke tests. In experiments with the least volatile PAH (group 3), the pyroprobe and mixing unit were replaced by a straight length of glass tubing (90 mm X 5 mm i.d., I piece) wrapped with heating tape. Group 3 PAH were vaporized directly from the tubing into an air stream. Each mixing unit was attached directly to the inlet of a 37-mm closed-face cassette containing a glass fiber filter, backed by a silver membrane fiiter and support pad. Normally, a sorbent tube containing Tenax followed by a tube containing XAD-2 resin was coupled in series to the outlet of the cassette. The combined flow rate for nitrogen and air through the sampling train was ca. 2 L/min (10). For 0.1 X TLV concentration, laboratory air was purified by passage through ca. 50 cm3 Tenax in a glass tube (10). Extraction. Extraction of sorbents, filters, and support pads and preparation of the extracts for analysis were done by using cyclohexane instead of 2-propanol as previously described for aromatic amines (IO). The vaporization and mixing apparatus components were also rinsed with cyclohexane, and the extracts were concentrated by evaporation prior to analysis. Appropriate amounts of n-tridecane internal standard were added to all extracts. In most cases, GC analysis was completed the same day as PAH vaporization and collection. Extraction recoveries were determined in triplicate (10). Aliquots of PAH standard solutions (25 MLfor group 2,50 pL for group 3) were injected onto the filters and support pad in a 50-mL beaker and the ether was allowed to evaporate for 1 min before extraction. Group 1 or 2 PAH solution (25 pL) was also injected onto Tenax or XAD-2 resin in the front sections of sorbent tubes. The tubes were capped immediately and stored a t 2 "C for 16 to 24 h, then the resin sections were extracted separately, as was done for the sampling runs. Tests for contamination were done periodically by analysis of solvents and rinses and extracts from

70 recovery amt added, pg Tenax XAD-2 filters

1.6 8 16 1.6 8 16 1.6 8 16 3.2 16 32 1.6 8 16 1.6 8 16 2.5 8 25 2.5 8 25 2.5 25 2.5 25

95.9 103.0 89.6 98.4 86.8 85.2 97.7 86.8 83.9 98.8 85.7 81.9 99.6 86.5 81.8 98.5 90.6 84.6 91.5 88.0 90.2 93.1 89.2 91.5 86.5 86.5 85.9 87.3

95.2 90.5 94.2 87.0 89.5 84.1 85.0 80.0 84.1 80.4 89.9 84.1 81.9 99.5 (93.1)b 82.7

2

20 chrysene

2

20 benzo [a]pyrene

2

20 dibenz [a,h]anthracene

2

20 benzo[ghi]perylene

2

20

101.0 (93.6)b 103.0 (94.0)b 102.0 (92.9)b 94.3 97.5 97.3 92.0 94.8 98.8 103.7 98.5 101.8 99.3

Averages for triplicate determinations. Values in parentheses are for benzene extraction. the pyroprobe assembly, filters, support pads, and sorbents in blank determinations using only nitrogen and laboratory air. The concentration and size of suspended particulate matter in laboratory air used to dilute PAH vapors were measured with a calibrated Berkeley Telonics QCM particle size classifier. Sampling periods were 3 to 4 h at an air flow of 250 mL/min through the instrument.

RESULTS AND DISCUSSION Boiling points, relative retention times, and equilibrium vapor concentrations (EVC) for the 15 PAH are listed in Table I. The nominal GC quantitation limits were 1 ng injected for each PAH. For calibration solutions containing individual PAH at nominally 2 and 20 pg/mL and with quantitation based on the internal standard GC response, results from all injections were within *5% of the calibrated concentration value. Group 3 compounds at 2 pg/mL were the exceptions in that the analytical precision sometimes ranged t o *lo%. Since the use of benzene can pose a health hazard and cyclohexane gave somewhat better recoveries in initial tests (Table 11), the latter solvent was used for all extractions. Mean extraction efficiencies for group 1and 2 PAH from Tenax with cyclohexane ranged from 82 t o 103% (Table 11). Best recoveries were obtained at loadings corresponding to 0.1 x TLV concentrations (overall mean recovery of 94.6%). For XAD-2, extraction efficiencies for group 1 and 2 PAH ranged from 80 t o 95% at loadings corresponding t o 1 X TLV concentrations. Since they were not found on the corresponding col-

ANALYTICAL CHEMISTRY, VOL. 59, NO. 13, JULY 1, 1987 100

s I100 %

E W 5 0

u W LT

Figure 1. Mean recoveries of PAH (see Table I) from filters (striped), Tenax (clear), and XAD-2 (dotted): top, 0.1 X TLV, 0.5 s; middle, 1 X TLV, 0.5 s; bottom, 1 X TLV, 37 s.

lection media during sampling tests, extraction efficiencies for group 3 PAH on Tenax and XAD-2 and group 1PAH on filters were not determined. Mean extraction efficiencies of group 2 PAH from filters ranged from 99 to 103% at a loading corresponding to 1 X TLV. For group 3 PAH on filters, mean extraction efficiencies were 92-99% and 94-104%, respectively, at 1 x TLV and 0.1 x TLV loadings. The precision of each triplicate extraction efficiency determination was better than 7% relative standard deviation. Blank determinations on the complete experimental system using different types of mixing apparatus, and on the various components separately, showed no contamination from laboratory air, from PAH residues on the equipment, or from filters, sorbents, or solvents under the conditions used for experiments a t 1 X TLV. At the higher sensitivity required for experiments at 0.1 X TLV, it was necessary to use Tenax sorbent to remove trace contaminanb from the laboratory air used for dilution and to provide clean blanks. Impurities from XAD-2 resin, as received, interfered with analyses a t 0.1 x TLV, and it was necessary to extract the resin with cyclohexane and diethyl ether prior to use. Calculated average total (Le., sum of individual PAH) concentrations of airborne PAH during 3-4.5 h of sampling ranged from 0.16 to 0.27 mg/m3 in experiments targeted a t 1X TLV concentration (0.2 mg/m3);70% of the runs averaged 0.18-0.22 mg/m3. For experiments a t ca. 0.1 X TLV, mean total PAH concentrations were 0.017-0.021 mg/m3. The volume of air sampled was usually 0.45-0.55 m3. Particulate concentrations in laboratory air used to dilute the PAH vapors ranged from 1.3 to 14.4 pg/m3. The principal size fractions had mass median diameters of 0.1-0.4 and 2.1-9.0 pm. The most volatile PAH (group 1)were not retained by the filters but were collected by backup sorbent resins (Figure 1). At 1 x TLV concentration (0.2 mg/m3 total PAH) a second sorbent tube (containing XAD-2) was necessary for complete recovery of indane, naphthalene, and acenaphthene (Figure l),but biphenyl, fluorene, and 9,lO-dihydrophenanthene were retained by the first tube. A second tube containing XAD-2 instead of Tenax was used in the attempt to obtain some information on the breakthrough characteristics and potential use of XAD-2 as an alternative sorbent. Group 1PAH were not retained by the filters even with a mean airborne residence time of 37 s (Figure 1). The filter surface temperature was approximately 22 "C. Overall recoveries tended to decrease with decreasing volatility of group 1 PAH a t 1 X TLV concentration (Figure 1). At 0.1 X TLV (0.02 mg/m3 total PAH)

1703

there was also no detectable retention of group 1 compounds by the filters and support pad (Figure 1). More than 90% was recovered from the initial Tenax tube. However, 25-50% of the indane was in the back section of the tube. PAH in group 2 were partly retained on the filters and support pad but mostly in the front section of the first sorbent tube (Figure 1). In experiments a t 1 x TLV, 0.5-s airborne time, mean retention by the filters was 1.5% for phenanthrene and anthracene and approximately 8% for fluoranthene and pyrene. However, results were variable, with recoveries by the filters and support pad ranging from 0 to 6% for phenanthrene and anthracene, and 0 to 25% for fluoranthene and pyrene. The filter surface temperature was approximately 25 "C as a result of heat applied to the mixing apparatus to minimize condensation on the walls. With a mean airborne residence time of 37 s, the proportion of group 2 PAH collected on the filters increased with decreasing volatility, from 0.7% for phenanthrene to approximately 16% for pyrene. The filter surface temperature was 22 "C. At 0.1 x TLV, all the detectable phenanthrene and anthracene was collected by the first tube front section; none was present on the filters and support pad (Figure 1).Higher portions of fluoranthene (mean 26%, range 11-34%) and pyrene (mean 28%, range 11-42%) were retained by the filters and pad than a t 1 x TLV. The least volatile group of PAH condensed in the outlet of the normal apparatus used for generation of airborne PAH and therefore were vaporized from a heated glass tube directly into a stream of dilution air. Group 3 PAH (Figure 1)were collected entirely on the filters at 1 X TLV and 0.1 X TLV concentrations even after an airborne residence time of only 0.5 s. The filter surface temperature was 40 OC as a consequence of the 160-180 "C apparatus temperature needed to prevent condensation of the compounds. Overall recoveries of 15 PAH on filters and sorbents, corrected for extraction efficiencies, averaged 94.4% (range 79.5-105.0%) a t 1 x TLV concentration, 0.5-s airborne time (Table 111). At 0.1 x TLV concentration under the same conditions, mean recovery of individual PAH was 92.4% (range 82.5-105.3%; Table 111). When the mean airborne time was ca. 37 s, recoveries of group 1and 2 PAH averaged 97.6% (86.5-105.7%) a t 1 x TLV concentration (Table 111). Sampling limitations of the NIOSH filtration method P & CAM 183 were thus defined for airborne PAH generated in a flow-through apparatus. The PAH fell into three categories: those that were too volatile to be collected by filters at normal ambient temperatures and were best sampled with a sorbent (i.e., indane, naphthalene, biphenyl, acenaphthene, fluorene, 9,10-dihydrophenanthrene, phenanthrene, and anthracene); those that were quantitatively collected by filters even after a brief airborne residence time (i.e., benz[a]anthracene, chrysene, benzo[a]pyrene, dibenz[a,j]anthracene, and benzokghi]perylene);and those that partitioned between filter and sorbent (Le., fluoranthene and pyrene). Similar distributions of particu1ate:vapor-phase PAH have been reported in the sampling of emissions from automobiles (5, 12),from coke ovens (9, 13), in urban air (6, 8) and in office air (14). The airborne PAH generation system used in this study provided a means to calibrate and standardize collection efficiencies of vapor phase and condensable PAH on a combination glass fiber/silver membrane filter backed by sorbent tubes. Recoveries of 82-105% were obtained from airborne concentrations of 0.003-0.005 mg/m3 individual PAH-total PAH concentration 0.02 mg/m3 (0.1 X TLV)-at a sampling air flow rate of 2 L/min over 3-4.5 h. Similar recoveries were obtained a t 10 times these concentrations, Le., at the TLV. Compounds with intermediate volatility, such as fluoranthene and pyrene, were sensitive to sampling conditions and gave a lower distribution precision between filter and sorbent

1704

ANALYTICAL CHEMISTRY, VOL. 59, NO. 13, JULY 1, 1987

___.

Table 111. Mean Recoveries of PAH 1 X TLV, 37 so

.

PAH

indane naphthalene biphenyl

acenaphthene fluorene 9,lO-dihydrophenanthrene

no. of tests

recovery, 9G

3 3 3 3 3 3

89.4 105.7 103.5 105.0 100.2

3

96.5 96.1 92.4 86.5

phenanthrene anthracene fluoranthene

3

pyrene

3

3

benz[a]anthracene chrysene benzo[a]pyrene dibenz[a,h]anthracene benzo[ghi]perylene

101.1

nab na na

na na

1 X TLV, 0.5 s"

0.1

recovery,

RSD, 90 no. of tests 3.1

4

3.8

4

7.6 4.9 8.3 8.6 6.7 6.1

4 4 4 4 4 4

5.2

4

7.0 na

4 4 4 4 4 4

na na

na na

%

82.7 105.0 100.0 97.2 84.9 79.5 91.2 97.1 97.3 99.1 90.6 91.2 96.6 102.1 101.8

X

TLV, 0.5 s" recovery,

RSD, 90 no. of tests 6.7 7.2 5.2 8.5 7.8 4.4 2.8 4.7 4.8 3.8 7.0 5.2 4.0 2.2 2.8

3 3 3 3 3

3 4 4 4 4

3 3 3 3 3

%

96.1 91.9 100.6 97.5 91.3 92.1 100.0

95.5 96.4 91.0 96.2 105.3 86.8 82.5 88.3

RSD,

%

4.8 3.0

1.6 1.1

4.5 4.9 1.4 2.5 5.4 6.3 5.7 7.8 4.0 7.4 2.1

"Airborne time. *na,not analyzed. than other compounds with greater or less volatility. However, there were no evident relationships between collection patterns and sampling conditions over the relatively narrow range of temperature and humidity prevailing during the tests. Minor differences in the ratios of polynuclear aromatic hydrocarbons on filters and sorbents in other studies (5,6,8-12) are probably attributable to differences in sampling and environmental conditions. Careful control and recording of sampling variables are necessary to ensure consistency in interpretation of airborne PAH measurements. A detailed examination (16) of sorbent breakthrough characteristics by use of chromatographic principles was not attempted partly due to the small number and nonuniform length and composition of the sorbent sections in the two types of tubes. However, the results showed that, under the test conditions, a single tube containing Tenax was sufficient for quantitative collection of the PAH in groups 2 and 3, as well as biphenyl, fluorene, and 9,lO-dihydrophenanthrenein group 1. In fact, more than 85% of the amounts of each of these PAH collected on Tenax was found in the 100 mg front section. A significant percentage of indane (27% a t 0.1 X TLV, 42% a t 1 x TLV) and naphthalene (11%a t 1 X TLV), and some acenaphthene (4% at 1 X TLV), was found in the backup tube containing XAD-2. Only 50% of the indane was retained on the Tenax front section in tests with group 1 PAH a t 0.1 and 1 X TLV. The Tenax front section retained 70% naphthalene and 78% acenaphthene in tests a t 1 X TLV. I t was estimated that an air sample volume of ca. 0.3 m3 would allow quantitative collection of indane, naphthalene, and acenaphthene in a single tube containing Tenax. The maximum amount of total PAH found in Tenax front sections was 54 pg. This amount was determined in several tests where there was breakthrough of a t least one group 1 or 2 PAH and is similar to the Tenax load capacity for aromatic amines (70 pg) determined under similar conditions (IO). Indane and naphthalene were found in the 50-mg back section of the backup tube containing XAD-2 in a few of the tests where up to 8 Fg of total PAH was found on the 100-mg front section. On the basis of these results, Tenax appeared to be a t least as effective as XAD-2 resin for collection of PAH vapors, in contrast to observations in other reports (13). The physical state of airborne PAH and factors such air flow rate (blow off effect), temperature, and humidity will influence their collection by particulate filters (7, 17). An estimate of the distribution of a PAH between vapor and particulate phases can be obtained from its equilibrium vapor concentration (EVC). The EVC will only represent the maximum losses possible during collection of airborne PAH

on filters since kinetic and adsorption effects also influence collection efficiency (17, 18). In addition, different methods for determining vapor pressure and EVC can give values that are quite different (16,17, 19). Table I lists some EVC values for comparison with total airborne concentrations and collection results for PAH in this study. The EVC exceeded the maximum airborne concentration of individual group 1 PAH in tests a t 0.1 X TLV (3 pg/m3) and 1 X TLV (34 fig/m3) and no significant amounts of these compounds were found on the filters. Conversely, EVC values for group 3 PAH in tests a t 0.1 X TLV (3.8 bg/m3) and 1 X TLV (54 pg/m3) were less than the maximum individual PAH concentrations and no significant amounts of these compounds were found in the sorbent. Group 2 PAH had EVC values near or greater than the individual PAH concentration and were found chiefly in the sorbent in tests a t 0.1 X TLV (4.8 pg/m3) and 1 X TLV (50 wg/m3). These results confirmed that the EVC calculated from the vapor pressure of a compound could be used to estimate its collection by NIOSH method P & CAM 183 under the test conditions. However, the distribution between filters and sorbent in these tests does not necessarily correspond to the ratio of vapor to particulate PAH in the air stream (16). The amount of airborne particulate material can also influence the proportions of material collected on filters and sorbent. Effects of suspended particulate matter could not be assessed because of very low concentrations of particulate in the laboratory air used for dilution. Even a 74-fold increase in airborne residence time did not measurably alter the collection pattern of PAH, possibly because of the low levels of aerosol present. The 37-s contact time was based on the calculated residence period for vapors in workplace air near the B.C. Research continuous coal liquefaction unit (1). Retention of material from the vapor phase by filters was probably due to condensation or adsorption on the filter surface rather than filtration of particulate. The least volatile group of hydrocarbons was collected entirely on the filters in spite of an airborne time of only 0.5 s and a filter surface temperature of 40 "C. There was no indication that significant degradation of PAH had occurred on the filters, although this has been cited as a problem in other studies (13, 19).

ACKNOWLEDGMENT Helpful comments on the manuscript by D. T. Williams and F. M. Benoit are gratefully acknowledged. Registry No. Tenax, 24938-68-9; XAD-2, 9060-05-3; Ag, 7440-22-4; indane, 496-11-7; naphthalene, 91-20-3; biphenyl, 9252-4; acenaphthene, 83-32-9; fluorene, 86-73-7; 9,lO-dihydrophenanthrene, 776-35-2; phenanthrene, 85-01-8; anthracene,

Anal. Chem. 1987, 59, 1705-1708

120-12-7; fluoranthene, 206-44-0; pyrene, 129-00-0; benz[a]anthracene, 56-55-3;chrysene, 218-01-9; benzo[a]pyrene, 50-32-8; dibenz[a,h]anthracene, 50-70-3; benzo[ghi]perylene, 191-24-2. LITERATURE CITED Otson, R.; Armstrong, V.; Leach, J. M. Am. Ind. Hyg. Assoc. J . , in press. Manual of Analytical Methods, 2nd ed.; Publication No. 77-157A: DHEW (NIOSH): Cincinnati, OH, 1977; Vol. 1. Stenberg, U.; Alsberg, T.; Westerholm. R. EHP, Environ. Health Perspect. 1983, 4 7 , 43-51. Stenberg, U.; Alsberg, T.; Westerholm, R. EHP, Environ. Health Perspect. 1983, 4 7 , 53-63. Scheutzle, D. EHP, Environ. Health Perspect. 1983, 4 7 , 65-60. Cautreels. W.; Van Cauwenberghe, K. Atoms. Environ. 1978, 12, 1133-1 141. Yamasaki, H.; Kuwata. K.; Miyamoto, H. Environ. Sci. Technol. 1982, 18, 189-194. Thrane. K. E.; Mikalsen, A. Atoms. Environ. 1981, 15, 909-916. Andersson, K.; Levin, J.-0.; Nilsson, C . 4 . Chemosphere 1983, 12. 197-207. Otson, R.; Leach, J. M.; Chung, L. T. K. Anal. Chem. 1987, 5 9 , 58-62.

1705

(11) TLVs Threshold Limit Values for Chemical Substances and Physical Agents in the Work Environment and Biological Exposure Indeces with Intended Changes for 1984-85; American Conference of Governmental Industrial Hygienists: Cincinnati, OH, 1984. Lee, F. S.-C.; Prater, T. J.; Ferris, F. I n Polynuclear Aromatic Hydrocarbons; Jones, P. W., Leber, P., Eds.; Ann Arbor Science Publishers: Ann Arbor, MI, 1979; pp 83-110. Jones, P. W. VDI-Ber. 1980, No. 358, 23-38. Otson, R.; Hung, I.-F. Polynuclear Aromatic Hydrocarbons: Mecha nism, Methods and Metabolism; Cooke, M., Dennis, A. J., Eds.; Battelle Press: Columbus, OH, 1984; pp 999-1012. You, F.; Bidleman, T. F. Environ. Scl. Technol. 1984, 18, 330-333. Bidleman, T. F.; Billings, W. N.; Foreman, W. T. Environ. S d . Technol. 1986, 20, 1038-1043. Pupp, C.; Lao, R. C.; Murray, J. J.; Pottie, R. F. Amos. Environ. 1974, 8 , 915-925. Sonnefeld, W. J.; Zoller, W. H.; May, W. E. Anal. Chem. 1983, 5 5 , 275-280. Spitzer, T.; Dannecker, W. Anal. Chem. 1983, 5 5 , 2226-2228.

-

RECEIVED for review November 5, 1986. Accepted March 4, 1987* This work was supported by the Canadian Intel&partmental Panel on Energy Research and Development.

Extraction and Recovery of Polycyclic Aromatic Hydrocarbons from Environmental Solids Using Supercritical Fluids Steven B. Hawthorne* and David J. Miller

University of North Dakota Energy Research Center, Grand Forks, North Dakota 58202

The use of supercritical fluids for the extractlon and recovery of polycycllc aromatic hydrocarbons (PAH) from environmental sollds has been developed and tested by uslng urban dust, fly ash, and river sedlment. Supercrltlcal N,O with 5 % methanol modlfler gave the best recoveries of PAH from all three samples when compared to supercrklcal CO, wkh 5 % methanol, N,O, CO,, and ethane. Quantltailve recovery of PAH from National Bureau of Standards SRM 1649 (urban dust) and of deuterlated PAH spikes (phenanthrene-d,,, pyrene-d,,, and perylene-d,,) from the river sedlment was obtalned wlth supercritical fluid extractlons In as little as 30 mln. In most cases, 30-60 mln extractions of the river sedlment and fly ash with supercritical N,0/5 % methanol gave better recovery of the deuteriated PAH spikes than the recoveries obtalned by uslng 4 h of sonication or 8 h of Soxhlet extractlon wlth either benzene or methylene chloride. Supercritical fluid extractions yield good PAH recoverles, requlre only small amounts of sample, minimize analyte concentration steps, and are simple and rapid to perform.

The extraction and recovery of organic species from environmental solids are critical and often the limiting steps in analysis schemes used to identify and quantitate organic pollutants. Extractions of polycyclic aromatic hydrocarbons (PAH) from solid samples such as airborne particulates, soil, and fly ash are usually performed with liquid solvents either in a Soxhlet apparatus or with sonication, although variations such as extraction with liquid COz have been described ( I ) . These traditional liquid solvent extraction methods generally take several hours to perform, require relatively large volumes of ultrapure solvent (often making subsequent concentration steps necessary), and may yield incomplete recovery and/or degradation of sample species (2-5).

Besides their strong solvating ability, supercritical fluids have several characteristics that may make them useful for the rapid and quantitative extraction and recovery of organic pollutants from environmental solids. The solvent strength of a supercritical fluid is directly related to its density as can be described by the empirical correlation 6 = 1.25Pc1/2(p,/pl) where 6 is the Hildebrand solubility parameter, P, is the critical pressure of the fluid, pg is the density of supercritical fluid, and p1 is the density of the fluid in its liquid state (6). Thus, the solvating ability of a particular supercritical fluid toward a particular species can easily be modified by changing the extraction pressure (and, to a lesser extent, the temperature). Supercritical fluids with widely varying polarities are available (i.e., pentane, COz, and ammonia have Hildebrand solubility parameters a t liquid densities of 7.0, 10.6, and 13.2 ( c a l / ~ m ~ ) 'respectively), /~, and the polarity of a supercritical fluid can also be controlled by the addition of solvent modifiers. The low viscosities and high solute diffusivities of supercritical fluids should yield good mass transfer during extraction. Supercritical fluids that have low critical temperatures (e.g., COP,nitrous oxide, and ethane have critical temperatures of 31,36.5, and 32 "C, respectively) can be used to avoid decomposition of thermally labile compounds. Finally, many supercritical solvents are gases at room temperature which should simplify analyte concentration steps and help reduce the loss of volatile sample species. Recent studies have demonstrated analytical uses for supercritical fluid extractions including the extraction of diesel fuel components from silica gel and marine sediments (7),the coupled supercritical fluid extraction/supercritical fluid analysis of caffeine from coffee beans (8),the silica column fractionation of oil residuals (9) and of polycyclic aromatic hydrocarbons (PAH) (IO), and the direct mass spectrometric analysis of supercritical fluid extraction products (7, 2 1 ) . The ability of supercritical COz extractions to yield quantitative extraction and recovery of PAH from diesel exhaust partic-

0003-2700/87/0359-1705$01.50/0 0 1987 American Chemical Society