Environ. Sci. Technol.

2003, 37, 597-605

Recovery of Semivolatile Organic Introduction

Accurate determination of airborne semivolatile organic
Compounds during Sample species presents sampling and analytical challenges because
Preparation: Implications for their lability complicates both collection and quantification.
Improved collection is possible with new air sampling
Characterization of Airborne technology such as the Integrated Organic Gas and Particle
Sampler (IOGAPS) that incorporates extractable sorbent-
Particulate Matter coated diffusion denuders and filters (1, 2). However, accurate
quantification of PM-associated organic compounds also
requires validation of the laboratory procedures for recovery
ERICK SWARTZ AND (e.g., solvent extraction and reduction of volume) over wide
LEONARD STOCKBURGER* volatility, molecular weight and polarity ranges. It is necessary
U.S. Environmental Protection Agency, Office of Research and to account for losses of both a) semivolatile compounds that
Development, National Exposure Research Laboratory, can evaporate during the reduction of volume procedure
Research Triangle Park, North Carolina 27711 and b) high molecular weight species that may precipitate
during steps that were intended to lower limits of detection.
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Careful attention must be paid to sample preparation

LARA A. GUNDEL
techniques to ensure maximum recovery of the wide variety
Environmental Energy Technologies Division, of organic compounds that are associated with ambient
Downloaded via JOHNS HOPKINS UNIV on July 12, 2019 at 14:51:39 (UTC).

Ernest Orlando Lawrence Berkeley National Laboratory, particulate matter.

University of California, Berkeley, California 94720 Previous work has led to the widespread use of organic
solvent mixtures or a series of solvents to optimize extraction
of airborne particulate organics. Grosjean (3) showed that
mixtures of nonpolar and polar solvents extracted more
Semivolatile compounds present special analytical organic carbon from PM than could be removed by individual
challenges not met by conventional methods for analysis solvents. Various mixtures have been evaluated in the last
of ambient particulate matter (PM). Accurate quantification two decades: Appel et al. (4) recommended sequential use
of PM-associated organic compounds requires validation of benzene followed by a methanol-chloroform mixture.
of the laboratory procedures for recovery over a wide volatility Cadle and Groblicki (5) determined that a benzene-ethanol
and polarity range. To meet these challenges, solutions mixture minimized residual nonextractable carbon in samples
of n-alkanes (nC12-nC40) and polycyclic aromatic of particles, compared to a variety of solvents and mixtures.
Sequential extractions with dichloromethane and acetone
hydrocarbons PAHs (naphthalene to benzo[ghi]perylene)
have been used for studies of the genotoxicity of ambient
were reduced in volume from a solvent mixture (equal volumes PM (6, 7). Mazurek et al. (8, 9) developed a multistep solvent
of hexane, dichloromethane and methanol), to examine extraction protocol (using hexane, followed by a benzene-
recovery after reduction in volume. When the extract solution 2-propanol mixture) that has been used extensively for
volume reached 0.5 mL the solvent was entirely methanol, speciation of organics in ambient and source particulate
and the recovery averaged 60% for n-alkanes nC12- matter (10-17).
nC25 and PAHs from naphthalene to chrysene. Recovery The recognition of the important role that SVOC play in
of higher MW compounds decreased with MW, because of PM atmospheric behavior and health effects has led to
their insolubility in methanol. When the walls of the development of improved technology, such as extractable
flasks were washed with 1 mL of equal parts hexane and diffusion denuders, to reduce sampling artifacts while
dichloromethane (to reconstruct the original solvent allowing determination of both gas and particulate phase
composition), the recovery of nC18 and higher MW compounds SVOC concentrations. Extraction of the multichannel XAD-
increased dramatically, up to 100% for nC22-nC32 and coated diffusion denuders of the IOGAPS required a solvent
mixture that would provide maximum solubility of adsorbed
then slowly decreasing with MW due to insolubility. To
semivolatile compounds without dissolution of the epoxy
examine recovery during extraction of the components of joints of the glass denuders. Based on the previous work of
the High Capacity Integrated Gas and Particle Sampler, other investigators, experience with the IOGAPS (1), and the
the same standards were used to spike its denuders and long-term goal of determining polar as well as nonpolar
filters. For XAD-4 coated denuders and filters, normalized organic compounds, we chose an extraction mixture of
recovery was >95% after two extractions. Recovery from hexane, dichloromethane and methanol.
spiked quartz filters matched the recovery from the This study began as an effort to validate sample prepara-
coated surfaces for alkanes nC18 and larger and for tion procedures for characterization of gas and particle
fluoranthene and larger PAHs. Lower MW compounds samples that had been collected with the high capacity (HiC)
evaporated from the quartz filter with the spiking solvent. IOGAPS and related samplers. However, a more general
This careful approach allowed quantification of organics purpose emerged with our growing recognition that insuf-
by correcting for volatility- and solubility-related sample ficient attention to analyte recovery could undermine the
efforts of any research team, with any sampling technology,
preparation losses. This method is illustrated for an ambient
to characterize the wide range of organics associated with
sample collected with this sampler during the Texas Air PM. Thus, the broader objective is to describe the implications
Quality Study 2000. of these recovery results for past and future efforts to speciate
airborne semivolatile and particulate organics. The results
of our quality control and quality assurance efforts broadened
* Corresponding author e-mail: [email protected]. beyond devising procedures to optimize recovery of specific
10.1021/es011128z CCC: $25.00  2003 American Chemical Society VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 597
Published on Web 12/19/2002
FIGURE 1. Schematic of the high capacity integrated organic gas
and particle sampler (HiC IOGAPS).

compounds to (a) exploring the implications of our results

for interpretation of past efforts to characterize particulate
organics and (b) developing recommendations for future FIGURE 2. Flow diagram of sample preparation for the HiC IOGAPS.
work by any investigators.
First, we examined how to optimize recovery of a group restored to the original solvent composition as described
of nonpolar alkanes and PAHs from a solvent mixture of below. The 1.5 mL extracts were stored at -30 °C for
hexane, dichloromethane and methanol after reduction of subsequent analysis. The filters were stored at -30 °C in
solvent volume. Next, we measured the recovery of this group Teflon-sealed 50 mL prebaked glass centrifuge tubes. The
of compounds from spiked HiC IOGAPS components (XAD-4 filters were thawed, extracted and analyzed at the US EPA’s
coated diffusion denuders, quartz filters and XAD-4 impreg- National Exposure Research Laboratory in RTP, NC.
nated quartz filters) and optimized the extraction procedures.
We then determined this group of compounds in a field Extraction Details. Before use, each denuder had been
sample from the HiC IOGAPS that operated at La Porte given a unique identification number, coated with fine
Airport, near Houston, TX, during the Texas Air Quality Study particles of XAD-4 (18-20), and spiked with 0.1 mL of a
- 2000 (TexAQS-2000). Finally the implications of our results deuterated PAH mixture as field surrogates to monitor
are discussed by comparison to other quantitative speciation processing losses during sample extraction and reduction of
studies, along with recommendation for future work. volume. The surrogate recovery was evaluated for acceptance
by determining whether the measured concentration fell
Experimental Section within the acceptance limits of 80-120%. The deuterated
PAH solution contained anthracene-d10 (0.04 µg mL-1),
Overview of Sample Preparation for HiC IOGAPS. For phenanthrene-d10 (3.3 µg mL-1) and fluoranthene-d10 (1.0 µg
illustration of the sample preparation method we analyzed mL-1), all in hexane, at concentrations estimated to be similar
a HiC IOGAPS (Model 3000DB, URG Corp., Chapel Hill, NC) to the native airborne PAH. The denuders were allowed to
air sample that was collected on August 31, 2000, between dry for a few minutes until the solvent evaporated and then
00:10 and 11:30 during the TexAQS-2000 at the LaPorte extracted to obtain a pre-sampling blank. After collection of
Airport. Figure 1 shows a schematic diagram of the sampler.
the ambient sample, and before extraction, each denuder
The HiC IOGAPS had a 2.5 µm cyclone for particle discrimi-
was spiked again with the same deuterated PAH mixture.
nation, followed by two 8-channel XAD-4 denuders in series
(each had 52 mm o.d. and 285 mm length), one prebaked The procedure for denuder extraction follows: The
quartz and three XAD-4 impregnated quartz filters (90 mm denuder was sealed at one end with a clean Teflon end cap,
diameter). The sample was collected at 85 L min-1 for 11.5 half-filled with 125 mL of Hex:MeCl2:MeOH solvent, and
h and yielded three extracts of each denuder and one extract capped with another Teflon end cap that had a hand-tight
for each filter, for a total of 10 extracts, not counting filter Swagelok fitting for pressure release. (This is an important
blanks. Figure 2 shows a flow diagram of sample preparation. safety measure, as freshly prepared Hex:MeCl2:MeOH can
Before extraction, a mixture of three deuterated PAH release dissolved air unless it is sonicated before use.) A rolling
standards was added to each denuder. The denuders were rinse technique was used; the capped denuder was rolled
extracted in a solvent mixture of equal volumes of pesticide back and forth along the laboratory benchtop for 30
grade hexane, dichloromethane and methanol (Hex:MeCl2: revolutions. To exclude any XAD-4 that may have shed from
MeOH) in a laboratory at the University of Houston, Clear the denuder coating, each extract was passed through a Teflon
Lake, where the extracts were also reduced in volume and membrane (unlaminated Fluoropore, 0.45 µm pore size,

598 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003

FHUP04700, Millipore Corp.) before volume reduction. The to 320 °C, and held at 320 °C for 15 min. The MS was operated
filtrate was transferred to clean 125 mL wide mouth brown under the following conditions: trap temperature 225 °C,
bottles for temporary storage. (Glassware had been rinsed emission current 15 µA, scan rate 50-350 Da in 0.5 s and
with Hex:MeCl2:MeOH solvent and heated to 500 °C for 2 h.) A/M amplitude was 3.0 V. The nonpolar organic compounds
After ambient sampling, each denuder was extracted three were identified by comparison with retention times and mass
times, and each extract was processed separately, with the spectra of authentic standards. The standard deviation for
third extract serving as the denuder blank for the next use repetitive GC-MS runs was <4%. The concentrations were
of that denuder. Quartz filters and XAD-4 impregnated quartz normalized to the p-terphenyl-d14 response for each injection
were each extracted twice by sonication for 1 h at ambient such that
temperature in 50 mL of Hex:MeCl2:MeOH solvent and
filtered by the same procedure as used for denuder extracts. Ax‚RFx C optp
In some cases it was necessary to refilter the extracts with Cx ) ‚ (1)
a 0.2 µm syringe filter (Acrodisc CR PTFE) before GC-MS V Aptp‚RFptp
analysis. For ongoing efforts we recommend that deuterated V
surrogates be added to each filter before extraction, in the
same approach as the denuders. where A is the area count from the GC-MS, RF is the response
Reduction of Volume. During the TexAQS field study, we factor (µg/area counts), V is the total volume of the solution
used a Labconco RapidVap Vacuum Evaporation System (mL), C o is the initial concentration (µg/mL), and C is the
(model 79000-00), which accommodated 8 (170 mL) samples, final concentration (µg/mL). The subscript x refers to any
instead of a conventional rotary evaporator, because of the given species within the solution, and ptp refers to the
large number of extracts (>500). The evaporation tubes had p-terphenyl-d14. The response for each n-alkane and PAH
1.5 mL endpoint stems with volume markings. The vacuum was found to be linear over the concentration range used.
model was chosen rather than the N2 model because
preliminary results with the N2 model yielded extracts that
contained significant amounts of water. The water had Results
condensed from room air during the procedure. With careful Optimizing Recovery of Alkanes and PAHs after Solvent
external baffling to prevent entry of ambient air, it has been Reduction and Reconstitution. The possibility of cross
possible to evaporate the Hex:MeCl2:MeOH solvent with the contamination during simultaneous evaporation of multiple
N2 model, without water condensation (Y. Pang, private extracts was examined by processing four vials of the standard
communication). mixture of alkanes and PAHs (Table 1) in Hex:MeCl2:MeOH
Before the field study, the sample preparation procedures along with four solvent blanks. Concentrations of the target
were optimized in a stepwise fashion, with assessment of compounds in the blanks were indistinguishable from solvent
the recovery of alkanes and PAHs at each step. What follows blanks, and therefore we concluded that no appreciable cross
is the optimized protocol. To test the recovery after the solvent contamination occurred.
reduction procedure, 1 mL aliquots of an n-alkane-PAH Figure 3 shows the results for recovery of alkanes and
mixture (Table 1) were diluted to 100 mL in Hex:MeCl2:MeOH PAHs for several different procedures. When the standard
solvent. The diluted standard mixture (later, denuder or filter mixture in Hex:MeCl2:MeOH was reduced to 0.5 mL without
extract) was also spiked with 1.58 µg of the nonvolatile nC24D50 reconstitution, (open squares) recoveries varied for each
(in Hex:MeCl2:MeOH) as a laboratory surrogate to monitor species, depending on volatility, and solubility in methanol.
processing losses during reduction of volume. The surrogate The most labile species nC12 had 30% recovery. The recovery
recovery was evaluated for acceptance by determining increased with MW and plateaued at 60% between nC16 and
whether the measured concentration fell within the ac- nC24. The recoveries for the n-alkanes above nC24 then
ceptance limits of 80-120%. The solution was then filtered decreased steadily until there was no recovery of nC34 and
through a Teflon membrane, transferred into a Labconco greater. The addition of 1 mL of equal volumes of hexane:
evaporation tube, and reduced to ∼0.5 mL using the dichloromethane (Hex:MeCl2) at the bottom of the evapora-
programmed stepwise process shown in Table 2. The stepwise tion tube, without using this mixture to rinse its walls, showed
process was necessary to prevent bumping of the solvent increased recoveries for nC24 and higher (open diamonds).
mixture. However, the recoveries were significantly higher when the
When the volume reached 0.5 mL, the concentrated walls were rinsed. A methanol wall rinse (open triangles)
standard solution or sample extract was cooled and weighed improved recovery of nC27 and higher, but a more dramatic
while still in the evaporation tube. Density measurements improvement occurred when the walls were washed instead
showed that the solvent mixture was primarily methanol with equal volumes of hexane: dichloromethane (closed
(>99%) at this point, so that exact volumes were determined circles). Using this improved procedure, the recovery of the
gravimetrically. Finally, to ensure dissolution of any com- internal standard nC24D50 averaged 90%, which is consistent
pounds coating the walls, before transfer, the walls of each with the recovery of nC24H50 from the standard mix. The
tube were rinsed with 1 mL of equal volumes of hexane:di- critical step to optimum recovery of the nonpolar hydro-
chloromethane (Hex:MeCl2) containing the internal standard carbons proved to be the wall rinse with the less polar
p-terphenyl-d14 (1.84 µg mL-1) that was used later to account components of the solvent mixture, to reconstitute the
for variation in GC-MS injection volume. The 1.5 mL extracts original extraction solution.
were thus reconstituted into the original Hex:MeCl2:MeOH The low recoveries of labile species nC12-nC19 can be
solvent mixture before quantitative transfer to 1.8 mL clean accounted for as irreversible evaporative loss of these species
glass autosampler vials with Teflon-lined caps. during the reduction of volume step. The increased recovery
Determination of Alkanes and PAHs. Determination of of the higher MW hydrocarbons when Hex:MeCl2 was added
the alkanes and PAHs was performed using a Varian Saturn to the reduced extract could be attributed to their higher
IV gas chromatograph-ion trap mass spectrometer (GC-MS) solubility in the Hex:MeCl2:MeOH mixture than in methanol
with a SPI injection system. The GC had a 30-m Rtx-5Sil MS alone. However, the results suggest that the wall rinse
column (0.28 mm i.d., 0.25 µm film thickness) and a 10-m resolubilized compounds that had been deposited on the
Integra Guard column. The carrier gas was helium at a flow surfaces of the tubes, and rinsing with the original solvent
rate of 1 mL min-1. The oven temperature was held at 60 °C mixture was necessary to optimize recovery. These results
for 5 min, heated at 10 °C min-1 to 140 °C, then at 5 °C min-1 illustrate the need to return the extract solution to the original

VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 599

TABLE 1. n-Alkane-PAH Spiking Mixture

Hex:MeCl2:MeOH solvent and wash the walls of the glassware, recoveries were determined using the n-alkane-PAH mixture
to redissolve the nonpolar organics that have precipitated with 10 times lower concentrations than the original. The
during the reduction of volume step. recoveries were still less than 100% for n-alkanes > nC32. The
For the higher MW n-alkanes with <100% recoveries results suggest that the concentrations of the higher MW
(>nC32), lower concentrations could lead to higher recoveries n-alkanes were low but not enough to redissolve completely
because of less precipitation from the solution. To test this, in the solvent mixture. We are finding that most of the extracts

600 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003

TABLE 2. Program for Reduction of Volume with Labconco RapidVap Vacuum Evaporating System
temp (°C) pressure (mbar) vortex speed time (min) comment
35 preheat system
35 750 28% 5 prevent bumping and evaporate dichloromethane
35 650 28% 5 prevent bumping and evaporate dichloromethane
40 550 28% 15 evaporate hexane
45 400 30% 20 evaporate hexane and methanol
50 250 34% as needed reduce to final volume of 0.5 mL (methanol)

FIGURE 3. Percent recovery of n-alkanes (a) and PAHs (b) from various reduction of volume procedures. A reduced extract only (0.5 mL)
is shown as open squares; the same 0.5 mL reduced extract with addition of 1 mL of Hex:MeCl2 is shown as open diamonds; a reduced
extract with the Labconco glassware rinsed with 1 mL MeOH is shown as open triangles; a reduced extract with the Labaconco glassware
rinsed with 1 mL of Hex:MeCl2 are shown as closed circles. The open circle markers (at carbon 24.5) show the recovery of the laboratory
surrogate nC24D50. The error bars represent the uncertainty at the 1σ level.

from the LaPorte, TX site have concentrations above those earlier. The extracts were reduced in volume and each extract
used in the low concentration standard mixture. was analyzed separately. Each extraction step removed >85%
The recoveries of the PAHs also showed similar trends, of the remaining spiked material from the denuder. Figure
based on their lability and solubility. The PAH concentrations 4 shows the comparison of the recovery from the denuder
used in the spiking mixture (and found in field samples) (sum of all three extracts) with the optimized procedure for
were lower than the n-alkanes. The lower concentrations, as the standard compounds (above). A t-test found no difference
well the chromatographic properties of the PAHs, led to larger at the 95% confidence interval between the means of
error bars associated with these compounds. However, the recoveries. Therefore, when compound-by-compound nor-
recoveries of phenanthrene (PAH 6) and fluoranthene (PAH malizations were made for losses during the reduction of
8) were 70 and 80%, respectively, consistent with values of volume procedure, the extraction efficiency of the hydro-
70% for deuterated phenanthrene and fluoranthene previ- carbons was determined to be >95 ( 10%, overall, for the
ously observed using rotary evaporation after cleanup on alkanes and PAHs.
silica (2). Recovery of hydrocarbons from filters (quartz and
Because the two classes of hydrocarbons showed similar XAD-4 impregnated quartz) was determined by spiking
patterns of recovery, for convenience they could be grouped 1 mL of the n-alkane-PAH mixture to representative filters
for preparation of future surrogate mixtures, tailored to of each type. Each filter was extracted twice in Hex:MeCl2:
specific detection requirements or availability. For each group MeOH, using sonication, as described earlier. The extracts
the recovery varied similarly with MW or ring size and were reduced in volume and analyzed separately. Each
chromatographic retention time. These groups have been extraction step removed >85% of the remaining spiked
identified in Table 1: (1) hydrocarbons nC14-nC17 showed material from the filters. Figure 4 shows the comparison of
recovery similar to naphthalene-fluorene; (2) nC18 to the recovery from the filters (sum of the two extracts) with
phenanthrene-anthracene; (3) nC19-nC24 to fluoranthene- the recovery of standards. Significantly lower recoveries of
pyrene; (4) nC24-nC32, to benz[a]anthracene-benzo[k]fluor- the lower molecular weight n-alkanes-PAHs (<nC18 and
anthene; and (5) nC34-nC36, to indeno[cd]pyrene-benzo- associated PAHs) were observed for the spiked quartz filters,
[ghi]perylene. but not for the XAD-4 impregnated filters. The results for the
Recovery of Hydrocarbons from XAD-4 Coated Denud- XAD-4 impregnated filter were similar to the XAD-4 coated
ers, Quartz Filters and XAD-4 Impregnated Quartz Filters. annular denuder, as expected for recovery from the same
To test recovery from the denuders, we spiked an XAD-4 sorbent. When compound-by-compound normalizations
coated denuder with the n-alkane-PAH mixture and extracted were applied to account for losses during the reduction
three times using the ‘rolling rinse’ method, as described of volume, the extraction efficiency of the hydrocarbons

VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 601

FIGURE 4. Comparison of the percent recovery of n-alkanes (a) and PAHs (b) from the reduction of volume procedure, the extraction of
an 8-channel XAD-4 denuder, the extraction of a quartz filter and the extraction of a XAD-4 impregnated quartz filter. The open markers
are the recovery of the internal standard nC24D50. The error bars represent the uncertainty at the 1σ level.

FIGURE 5. Average concentration of n-alkanes for LaPorte, TX on 8/31/2000 from 00:00-11:30 (local time), including the contribution from
each HiC IOGAPS component. The error bars represent the uncertainty at the 1σ level.

averaged >95% (10 for all standard compounds for XAD-4 concentrations for the hydrocarbons have been corrected
impregnated quartz filters and >95% ( 10 for >nC18 and only for the losses during the reduction of volume procedure.
associated PAHs for quartz filters. Furthermore, the normalized recovery was 104% ( 8 for
The reduced recovery from quartz filters can be attributed phenanthrene-d10 and 97% ( 6 for fluoranthene-d10; an-
to evaporation of the more volatile hydrocarbons (<nC18 and thracene-d10 was below the detection limit. (While the spiked
associated PAHs) along with the solvent, from the filter amount of anthracene-d10 could not be quantified by GC-MS,
surfaces before extraction. The extraction efficiency for it can be quantified using HPLC with fluorescence detection.
remaining amounts of these compounds from the quartz The amount used was chosen to avoid swamping the native
filters was also assumed to be >95% ( 10. This assumption anthracene during HPLC analysis.)
is supported by observation that (1) the extraction efficiency Figure 5 shows the ambient concentrations of the n-
of the less volatile species on the quartz filter was >95%, and alkanes. Most of the mass of the lighter molecular weight
(2) the extraction efficiency for <nC18 and associated PAHs n-alkanes < nC19 was on the denuders, consistent with their
from the XAD-4 impregnated quartz filters was also >95%. higher volatility. No evidence was seen for particle loss in
Ambient Sample. The optimized procedure was used to the first or second denuders. For n-alkanes g nC19 a significant
analyze an entire HiC IOGAPS sample that was collected fraction of the mass was on the quartz filters, but even more
during the TexAQS-2000 at the LaPorte Airport, Houston, TX was found on the XAD-4 impregnated after-filters. This figure
(21). The mean ambient temperature was ∼28 °C during illustrates the need for back-up sorbent substrates, such as
this period. Since the extraction efficiency was >95% for the XAD-4 impregnated quartz filters, to trap SVOC desorbed
alkanes and aromatic hydrocarbons, the reported ambient from the collected particles. Figure 6 shows concentrations

602 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003

FIGURE 6. Average concentration of PAHs for LaPorte, TX on 8/31/2000 from 00:10-11:30 (local time) including the contribution from each
HiC IOGAPS component. The error bars represent the uncertainty at the 1σ level.

of the PAHs that could be detected by the GC-MS. Most of semivolatile and nonvolatile organic compounds from air
the mass of the PAHs was on the first denuder, and the samples. The results showed how to minimize and account
concentrations generally decreased as the MW and ring size for losses of both semivolatile hydrocarbons that can
increased and the lability decreased. Naphthalene was the evaporate, and high molecular weight species that can
only PAH found in appreciable concentration on the second precipitate, during concentration steps that are intended to
(downstream) denuder, and it also broke through to the lower limits of detection. Quantitation of polar compounds
XAD-4 impregnated filters. From these data, the volumetric will require meeting these objectives with suitable laboratory
capacity of the two denuders for naphthalene in ambient air and field surrogates, internal standards and detection
at 28 °C was less than the 58 m3 of air that passed through methods, while accounting for additional causes of decreased
them. The apparent volumetric capacity for the other PAHs recovery such as lower extraction efficiency and analyte
was greater than 58 m3 but could not be established more degradation during sample preparation and analysis.
precisely because they did not break through to the second Our results point to the need for inclusion of a range of
denuder, or they were below limits of detection for the standards for recovery. Even for relatively inert compounds
GC-MS. such as hydrocarbons, non-normalized recovery depended
on volatility, MW, ring size and solubility. We recommend
Discussion that a series of deuterated surrogates be used to account
adequately for the preparation-related losses of organic
In other studies involving only one class of nonpolar species that are associated with ambient particulate matter.
compounds, or classes with similar polarity and solubility, These surrogates should span the ranges of both volatility
single solvents led to higher and more uniform non- and polarity that are expected in an environmental matrix
normalized (uncorrected for reduction of volume losses) that could also contain multifunctional and polymeric
recovery of deuterated PAH from polyurethane foam (PUF) species. Investigators should adequately describe the nor-
and filters than we found for the solvent mixture of Hex: malization algorithm(s) for each compound or group of
MeCl2:MeOH. For example, Hawthorne et al. (22) reported compounds. A good example is provided by McDonald et al.
non-normalized recoveries of over 95% for PAHs from (27) who used a suite of deuterated PAH from naphthalene-
naphthalene-d8 to benzo[ghi]perylene-d12 and several deu- d8 to coronene-d12. They reported that the recovery of each
terated methoxyphenols. Fernandez et al. (23) reported better analyte was normalized to the recovery of the deuterated
recoveries of naphthalene-d8-benzo[ghi]perylene-d12 ex- PAH with the closest chromatographic retention time and
tracted with hexane from PUF (82-91%), than of anthracene- stability.
d10-benzo[ghi]perylene-d12 from filters that were extracted Although we found no other reports in the literature that
in a mixture of MeCl2 and methanol (53-126%). Liang et al. included non-normalized recovery data for such a wide range
(24) and Mader and Pankow (25, 26) also found non- of alkanes and PAHs, our results for individual native
normalized recoveries of several deuterated PAH recoveries compounds were similar to published recoveries of the same
close to 100% from PUF and filters that were extracted only individual deuterated compounds added to filters and
in MeCl2. extracted with the solvents used by Mazurek et al. (8, 9):
However, more comprehensive characterization of air- hexane, followed by a benzene-2-propanol mixture. Ad-
borne semivolatile and particulate organics frequently ditionally, the recovery data of Schauer et al. (28-29) for
requires sample preparation procedures for compounds that nC15D32 and nC24D50 fit nicely with the data for our optimized
span wide volatility, molecular weight and polarity ranges. procedure (Figures 3 and 4). Although the Mazurek et al. (8,
In this paper we concentrated on the recovery of nonpolar 9) procedure has been widely cited, there are few reports in
hydrocarbons from a solvent mixture that was chosen to the literature of non-normalized recoveries of deuterated
optimize dissolution of both nonpolar and polar organics, standards. The literature contains even fewer descriptions
without the need for multistep extractions. Table 3 sum- of whether and how analyte concentrations were corrected
marizes our recommendations for optimal recovery of both for variation of losses with MW or ring size.

VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 603

TABLE 3. Recommendations for Optimal Recovery of Airborne Semivolatile and Particulate Organic Compounds from Sorbent and
Filter Media
issue action concerns
collection media choose high-capacity extractable sorbent for gas phase; few options
inert extractable filter or surface for particles
solvents choose efficient solvent mixture polarity
reactivity
solubility
MW
concentration optimize with complete range of target analytes; irreversible volatility losses;
choose several surrogates for range of functional groups, precipitation as solvent composition
volatility, solubility and molecular weight; changes
resolubilize analytes by reconstituting extract in original
(or improved) solvent mixture, including wall rinse;
choose appropriate internal standards for
detection technique
validate with standard mixture, blanks and SRM before
sample preparation;
track and report recovery data
extraction validate before routine sample preparation
gaseous SVOC validate by adding standards and surrogates to clean standards on sorbent may not be representative
sorbent media, analyze; of behavior during sampling
add recovery surrogates to collection media
particles add standards and surrogates to clean filter and sorbent dry, particle-free quartz filter has limited
impregnated filters, analyze; capacity for SVOC
extract, concentrate, analyze;
validate with particulate SRMs when available;
add recovery surrogates to sample filters immediately
before extraction, without drying
detection choose best analytical method(s);
determine response factors with authentic standards
quantitation choose appropriate surrogate for extract volume change
add internal standard(s) for instrument response
data reduction track and report surrogate recovery data confusion about standard conditions;
report target analyte concentrations per unit air volume inadequate QA/QC
at ambient T and P
report measurement uncertainties
document procedures
storage monitor sample extracts periodically for losses

Our results have several implications for ongoing efforts nonpolar compounds, concerns may remain about inter-
to characterize airborne PM and associated semivolatile pretation of these recent studies (28, 29, 31-34). We found
compounds. First, caution must be applied to generalizing two and a half times lower recovery for nC15H32 than for n
the recoveries of SVOC from clean filters in the laboratory. C24H50. Quantitative comparisons among the results of many
The evaporation of labile species from filters may not investigators will be affected by unclear explanation of
represent the true behavior of the SVOC in PM because of normalized algorithm and inconsistent use of surrogates.
their stronger adsorption to PM than to the filter substrates.
Overestimates of their ambient PM-associated concentrations
would result. Second, caution must also be applied when Acknowledgments
interpreting results based on one surrogate for recovery of
We would like to thank our colleagues Douglas A. Lane and
many compounds. Basing ambient concentrations on re-
Cristian Mihele from Environment Canada and Yanbo Pang
covery of a midrange (volatility and MW) surrogate would
lead to worsening underestimation of ambient concentra- from Washington State University and the University of
tions of compounds the more different in volatility and MW Washington, for their inspiration to use the multiple-sample
they were from the surrogate. Problems could arise concern- programmable evaporation technique, and for their as-
ing mass closure estimates from such studies and they should sistance with the field study. We thank Robert Seila from US
be interpreted with caution. This may have been the case for EPA, Jenny Sprung from the Labconco Corporation, and
the extensive characterization of PM sources by Rogge et al. Raymond L. Dod of Lawrence Berkeley National Laboratory
(10-17) who used only nC24D50 for recovery of all non- for their contributions. We also gratefully acknowledge access
derivatized compounds. Our results suggest that using only to the laboratory facilities of Prof. Jim Lester at the University
nC24D50 could underestimate PM concentrations for both of Houston, Clear Lake. The United States Environmental
lighter and heavier hydrocarbons outside the nC19-nC34 Protection Agency through its Office of Research and
range. Similarly, Fraser et al. (30, 31) used nC24D50 as the Development partially funded and collaborated in the
recovery surrogate for PM, but also added it to a PUF, without research described here under Interagency Agreement DW-
clearly detailing the normalization algorithm for the semi- 898164-01 to E. O. Lawrence Berkeley National Laboratory
volatile compounds. There is good agreement with our data (LBNL), through the U.S. Department of Energy, under
for semivolatile compounds. However, significant bias could Contract DE-AC03-76SF00098. The project was also sup-
exist even when a few surrogates for recovery are used, unless ported by the University of Texas under an intergovernmental
recovery has been measured and normalized through a series personal act and a contract with LBNL. It has been subjected
of related compounds. Despite improved procedures that to Agency Review and approval for publication. Mention of
use solvent mixtures (or sequential solvent extraction) and trade names or commercial products does not constitute an
two deuterated species (nC15D32 and nC24D50) for recovery of endorsement or recommendation for use.

604 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 3, 2003

Literature Cited (20) Gundel, L. A.; Daisey, J. M., Stevens, R. K. Method for fabricating
a quantitative integrated diffusion vapor-particle sampler for
(1) Gundel, L. A.; Lane, D. A. In Gas and Particle Phase Measure- sampling, detection and quantitation of semi-volatile organic
ments of Atmospheric Organic Compounds. In Advances in gases, vapors and particulate components, 2001, U.S. Patent
Environmental, Industrial and Process Control Technologies; 6,226,852.
Lane, D. A., Ed.; Gordon and Breach: Newark, 1999; Vol. 2, pp (21) Swartz ,E.; Stockburger, L.; Michel, C.; Lane, D. A.; Pang, Y.;
287-332. Gundel, L. A. Characterization of Ambient Aerosols at LaPorte,
(2) Gundel, L. A.; Lee, V. C.; Mahanama, K. R. R.; Stevens, R. K.; Texas, during TEXAQS-2000, presented at the 20th Annual
Daisey, J. M. Atmos. Environ. 1995, 29, 1719-1733. Conference of the American Association for Aerosol Research,
(3) Grosjean, D. Anal. Chem. 1975, 47, 797-805. Portland, OR, Oct 15-19, 2001.
(4) Appel, B. R.; Hoffer, E. M.; Kothny, E. L.; Wall, S. M.; Haik, M.;
(22) Hawthorne, Miller, D. J.; Langenfeld, J. J.; Krieger, M. S. Environ.
Knights, R. L. Environ. Sci. Technol. 1979, 13, 98-104.
Sci. Technol. 1992, 26, 2251-2262.
(5) Cadle, S. H.; Groblicki, P. J. In Particulate Carbon: Atmospheric
(23) Fernandez, P.; Grimalt, J. O.; Vilanova, R. M. Environ. Sci. Technol.
Life Cycle; Wolff, G. T., Klimisch, R. L., Eds.; Plenum Press: New
2002, 36, 1162-1168.
York, 1982; pp 89-109.
(6) Daisey, J. M.; Kneip, T. J.; Hawryluk, I.; Mukai, F. Environ. Sci. (24) Liang, C.; Pankow, J. F.; Odum, J. R.; Seinfeld, J. H. Environ. Sci.
Technol. 1980, 14(12), 1487-1490. Technol. 1997, 31, 3086-3092.
(7) Gundel, L. A.; Daisey J. M.; de Carvalho, L. R. F.; Kado, N. Y.; (25) Mader, B. T.; Pankow, J. F.; Atmos. Environ. 2000, 34, 4879-
Schuetzle, D. Environ. Sci. Technol. 1993, 10, 2112-2119. 4887.
(8) Mazurek, M. A.; Simoneit, B. R. T.; Cass, G. R.; Gray, H. A. Int. (26) Mader, B. T.; Pankow, J. F.; Atmos. Environ. 2001, 35, 1217-
J. Environ. Anal. Chem. 1987, 29, 119-139. 1223.
(9) Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. (27) McDonald, J. D.; Zielinska, B.; Fujita, E. M.; Sagebiiel, J. C.; Chow,
Technol. 1991, 25, 684-694. J. C.; Watson, J. G. Environ. Sci. Technol. 2000, 34, 2080-2091.
(10) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (28) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T.
Simoneit, B. R. T. Environ. Sci. Technol. 1991, 25, 1112-1125. Environ. Sci. Technol. 1999, 33, 1566-1577.
(11) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (29) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T.
Simoneit, B. R. T. Environ. Sci. Technol. 1993, 27, 636-651. Environ. Sci. Technol. 2002, 36, 567-575.
(12) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (30) Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T.; Rasmussen, R. A.
Simoneit, B. R. T. Environ. Sci. Technol. 1993, 27, 1892-1904. Environ. Sci. Technol. 1997, 31, 2356-2367.
(13) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (31) Fraser, M. P.; Cass, G. R.; Simoneit, B. R. T.; Rasmussen, R. A.
Simoneit, B. R. T. Environ. Sci. Technol. 1993, 27, 2700-2711. Environ. Sci. Technol. 1998, 32, 1760-1770.
(14) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (32) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T.
Simoneit, B. R. T. Environ. Sci. Technol. 1993, 27, 2736-2744. Environ. Sci. Technol. 2002, 36, 1169-1180.
(15) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (33) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T.
Simoneit, B. R. T. Environ. Sci. Technol. 1994, 28, 1375-1388. Environ. Sci. Technol. 2001, 35, 1716-1728.
(16) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.; (34) Schauer, J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T.
Simoneit, B. R. T. Environ. Sci. Technol. 1997, 31, 2726-2730. Environ. Sci. Technol. 1999, 33, 1778-1587.
(17) Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Gass, G. R.;
Simoneit, B. R. T. Environ. Sci. Technol. 1997, 31, 2731-2737.
(18) Gundel, L. A. Standard operating procedures for coating annular
denuders with XAD-4 resin; U.S. EPA Chemical Speciation Received for review July 11, 2001. Revised manuscript re-
Program, 1998, LBNL PUB-3143. ceived September 3, 2002. Accepted October 7, 2002.
(19) Gundel, L. A.; Daisey, J. M.; Stevens, R. K. Quantitative Organic
Vapor-Particle Sampler, 1998, U.S. Patent 5,763,360. ES011128Z

VOL. 37, NO. 3, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 605