United States
Environmental Protection
Agency
Environmental Monitoring Systems
Laboratory
Las Vegas NV 89114
Research and Development
EPA/600/S4-85/078 Jan. 1986
Project Summary
GC/FT-IR and GC/FT-IR/MS
Techniques for Routine
Environmental Analysis
Peter R. Griffiths, Charles L. Wilkins, and Donald F. Gurka
This report documents progress on
three tasks related to the design and
testing of procedures and techniques
for analyzing volatile and semivolatile
components of environmental samples.
The tasks include (1) develop and test a
procedure to use infrared molar absorp-
tivities and internal standards for the
routine quantification of environmental
contaminants, (2) prepare and test
computer software to use for the on-
the-fly analysis (both qualitative and
quantitative) of mixtures of volatile
components by direct-linked gas chrom-
atography/Fourier transform infrared/
mass spectrometry, and (3) develop,
test, and construct a high-sensitivity
gaschromatography/Fouriertransform
infrared system, and retrofit the Fourier
transform infrared spectrometer at the
Environmental Protection Agency's
Environmental Monitoring Systems
Laboratory in Las Vegas, Nevada.
In response to Task 1, a method to
estimate the quantity of each compo-
nent separated by gas chromatography
based on the results of a spectral search
program is described, and an approach
to improve the quality of the infrared
vapor-phase spectral data base is sug-
gested.
In response to Task 2, an interface
between a gas chromatograph, a Fourier
transform infrared spectrometer, and a
Fourier transform mass spectrometer is
described. An algorithm to improve the
certainty of identifying materials
through both their infrared and mass
spectra is discussed. A quadrupole mass
spectrometer has been purchased and
tested, and a gas chromatograph/
Fourier transform infrared/mass spec-
trometer system based on this instru-
ment is being constructed.
In response to Task 3, methods of
improving the sensitivity of the interface
between a gas chromatograph and a
Fourier transform infrared spectrometer
have been investigated. A method of
decreasing the nonlinearity of the de-
tector response introduced by unmodu-
lated emission from the hot lightpipe
through cooled apertures is described.
Techniques to simultaneously increase
the sensitivity of gas chromatography/
Fourier transform infrared measure-
ments and to decrease the nonlinearity
of detector/amplifier response through
the use of optimal optics and small focal
area detectors are discussed.
This Project Summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Las Vegas, NV. 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
Currently the U.S. Environmental Pro-
tection Agency (EPA) screens the gas
chromatographicable portion of sample
extracts for a few hundred target organic
compounds. Since over sixty thousand
manufactured chemicals are currently
regulated under theToxic Substance Con-
trol Act, it is apparent that many regulated
compounds are not identified (let alone
determined) in environmental samples.
The present protocol for the analysis of
-------
volatile and semivolatile compounds by
the EPA involves a separation by gas
chromatography (GC) and measurement
of the mass spectrum of each separated
component. Mass spectrometry (MS) is a
fast, sensitive, instrumental analytical
technique, and the linkage of gas chroma-
tography and mass spectrometry (GC/MS)
represents a good first step towards the
total characterization of environmental
samples. Nevertheless, GC/MS does
have several drawbacks.
First, the mass spectra of many isomeric
compounds are very similar, yet the
toxicity and/or carcinogenicity of the
individual isomers may be very different.
For certain compounds, it may be difficult
to produce a discernable molecular ion
when ionization is initiated by electron
impact (El), and, even when chemical
ionization (Cl) methods are used, the
molecular ion (M+) of certain compounds
may be present at a level considerably
below the intensity of other ions in the
mass spectrum. Under these circum-
stances, a priori interpretation of the
mass spectrum may be difficult if not
impossible. Finally, even a semiquantita-
tive determination of all peaks in the gas
chromatograms of complex environmen-
tal samples is usually impossible without
an unambiguous identification of each
peak component and the availability of
calibration standards for each component.
The application of Fourier transform
infrared spectrometry (FT-IR) in place of
mass spectrometry, as well as the ap-
plication of FT-IR in addition to mass
spectrometry for the identification of
components separated by gas chroma-
tography (GC/FT-IR and GC/FT-IR/MS,
respectively) has been proposed as an
alternative technique to GC/MS. Al-
though GC/FT-IR measurements may
never have superior sensitivity to the
corresponding GC/MS measurements,
the capability of infrared spectrometry to
distinguish between isomers and the
reproducibility of GC/FT-IR spectra from
one instrument to another should en-
hance the potential for unambiguous
structural assignments of components of
complex mixtures separated by GC. Be-
cause of the greater reproducibility of
infrared spectra, relative to GC/MS
spectra (reproducibility requires precise
quadrupole tuning), GC/ FT-IR also has
the potential to improve the capability of
obtaining estimates of the quantity of
each component eluting from the chro-
matographic column, whether or not this
molecule has been unequivocally identi-
fied or has merely been assigned to a
certain chemical class.
Results and Discussion
Quantitation Without
Identification (Task No. 1)
This task involves the development of a
technique for obtaining an estimate of the
quantity of each GC eluate from its
GC/FT-IR spectrum. An initial approach
based on the assumption of a constant
molar absorptivity was eventually aban-
doned in favor of a technique that appears
to have the potential of giving a far
superior result. Through an internal
search of EPALIB, it was recognized that
the best matches to the spectrum of a
selected probe molecule often had similar
structures. The molar absorptivities of
these compounds also were often found
to be quite similar. For a given peak in the
chromatogram, a good match will be
obtained if the reference spectrum of that
compound is in the data base. In this case,
the amount of sample passing through
the lightpipe can be estimated from the
quantitative information given in the
EPALIB header and the lightpipe dimen-
sions and carrier-gas flow rate.
If the "unknown" is not represented in
the data base, a spectral search will lead
to the closest analogs to the correct
molecule appearing in the search output.
In fact, even if a reference spectrum of
this sample is in the data base, it can
often be difficult to unequivocally assign
its chemical structure because of the
effect of spectral baseline noise, coelution
peaks, or the presence in the data base of
reference spectra of compounds with a
similar chemical structure. We have,
therefore, developed a method for esti-
mating sample quantity based on the hit
quality index (HQI) of a spectral search.
The quantitative information listed in the
header of each reference spectrum in
EPALIB is weighted by the reciprocal of
the HQI to derive a quantitative estimate
of the amount of sample present in the
lightpipe. The HQI is a measure of the
difference between the spectrum of the
unknown and of each reference spectrum
in the data base. The procedure we are
developing takes the quantitative value
afforded by each of the top search hits
and weights them by the reciprocal of the
HQI.
For some compounds, the results of
this approach were excellent. For exam-
ple, the value of sample quantity calcu-
lated for phenol from the five closest
matches was only in error by 3 percent
from the value listed in the EPALIB
header for phenol. Other compounds,
especially those with exceptionally high
peak absorptivities, gave poorer results.
Nevertheless, for the great majority of
compounds tested, the error was less
than a factor of two. In the second year of
this project, further investigations of this
technique will be made. This approach is
most promising for the nontarget com-
pound analytical situation in which the
unknown analyte's spectrum is rarely in
the search library.
The GC/FT-IR/MS Interface
(Task No. 2)
A linked analytical technique compris-
ing capillary GC/FT-IR/MS can provide
chemical information about mixtures of
volatile and semivolatile compounds to
aid in component elucidation. The infrared
and mass spectral data generated are
complementary and can be combined to
give a less ambiguous chemical identifi-
cation. In the initial phase of this work a
Fourier transform mass spectrometer
(FT-MS) was used to measure the mass
spectra. At the end of the first year a
quadrupole mass spectrometer (Hewlett-
Packard Mass Selective Detector) was
installed.
There are several advantages to using
an FT/MS in this system. Besides the
high-split ratio, which allows almost all of
the GC effluent to pass to the less
sensitive FT-IR, several types of mass
spectral data can be obtained during one
chromatographic run. Due in part to the
rapid data acquisition of the FT-MS,
electron impact (El) and chemical ioniza-
tion (Cl) mass spectra can be acquired
alternately. To prevent Cl reagent gas
(methane) from interfering in El acqui-
sition, the gas is pulsed into the FT-MS
only during the Cl experiment. To demon-
strate that the interface is amenable to
the study of complex mixtures, a pepper-
mint oil was analyzed. This sample
contains potentially difficult compounds
to identify. This oil, however, has been
characterized by a chromatographic sup-
plier and can be considered as a "known"
sample. Peppermint oil contains isomers
of cyclic hydrocarbons as well as isomers
of cyclic ketones. The concentration dy-
namic range was also a concern, because
95 percent of the sample is represented
by only a few components.
The spectral library search results from
the GC/FT-IR/MS experiment are very
informative in themselves (Table 1). With
both sets of data, a very powerful identifi-
cation tool is possible. One may ask,
however, whether one set of results
should be weighted against the other. An
algorithm to accept or reject search-result
-------
Table 1. FT-IR Library Search Results for First Peak of Peppermint Oil Sample
EPALIB'
Number
2244
3303
2433
3013
2925
3304'
327
3305
121
457
HOI
1318
1319
1338
1339
1352
1353
1355
1356
1360
1363
Identification
Bicyclo/3.1 .1 /Hept-2-ene. 2.6,6-(C,0 H^f
Alpha Pinene (C\0Hts)
Pinene, 2/10/-. (C,0H-,e)
Bicyclo/3.1. 1/Hept-2-ene-2-ethano fCn H,&
Ethane ICzHe)
Beta-Pinene (C,oH,e)
Cyclohexane, 1 ,4-Dimethyl-. (CaHtei
Beta-Pinene (C,o HM)
Cyclohexane. Cis-1, 3 -Dimethyl-, (CaH-,ei
2-Butene, 2,3 -Dimethyl-, (CeH12)
CAS" Number
80-56-8
80-56-8
18172-67-3
m ) 128-50-7
74-84-0
127-91-3
589-90-2
127-91-3
638-04-0
563-79-1
^Entries with an EPALIB number greater than 3300 have been added to the data base at the
University of California, Riverside. The similarity of the HQI values given by the first two hits (both
a-pinene but measured at different locations) is noteworthy.
"Chemical Abstracts Service.
"Elemental composition.
combinations has recently been pub-
lished. The algorithm utilizes accurate
mass measurement (AMM) FT-MS results
to establish the most accurate molecular
formula possible. The mass error for such
a determination was typically less than
10 ppm at mass 250. An error of less than
10 ppm has been found to lead to
unambiguous determinations of molecu-
lar formula.
Forty-five model compounds of various
group-types and polarities were used at
concentrations that might typically be
found in a thick-film capillary GC separa-
tion. Infrared and mass spectral search
results were incorporated so that if the
calculated molecular formula is not rep-
resented by any of the first five MS search
results, the FT-MS search results are not
used. If none of the first five matches of
the IR search results coincide with the
determined molecular formula, then the
compound is considered unidentified. No
compounds were incorrectly identified,
such that the MS sampling conditions
appear to be adequate. Table 2 summar-
izes the performance of the various
algorithm combinations possible with the
data collected. The accurate mass FT-
IR/MS variation shows no incorrect
identification and more correct identifica-
tions than FT-IR/MS without accurate
mass data.
Part I: Single-Beam Studies
We have taken two approaches towards
optimizing single-beam GC/FT-IR meas-
urements. In the first, ways of modifying a
standard commercial GC/FT-IR system
have been investigated, while in the
second, a completely new GC/FT-IR
system is being designed and built.
Optimizing the GC/FT-IR
Interface (Task No. 3)
Standard System Modifications
Methods are needed to reduce the
signal loss that takes place at the detector
when the lightpipe is heated. This de-
crease in signal leads to an increase in
the baseline noise level in the ratio-
recorded spectra and a consequent re-
duction in signal-to-noise (SNR). Gurka,
Laska, and Titus (J. Chromatogr. Sci.
1982,20:145) have shown that this effect,
for temperatures ranging from ambient to
Table 2. Comparison of Various AMM Algorithms for the 45 "Unknown" Compounds
Number Identified
Algorithm
FT-IR
FT-IR/AMM
FT-MS
FT-MS/AMM
FT-IR/MS
AMM, FT-IR/MS
Correct
32
33
26
30
32
35
Incorrect
13
8
19
12
0
0
Eliminated
0
4
0
3
13
10
250°C, leads to a threefold loss in sensi-
tivity. A similar profile was observed in
our laboratory for measurement using a
Nicolet 60-SX GC-FT-IR spectrometer.
The best explanation of this effect is that
the detector (and/or preamplifier) become
saturated and are driven into a nonlinear
response by the unmodulated infrared
radiation emitted by the hot lightpipe.
This heat is focused on the detector along
with the modulated mid-IR radiation,
causing the response of the detector to
the modulated signal (the interferogram)
to decrease.
We found that if a short length of
lightpipe, the internal diameter (i.d.) of
which was slightly bigger than the i.d. of
the lightpipe, was located after the exit
aperture of the lightpipe (detector end),
an immediate increase in interferogram
signal resulted for the Nicolet 60-SX
GC/FT-IR system. The use of such an
aperture, even cooled by water, made no
difference at all for the older 7199 system.
We also noted that when a water-cooled
cone was placed over the end of the short
lightpipe, a signal increase resulted even
at ambient temperature. The increase in
signal appears to result from a combina-
tion of the lightpipe acting as a cold
shield, thus preventing unmodulated
radiation from the end of the GC/FT-IR
lightpipe from reaching the detector, and
the cone collecting modulated radiation
transmitted by the GC/FT-IR lightpipe but
scattered by the heat-shield lightpipe,
then passing it to the detector.
Polished aluminum cones were then
fabricated to determine an optimum cone
angle. The reflective cones were collect-
ing the component of the modulated
infrared signal that had been scattered by
the lightpipes. A 20-percent increase in
ambient signal was achieved for the 60-
SX GC/FT-IR interface and as much as 38
percent increase was observed for the
7199 system with a lightpipe of same i.d.
(1 mm) and similar length. We noted that
positioning the cone near the lightpipe
still resulted in a signal loss upon heating.
A short lightpipe extension was used to
move the cone further from the end of the
hot lightpipe.
Design of a New Single-Beam
GC/FT-IR System
To determine the optimum for GC/FT-
IR lightpipes, we found it necessary to
design and build an interface capable of
supporting lightpipes of various dimen-
sions while keeping set-up time to a
minimum. An investigation of the opti-
mum optical configuration for collecting
-------
the beam emerging from the lightpipe is
being made, using lightpipes held in this
variable-length oven. The goal of this
study is to determine the combination of
optics and detector that will result in the
highest possible interferogram SNR being
measured for a given lightpipe. In the
experiment we are performing, a KBr lens
and a detector are mounted upon a track.
A lens follows the equation:
1 /f = 1 /x + 1 /y
(DJ
where f is the focal length of the lens and
x and y are distances of the object
d, = d0y/x
(2)
where d, is the diameter of the image at
the detector and d0 is the diameter of the
object (in this case, the internal diameter
of the lightpipe). By varying the position of
the lens and detector along the track, the
tradeoffs between the solid angle of
radiation collected versus detector size
[and hence its noise-equivalent-power
(NEP)] may be determined. The results of
this experiment are very much dependent
upon the lightpipe employed but should
be completed in about 2 months time
(September 1985). Shortly thereafter, the
EMSL-LV GC/FT-IR system will be retro-
fitted to realize the increased sensitivity
resulting from this optimized single-beam
system.
Dual-Beam GC/FT-IR
The complete design of a dual-beam
system is necessarily dependent upon
the results of the investigation into
single-beam GC/FT-IR systems. For this
reason, much of the design is yet to be
completed. One area in which we have
been involved during the first year of this
project is the construction and evaluation
of a dual-element detector. By utilizing a
detector with dual elements housed
within the same dewar, we believe that
nonlinear conditions (observed at high
levels of radiation flux incident upon the
large single detector used in a dual-beam
FT-IR system) may be avoided by dividing
the energy between two elements rather
than concentrating it on one. These ele-
ments must be closely matched to opti-
mize the signal-addition step, which is
carried out electronically before the re-
sultant interferogram is digitized.
Conclusions and
Recommendations
Taskl
a. Measurements of the absorbance of
the most intense peak in the spectrum
or that of certain functional groups
does not permit the quantity of a
given component present in the light-
pipe to be determined accurately.
b. A method based on a weighted aver-
age of the quantities of the samples
given as the top few matches by
spectral searching routines appears
to have great potential for allowing
the quantity of an unidentified peak
component to be estimated.
c. A major limitation to this method
appears to be the quantitative accu-
racy of the EPA library of vapor-phase
infrared spectra (EPALIB).
d. We recommend that efforts to im-
prove the quality and size of GC/FT-
IR reference data bases be renewed,
possibly through the initiation of a
collection of reference spectra sub-
mitted to a central computer by
current GC/FT-IR users. This is an
economic approach to increasing the
data base because reference spectra
would be donated.
Task 2
a. A parallel interface between a gas
chromatograph, an FT-IR spectrom-
eter, and a Fourier transform mass
spectrometer was constructed and
was applied to the identification of
the components of several complex
mixtures.
b. Accurate mass measurements have
been used to establish the molecular
formula of GC eluates and when
combined with infrared spectral
search data, accurate mass measure-
ments appear to present a powerful
method for compound identification.
c. A low-cost quadrupole mass spec-
trometer has been installed and will
be interfaced first to a commercial
GC/FT-IR interface and then to an
optimized GC/FT-IR interface being
constructed under Task 3.
Task 3
a. Methods of reducing the effect of
detector nonlinearity because of un-
modulated radiation emitted by the
lightpipe are being developed.
b. The use of a cooled lightpipe in
conjunction with a cone to collect
scattered radiation has been shown
to improve the sensitivity of a com-
mercial GC/FT-IR system.
c. An optical configuration has been
designed to reduce the size of the
infrared detector below the size of
detectors now being used in com-
mercial GC/FT-IR systems. This will
result in an increased system sensi-
tivity.
In conclusion, prospects are high for
the construction of a directly linked
GC/FT-IR/MS system in which the FT-IR
sensitivity is over an order of magnitude
more sensitive than are currently mar-
keted FT-IR systems. Computer software
is being prepared to aid in evaluating the
large quantities of analytical data that will
be generated by such a system. Tech-
niques are being developed in an attempt
to semiquantitate by FT-IR without com-
plete analyte identification, and to quanti-
tate FT-IR identified compounds without
quantitation calibration curves by means
of infrared absorption coefficients. These
quantitation techniques cannot be achiev-
ed by conventional GC/MS and should
result in reduced analytical costs because
of reduced spectrometer data acquisition
time. The optimized GC/FT-IR/MS sys-
tem should provide a much larger amount
of analytical information than that ob-
tainable from GC/MS used alone.
U. S. GOVERNMENT PRINTING OFFICE:1986/646 116/20810
-------
Peter R. Griffiths and Charles L Wilkins are with University of California-
Riverside. Riverside. CA 92521; and the EPA author. Donald F. Gurlta fa/so the
EPA Project Officer, see below), is with Environmental Monitoring Systems
Laboratory, Las Vegas, NV 89114.
The complete report, entitled "GC/FT-IR and GC/FT-IR/MS Techniques for
Routine Environmental Analysis," (Order No. PB 86-128 584; Cost: $9.95,
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 Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas. NV 89114
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
-
------- |