United States
Environmental Protection
Agency
Environmental Monitoring and
Support Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S4-88/030 Sept. 1988
x°/EPA Project Summary
Capillary Column GC-MS
Determination of 77 Purgeable
Organic Compounds in Two
Simulated Liquid Wastes
M. F. Yancey, R.A. Kornfeld, and J. S. Warner
The suitability of purge-trap-desorb
(PTD) procedures for determination of 84
volatile organic compounds with
capillary column gas chromatography
(GC) and mass spectrometry (MS) was
evaluated. After collecting GC-MS data
not previously available for some
analytes, 7 of the 84 compounds were
eliminated from further consideration
because of poor purging efficiency or
analyte stability problems.
For each of the remaining 77 com-
pounds, the linear concentration range
and detection limit were determined with
data obtained by PTD GC-MS analysis of
spiked reagent water. A relative standard
deviation (RSD) of <25% for the average
response factor (RF) was chosen as the
acceptance criterion for determining the
linear range. This criterion was met over
a concentration range of at least two
orders of magnitude for 56 of the 77
analytes, 1.5 orders of magnitude for 12
analytes, and 1 order of magnitude for
6 analytes. The criterion was not met for
acetone, trichlorofluoromethane, and
2-chloro-1,3-butadiene.
Method performance was assessed by
analyzing eight replicate aliquots of each
of two simulated liquid waste samples (a
municipal sewage sludge leachate and
reagent water containing fulvic acid)
containing analytes spiked at two con-
centrations. For > 80% of the analytes,
bias of measured concentrations was
<30%. For most other analytes ac-
curacy was > +30%. The observed high
positive bias was attributed to enhanced
sensitivity caused by high concentra-
tions of ions in the MS source. Calibra-
tion data showed that short term (daily)
and long term (two weeks) precision was
very good.
This Project Summary was developed
by EPA's Environmental Monitoring and
Support Laboratory, Cincinnati, OH, 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 Resource Conservation and
Recovery Act (RCRA) specifies over 200
toxic organic compounds (in Appendix IX
to 40 CFR, Parts 264 and 270) to be used
to screen for suspected ground water con-
tamination at land-based hazardous waste
treatment, storage, and disposal facilities
(Federal Register 52, July 9, 1987, pp.
25942-25953). Analytical methods for most
of these analytes are published in SW-846,
"Test Methods for Evaluating Solid Wastes,
Physical/Chemical Methods," Third Ed.,
November, 1986).
The SW-846 method recommended for
determining volatile, relatively water insolu-
ble, organic analytes is Method 8240, which
involves purge-trap-desorb (PTD) analyte
extraction followed by packed column GC
separation and MS detection and measure-
ment. Advances in GC column technology
now permit determination of a wider range
of compounds in a shorter time with greater
sensitivity using a fused silica or glass
capillary column. In this study, a 0.75 mm
i.d. glass capillary column was used to
-------
evaluate Method 8240 procedures for 84
analytes. The compounds considered for
inclusion in this study include all USEPA
Method 524.2 analytes and all compounds
on the Appendix IX list (Federal Register,
57, 26639, July 24, 1986) that might be
amenable to determination by room
temperature PTD extraction followed by
GC-MS analysis using a 0.75 mm i.d. glass
capillary column.
Experimental
PTD-GC-MS Analyses
Analyses were performed with Method
8240 procedures using a Tekmar Model
LSC-2 PTD system, a Carlo Erba Model
4160 GC, a Finnigan Model 3200 MS fitted
with a glass jet separator, and an Incos data
system with Revision 5.5 software. The PTD
system was fitted with a 5-mL fritted glass
purge tube and a 305 mm x 4 mm i.d.
stainless steel trap containing 10 mm of 3%
SP-2100 on Supelcoport, 77 mm of Tenax,
77 mm of silica gel, and 77 mm of coconut
charcoal. The system was operated with a
helium purge for 11 min at 26 mL/min at
room temperature (23-25°C), desorption for
4 min at 15 mL/min at 180°C, and a trap
bake for 7 min at 26 mL/min at 180°C. The
GC was fitted with a 60 m x 0.75 mm i.d.
Supelco VOCOL column coated with a 1.5
urn film and operated with helium carrier
gas flow of 15 mL/min. The column
temperature was maintained at 10°C dur-
ing the desorb cycle, programmed to 200°C
at 10°C/min at the end of the desorb cycle,
and maintained at 200°C for 10 min. The
MS was tuned daily to meet bromofluoro-
benzene (BFB) criteria daily and was
operated with a scan time of 1 sec over a
mass range of 35-325 amu. The emission
current was selected to achieve acceptable
tuning and to stay within the emission cur-
rent range recommended by the manufac-
turer. For maximum dynamic range, the
electron multiplier voltage was set to per-
mit analytes to be detected without satura-
tion of the multiplier at concentrations up
to four times the internal standard (IS) con-
centration of 50 ng/L specified by Method
8240.
The system met all daily performance
criteria specified by Method 8240. In addi-
tion to BFB tuning criteria, these criteria in-
clude (1) minimum RF of 0.30 for each of
the five system performance check com-
pounds (chlorobenzene, chloromethane,
1,1-dichloroethane, 1,1,2,2-tetrachloro-
ethane, and tribromomethane); (2) RF dif-
ference of <25°/o for the six calibration
check compounds (chloroform, 1,1-dichlor-
oethene, 1,2-dichloropropane, ethyl-
benzene, toluene, and vinyl chloride); (3) IS
retention time changes of < 30 sec; and (4)
IS peak area changes of <50°/o.
Method Range Studies
In method range studies, 5-mL aliquots
of reagent water were spiked with com-
posite spiking solutions to achieve 13 con-
centrations ranging from 0.1 to 550 Ig/L for
most analytes, but 15 analytes expected to
be poorly purged were spiked at 10-fold
higher concentrations. Eight replicate
samples were analyzed for each of the 13
spike levels.
A reverse library search of the data was
performed using a project-specific mass
spectral library containing the retention time
and quantitation ion of each analyte and IS.
The quantitation ion was chosen for max-
imum sensitivity while attempting to avoid
interferences from coeluting materials. For
the majority of analytes, the quantitation ion
selected was the base peak. For Method
8240 analytes, the primary ion specified in
Method 8240 was used as the quantitation
ion. When the quantitation ion was detected
above the background, an RF was calcu-
lated for each analyte using the area
responses and concentrations of the
analyte and the appropriate IS.
To determine an estimated detection limit
(EDL) for each analyte, a trained analyst in-
spected the mass spectrum from one of the
replicate samples at the lowest concentra-
tion at which the computer detected the
quantitation ion in at least four replicates.
The analyst examined extracted ion current
profiles of 2-5 major ions, including the
quantitation ion, selected from the
reference mass spectrum. The analyte was
considered to be present if the major ions
comaximized and had relative intensities
within 20% of those in the reference mass
spectrum (as specified in Method 8240),
and if the quantitation ion gave an area
response greater than 1000 or a signal-to-
noise ratio of at least 3:1. If, in the opinion
of the analyst, the mass spectrum indicated
the presence of the analyte in question, that
concentration was considered the
estimated detection limit. If the mass spec-
trum did not indicate the presence of the
analyte, the inspection process was
repeated at the next higher concentration.
Mean RFs and RSDs of measured RFs
were calculated at each concentration as
the first step in determining the linear range
of the method for a given analyte. The high
concentration data were evaluated for
system saturation by plotting and visually
evaluating the RF as a function of analyte
concentration. When an RF obviously
decreased with increasing concentration,
appropriate concentrations were eliminatec
from the linear range. For each analyte ar
overall average RF was calculated usinc
RFs from all concentrations other than re
jected high concentrations. If the RSD foi
an overall average RF was > 25% (an ac
ceptable threshold value selected with
USEPA personnel concurrence), the con
centration range was narrowed in ar
attempt to achieve <25% RSDA. A concen
tration range was, however, never reduce<
to less than one order of magnitude.
The lowest concentration at which the
analyte was identified and measured in a
least four of eight replicates was considerec
the EDL. Data obtained at the EDL were
used to calculate the method detectior
limits (MDLs) using the USEPA procedure
described in Appendix B to 40 CFR Par
136 (Federal Register 49 198, October 26
1984).
Matrix Validation Studies
Two simulated liquid waste samples wer<
prepared for further method evaluation
One sample was a municipal sludge
leachate prepared using a modification o
the USEPA toxicity characteristic leachim
procedure (Federal Register 57, 21685
June 13, 1986). The other sample was ar
artificial ground water prepared by spikm<
reagent water with fulvic acid (Suwannei
Stream Reference, U.S. Geological Survey
International Humic Substance Society) a
a concentration of 1 mg/L.
Eight replicates at each of two analyti
spike concentrations (20 and 200 ng/L fo
most analytes; 200 and 2000 /ig/L for thi
poorly purged analytes) and eight unspike<
replicates were analyzed for each of the twi
samples. Calibration standards prepared b;
spiking reagent water with each analyte a
a concentration of 50 ^.g/L for most analyte!
and 500 /ig/L for the poorly purged analyte:
were analyzed at the beginning, middle
and end of each day. Measured analyti
concentrations were calculated using dai
ly average RFs. Precision (RSD) and ac
curacy (bias) of measured concentration:
were calculated for each analyte at eacl
spiking concentration in each sample.
Results and Discussion
Method Range Studies
For three compounds (acetone, 2-chlorc
1,3-butadiene, and trichlorofluoromethane]
RSDs of measured RFs were > 25%, evei
with a concentration range of one order c
magnitude. Acetone might yield mon
reliable data using m/z 58 as the quantita
tion ion rather than m/z 43, which i
specified by Method 8240. A decreasing Rl
-------
with increasing concentration was evident
for 2-chloro-1,3-butadiene, which is known
to polymerize readily. The degree of poly-
merization, which would result in loss the
monomer, would be expected to be higher
at higher concentrations and could account
for lower observed RFs at higher concen-
trations. Trichlorofluoromethane was par-
ticularly sensitive to the effects of methanol
and water on GC peak shape. Other early
eluting compounds such as dichlorodi-
fluoromethane (20% RSD), chloromethane
(23% RSD), and vinyl chloride (20% RSD)
also produced average RFs that were less
precise than those of most other analytes.
The range for a fourth compound (hex-
achloropropene) that produced an average
RF with >25% RSD was not narrowed
because the greatest deviation from the
average RF occurred at 300 /tg/L, near the
middle of the concentration range.
Of the 74 analytes having a satisfactory
method range and average RF precision,
the linear range for 56 was at least two
orders of magnitude and for 12 others was
at least 1.5 orders of magnitude. For the re-
maining six analytes, the linear range was
only one order of magnitude. Three of those
analytes (dichlorodifluoromethane, chloro-
methane, and chloroethane) were highly
volatile and two (trans-1,4-dichloro-2-butene
and hexachloropropene) were poorly
purged.
The linear range, EDL, and MDL for each
analyte are given in Table 1. For all but 10
analytes, the EDL (the lowest concentration
at which the analyte was detected and
quantified in at least four of eight replicates)
was the same as the lowest concentration
in the linear range. Lower EDLs could un-
doubtedly have been achieved for most
analytes if MS operating conditions had
been selected to achieve maximum sen-
sitivity instead of a wide dynamic range and
measurement of high anatyte concentra-
tions.
For all analytes except acetone,
calculated MDLs were considerably lower
than EDLs. For all but seven analytes each
calculated MDL was even lower, usually by
a factor of two to five, than a concentration
at which the analyte could be detected ex-
perimentally. In all cases, the highest con-
centration at which an analyte was not
detected in any of eight replicates (Table 1)
was within a factor of three of the
associated EDL. Low calculated MDLs
reflect excellent measurement precision at
the EDL rather than excessively high
signal-to-noise ratios. The data indicate that
calculated MDLs may be misleading.
Matrix Validation Studies
The 74 analytes studied included 29 of
the 30 compounds listed in Method 8240
Table 6, which specifies acceptance criteria
for data obtained from analysis of a quality
control check sample. For 28 of those 29
compounds, Method 8240 acceptance
criteria were achieved in both matrices at
both high and low concentrations. The one
exception was ethylbenzene spiked at the
high concentration into reagent water con-
taining fulvic acid; a bias of +77% was
observed while +62% is acceptable.
The acceptability of measured concen-
trations for all 74 analytes was evaluated by
selecting a bias of +30% as an acceptance
limit. (This limit is much more stringent that
Method 8240 analyte-specific criteria,
which are generally +50% or greater.) With
a +30% bias limit, measured concentra-
tions were acceptable for 61 of the 74
analytes spiked into the POTW sludge
leachate at the high concentration and for
63 analytes at the low concentration. Ac-
ceptable concentrations were measured for
50 analytes added to the fulvic acid spiked
water at the high concentration and for 70
at the low concentration.
In nearly 90% of the cases in which the
bias of measured concentrations was
>30%, the bias was positive rather than
negative. A possible explanation of the high
positive biases is an increased MS sen-
sitivity when ion concentrations are
unusually high. This effect would be ex-
pected to be much more noticeable when
a capillary column is used rather than a
• packed column, because a capillary col-
umn produces much sharper peaks and
higher momentary analyte concentrations
than a packed column. The high positive
bias was more prevalent at the high spike
concentration than at the low spike concen-
tration, especially for the fulvic acid spiked
water. The high spike concentration of 200
i^g/L provides 1000 ng of analyte in the 5
ml_ of sample purged. The increased sen-
sitivity at high concentrations was not as
apparent in the method range study as in
the matrix validation study, possibly
because the ion source had been cleaned
immediately before beginning the method
range study. Decreased source cleanliness
may enhance the effect.
For all but two of the cases in which the
bias was > -30%, the low spike concen-
tration was involved and the analyte was
one of the 14 poorly purged analytes spi-
ked at a 10-fold higher concentration than
other analytes. For those analytes, the
calibration standard provided 2500 ng,
which could have produced a high ion con-
centration and high RF; that could account
for a negative bias at the low spike concen-
tration. Thus, an increased sensitivity
resulting from high ion concentrations can
account for essentially all biases, high and
low. No evidence for a matrix effect, which
would be expected to give a negative bias,
was observed.
Conclusions and
Recommendations
The following conclusions and recom-
mendations are based on the results of this
study:
• The use of methanol as a solvent in-
terferes with the chromatographic per-
formance of a nonpolar capillary col-
umn for the determination of polar
volatile compounds such as acetoni-
trile, isobutyl alcohol, and propargyl
alcohol.
• Methanol and water desorbed from a
trap containing Tenax, silica gel, and
charcoal, interfere with the chromato-
graphic performance of a nonpolar
capillary column for the determination
of gaseous and very low boiling non-
polar compounds by a PTD procedure.
• Of the 84 volatile compounds studied,
74 can be determined satisfactorily by
SW-846 Method 8240 using a VOCOL
capillary column.
• With MS conditions that permit Method
8240 performance criteria to be met
using a capillary column and 250 ng
of IS, an increased sensitivity may be
observed at high ion concentrations in
the mass spectrometer source. The ef-
fect of MS source tuning parameters
and cleanliness on changes in RF with
concentration should be evaluated.
• Calculated MDLs en be considerably
lower than the lowest concentrations at
which analytes can be detected
experimentally.
• Cryofocusing or other means to focus
early eluting compounds to minimize
peak broading and improve quantia-
tion, especially at low concentrations,
should be investigated.
• A non-volatile, water soluble solvent
should be used for spiking solutions to
avoid deleterious chromatographic ef-
fects of methanol on early eluting
analytes.
• Differences between calculated and
observed detection limits should be in-
vestigated to establish a protocol for
obtaining more meaningful MDLs.
-------
TABLE 1. Linear Range and Detection Limits Obtained From Method Range Study
Applicable Experimentally
Linear Range, Determined
Analyte ^g/L EDL, ^g/La
1. Acetone
2. Acrolein
3. Acrylonitrile
4. Ally! chloride
5. Benzene
6. Bis-(2-chloroethyl) ether
7. Bromobenzene
8. Bromodichloromethane
9. Bromomethane
10. 2-Butanone
11. n-Butylbenzene
12. sec-Butylbenzene
13. tert-Butylbenzene
14. Carbon disulfide
15. Carbon tetrachlonde
16. Chlorobenzene
17. 2-Chloro-1 ,3-butadiene
18. Chlorodibromomethane
19. Chloroethane
20. 2-Chloroethyl ethyl ether
21. Chloroform
22. 1-Chlorohexane
23. Chloromethane
24. 2-Chlorotoluene
25. 4-Chlorotoluene
26. 1 ,2-Dibromo-3-chloropropane
27. 1 ,2-Dibromoethane
28. Dibromomethane
29. 1 ,2-Dichlorobenzene
30. 1,3-Dichlorobenzene
31 . 1 ,4-Dichlorobenzene
32. trans-1 ,4-Dichloro-2-butene
33. Dichlorodifluoromethane
34. 1,1-Dichloroethane
35. 1,2-Dichloroethane
36. 1,1-Dichloroethene
37. cis-1 ,2-Dichloroethene
38. trans-1 ,2-Dichloroethene
39. Dichloromethane
40. 1 ,2-Dichloropropane
41 . 1 ,3-Dichloropropane
42. 1,1-Dichloropropene
43. cis-1 ,3-Dichloropropene
44. trans-1 ,3-Dichloropropene
45. 1 ,2-Dimethylbenzene
46. 1 ,4-Dimethylbenzene
47. Ethyl methacrylate
48. Ethylbenzene
49. Hexachlorobutadiene
50. Hexachloroethane
51. Hexachloropropene
52. 2-Hexanone
53. lodomethane
54. Isopropylbenzene
55. p-lsopropyltoluene
56. Methacrylonitrile
57. Methyl methacrylate
58. 4-Methyl-2-pentanone
59. Naphthalene
60. Propionitrile
61. n-Propylbenzene
62. Styrene
63. 1,1,1,2-Tetrachloroethane
64. 1,1,2,2-Tetrachloroethane
65. Tetrachloroethene
5500-170
5500-100
5500-170
550-10
550-3.0
5500-55
170-3.0
550-3.0
550-5.5
5500-100
170-1.0
170-1.0
170-1.0
550-3 0
550-3.0
300-3.0
550-5.5
550-3.0
550-30
5500-30
550-3.0
550-3.0
550-30
550-3 0
550-3.0
550-10
550-3.0
550-3 0
300-3 0
300-3 0
300-3.0
550-170
55-5.5
550-5 5
550-5.5
550-5.5
550-5.5
550-5.5
550-5 5
550-5.5
550-3 0
550-5.5
550-5.5
550-5.5
550-1.0
550-10
1000-10
550-1.0
300-3.0
550-5.5
3000-170
5500-30
550-5.5
300-3.0
300-10
5500-30
3000-30
1700-30
300-3.0
5500-170
170-1.0
300-3.0
300-3.0
550-5.5
300-3.0
170
55
170
5.5
3.0
55
3.0
3.0
5.5
30
1.0
1.0
1.0
30
30
30
5.5
3.0
17
30
3.0
3.0
17
30
30
10
3.0
30
30
30
30
55
55
5.5
55
5.5
5.5
55
5.5
5.5
30
55
55
5.5
1.0
1.0
10
1.0
3.0
5.5
170
30
5.5
3.0
10
30
30
30
3.0
170
1.0
3.0
3.0
5.5
3.0
Calculated Nondetection
MDL, tg/L" Limit, ^g/Lc
200
10
40
2
0.2
10
0.7
02
2
10
0.4
0.4
0.8
0.3
0.2
0.3
2
0.2
5
9
01
0.1
10
0.3
03
2
0.2
0.2
0.5
0.4
0.5
30
1
0.7
0.4
1
1
0.7
0.8
0.7
0.3
0.5
0.9
1
0.2
0.6
7
04
0.7
2
50
6
2
0.4
0.7
9
5
6
0.5
40
0,2
0.4
0.2
1
0.2
100
30
100
3.0
1 0
30
1.0
1.0
3.0
10
0.3
0.3
0.3
1.0
1.0
1.0
3.0
1.0
10
10
1.0
1 0
10
1 0
1.0
5.5
1.0
1 0
1 0
1 0
1.0
30
30
3.0
3.0
3.0
3.0
30
30
3.0
1 0
30
30
3.0
03
0.3
30
0.3
1.0
3.0
100
17
3.0
0.3
5.5
17
3.0
17
1.0
100
0.3
1 0
1.0
3.0
1.0
-------
TABLE 1.
Continued
Analyte
66. Toluene
67. Tribromomethane
68. 1 ,2,4-Trichlorobenzene
69. 1,1,1-Trichloroethane
70. 1.1,2-Trichloroethane
71. Trichloroethene
72. Trichlorofluoromethane
73. 1,2,3-Trichloropropane
74. 1,2,4-Trimethylbenzene
75. 1 ,3,5-Trimethylbenzene
76. Vinyl acetate
77. Vinyl chloride
Applicable
Linear Range,
iig/L
550-3.0
300-3.0
300-3.0
550-3.0
550-5.5
550-3.0
300-30
550-17
300-3.0
300-3.0
3000-55
550-5.5
Experimentally
Determined
EDL, ng/L"
3.0
3.0
3.0
3.0
5.5
3.0
30
17
3.0
3.0
55
5.5
Calculated Nondetection
MDL, ng/L" Limit, pg/Lc
0.2
0.3
0.3
0.2
0.3
0.1
10
2
1
0.3
4
0.6
0.3
1.0
1.0
1.0
3.0
0.3
17
10
1.0
1.0
30
3.0
"Experimentally determined estimated detection limit.
"Calculated method detection limit.
cNondetection limit is the highest concentration studied at which the analyte was not detected.
M. F. Yancey, R. A. Kornfeld, andJ. S. Warner are withBattelle—Columbus Division.
Columbus, OH 43201-2693.
Thomas Pressley is the EPA Project Officer (see below).
The complete report, entitled "Capillary Column GC-MS Determination of 77
Purgeable Organic Compounds in Two Simulated Liquid Wastes," (Order No.
PB 88-245 881/AS; Cost: $14.95, 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 Monitoring and Support Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
U
oa;ncs i-
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
POSTAGE & FEES
PERMIT No. G-3
Official Business
Penalty for Private Use $300
EPA/600/S4-88/030
OOOC329 PS
------- |