United States
 Environmental Protection
 Agency
 Environmental Sciences Research
 Laboratory
 Research Triangle Park NC 27711
                                                                   s~
                                                                  (
                                                                  \  o
 Research and Development
 EPA-600/S2-80-194  Apr 1981
 Project Summary
 GC/MS  Analysis  of
 Ambient  Aerosols  in  the
 Houston,  Texas,  Area
 Catherine H. Skintik
   This study was designed to provide
 information on the organic aerosol
 pollutants in the Houston, Texas, area.
 The Research Triangle Institute (RTI)
 was responsible for qualitative and
 quantitative analysis of ambient air
 aerosols and vapor samples collected
 in the Houston area by the Radian
 Corporation of Austin,  Texas.  Three
 different types of samplers were used:
 high volume (hi-vol) and dichotomous
 samplers were employed to collect
 particulates, and resin-trap samplers
 (developed  at the Illinois Institute of
 Technology Research Institute and
 known as IITRI  samplers) were em-
 ployed to collect vapor-phase organic
 samples. Filters and sorbent resins
 were shipped to RTI following field
 sampling. A total of 23 samples were
 analyzed: 12 filter and adsorbent trap
 samples, 5 dichotomous samples, and
 6 hi-vol samples.
  Hi-vol samples were extracted and
 solvent-partitioned prior to analysis by
 glass capillary gas chromatography/
 massspectrometry (GC/MS). Dichot-
 omous filters were  extracted and the
 concentrated extracts analyzed directly
 by GC/MS.  Vapors were analyzed by
 thermal desorption from the resin bed
 followed by  GC/MS.
  Analyses  showed the presence of
 many saturated and unsaturated ma-
 terials  in  all samples. The extracts
 from hi-vol  filters  contained small
 quantities of organics, and their anal-
ysis was complicated by the presence
 of background contaminants tenta-
 tively identified as siloxanes. Results
 from the dichotomous samples indi-
 cated insufficient collection of ma-
 terial for comprehensive air analysis.
 Vapor-phase organics  collected  in
 IITRI samples consisted predominantly
 of hydrocarbons, alkylated benzenes,
 and some chlorinated compounds.
 Quantitative analysis was carried out
 for selected components from each
 sample type.
  This Project Summary was devel-
 oped by EPA's Environmental Sciences
 Research Laboratory, Research Tri-
 angle Park,  NC, to announce key
 findings of the research project that is
 fully documented in a separate report
 of the same title (see Project Report
 ordering information at back).

 Introduction
  The 1977 Clean Air Act Amendments
 called for EPA research studies in the
 Gulf Coast area. Priority  was given to
 the problems of aerosol pollutants, and
 Houston,  Texas,  was chosen as the
 initial study site. The EPA issued a
 procurement request to obtain gas
 chromatography/mass  spectrometry
 (GC/MS) evidence on the identity of the
 organic particulate pollutants present in
the atmosphere above Houston and
surrounding areas. The evidence ob-
tained was intended to guide studies of
 health effects from ambient aerosols
and to determine sources and causes of
aerosol pollution in the Houston area.

-------
Sampling
  The major sampling effort for the
Houston air study was provided by
Radian  Corporation,  Austin, Texas,
under a  separate contract. All samples
analyzed by RTI were collected by
Radian during September and October
1978 using filter and  adsorbent trap,
dichotomous, or hi-vol  samplers.
Filter and Adsorbent Trap
(IITRI) Sampler
  The filter and adsorbent trap sampler
consisted of a conventional high-volume
sampler modified to allow for sampling
of vapor-phase material downstream of
the aerosol filter. The  vapors  were
collected by passing a portion of the
post-filter airstream through cartridges
containing a trapping medium consisting
of polymeric  sorbents. Two sorbent-
containing  cartridges were connected
in series, and four such cartridge pairs
comprised the vapor sampling unit. This
sampler was designed and constructed
by researchers at the Illinois Institute of
Technology Research Institute (IITRI)
and is referred to in this report as the
IITRI sampler. A total of three IITRI
samplers (12 cartridge pairs) was used
for this program.
  The  sorbent selected for use in the
samplers was Tenax® GC, a phenylene
oxide polymer with well-tested vapor
trapping properties.
  All sorbed organic species had to be
removed from the Tenax® GC before it
was  released for field use. These ma-
terials were removed by thermal de-
sorption, after which the levels of back-
ground materials remaining were as-
sessed by glass capillary gas chroma-
tography /flame ionization detect ion
(GC/FID) of the desorbed vapors.
  Once the cartridge pairs were cleaned
by thermal desorption, all U-tubes were
reassembled and each sealed sampler
was  shipped to Houston. The samplers
were designated by number (samplers
1, 2, and 3),  and each of the four
cartridge pairs for each sampler was
numbered 1 through 4. In addition, the
individual sorbent beds were labeled as
either upstream (U) or downstream (D).
Thus, each  sample could be described
by a single designation preceded by "IT"
for IITRI. For example,  the upstream
Tenax® GC sample from U-tube 3,
sampler 1, was referred to as IT1-3U.
  All sampling information was recorded
on data sheets supplied by RTI to Radian
and  returned with each  sampler. This
information is detailed in Table 1. One
cartridge pair  for each  sampler was
utilized as a field blank.  The blank was
never exposed to the sampling stream,
but experienced the same shipment,
storage, and handling conditions as the
sample cartridges.
Dichotomous Sampler
  The dichotomous sampling effort for
the Houston study was carried out by
the EPA (Inorganic Pollutant Analysis
Branch, Atmospheric Chemistry and
Physics Division). A portion of the many
dichotomous  filters used for aerosol
collection was shipped to RTI as specified
in the  sampling protocol. Five "fine"
filters were designated for extraction
and analysis by RTI. The  sampling
information for the dichotomous sam-
ples is  shown in Table 2.
Hi-Vol Sampler
  The high-volume samples were col-
lected on quartz fiber filters (Gelman
Micro Quartz).  Prior to field studies,
investigations were conducted to deter-
mine the extent of background organics
contained  in the quartz fiber filters
(OFF).  Each filter was subjected to
solvent washing experiments to deter-
mine  the results  of  such clean-up
procedures and the best wash medium
if clean-up  was shown to be effective.
Filter extracts and solvent washes were
analyzed using packed column GC/FID.
Results showed that both methanol/
chloroform  and diethyl ether removed
significant  amounts of contaminants
but that methanol/chloroform was
more efficacious than ether. A variation
in contaminant levels for filters from the
Table 1.    Samp/ing Data—IITRI Samples
IITRI
Sample No.
IT1-1 (blank)
171-2

IT1-3

IT1-4

IT2-1 (blank)
IT2-2

IT2-3

IT2-4

IT3-1

IT3-2

IT3-3

IT3-4 (blank)
Houston
Site
	
#21-Alief

U23-Stude Park

#23-Stude Park

—
#23-Stude Park-
Central
#23-Stude Park-
Central
#21 -A lief -West

#22-Baytown Area-
North
#22-Baytown Area-
North
#22-Baytown Area-
North
—
Sampling
Data
—
10-08-78
(day}
J 0-06 -78
(day)
10-06-78
(night)
—
10-11-78
(day)
10-11-78
(night)
10-10-78
(day)
10-10-78
(day)
10-11-78
(night)
10-11-78
(day)
—
Sampling
Rate (t/m/nj
—
2.84

2.84

2.84

—
2.84

2.84

2.84

2.84

2.84

2.84

—
Sampling
Time (min)
—
710

720

595

—
585

720

720

690

660

720

—
Sample
Volume (L)
—
2,0/6

2,045

1,690

—
1,661

2,045

2,045

1,960

1,874

2.045

—

-------
 Table 2.    Sampling Data—Dichotomous Samples.

 Dichotomous  Houston    Sampling     Sampling    Sampling    Sampling
   Filter No.      Site        Date     Rate (l/min)   Time (hr)   Volume (m3)
4F-011

4F-012

4F-018

4F-021

4F-022

#19

#19

#17

#21

#21

9-27-78
(night)
9-28-78
(day)
10-6-78
(day)
10-9-78
(day)
10-10-78
(day)
16.7

16.7

16.7

16.7

16.7

12.0

12.0

11.8

12.0

12.0

12.0

12.0

11.8

12.0

12.0

 same lot was noted, and these results
 were conveyed to Radian prior to field
 sampling.
   Data sheets were supplied to Radian
 for  recording pertinent sampling  in-
 formation and were returned to RTI.
 Sample Analysis
   To minimize problems associated
 with sample decomposition, all ana-
 lytical work was carried out in a labora-
 tory fitted  with yellow lights (trans-
 mission >500 nm); samples and sample
 extracts were never allowed to exceed
 50°C, and were stored under conditions
 of refrigeration and darkness.
 IITRI Samples
   The Tenax®  GC was removed from
 the IITRI samplers, and a measured
 portion of the resin was transferred to
 clean glass cartridges. The cartridge
 contents were  then analyzed by glass
 capillary GC/MS. The whole sample
 was introduced to the analytical system
 by thermal desorption  into a cryogenic
 capillary trap and then rapidly swept
 onto the chromatographic column. Each
 cartridge was loaded with a known
 amount of perfluorobenzene  and per-
 fluorotoluene for use  in quantitation.
 Qualitative data interpretation was
 done by visual inspection of the spectra
 and comparison with standard reference
 spectra. Quantitation was achieved by
 determining the peak intensity of a
 characteristic mass for a compound
 from the chromatographic peak. The
 amount of compound in the sample was
Ithen calculated using the relative molar
 response (RMR) factor and the integrated
 area for a standard. The concentration
 in ambient air was calculated using the
 volume sampled.
Dichotomous Samples

Preliminary Experiments
  The dichotomous sampler was de-
signed to collect aerosols for size frac-
tionation, mass determination, and
compositional characterization. The
sample  rate (16.7  l/min) is low com-
pared to the commonly used hi-vol
sampler (—600 l/min),  and thus the
amount of aerosol collected is smaller.
Since organics typically comprise 5 to
15  percent of the total mass  of air
aerosols, and since only a limited fraction
of a sample extract can  be introduced
into the GC/MS system, experiments
were undertaken prior to field sample
analysis to optimize the organics recov-
ery procedure.  Efforts were directed
toward  reducing  the extracts to the
smallest practical  volume, since the
maximum aliquot that could be removed
for  GC/MS injection  was fixed at ap-
proximately 1 fj\.  Given the small a-
mounts  of organics likely to be present
and the  size of the containers available
for  use,  concentrations of an extract to
low /j\ volumes without  concomitant,
significant losses of sample components
presented some difficulty.
  Information obtained from EPA  led to
examination of Schenk-Bauer sedimen-
tation tubes as appropriate containers
for  sample concentration. Recoveries
were tested using 14C-7, 12-dimethyl-
benz(a)anthracene, a compound pre-
sumably representative of an important
class of aerosol organics. The  tests
succeeded in concentrating a significant
amount of material in a small volume of
solvent, from which aliquots could then
be removed for GC/MS analysis. Varia-
                                        tions in the precise extent of recovery
                                        could be compensated through the use
                                        of internally added standards.
Analysis of Field Samples
  Based on the preceding study, final
concentration of dichotomous filter
extracts was achieved using Schenk-
Bauer sedimentation tubes. Each di-
chotomous filter was extracted using a
custom-made Teflon® extraction block
provided by EPA. Two extractions were
carried out with approximately 4 ml of
chloroform and sonication in a  water
bath. The extracts were combined and
the  internal standard (anthracene-dio)
added. The extract was then transferred
in portions to the sedimentation tubes
for solvent removal. Just prior to GC/MS
analysis, the dried extracts were recon-
stituted in 10//I of chloroform. Approxi-
mately  1 //I was removed for injection
onto the GC column.
Hi-Vol Samples

Extraction/Partition
  Hi-vol  filters (OFF) were extracted
overnight with methanol using a Soxhlet
apparatus. Following solvent removal,
the residue was fractionated. The indi-
vidual fractions were then concentrated
for GC/MS/COMP analysis.
  The efficacy of the scheme was as-
sessed by subjecting a mixture contain-
ing known amounts of compounds to
the partition procedure. The mixture
consisted of benzoic acid, phenol, quin-
oline, hexadecane,  phenanthrene, and
ethylene glycol. These compounds were
chosen to represent the five classes of
materials produced by the partition
scheme (all these materials have been
found in air paniculate samples except
ethylene glycol). No information on the
composition of the polar neutral fraction
was available; therefore, ethylene glycol
was included as a likely component of
this fraction based on its known chemi-
cal properties.  The experiment was
conducted using  both large and small
mass samples. Recoveries were deter-
mined gravimetrically.
  As a further check on the procedure,
thin-layer chromatography (TLC) scans
were conducted on each fraction to
ascertain the extent, if any, of compound
spillover into  other  fractions.  No such
spillover was detected.

-------
 GC/MS/COMP Analysis
   Samples were analyzed using an LKB
 2091 GC/MS with a dedicated PDP-
 11/34 data system. The samples were
 chromatographed on an OV-101 capil-
 lary column  (25  m, WCOT, obtained
 from LKB). The column was  held at
 100°Cfor2 minutes after injection, and
 then heated to 265°C at a rate of 8°/min.
 Carrier gas flow rate was 1.8 ml/min.
 Injector temperature was 245°C. Mass
 spectral scans were taken every 3 sec,
 scanning from 5 to 492 amu. The ion
 source temperature was 210°; electron
 energy, 70 eV; trap current,  50 //A;
 accelerating voltage, 3500 V; and multi-
 plier setting, 400-500. Total  ion  current
 and mass spectra plots were generated
 for interpretation.
   Data were interpreted by comparison
 of the unknown spectra with standard
 reference collections. Where no refer-
 ence spectrum was available, or where
 spectral quality was not sufficient for
 positive identification, the compounds
 were labeled  "tentative." Further vali-
 dation of equivocal identification was
 made by consultation with other chem-
 ists and spectroscopists experienced in
 similar  research. This mass spectral
 interpretation protocol was established
 to extract  the maximum information
 from  the data  and to guard against
 misidentification. The specific compound
 identification criteria are as follows: (1)
 computer interpretation, (2) manual
 interpretation, (3) manual interpretation
 plus retention time/boiling point  of
 compound, (4) manual interpretation
 plus retention time of authentic com-
 pounds, and  (5) criterion number (4)
 plus independent  confirmation tech-
 niques.


 Results
   Within the limits  of the GC/MS
 technique, complete identifications of
 the organics isolated from each sampler
 medium were made. Of the compounds
 identified,  certain components were
 selected for quantitation. The selection
 was based on relative abundance and
 the known hazard potential of each
 compound, and was made after consul-
 tation with the project officer. The
 results for each of the three sample
 types are discussed individually in the
 following sections.

IITRI Samples
  Five quartz fiber filters were returned
for analysis of aerosol organics (a blank
filter was subsequently made available
for analysis), and 12 sorbent-containing
cartridge pairs (9 samples, 3 blanks)
were returned for analysis of volatiles.

Volatiles—Qualitative Analysis
  The 12 IITRI cartridge tubes contained
an upstream and a downstream Tenax®
GC bed (~2.5 g Tenax® GC/cartridge)
thus providing 24 samples for analysis.
Analyses of both the upstream and
downstream cartridges for all blanks
and for two sample cartridges were
carried out. For all these samples it was
evident that the analysis of the down-
stream cartridge represented a duplica-
tion of effort since a large number of
compounds (>50) were found on both
the upstream and downstream cartridge,
the same compounds were found on the
downstream as the upstream cartridge,
and no useful information was obtained
by virtue of analysis of the downstream
'cartridge. Therefore, only the upstream
cartridge was  analyzed for the  bulk of
the volatile samples. A summary of
those samples analyzed is  provided in
Table 3. For each sample a gas chro-
matogram, in the form of a total ion
current (TIC) plot (TIC versus mas!
spectral scan number) was obtained
Each peak of theTIC plot was numbered
and the identification of all compound;
corresponding to each peak was tabulated
  A  large number of compounds was
identified for each cartridge. With on<
exception, from 104 to 174 compounds
were characterized for each upstrearr
cartridge, with approximately two-
thirds of these components detected or
the corresponding downstream cartridge
The  appearance of relatively large
numbers of components on the down-
stream cartridge indicates significam
breakthrough from the upstream car-
tridge. In view of the volume of aii
sampled (Table 1), and the known break-
through volumes for similar components
using Tenax® GC  under similar condi-
tions, these findings are not surprising
The vast bulk of volatile species (—80 tc
85 percent) was identified as saturated/
monounsaturated hydrocarbonsand
alkylated aromatic hydrocarbons. For
the amount of Tenax® GC used and the
temperatures at which sampling was
carried out, the breakthrough volumes
for these compound types are  on the
Table 3.    Summary of Analysis of IITRI Volatiles

                           Qualitative Analysis
  IITRI Sample No. *             by GC/MS
           Number of Components
                 Identified
  IT1-1U (Blank)
  IT 1-1D (Blank)
  IT1-2U
  IT1-2D
  IT1-3U
  IT1-3D
  IT1-4U
  IT1-4D
  IT2-1U (Blank)
  IT2-2D (Blank)
  IT2-2U
  IT2-2D
  IT2-3U
  IT2-3D
  IT2-4U
  IT2-4D
  IT3-1U
  IT3-1D
  IT3-2U
  IT3-2D
  IT3-3U
  IT3-3D
  IT3-4U (Blank)
  IT3-4D (Blank)
                     82
                     46
                    174
                    112
                    147
                    104
                    167

                     89
                     69
                    165

                    153

                    149

                     68

                    104

                    107

                     81
                     70
*Sample Code: IT3-4U refers to IITRI Sampler #3, cartridge pair #4, upstream cartridge*
^-Indicates analysis performed.                                               \
-Indicates analysis not performed.

-------
order of 0 to 100 I and 100 to 1,000 I,
respectively. The amount of air sampled
for the IITRI cartridges averages about
2,000 I.
  The qualitative data from  the IITRI
cartridge analysis shows a considerable
number and amount (from TIC) of com-
pounds associated with the blank car-
tridges. These cartridges were never
intentionally exposed to ambient air,
and  thus  should have produced gas
chromatographic patterns similar to
those for components desorbed from
the IITRI cartridges. The intensity of the
chromatographic pattern obtained for
the blanks indicates exposure of these
cartridges to ambient air. This was
presumably due to leakage through
either the toggle valves controlling air
flow  to the IITRI U-tubes or the Swage-
lock fittings.
  The qualitative profile of the volatile
components collected by the IITRI sam-
plers is very similar to results obtained
by other workers at various urban sites.
Approximately 150  compounds were
identified from each cartridge. Of these
about  60 percent were saturated/
monounsaturated hydrocarbons, about
         20 to 25 percent were aromatics, about
         5 percent were halogenated hydrocar-
         bons, and the remaining components
         (~10 to 15 percent) consisted of oxygen-
         ated compounds (aldehydes, ketones,
         ethers).
           Sampling for vapor phase material
         was carried out at three sites represent-
         ing a suburban community, a downtown
         area, and a heavily industrialized  site.
         Based solely on the qualitative results,
         no major trends or characteristics
         regarding site specificity were noted.
         Although similar volumes of air were
         sampled at each site, the  average
         number of compounds  identified from
         the industrialized site was less (93) than
         the average number identified from the
         suburban site (162) or  the downtown
         area (158). While this observation may
         seem noteworthy, the limited number of
         samples analyzed from  each site and
         the excessive volume of  air sampled for
         the amount of Tenax® GC used detract
         from its significance.
           No significant differences were noted
         between results obtained for samples
         collected during  the day and those
         collected at night.
Volatiles—Quantitative Analysis
  The selection of compounds for quan-
titative  determination was based on
considerations of relative abundance
and toxicity. These criteria were adopted
following consultation with the project
officer.  Toxicity  information was ob-
tained from the "Handbook of Environ-
mental Data on Organic Chemicals."
  Using the method  of relative molar
response (RMR) factors, reference quan-
titation  standards were obtained and
their GC/MS response relative to an
internal standard was determined under
the same conditions as were used for
the analysis of field samples. In some
cases, duplication of effort was avoided
by the determination of RMRs for several
similar species using a single, represen-
tative standard.  It is known from past
experience that structurally similar
compounds generally produce similar
response factors.
  A total mass of each species was
divided by the volume of air sampled to
provide ambient air  concentrations.
These quantitation results are shown in
Table 4. In one case, methylene chloride,
the large amount of material introduced
Table 4.    Quantitation Results from IITRI Volatiles Analysis
         Compound
                                                             Concentration (ng/m3)
IT1-2U   IT1-3U   IT1-4U   IT2-2U  IT2-3U    IT2-4U    IT3-1U    IT3-2U    IT3-3U
  n-Hexane
  methylcyclohexane
  n-heptane
                                     5920
                                     1582
          998
          126
4115
788
n- octane
n-nonane
n-decane
n-undecane
n-dodecane
n-tridecane
benzene
toluene
ethylbenzene
xylene isomer
xylene isomer
1 ,2,3-trimethylbenzene
1 ,2,4- trimethylbenzene
1 , 3, 5 - 1 rimet h ylb enzen e
ethyltoluene isomer
styrene
naphthalene
chloroform
carbon tetrachloride
methylene chloride
1 ,2-dichloroethane
trichloroethylene
1, 1 , 1 -trichloroethane
dichlorobenzene
1059
1086
1418 1796
2495 2478
880 811
6636
2860 10443

3835
1659 1362
333
234 233
473
181
165
7169


4142 3624




4211
6049
4285

11214
8126
8064

1151
1383
1253
1650
8971



27817




4561
9631
9201
33454
3112

23698

21841
7763

2240

756
4078



79281




1052
1209
1963

14197

8239
1725

371


1480



13451

3172


2476


5336
20124
4781
337
4107
766
208
677



169982

6786
179


158
1319






429
104

685
388
149526
219
2543
79

203
3393
6242
1458
2475
643




721


1922
284218

1967
216


1169

99
47
5
22



64
22
236

15200

1414

15

-------
Table 5.    Summary of Qualitative Results from Dichotomous Filter Extracts
                                                                      Composition of Real Components
Filter
No.
4F-011
4F-012
4F-018
4F-021
4F-022
Total
Peaks
Identified
25
41
42
28
25
Background
Components
10
16
16
13
8
Real
Components
15
25
26
15
17
Sat'd
Hydrocarbons
8
13
11
6
12
Unsat'd
Hydrocarbons
2
5
8
3
1
Alkyl
Phenols
1
1
2
1
1
Unknowns
4
6
5
5
3
to the GC/MS produced analyzer satu-
ration.  For this compound a minimum
concentration is therefore reported.
However, as noted previously, the rela-
tively  large volume of air sampled
exceeded the known breakthrough
volumes (BTVs) for most components.
Thus all quantitative results must be
interpreted as minimum concentrations.

Nonvolatiles
  Since each IITRI sampler was used in
conjunction with a hi-vol  sampler,  a
quartz fiber filter (OFF) corresponding to
each of the IITRI cartridge pairs was
used at each sampling period. Five such
filters were returned to the RTI for
aerosol organics analysis.
  The analytical protocol for nonvolatile,
aerosol organics involved Soxhlet ex-
traction followed by fractionation of the
extracted organics into five major frac-
tions. These fractions were individually
analyzed using GC/MS. Earlier results
obtained for the analysis of the hi-vol
filters not used in conjunction with the
IITRI samplers showed very low levels of
organics in  each of the five fractions.
Consequently, for all but one of the hi-
vol filters used with the IITRI samplers,
the fractionation  scheme was not em-
ployed, and the GFF extract was analyzed
directly by  GC/MS. Diminished peak
intensities from the TIC plots  and the
relatively few compounds identified
even from the nonfractionated extracts
seem  to indicate that the levels of
nonvolatile aerosol organics from these
samples are quite low. The other obvious
result of these analyses is the presence
of substantial levels of background
components, principally phthalate esters
and siloxanes.

Dichotomous Samples

Preliminary Studies
  Prior to complete analysis of aerosol
organics from the five dichotomous
filters, the levels of background contam-
inants  and their sources were deter-
mined.  Control runs were carried out to
allow for the contribution of background
components from the extraction solvent,
from the dichotomous filter, and from
the extraction device utilized in  the
scheme. All operations for the control
runs were identical to those used for the
analysis of field samples.
  The  results of these control runs
indicated that the number of contami-
nants was  acceptably low. Not  unex-
pectedly, phthalate esters are present in
the solvent; the only other major con-
taminant is an unsaturated hydrocarbon
(octadecene) extracted from the extrac-
tion  block. The relative retention time
and  mass spectral fragmentation pat-
terns of all background  compounds
were noted for subsequent use during
qualitative analysis of the field samples.

Qualitative Analysis
  The number of components identified
from each extract ranged from 25 to 43,
of which a substantial portion was
background contaminants.  Phthalate
esters  and siloxanes were the  major
contaminants. Of those  compounds
identified as non-background materials,
saturated and unsaturated hydrocar-
bons accounted for nearly all species.
An alkylated phenol (Cs and/or do) was
also  identified in each extract, presum-
ably  indicating a motor vehicle exhaust
source  for this compound. A summary
of these results is presented in Table 5.

Quantitative Analysis
  The amount of aerosol  collected on
each dichotomous filter  is  quite low
(approximately 100 to  2,000 fjg) com-
pared to aerosol collection devices more
commonly used for nonvolatile organics
analysis. Thus, low-level components
were not present in quantities sufficient
for full qualitative/quantitative  treat-
ment.
  Of the materials characterized from
the dichotomous filter extracts, roughly
half were identified as non-background
components.  Of these,  5  to 6 were
present at levels that permitted quanti-
tation. These  compounds, all higher
molecular weight long chain hydrocar-
bons plus nonylphenol, are shown in
Table 6.
  The variation both qualitatively and
quantitatively between extracts is small,
and no trends were noted regarding
sampling site or time of sampling. The
levels of saturated hydrocarbons (Cis -
CIB) were comparable  to levels from
other urban sites and slightly higher
than levels found from nonurban sites. |
Hi-Vol Samples
  The quartz fiber filters designated for
analysis  under this program were re-
ceived approximately 3 months after
field collection. They were stored and
shipped  by Radian  Corporation per
protocol and were stored at -20°C in our
laboratories until analysis was initiated.
Although it had been anticipated that a
gravimetric determination of total col-
lected aerosols would be made, tearing
and loss of small but significant portions
of the OFF during unpacking precluded
such determinations. Following extrac-
tion, the total mass of aerosol organics
was determined; the results are shown
in Table  7. The organic loads ranged
from low  to average in comparison with
expected  levels from  urban aerosols.
  In contrast to the amount of aerosol
collected  by the dichotomous sampler,
the hi-vol  sampler is capable of efficiently
trapping  approximately 200 to 500
times as  much aerosol for a  24-hour
sampling  period. It is known from a large
body of work  that a wide variety of
organic compounds are adsorbed to th
air particulate.  As the direct analysis of

-------
 Table 6.
Quantitative Results from Dichotomous Filter Extracts
                                                                    Filter
Compound
Cg-alkyl phenol isomer
n-pentadecane
n-hexadecane
n-heptadecane
n-octadecane
octadecene & isomers
4F-001
60 ng( 5.0 ng/m3)
70 ng ( 5.8 ng/m3)
90 ng ( 7.5 ng/m3)
100 ng ( 8.3 ng/m3)
260 ng (21. 7 ng/m3)
4F-012
20 ng ( 1.7 ng/m3)
50 ng 1 4.2 ng/m3)
50 ng ( 4.2 ng/m3)
60 ng ( 5.0 ng/m3)
80 ng ( 6.7 ng/m3/
6620 ng (551.7 ng/m3)
4F-018
20 ng (1.4 ng/m3)
50 ng( 3.6 ng/m3)
100 ng (7.2 ng/m3)
100 ng (7.2 ng/m3)
4F-021
20 ng ( 1.7 ng/m3)
35 ng ( 2.9 ng/m3)
30 ng ( 2.5 ng/m3)
70 ng ( 5.8 ng/m3)
130 ng (10.8 ng/m3)
620 ng (51.6 ng/m3)
Table 7.    Total Organics Extracted from Hi-vol Filters

    Filter Number                        Weight of Organics Extracted (mg)
12
13
14
15
65
66
91 (blank)
8.5
5.2
7.3
14.1
23.1
24.3
3.4
such a mixture is not practical owing to
the large number of species present and
the generally unfavorable GC behavior
of some classes (e.g., acids, .bases), a
sample fractionation procedure was
proposed for the organic extracts ob-
tained from the hi-vol filters used in this
study


Qualitative Analysis
  Analysis by glass capillary GC/MS
was accomplished for each fraction
from each OFF extract. Firm characteri-
zation proved difficult for all  fractions
because of the unexpected presence of
large numbers and amounts of interfering
substances. Coelution of the interfering
species with the compounds of interest
produced complicated MS fragmentation
patterns, obscured sample components
in some cases, and effectively raised
sensitivity limits. The problem was
noted for virtually all hi-vol samples (no
such  interferences were seen for the
4ITRI  or  dichotomous  samples), and
investigations into the precise nature of
the substances were made.
  The identification of the components
from the sample filters indicated the
presence of a disproportionate number
of phthalate  esters  and siloxanes of
indeterminate structure. Both classes of
compounds possess single ions that are
characteristic of each compound type. It
is recognized that GC column bleed
from columns employing silicone-based
                             phases consists predominantly of si-
                             loxanes. This process is characterized
                             by a relatively slow increase  in GC
                             baseline and by the continuous presence
                             throughout a long series of recorded
                             mass spectra of background levels of
                             siloxane fragment ions. The siloxanes
                             found associated with the hi-vol samples
                             were identified as discrete and, for the
                             most part, intense peaks, possessing
                             mass spectra with fragment ion  inten-
                             sities much greater than is normally
                             observed for column bleed. Thus,  in
                             addition to  TIC chromatograms of the
                             filter blank  fractions, single ion plots
                             (m/z 73,149) were also obtained. When
                             compared with similar chromatograms
                             obtained from the GC/MS analysis of a
                             solvent control, where minimal m/z
                             73,149 responses were noted, it would
                             appear that the background components
                             are  associated  with the filters utilized
                             for sample  collection. Although each
                             OFF was cleaned prior to field use, as
                             noted earlier, a great deal of variation in
                             background contamination levels was
                             observed, even for filters from the same
                             lot.  Further, the filter clean-up proce-
                             dures as carried out in our laboratories
                             showed that different solvent systems
                             can  be  more effective than  others for
                             removing some contaminants.  It  is
                             possible that the ether washing proce-
                             dure used for the hi-vol filters in the
                             Houston study was not fully efficacious.
                              Other sources of background contam-
                             ination were considered, including the
 GC septa and silanized glass surface.
 Most septa are composed predominantly
 of silicone-based polymers. High tem-
 perature, exposure to solvent vapor
 clouds, and repeated piercing by syringe
 needles are among those processes that
 could conceivably lead to the production
 of siloxanes. Silanized glass surfaces
 are commonly  used in trace organic
 analysis systems, and are therefore also
 a likely source of siloxanes. Both mate-
 rials were examined and ruled out as
 siloxane sources.
  From  the  evidence obtained, within
 the limits of funding and effort provided
 for this  program, the background com-
 ponents, principally siloxanes and phthal-
 ate esters, appear to be associated with
 the QFF used in  this study.


 Quantitative Analysis
  The analysis of extracts from the hi-
 vol filters showed generally low levels of
 organics,  except for the above-men-
 tioned phthalate esters and siloxanes.
 At the direction  of the project officer,
 quantitation was carried out only for
 saturated  hydrocarbons. Based on the
 GC/MS behavior of Ci5 - Cia hydrocar-
 bon standards, the relative molar re-
 sponses (RMRs) of these compounds
 were shown to be very similar for m/z
 71, 99, and 113 (these ions are present
 in all saturated  hydrocarbons).  Thus
 average RMRs for each ion could be
 used for the quantitative determination
 of each compound possessing these
 chromatographic ions. Although any of
the three ions would suffice for quanti-
tation, calculations based on all three
 ions were  made, where possible, to
 detect  interference from other  com-
pounds,  and as a check on the overall
 method. To  calculate amounts of the
 individual  compounds, determinations
based on all three ions were averaged. If
one determination was clearly deviant
(i.e., if it differed by more than a factor of
2 from the other values), it was not used

-------
 in the calculation of the average. It is
 recognized that the use of average
 RMRs for all hydrocarbons produces
 some uncertainty in final  tabulations,
 and the figures obtained are best viewed
 as approximate. The paucity of material
 for filters 0014 and 0065, in  part because
 of the lower volume of air samples
 (approximately 850 m3 as compared to
 about 1,050 m3 for filter 0012), allowed
 for the quantitation of only five  com-
 pounds.
  For purposes of comparison, concen-
 trations of selected aliphatic hydrocar-
 bons were obtained from two sources,
 one representing results of "clean" air
 and  the other representing results
 obtained from urban air. Average hydro-
 carbon concentrations (for  the alkanes
 chosen) from Houston aerosols are
 comparable to other  urban levels, and
 are significantly higher than values
 obtained for  "clean" air.  It must be
 noted that the Houston data presented
 here are somewhat suspect because of
the background contamination problem,
the limited number of filters examined,
and the method  of estimation utilized.
 More  rigorous methods of quantitation
 are clearly desirable.
Conclusions and
Recommendations
  Overall results indicate that the
organic component of Houston ambient
air aerosol  is predominantly saturated
hydrocarbons, with lesser but significant
amounts of alkyl aromatics and lower
molecular weight (Ci - C?) halogenated
hydrocarbons. The levels of these mate-
rials are comparable to those found in
other urban atmospheres and are signi-
ficantly higher than  those found in
nonurban "clean" air. No clear relation-
ship was observed  between levels or
number of components found, and time
or location of sampling site.
  The  results of this study indicate
several areas where further investiga-
tions could provide useful results.
Specific recommendations include:
  1) Modification of the IITRI sampler.
     Leakage through fittings and valves
     was evident from cartridge analy-
     sis; redesign should include vacu-
     um tight  valves and fittings. The
     ease of handling of the sampler
     could be improved by construction
     of an all-glass or glass/stainless
     steel unit.
2) Further  research into the use of
   the dichotomous sampler for col-
   lection and subsequent analysis of
   aerosol  organics. The automated
   nature of this sampler, its high
   collection efficiency, sizing capa-
   bility, and use of standard Teflon®
   filters for collection are attractive
   features that could well be taken
   advantage of for aerosol organics
   analysis. Research should  be di-
   rected toward improving analytical
   sensitivity to a point that is com-
   patible with the amount of aerosol
   collected by the dichotomous sam-
   pler.
3) Investigation  of the source  and
   nature of glass or quartz fiber filter
   (GFF and OFF, respectively) back-
   ground contamination. The devel-
   opment  of suitable clean-up-pro-
   cedures and a determination of
   the precise nature of the contami-
   nants associated with such filters
   would allow for greater confidence
   in analytical results obtained from
   samples collected via GFF or OFF.
4) Development of methodology for
   accurate nonmethane hydrocarbon
   determinations. Software devel-
   opment (GC/MS) could provide for
   the group quantitation of paraffinic
   hydrocarbons  as  obtained  from
   the routine analysis of GFF extracts
   by  glass  capillary GC/MS/COMP.
   Such data could thus be available
   in  a reliable and relatively rapid
   fashion as an adjunct to compre-
   hensive  qualitative results from
   ambient  air  aerosol analysis.
This Project Summary was authored by Catherine H. Skintik of WAPORA, Inc.,
  Cincinnati, OH 45233.
Kenneth Krost is the EPA Project Officer (see below).
The complete  report, entitled "GC/MS Analysis of Ambient Aerosols in the
  Houston, Texas, Area," was authored by Charles M. Sparacino of Research
  Triangle Institute. Research Triangle Park,  NC 27709.
The above report (Order No. PB 81-126 377; Cost: $20.00. subject to change) will
  be available  only from:
        National Technical Information Service
        5285  Port Royal Road
        Springfield. VA 22161
        Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
        Environmental Sciences Research Laboratory
        U.S. Environmental Protection Agency
        Research Triangle Park, NC 27711
                                  8
                                                                                i US GOVERNMENT PRINTING OFFICE 1981-757-012/7083

-------
3J
m
C
3
TJ O
(B  =*

3  2?
03  O
3   s£
3   ?i
>
3>
(0



(/>
0>


«>

CO
4
ry.


n
> m


                    00
            m > TJ m Ti
            •DtQ  3 3  »
       CO O O O -ne
       CO < ^.3 Q)

       01   Ms
               3   ^

-------