EPA-600/4-75-007
September 1975
Environmental Monitoring Series
ANALYTICAL QUALITY ASSURANCE
FOR TRACE ORGANICS ANALYSIS
BY GAS CHROMATOGRAPHY/
MASS SPECTROMETRY
Environmental Monitoring and Support Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
EPA-600/4-75-007
September 1975
ANALYTICAL QUALITY ASSURANCE
FOR TRACE ORGANICS ANALYSIS
BY GAS CHROMATOGRAPHY/
MASS SPECTROMETRY
by
James W. Eichelberger
William M. Middleton
and William L. Budde
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
ENVIRONMENTAL MONITORING AND SUPPORT LABORATORY
CINCINNATI, OHIO 45268
-------
This document is available to the public through the National Technical Infor-
mation Service, Springfield, Virginia 22151.
ii
-------
REVIEW NOTICE
This report has been reviewed by the Environmental Monitoring and
Support Laboratory, U.S. Environmental Protection Agency, and approved for
publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
ill
-------
INTRODUCTION
The importance of analytical quality assurance
in trace organic pollutant analysis cannot be
overestimated. Data generated in surveys are
being used to set standards for drinking water,
surface water quality, and effluents. Possible
correlations between the presence of organic
contaminants in drinking water and human health
effects are under widespread study. In the past
many carefully conducted measurements were not
documented with sufficient data to support their
reliability. This caused doubt about the validity
of the measurements and concern for the cor-
rectness of correlations and proposed standards.
The purpose of this report is to describe the
application of analytical quality assurance (AQA)
concepts to the qualitative analysis of water
samples for trace organics. The concentration,
isolation, and identification procedures used in
this work were liquid-liquid extraction and gas
chromatography — mass spectrometry (GC/MS).
However, some of the AQA techniques that are
described also have applicability in other methods
of trace organic analysis including: the entrain-
ment of volatiles in an inert gas stream followed
by trapping and GC/MS (1); and carbon or
resin adsorption, extraction, and GC/MS (2).
The data used to illustrate the AQA were
obtained from five drinking water samples taken
during January and February 1975. The samples
were collected from Miami, Florida, Seattle,
Washington, Philadelphia, Pennsylvania, Cincin-
nati, Ohio, and Ottumwa, Iowa. The results of
these analyses are a part of a larger survey of
drinking water supplies that was conducted by
several U.S. Environmental Protection Agency
(EPA) laboratories during early 1975. Some
EPA facilities applied different methodologies of
isolation and concentration of the organic con-
taminants, but GC/MS was always applied for
identification of individual pollutants. The dif-
ferent methodologies are effective with different
classes of pollutants, but there is some overlap
between classes, which serves as an excellent
verification of certain results.
The overall philosophy of the survey was to
analyze for all organic compounds present in the
samples. This is in sharp contrast to many
previous studies where the approach was to
"analyze for" specific materials of interest. Within
this context, the emphasis of the survey was
qualitative, i.e., the identification of individual
organic compounds in the water. Precise meas-
urement of concentration was not a goal of the
survey. A comprehensive report on the results
from all methodologies was prepared (3).
EXPERIMENTAL
Glassware
All laboratory glassware was washed with
detergent, rinsed with tap water, rinsed with dis-
tilled water, and air dried. It was then heated
at 400 °C for 1 hour in a muffle furnace. Samples
were collected in 1-gallon glass jugs supplied with
Teflon cap liners. Sample jugs were washed with
detergent, rinsed with tap water, air dried, and
heated to 400 °C for 15 minutes.
Materials
Anhydrous sodium sulfate (Mallinckrodt Ana-
lytical Reagent) was used as received. In one
experiment a batch of sodium sulfate was placed
in a large soxhlet extractor and extracted with 2
liters of hexane/acetone (1:1 v/v) for 24 hours.
The sodium sulfate was dried at 120°C and used
in normal sample preparation and in the blanks.
No reduction in the number or size of the con-
taminant peaks was observed. It was concluded
that no significant amount of contamination was
contributed by the sodium sulfate. Acetone,
methylene chloride, and diethyl ether were Bur-
dick & Jackson "distilled in glass" and were used
as received.
Instrumentation
Mass spectra were measured with a Finnigan
1015 quadrupole mass spectrometer. The inlet
system was a Varian Series 1400 gas chromato-
graph that was interfaced to the.spectrometer by
an all-glass jet-type enrichment device and an
all-glass transfer line. Control of the quadrupole
rod mass set voltages, data acquisition, data
-------
reduction, and data output was accomplished
with a System Industries data system that
employed a Digital Equipment Corporation
PDP-8/E mini-computer and a 1.6-million-word
Diablo disk drive. Data were displayed on a
Tektronix 4010 cathode ray tube (CRT) or a
Houston plotter.
The GC column used in this study was a 6-ft
(2 mm ID) coiled glass tube packed with Supel-
coport (80/100 mesh) coated with 1.5% OV-17
and 1.95% QF-1. The initial column tempera-
ture of 60 °C was held for 1.5 minutes, then the
temperature was programmed at 8°C per minute
to a final temperature of 220°C, which was held
for 15 minutes. The total run time was approxi-
mately 35 minutes. Conditions that were held
constant throughout the analyses were: helium
carrier gas at a flow rate of about 30 ml per
minute; temperature of the GC injection port at
190°C; the temperatures of the interface and
transfer line at 210°C; spectrometer manifold
temperature at 100°C; pressure in the MS of 10~5
torr; ionizing voltage of 70 eV; a filament current
of 500/namp; electron multiplier at 3000 volts;
mass range scanned from 33-450 amu at an
integration time of 8 msec/amu; and sensitivity
at 10~T amps/volt.
SCOPE AND LIMITATION OF THE METHOD
The method used applied to all organic com-
pounds present that are extracted partially or
completely into the methylene chloride — diethyl
ether solvent. All compounds originally present
in water at a concentration of approximately
10 ng per liter (0.01 ppb) or greater that elute
from the GC column without decomposition
within 35 minutes will be observed. Very volatile
compounds, e.g., chloroform, vinyl chloride, etc.,
will not be observed as they are either lost during
extract concentration or masked during solvent
elution from the GC. For example, compounds
that are observed include the following: aliphatic
hydrocarbons — Ca and larger; aromatic hydro-
carbons — benzene derivatives, biphenyls, alkyl
benzenes, polynuclears, etc.; pesticides — chlori-
nated, organophosphorus, some carbamates; phe-
nols of all types; PCB's; plasticicers — phthalates,
adipates, and sebacates; and various other types
of compounds including sulfur compounds,
amines, alcohols, aldehydes, ketbnes, and some
carboxylic acids.
PROCEDURE
1. After measuring the pH of the gallon sample
(generally pH = 6-8), 3 liters were transferred to
a 6-liter separatory funnel. Fifty milliliters of
ethyl ether were added, and the mixture was
shaken for 1 minute. The sample was then
extracted three times with 75-ml portions of
methylene chloride, and the extracts were com-
bined in a 300-ml Erlenmeyer flask. The pur-
pose of the ethyl ether was to improve the
extraction efficiency of the more polar com-
pounds like phenols and acids.
2. The combined extract was poured through
2 inches of anhydrous sodium sulfate in a 19-mm
ID glass column. The dried extract was collected
in a 500-ml Kuderna-Danish (K-D) flask fitted
with a 10-ml ampul graduated in 0.1-ml incre-
ments. As an added precaution, the anhydrous
sodium sulfate was prerinsed with 100 ml
methylene chloride to remove soluble impurities.
For each water sample, a parallel experiment
was conducted with all quantities of materials
and procedures exactly the same as the sample
except that no water or aqueous solution was
employed. This reagent blank was initiated in
this step by pouring 250 ml of methylene chloride
— ethyl ether (4:1 v/v) through 2 inches of
anhydrous sodium sulfate in a separate 19-mm
ID glass column.
3. After the combined extract or blank had
filtered through the sodium sulfate, the sodium
sulfate was rinsed with 50 ml of acetone. This
was done for two reasons: to rinse any residual
sample components from the sodium sulfate, and
to introduce a nonchlorinated solvent into the
sample for GC/MS injection.
4. The pH of the water layer was then adjusted
to 2.0 using concentrated HC1 and steps 1, 2,
and 3 were repeated. In step 1, it was not neces-
sary to add the ethyl ether a second time.
5. When the second extraction was completed,
the pH of the water layer was adjusted to 12.0
using a saturated NaOH solution. Again steps
1, 2, and 3 were repeated ignoring the addition
of ethyl ether. The three sample extracts were
now contained in three K-D flasks: the neutral
compounds extracted from a solution of approxi-
mately pH 7, the acid compounds extracted from
a solution of pH 2, and the basic compounds
extracted from a solution of pH 12. The reagent
-------
blank was in a separate K- D flask.
6. A Snyder column was fitted to each K-D
flask, and the extracts were concentrated on a
steam bath to approximately 5 ml. After con-
centration the methylene chloride (bp = 39.8°C)
was largely removed and the sample was con-
tained in acetone (bp = 56.1°C). The acetone
was used because several microliters of methylene
chloride will cause an excessive increase in the
pressure in the mass spectrometer and cause
automatic shut down of the system. Up to 8 /j.1
of acetone will not cause this undesirable situa-
tion as it is removed more efficiently by the en-
richment device.
7. The extracts were further concentrated in the
ampul in a warm water bath under a stream of
clean, dry nitrogen to 100 ^1 with repeated rins-
ings of the inside of the ampul.
8. Five-microliter injections were made into the
GC/MS, with data acquisition started after a
delay of 90 seconds.
RESULTS AND DISCUSSION
Sampling information and results of the water
analyses are given in Table I. Table II is a sum-
mary of the application of AQA techniques to the
analyses reported in Table I.
Most of the concentration values in Table I
are estimates that were based on conservative
extraction efficiencies and average response fac-
tors. These estimates are probably accurate to
within a factor of ten. In a few cases authentic
samples were available and extraction efficiencies
and response factors were determined. This per-
mitted better estimates of concentration, and these
results are probably accurate to better than
±50%. In no case was a precise concentra-
tion measurement attempted by development of
a calibration curve with several standards and
careful measurement of integrated instrument
signals. High precision measurements were be-
yond the scope of this survey.
Analytical Quality Assurance
The AQA that applies to qualitative organic
trace analysis may be conveniently divided into
four categories:
(a) reagent and glassware control;
(b) instrumentation control;
(c) supporting experiments; and
(d) data evaluation.
Reagent and glassware control is required to
minimize the introduction of contamination from
the materials used in the liquid-liquid extraction
procedure. Glassware cleaning procedures have
been developed and they are effective. High-
quality commercial reagents and solvents are
available, but quality is still somewhat variable
and usually unpredictable. In solvents that are
used for extractions, impurities are amplified by
about a factor of 2000 during extract concen-
tration. Clearly, if background contaminants that
are introduced from reagents or solvents seriously
obscure compounds in the sample, purification of
these materials is required.
Instrumentation control is required to ensure
that the total operating instrumentation system
is calibrated and in proper working order. If
a computerized GC/MS system is used to collect
data, the computer data system must be included
in the performance evaluation. The recommended
instrumentation control procedure employs a
standard reference compound and a set of refer-
ence criteria to evaluate the performance of the
overall system (4). This evaluation should be
performed on each day the GC/MS system is
used to acquire data from samples or reagent
blanks. The records from the performance
evaluations should be maintained with the sample
and reagent blank records as permanent docu-
mentation supporting the validity of the data.
The reagent blank is a supporting experiment
required for all samples. This is true even when
contamination from glassware and reagents is
well controlled. The reagent blank result is the
documentation that proves that good control
was exercised, and it defines precisely the level
of background that was beyond control.
The reagent blank evaluation may be a straight-
forward comparison of corresponding peaks and
mass spectra in the reagent blank and sample.
A more rigorous procedure is required to make
objective judgments in situations that are not
obvious. An effective technique for comparing
blanks and samples employs the extracted ion
current profile (EICP) of one or several ions.
An EICP is denned as a plot of the change in
relative abundance of one or several ions as a
function of time (5). The data for this plot are
extracted from all the ion abundance measure-
ments made over the mass range observed dur-
ing the eluu'on of the separated components from
the GC. The EICP produces an apparent increase
in sensitivity by subtracting from the total ion
current profile all the ion abundance data con-
tributed from background, unresolved compo-
nents, and other irrelevant ions. The EICP
-------
TABLE I. RESULTS OF ANALYSES USING LIQUID-LIQUID EXTRACTION AND GC/MS
Location of
sampling
Miami, Fla.
Miami-Dade Water
& Sewer Authority
Seattle, Wash.
Seattle Water Dept.
Philadelphia, Pa.
Philadelphia Water
Dept.
Cincinnati, Ohio
Cincinnati Water
Works
Ottumwa, Iowa
Ottumwa Water Works
Date
collected
1/20/75
1/27/75
2/3/75
2/11/75
2/17/75
Date
received
1/29/75
2/5/75
2/12/75
2/18/75
2/26/75
Date
extracted
1/31/75
2/10/75
2/18/75
2/19/75
2/28/75
Compounds
identified
bromoform
hexachloroethane
di-n-octyl adipate
nicotine
none
1,2-bis (2-chloroethoxy)-
ethane
dibromochloromethane
isophorone
trimethyl isocyanurate
benzole acid
phenylacetic acid
Approximate
concentration*
/xg/1 (ppb)
0.2
0.07
20.
3.
0.03
0.05
0.02
0.02
15.
4.
^Concentrations are estimated as accurate to within a factor of ten; with d-«-octyl adipate, nicotine, and ben-
zoic acid, authentic samples were available and the concentrations of these are probably accurate to within ±50%.
TABLE II. ANALYTICAL QUALITY ASSURANCE IN THE IDENTIFICATION OF
ORGANICS
Compounds
identified
bromoform
hexachloroethane
di-n-octyl adipate
nicotine
l,2-bis(2-chloro-
ethoxy) ethane
dibromochloro-
methane
isophorone
trimethyl
isocyanurate
benzoic acid
phenylacetic acid
In
blank
EICP
no
no
no
no
no
no
no
no
no
no
Spectrum*
matched
NIH
NIH
NIH&
standard
NIH&
standard
NIH
NIH
NIH
NIH
NIH&
standard
NIH
GCt
retention
time
very short
short
matched
standard
matched
standard
not
applicable
very short
not
applicable
not
applicable
matched
standard
not
applicable
Extracted Molecular
fraction ion observed
neutral
neutral
acid same
as standard
base same
as standard
neutral
neutral
neutral
neutral
acid
acid
yes
no
no
yes
no
yes
yes
yes
yes
yes
M-f- 1 ion isotope t
accuracy
Calcd % Found %
Bra pattern
CL> pattem§
not applicable
10.8 12
not applicable
BrzCl pattern
9.7 9.1
6.5 7.2
7.6 7.6
8.6 7.3
Spectrum
checked for
consistent
major
fragments
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
*The computerized mass spectrum matching system used was developed by the National Institutes of Health (NIH)
and EPA.
tPrecise measurements of retention times were not made because of the difficulty of reproducing the initial tem-
perature conditions exactly.
tin non-halogen containing compounds, the M-j-1 ion abundance is expressed as a percentage of the molecular ion
abundance in the calculated and experimental values.
§The CU pattern was observed in the M-C1 ion.
-------
generator is a standard data reduction program
on all modern computerized GC/MS systems. A
fast graphics display device is required to facili-
tate reviewing a large number of EICP plots.
It is emphasized that it is not necessary to have
even a tentative identification of a compound to
apply this technique to reagent blank evaluation.
To conduct an EICP comparison, the mass
spectra of all peaks in the sample are examined.
One or several ions that are prominent in a
spectrum from each peak are selected, and the
sample and reagent blank EICP's are generated
on the CRT. Comparison of these permits, in
most cases, straightforward judgments concern-
ing the presence of compounds in the sample and
the reagent blank. In Figure 1 is the EICP for
mass 171 from a sample and the corresponding
reagent blank. Clearly there is a compound
having a mass 171 ion in the sample, but there
is no corresponding peak above the noise level
in the reagent blank.
If a corresponding peak is observed in the
reagent blank and its concentration, as judged
from the peak height, is approximately the same
as or exceeds the sample concentration, the deci-
sion is clear and the compound must not be
reported. A far more difficult judgment must
be made when the concentration of a component
in the sample exceeds its concentration in the
reagent blank. The material in the sample could,
of course, be a true sample component. Alterna-
tively, it has been observed empirically that com-
pounds in the blank sometimes merely appear
to be at lower concentrations than the same com-
pounds in the corresponding sample. In Figure
2, total ion current profiles for a sample and a
corresponding reagent blank are shown. Careful
comparison of the profiles reveals a very similar
pattern of peaks and valleys in certain areas (e.g.,
spectrum numbers 170-190 and 235-245) yet a
significantly lower apparent concentration in the
reagent blank. There are several possible reasons
for this. One rationalization is that impurities
in the pure solvent are adsorbed more efficiently
onto the drying agent and other surfaces, which
creates a purifying effect. With extracts contain-
ing some water, the wetting effect of the water
precludes efficient adsorption on surfaces and the
impurities are carried on in the solvent. Alter-
natively, certain solvent impurities may be lost
more readily from the blank than from the
sample extract during the concentration step. This
may be caused by the general organic back-
ground matrix in the sample extract which acts
as a keeper for the solvent impurities. Both
explanations are reasonable but unproved. In
view of the uncertainties, any compound that is
observed in the sample should not be reported if
it is part of an overall pattern of peaks that is
repeated in the blank, although at a lower ap-
parent concentration.
Chemical ionization, field ionization, and high-
accuracy mass measurements are GC/MS tech-
niques capable of generating very strong evidence
in support of identifications. However, the pro-
duction of this evidence is restricted because
only a relatively few laboratories have developed
capabilities with these techniques. High-accuracy
mass measurements are further limited by sample
size, since some sacrifice in sensitivity is required
to achieve the high accuracy.
After a tentative identification is made, by
either interpretation or empirical spectrum match-
ing, several other types of supporting experiments
become possible. The retention time data from
the GC/MS of a pure compound (standard) may
be compared with analogous data from the sam-
ple component. Similarly the mass spectrum of
the standard, obtained under the same conditions
that were used for the sample, may be compared
with the spectrum of sample component. The
standard may be dissolved in water at an ap-
propriate concentration, extracted, and measured.
The recovery of this spike in the same fraction
that the suspected component appeared in and
the observation of the same mass spectrum as the
sample component gave is a strong confirmation
of the correctness of the identification.
The evaluation of the data must weigh the
available evidence in terms of its reliability and
determine the cost and benefits to be gained by
gathering additional information.
Clearly the most convincing evidence for an
identification is obtained by examining pure
standards that correspond to suspected sample
components. However, the existence of this evi-
dence is constrained by the availability of the
pure standard and the additional cost and time
required to examine it. Because it is not usually
possible to predict which compounds will be
found, some standards will not be available
immediately. There are many very practical
limitations imposed on the development and main-
tenance of a large library of pure authentic stand-
ards. Many compounds are obtainable from
fine chemical supply houses, but procurement time
is variable and may extend to weeks or months.
Some compounds are not available from any
supplier, frequently because they are by-products
or metabolites of industrial processes rather
than manufactured products. This same limita-
-------
8
5L
fe.
8_
MASS 171 IN REAGENT BLANK
MASS 171 IN SAMPLE
18 20 30
SPECTJU1
10 SO 60 70 90 30
100 HO 120 130
110 ISO 160 170 ISO
1302932102202303*3350360270280230300318330330
Figure 1. The extracted ion current profiles of mass 171 from the Cincinnati neutral fraction and the corresponding reagent blank.
-------
REAGENT BLANK
28 30 10
SflECIRf1fO«R
SB 6fl 70 88 90
100 110 120 130 110 ISO 160 170 180 ISO 200 210 238 230 210 250 260 278
Figure 2. The total ion current profiles for the Philadelphia neutral fraction and the corresponding reagent blank.
-------
tion of standard availability also precludes careful
concentration measurements in many cases.
Because of the problem of standard availability,
it is worthwhile to determine whether conditions
could exist that would lead to reliable identifica-
tions without standards. One criteria for a reliable
identification that might be used is a quantitative
measure of the goodness of match between an
experimental mass spectrum and a spectrum from
the printed literature or a computer-readable data
base. A similarity index (SI), calculated on a
scale from zero to one, has been described (6) and
used in one computerized mass spectrum match-
ing system (7). Experience with this SI indicates
that, in general, a value greater than about four
tenths corresponds to the existence of a reason-
able match between two mass spectra.
A reasonable match often does imply an
identification, but sometimes it does not. It is
well known that position isomers and members
of homologous series of compounds often give
very similar mass spectra. There is another unde-
termined number of compounds that are simply
not uniquely characterized by their mass spectra.
Figure 3 is the mass spectrum of an unidentified
compound and Figure 4 shows the spectrum of
the compound chloropicrin, C1:1CNO.>. The match
is clearly good by inspection, and a SI of
0.453 was calculated. Nevertheless, the com-
pound whose spectrum is in Figure 3 is not
chloropicrin as determined by the gas chromato-
graphic behavior of the unknown and pure
chloropicrin.
Another problem with identifications based on
empirical spectrum matching is that significant
differences in ion abundance measurements are
sometimes observed when the mass spectrum
of a compound is measured on two different
spectrometers. Most of these differences are
probably caused by non-uniform calibration pro-
cedures or a failure to use an ion abundance
calibration procedure. In addition, it is well
known that different types of inlet systems may
have significant effects on relative abundance
measurements. With a GC of batch inlet system
that is generally operated in the 100-250°C
temperature range, temperature-dependent frag-
mentations are promoted with frequent reduc-
tions in the abundances of molecular and other
higher mass ions. With a well-designed direct
inlet system, these temperature effects may be
largely precluded. As a result of these factors,
it is quite common for two spectra of the same
compound, measured with different inlet sys-
tems or spectrometers, to give a rather low SI.
A low SI may also be caused by unresolved or
partly resolved components which generate mass
spectra containing extraneous ion abundance
measurements.
It is concluded that the SI must be used with
caution. A relatively high SI may be regarded
as an indication of a reasonable match, but only
as suggestive of the probability of an identifica-
tion. A relatively low SI cannot be regarded as
complete rejection.
Another criterion for reliable identifications
when standards are not available is based on an
assessment of the quality of the ions in the experi-
mental and reference mass spectra. In the SI
calculation (6), molecular ions (M+), molecular
ions having naturally occurring isotopes (typically
M+l), and all key fragment ions are weighted
the same as many very common fragment ions.
However, the M+, for example, is unique in
every mass spectrum and has significance to an
identification that far outweighs most other ions.
Mass spectra may be categorized according to
the quality of the ions observed, and a quality
index (QI) can be calculated that is a weighting
factor for the SI. Several categories of quality are:
(a) The highest quality spectrum is one in
which the molecular ion is observed and
the observed distribution of abundances for
it and its isotope-containing species is within
10% of the expected distribution. For this
spectrum, the QI is 1.0 and the full SI
value is retained with considerably enhanced
significance.
(b) If a molecular ion is observed but the
isotope data are not within 10% of the
expected value, lower confidence is assigned
by a QI of 0.75.
(c) Failure to observe a molecular ion but
observation of key fragments that account
for all the atoms of the molecular ion sug-
gests a QI of 0.5. This index may be raised
or lowered between 0.4 and 0.6 depending
on the observation of consistent isotope data
in the key fragment ions.
(d) Finally, the lowest confidence is placed
on spectra which do not contain adequate
fragment ions to account for all the atoms
of the molecular ion. A QI of 0.1 is as-
signed to these spectra.
It is recognized that position isomers may not
be distinguishable under any circumstances, but
this is often true even when pure standards are
available.
The QI is amenable to additional positive
adjustments by 0.1-0.2 QI units if all major
fragment ions are scrutinized and found to be
reasonable and compatible with the assigned
-------
SPECTRUM NUMBER 161 - 159
g
8.
CINCINNATI NEUTRflL
R-
o
i*
4JJ
20 30 40 SO
M/ E
•i" i I— • |- I— | i
60 70 80 90
100 110 120 130 140 150 160 170 180
Figure 3. The mass spectrum of a compound found in the Cincinnati neutral fraction.
-------
g
OLOROPICRIN STfNHD
8.
8.
jJO
flcD-
feS.
h
e8-
8.
e>
0
1
y,
i
i
Illlllllllilllllllllllllllllll 1
NimlilmiiiiMiiiiiiiMiilimlmiimimlimliiimimiiiii
LO
j
.8
1
:
-LO
- E)
20 30 10 SO
M/ E
60 70 80 90 100 110 120 130
ISO 160 170 180 190
Figure 4. The mass spectrum of the compound chloropicrin, C1.{CNO2.
-------
structure. Reasonableness should be based on
compatibility with the accepted principles of
fragmentation of organic ions in the gas phase.
With magnetic deflection spectrometers, addi-
tional fractional quality points may be added if
fragmentations are supported by the observation
of ions from the decomposition of metastable
ions.
Good spectrum matches that have a QI of
0.75-1.0 are considered reliable identifications
when pure standards are not available.
Application of Analytical Quality Assurance
Concepts
These AQA concepts were applied to the
analyses reported in Table I. Table II summarizes
the results of the tests. Bromoform, nicotine,
dibromochloromethane, isophorone, trimethyl
isocyanurate, benzoic acid, and phenylacetic acid
were not found by EICP in the corresponding
reagent blanks and each gave spectra that were
good matches by inspection with spectra in the
National Institutes of Health data base (7). In
each spectrum the molecular ion was observed
and the M++1 isotope accuracy or halogen iso-
tope abundance distribution pattern was within
the expected experimental error (4). In addition
each compound was extracted in the expected
fraction, the most volatile of the set gave short
or very short retention times, and all the frag-
ments in the mass spectra were reasonable and
consistent with the assigned structure. On the
basis of this evidence, these seven identifications
were considered firm without recourse to authen-
tic standards. Standards of two of the compounds
were readily available, and these were used to
supply additional supporting evidence.
Hexachloroethane gave a very good spectrum
match by inspection and an expected short
retention time; also, it was extracted in the
appropriate fraction. The molecular ion was
not observed, but ions were observed with
halogen isotope distribution patterns that cor-
responded to the C2C15, C,C14, C2C1.<, C,C12,
and CCl.t ions. Therefore, a consistent set of
fragment ions was observed and these account
for all the atoms of the proposed structure. The
only other reasonable possibility for this peak
was pentachloroethane, and the recorded spec-
trum of this in the NIH data base did not exhibit
a C,HC1S molecular ion.
The compound di-n-octyl adipate was tenta-
tively identified by the empirical matching pro-
cedure. However, no molecular ion was observed.
and the complexity of the fragmentation pattern
precluded a rapid determination of its consistency
with the proposed structure. This is a clear
example of a spectrum that contains inadequate
information to permit an accurate identification
without a pure reference standard. The com-
pound was found in the acid fraction, and this
also aroused suspicion about its identity since
dioctyl adipate might be expected to appear in
the neutral fraction.
A sample of octyl decyl adipate containing
di-w-decyl and di-«-octyl-impurities was avail-
able in the laboratory. This was dosed into
laboratory tap water and extracted according to
the method. All the adipates were found in the
acid fraction as was the adipate that was found in
the Miami water sample. The retention time
for dioctyl adipate was within experimental error
of the retention time of the compound in the
Miami water, and the observed spectrum was the
same as the sample spectrum. This evidence
strongly supports the identification as being that
of an authentic contaminant. Its origin may be
from vinyl plastic garden hose and similar mate-
rials that are in widespread use.
The compound l,2-bis(2-chloroethoxy)ethane
(C,,H,.,O,C1,) gave a spectrum that was in ex-
cellent agreement with the spectrum in the data
base. Again, no molecular ion was observed and
a reference standard was not available. How-
ever, the fragment ions at masses 63, 65 (C.,H4C1),
93, 95 (C,HHOC1), and 107, 109 (QH.OCl) ac-
count for all the atoms in the compound and are
consistent with the assigned structure. Two addi-
tional ions at masses 137 and 139 correspond
to the loss of a CH,-C1 group from the molecular
ion. The overall evidence strongly supports the
identification.
Seven other peaks were detected, but no com-
pounds are reported because inadequate evidence
was available to permit reliable identifications.
Three of these appeared to be the same com-
pound, and the available information about them
is an excellent example of the application of
the AQA concepts. Peaks were found at the
same retention time near the detection limit of
10 ng per liter (ppt) in the neutral fractions from
Miami, Cincinnati, and Philadelphia. The three
mass spectra were essentially the same except
for variations in the abundance of the common
background ion at mass 43 (C;H7) and several
other weak ions. The spectrum of the compound
in the Cincinnati fraction is shown in Figure 3.
As previously discussed, this spectrum is a good
match of the spectrum of chloropicrin in Figure
4. However, since the molecular ion was not
observed, interpretation of the match must be
made with caution. The ions at masses 117.
119, and 121 clearly indicate the presence of a
11
-------
CC13 group, and this is supported by the CCL
ions at masses 82, 84, and 86 and the CCl ions
at masses 47 and 49. But there is no ion that
clearly points to an N(X group, and the spectrum
therefore fails to account for all the atoms of
the proposed structure. Under these circum-
stances an authentic standard was required to
obtain additional information. Pure chloropicrin
was shown to elute much earlier than the com-
pounds in the three neutral fractions. Therefore,
although this compound contains a CCl, group,
it remains unidentified at this time. There are
small ions at masses 103 and 145 in all three
spectra; these suggest the saturated oxygenated
hydrocarbon ions C5H1:tOj, and CRH17O2. This
compound appears similar to the l,2-bis(2-chloro-
ethoxy)ethane found in the Philadelphia neutral
fraction: it has a longer chain and more chlorine
and may be an intermediate in the formation of
the ubiquitous chloroform (1). A chemical ioni-
zation mass spectrum may provide a valuable
insight into the identity of this compound.
CONCLUSION
AQA is required in identifying as well as
measuring the concentration of trace organics.
The guidelines based on spectrum similarity and
the quality of the ions found in the measured
mass spectrum are a reasonable basis for eval-
uating the reliability of an identification.
REFERENCES
(1) T. A. Bellar and J. J. Lichtenberg, J. Amer.
Water Works Ass., 66, 739 (1974).
(2) G. A. Junk, H. J. Svec, R. D. Vick, and
M. J. Avery, Environ. Sci. Technol., 8, 1100
(1974).
(3) USEPA, Office of Toxic Substances. Pre-
liminary assessment of suspected carcinogens
in drinking water. Interim Report to Con-
gress, June 1975.
(4) J. W. Eichelberger, L. E. Harris, and W. L.
Budde, Anal. Chem., 47, 995 (1975).
(5) R. A. Hites and K. Biemann, Anal. Chem.,
42, 855 (1970).
(6) H. S. Hertz, R. A. Hites, and K. Biemann,
Anal. Chem., 43, 681 (1971).
(7) S. R. Heller, J. M. McGuire, and W. L.
Budde, Environ. Sci. Technol., 9, 210 (1975).
12
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/4-75-007
3. RECIPIENT'S ACCESSIOWNO.
4. TITLE AND SUBTITLE
ANALYTICAL QUALITY ASSURANCE FOR TRACE ORGANICS
ANALYSIS BY GAS CHROMATOGRAPHY/MASS SPECTROMETRY
5. REPORT DATE
September 1975 ("Issuing Date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
James W. Eichelberger, William M. Middleton, and
William L. Budde
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Monitoring and Support Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
10. PROGRAM ELEMENT NO.
1BA027; ROAP 09ABZ; Task 001
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Same as above
13. TYPE OF REPORT AND PERIOD COVERED
In-house
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Analytical Quality Assurance concepts are applied to the qualitative analysis of
drinking water supplies for trace organics by liquid-liquid extraction and gas
chromatography - mass spectrometry. Some of these concepts are also applicable to
other methods of analysis.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Potable water
Quality assurance
Solvent extraction
Gas chromatography
Chromatographic analysis
Mass spectroscopy
Qualitative analysis
Finished water analysis
Analytical procedures
Reagent blank interpre-
tation
Finished water
Trace organics
Gas chromatography -
mass spectrometrv
13B
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
17
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
13
ftUSGPO: 1975 — 657-695/5307 Region 5-11
------- |