EPA-660/2-73-034
January 1974
                        Environmental Protection Technology Series
     Infrared  Fourier  Transform
     Spectrometry of Gas  Chromatography
     Effluents
                                National Environmental Reaserch Center
                                Office of Research and Development
                                U.S. Environmental Protection Agency
                                Corvallis, Oregon  97330

-------
            RESEARCH REPORTING SERIES
Research reports of the  Office  of  Research  and
Monitoring,  Environmental Protection Agency, have
been grouped into five series.  These  five  broad
categories  were established to facilitate further
development  and  application   of   environmental
technology.   Elimination  of traditional grouping
was  consciously  planned  to  foster   technology
transfer   and  a  maximum  interface  in  related
fields.  The five series are:

   1.  Environmental Health Effects Research
   2.  Environmental Protection Technology
   3.  Ecological Research
   4.  Environmental Monitoring
   5.  Socioeconomic Environmental studies

This report has been assigned to the ENVIRONMENTAL
PROTECTION   TECHNOLOGY   series.    This   series
describes   research   performed  to  develop  and
demonstrate   instrumentation,    equipment    and
methodology  to  repair  or  prevent environmental
degradation from point and  non-point  sources  of
pollution.  This work provides the new or improved
technology  required for the control and treatment
of pollution sources to meet environmental quality
standards.

-------
                                  EPA-660/2-73-034
                                  January 1974
            INFRARED FOURIER TRANSFORM
   SPECTROMETRY OF  GAS CHROMATOGRAPHY EFFLUENTS
                         by

                  Leo V. Azarraga
                   Ann C. McCall
    Southeast Environmental Research Laboratory
               College Station Road
               Athens, Georgia 30601
                    ROAP 16ADN-26
              Program Element  1BA027
      NATIONAL ENVIRONMENTAL RESEARCH CENTER
         OFFICE  OF RESEARCH & DEVELOPMENT
       U.S. ENVIRONMENTAL PROTECTION AGENCY
              CORVALLIS, OREGON  97330
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price $1.10

-------
                        ABSTRACT

A study was made on vapor phase GC/IR analysis with
the Digilab FTS-14D/IR, GC/IR, spectrophotometer in
order to determine its performance.  The chromatographic
equipment consists of a P.E. 990 GC equipped with a 15.2
m SCOT capillary column, a flame ionization detector, a
14:1 splitter and a heated transfer line.

The maximum signal-to-noise ratio of a single scan,
non-ratioed, spectrum from a GC/IR cell with 25% trans-
mission is ea.  250.  The detection limits for aromatic,
aliphatic and other compounds with C=0 functional groups
are from 0.2 to 1 yg.  Usable spectra may be obtained
for samples as small as 2 to 5 yg.  Separate spectra
may be obtained for GC peaks that are resolved by twice
their peak widths.  Effective signal averaging for
trapped samples is limited to 100 scans by trap leakage.
The Every Scan Mode is most useful for programmed
temperature GC/IR runs.  However, a cooled detector,
additional software, and data storage capacity are
needed to apply this mode more effectively for GC/IR
analysis.
                           ii

-------
                         CONTENTS


                                                   Page


Abstract                                            ii


List of Figures                                     iv


Sections


I        Conclusions                                 1


II       Recommendations                             3


III      Introduction                                4


IV       Experimental                                8


V        Results and Discussion                     21


VI       References                                 61
                          • • *
                          111

-------
                         FIGURES

No.                                                 Page

 1    Block Diagram of the GC/IR System               9

 2    Non-Ratioed Single Scan Background Spectrum    22

 3    The Region (2020-2070 cm-1) of Maximum
      Spectral Intensity in Fig. 2 Examined for
      the Maximum Signal-to-Noise Ratio              23

 4    Transmission Baseline Using Equal Number of
      Scans for Background and Sample Between 500
      and 3800 cm-1                                  24

 5    The Noise Level on the Transmission Baseline
      Between 800 and 3500 cm-1                      26

 6    Transmission Baseline Using a 1024-Scan
      Background Spectrum                            27

 7    Time Delay Calibration Curve Showing the
      Inverse Correlation Between Flow Rate and
      the Values of the Time Delay Parameter         31

 8    The Resolution of the Eluates in the GC/IR
      System                                         33

 9    GC and GC/IR Profiles of the Same Peak         34

10    Plot of Absorbance and GC Peakwidth vs.  Flow
      Rate Showing the Change in Absorbance with
      the Corresponding Increase in the Concentra-
      tion of the Eluate in the Carrier Gas          36

11    Plot of Absorbance vs.  Time of a Trapped 5
      pg Naphthalene Eluate                          37

12    Signal to Noise Ratio as a Function of
      /NSS for a Trapped 5 yg Cyclohexanone Eluate   38

13    Plot of Concentration vs. GC Peak Area         42
                          IV

-------
                    FIGURES (con't.)

No.                                                 Page

14    Plot of Absorbance vs. GC Peak Area            47

15    Comparative Display of Reconstructed
      Chromatogram from IR Absorbance Data and the
      Original Chromatogram                          ^5

16    Single Scan Spectra of the Eluates in Fig.
      15 Plotted in the Absorbance Mode Between
      600 and 3800 cm-1                              56
                            v

-------
                        SECTION I
                       CONCLUSIONS

The optimum wave number range  for GC/IR analysis is
between 800 and 3500  cm"1, where the signal-to-noise
ratio does not vary by more than an order of magnitude.
The spectra above  3500 cm   and below  800 cm""1 are 10
to 20 times lower  in  signal-to-noise ratio.  The trans-
mission of the IR  cell in the  low frequency region is
inadequate for spectral analysis of yg quantities of
eluates.  This is  a definite disadvantage since spec-
tral information down to 600 a
qualitative IR identification.
tral information down to 600 cm   is important for
No degradation of GC resolution is observed in the GC/IR
system.  GC peaks that are resolved by twice their half-
width are also resolved in the system.

The absorbance did not increase as rapidly as expected
when the GC peakwidth for a given quantity of eluate was
reduced by increasing the carrier gas flow.  This indi-
cates that the eluate undergoes a considerable dilution
in the GC/IR cell and that the cell's volume is greater
than optimum for the spectral analysis of eluates from
a capillary column.  Effective signal averaging is
limited to ca. 100 scans on a trapped eluate.  The
signal-to-noise ratio deteriorates when signal averag-
ing is carried over a longer period of time because
leakage from the trap continuously attenuates the
concentration of the eluate.  Hence, each succeeding

-------
scan contains a progressively lower and lower signal-
to-noise ratio.

The detection limit of the current system is of the
                           —9
order of tenths of y g or 10   moles.  Reasonably good
spectra of substances are obtained for quantities one
order of magnitude greater than the detection limit.  An
increase of one order of magnitude in sensitivity would
insure the practical GC/IR analysis of organic pollutants
in the ppb level of concentration.

The Every Scan mode  (see last paragraph p. 11) is
potentially the most useful mode of GC/IR analysis.
Spectral scanning of each GC peak need not be synchro-
nized with the signal'from the GC detector.  Consequently
optimum spectral data on the eluates can be obtained
independent of the changes in the carrier gas flow dur-
ing the GC run.  Furthermore, the system may be operated
without the GC splitter to insure the maximum quantity
of eluate in. the GC/IR cell.  Hardware and software
additions to the present system are required to utilize
this mode of spectral analysis efficiently.
                           2

-------
                       SECTION II
                     RECOMMENDATIONS

The detection limit should be improved by at least a
factor of 5 so that spectra that are useful for quali-
tative identification may be obtained for submicrogram
quantities of samples.  A GC/IR cell with half the
volume and twice the optical path of the current one,
coupled with a high sensitivity, cooled  detector to
offset the loss in signal-to-noise ratio by the con-
current decrease in the cell transmission, will probably
suffice.

Additional software and hardware should be provided for
automatic sampling, GC/IR data collection, processing,
and storage to facilitate routine GC/IR analysis and
the collection of vapor phase spectra of known compounds,
The latter may be used with a spectrum or interferogram
correlation software to provide the system with spectral
matching capabilities for unique substance identifica-
tion.

The current GC/IR system should be used for pollutant
identification especially where it is practicable to
concentrate the components in the sample so that the
volume of the sample used for GC injection contains
from 5 to 10 yg of each component.

-------
                       SECTION III
                      INTRODUCTION

Identification of organic compounds in water at
concentrations as low as 1 vg/i. is necessary in enforc-
ing water pollution control legislation and evaluating
the environmental impact of organic pollutants.
Chromatography and mass spectrometry are currently the
most generally used methods for the analysis of those
substances.  The Environmental Protection Agency is
developing a gas chromatography-mass spectrometry
system and a computerized spectral matching library
for organic analysis.

In general, methods of organic compound identification
rely on the empirical correlation between molecular
properties and the observed data.  As such, ambiguity
in substance identification can be expected.  Confirma-
tion by two or more independent analytical methods is
preferable.

The infrared spectrum contains unique structural
information on the molecule.  This information when
combined with those derived from other analytical
methods should reduce to a high degree the uncertain-
ties in the identification of substances.

The. combination of gas Chromatography and infrared
spectroscopy (GC/IR system)  offers considerable poten-
tial for obtaining this desired information.  The infra-
red spectrum of a GC effluent may be obtained in two

-------
ways: one is the on-line measurement of the IR
spectrum of the eluates (GC/IR method) ; the other
involves the trapping or isolation of the eluted sub-
stance, which is then dispersed or dissolved in an
appropriate medium for IR spectral analysis.  The
latter process is tedious and time consuming.  Losses
of substances being isolated are very likely.  It is,
however, quite sensitive; quantities of substances of
the order of 0.1 pg are sufficient for spectral analy-
sis in most cases, and an extensive collection of
standard spectra is available for qualitative identifi-
cation by spectral matching.

On-line measurement of the infrared spectrum allows a
real time spectral analysis of the GC effluent.  In
conventional GC/IR systems, the spectrum obtained is
that of the substance in the gas phase.  The positions
and contours of the absorption bands in such a spectrum
will differ, generally, from those of the same sub-
stance in the condensed phase.  Qualitative identifica-
tion by matching with standard spectra, which are
mostly condensed phase spectra, is not a straight
forward process.

The lack of an extensive collection of standard vapor
phase spectra for qualitative identification purposes
is, however, not a serious disadvantage in GC/IR
analysis.

The simplicity and speed of the technique allow the
spectral measurement in a series of compounds to be

-------
performed without a great deal of difficulty.  Solutions
of known compounds suspected to include the unknown may
be injected into the GC.  The soectrum of each compound
may be recorded under exactly the same conditions used
for the unknown,and spectral matching for qualitative
identification may be performed.  These compounds need
not be rigorously pure so long as the impurities are
separable under the GC conditions used.  If the known
substances are also separated by the GC, a single solu-
tion containing them may be subjected to GC/IR analysis.

The spectrum of each compound can be recorded in a
single GC run with a considerable savings in time and
sample manipulations.  Each spectrum may be stored for
future reference.  Furthermore, matching of the vapor
phase spectra would lead to a more accurate means of
identification since the absorption frequencies and
intensities are free from solvent and matrix effects.
The sensitivity reported for vapor phase GC/IR analysis
is between 10 and 50 yg .  Good quality spectra were
also reported for .001 and .005 ]il quantities of pure
        2          '
liquids.   A recent innovation to the GC/IR technique
allows for the on-line recording of spectra of GC
eluates in solution.   A cholesteric liquid crystal
film in the GC/IR cell fractionates the eluted substances
from the carrier gas and enables the solution spectra
of 50 yg of the eluates to be recorded on-the-fly.

This report deals with vapor phase GC/IR measurements
using the Digilab FTS-14D/IR spectrophotometer with
its GC/IR accessory and a Perkin-Elmer 990 GC.  Its
purposes are to report the experimental conditions and
                          6

-------
instrumental operations for optimum spectral analysis
and to find out what modifications or additions to the
present system may be required to enhance its perfor-
mance and capabilities.

In view of these objectives, only synthetic samples
were used in our work.  Such factors as the signal-to-
noise ratio, sensitivity, optimum determination of
delay times, trap leakage rates, the time dependence
of the density of the eluted compound in the IR cell, and
the optimum use of the various GC/IR modes for analysis
were investigated.

-------
                      SECTION IV
                     EXPERIMENTAL

INSTRUMENTATION AND FUNCTIONAL DESCRIPTION

Figure 1 is a block diagram of the experimental setup.
A Digilab FTS-14D/IR spectrophotometer equipped with
the Digilab GC/IR accessory and controller was connected
with a Perkin-Elmer 990 Gas Chromatograph.  The latter
was equipped with a SCOT capillary column, a flame
ionization detector, and a 14:1 splitter.  The splitter
was derived from a standard 50:1 splitter by constrict-
ing the output port so that an adequate signal from the
FID was obtained for submicrogram quantities of the
compounds injected.  A Wilks heated transfer line con-
nects the output of the splitter to the IR cell.

The IR cell is a cylindrical cavity, 6 mm in diameter
and 50 mm long, bored through a heated metal block.
KBr windows are sealed to each open end of the cavity.
The inlet and exit ports of the cell are arranged in
such a manner that the gas stream sweeps the full
length of the cell.  A system of valves, which can be
manually or automatically actuated, allows the flow of
the carrier gas to be stopped or the eluted substance
to be trapped in the IR cell.

Infrared radiation is passed through the cell by means
of an elipsoidal mirror located at the input side of
the cell.  A plane mirror reflects the IR beam from
the interferometer into the ellipsoidal mirror.  The IR
                           8

-------
Figure 1.  Block diagram of the GC/IR system

-------
beam, after passing through the cell, is collected by
another ellipsoidal mirror at the opposite end of the
cell.  The collected radiation is directed into a plane
mirror,which reflects it into the optical system in the
spectrometer.  The optical system in turn directs the
radiation into the triglycerine sulfate  (TGS) pyro-
electric detector.  The cell has a 25% transmission
when the high frequency optical cut-off  filter is in the
light path.  Without the cut-off filter  the cell trans-
mits as much as 33% of the incident energy.

The Digilab FTS-14D/IR contains two IR systems.  System
1  (SSI) is for infrared measurements using conventional
IR procedures.  Spectra with resolutions of 8, 4, 2, 1
and 0.5 cm"  may be obtained with SSI.   System 3 (SS3)
is the GC/IR system.  All spectra are obtained at a
resolution of 8 cm   in this system.  Data collection
for both systems between 3950 and 400 cm"  at this reso-
lution is completed within 0.6 seconds.  The duty cycle
per scan is approximately 1 second.

Four different modes of spectral analysis of GC
effluents are available in SS3.  These are the On-The-
Fly (FL), Every Scan (EV) , Trapped (TR)/and Stop-Flow
(ST) modes, each of which is unique.  The desired mode
of operation may be selected by giving the GC mode
parameter, GCM, the appropriate value.   Thus, for an
On-The-Fly mode, GCM is set equal to FL.

When GCM = FL, the spectrum of each GC peak is obtained.
The number of scans taken for each GC peak is determined

                          10

-------
by the length of time the signal from the FID exceeds
the. threshold voltage of the scan threshold sensing
circuit of the GC/IR controller.  The data from the
scans taken are automatically coherently added (co-
added) and stored in the disk.  The location of the
data is designated by the parameter, PKN.  The storage
is sequential.  Thus, the data from the first peak is
at PKN=1, the next at PKN=2, etc.  The spectra of all
the peaks in the chromatogram may be obtained by
setting the parameter NPK equal to the number of GC
peaks.  As many as 80 spectra may be stored in the disk.

When GCM=EV, the spectrometer takes consecutive sets of
scans/ co-adds the data for the number of scans in the
set, and stores each co-added scan sequentially until
all the files are filled or the data collection is
terminated by the operator.  The number of scans per
set is specified by the parameter, NSS.

In the trap mode, GCM=TR, a particular peak in the
chromatogram may be trapped in the IR cell while the
rest of the effluent is allowed to bypass it.  If trap-
ping of the nth peak is desired, the PKN parameter is
set equal to that number.  NPK must be set >_ PKN.  The
number of scans taken on the trapped sample is deter-
mined by the value of NSS.  The trap automatically opens
after NSS scans are completed.

The Stop-Flow mode, GCM=ST, allows one to stop the flow
of the carrier gas while NSS scans are taken on a GC
                          11

-------
peak.  After NSS scans are completed the Stop-Flow
valve opens automatically.  Spectral analysis of each
peak in the chromatogram may be carried out in this
manner if no degradation of GC resolution occurs.
Switches are located on the GC/IR controller to allow
manual scan, trap, and stop-flow operations.  SS3
contains provision for automated operation with an
automatic GC injector.

Except for the EV mode of operation, data must be
collected at the instant when the GC component is in
the IR cell.  The eluted substance has to travel from
the GC column to the IR cell.  There is a time differ-
ence between the time the signal is received from the
FID by the peak detection circuit in the GC/IR controller
and the time the maximum quantity of the eluted compound
is actually in the IR cell.  A program, called the GC
timing setup, to measure this time lag is provided in
SS3.  It is brought into operation by the command GCT.
The parameter, MAR, is given a value equal to the per
cent change in the intensity of the center burst of
the interferogram due to instrumental variation.  After
the MAR value is set by the operators, the GCT command
is executed and a strong IR absorbing compound is
injected into the GC.  The peak detect circuit in the
GC/IR controller senses the maximum of the FID signal
and counts the number of scans as the intensity of the
interferogram decreases to a minimum.  The number of
scans is printed out by the teletype as the value of
the time delay parameter, TIM.  The value of TIM
                          12

-------
remains useful so long as the carrier gas flow rate
remains unchanged.

Data collection in all four modes is performed by
executing the command GCS.  The data stored at this
stage are interferograms.  The interferogram in a file
specified by PKN may be plotted by setting the plot
mode parameter, PLM, equal to its non-ratioed plot
mode value, E, and executing the GC plot command, GCP.

The interferogram is transformed into a non-ratioed
spectrum and plotted when the command to compute ,  GCC,
is executed.

The ratioed spectrum is not stored in the system.  It
may be plotted by executing the command GCP when PLM
is set equal to T, A, or L if a transmission, absor-
bance, or log absorbance plot is desired, respectively.
Stored spectra in any of the sample files may be
ratioed to any of the reference or background spectra
in the reference file.  The location of the latter is
defined by the parameter, KEF, which may be equated to
any interger from 1 to 5.  Thus, as many as five
separate background spectra may be stored.  A reference
spectrum may be taken before or after a GC run.  To
collect a reference spectrum a value for REF is desig-
nated, PKN  is set equal to zero, and NSR to the number
of scans desired.  Collection of reference interfero-
grams is performed by executing the command GCR.
Computing of the spectrum from the interferogram and
plotting it in the non-ratioed mode, PLM=E, is effected
                          13

-------
when the command GCC is executed.  In SS3, stored data
(interferogram or spectrum) are plotted by executing the
command GCP.  The value in wave numbers given for the
Start-of-Plot (STP) and End-of-Plot (ENP) parameters
determines the frequency range to be displayed.  A list
of fifteen commands containing aa. sixty-four parameter
updates may be executed.

MATERIALS

Biphenyl, Chem Service, Inc., HP grade; mp 69.5 - 70.5°
Chloroform, Burdick & Jackson, distilled in glass
Cyclohexanone, Chem Service, Inc., HP grade; Bp 154-156°
Diethyl ethylmalonate, Chem Service, Inc., HP grade;
  Bp 95-97°/15 mm
Diethyl malonate, Chem Service, Inc., Purif.; Bp 198-9°
Diethyl oxalate, Chem Service, Inc., Purif.; Bp 184-6°
Dimethyl adipate, Chem Service, Inc., HP grade; mp 9-10°
Dimethyl oxalate, Chem Service, Inc., HP grade; mp 53-
  55°
Naphthalene, J.  T. Baker Chem. Co., "Baker Analyzed,"
  mp 80°
Tetradecane, Pfaltz & Bauer
                          14

-------
PROCEDURES

General

The compounds were used as received without further
purification.  Stock solutions of each compound were
made in chloroform at concentrations of 25, 20, 15, 10,
and 5 yg/yl.  Solutions of lower concentrations were
prepared by diluting appropriate portions of the stock
solutions.  A multi-component solution containing the
five diesters, each at a concentration of 25 yg/yl, was
also prepared.  In these experiments unless otherwise
specified, a 1 yl injection was made using a 10 ul syringe,
GC/IR analyses of solutions containing only a single
solute were carried out under isothermal GC conditions.
Multi-solute solutions were analyzed under GC programmed
temperature.

The temperature of the GC injection ports, manifold,
transfer line, and the IR cell were set at least 50  C
above the column oven temperature.  The spectrometer
was purged with N2 at a rate of 2.5 H/min.

Determination of the Optimum Value of TIM

While the value of the time delay parameter may be
obtained by the GCT program, this value although pre-
cise is likely to be erroneous.  Its accuracy depends
to a large extent on the accuracy of the value of MAR,
which is not calculated directly in the program.  The
value for MAR is the best guess by the operator of the
per cent change in the intensity of the interferogram
due to instrumental variations.
                         15

-------
In our work the time delay parameters were obtained in
one of two different ways to eliminate this source of
error.  In one method the GCT program was used with MAR
set to a value ea. 5.  Acetone was injected into the GC
and the center burst of the interferogram for each scan
was displayed sequentially on a storage oscilloscope.
The horizontal sweep was displaced slightly to the
right for each scan so that a raster-type display was
obtained.  The number of the scan in the array giving
the minimum intensity was assigned to the parameter TIM.

In the other method, the actual solution of the
compound for GC/IR analysis was used.  A sixteen scan
background spectrum was collected using the following
entry :
                         REF/1
                         PKN/0
                         NSR/16
                         GCR
                         GCC
                         i
after the reference was obtained the EV mode of
operational analysis was used, i.e. ,
                        GCM/EV
                        PKN/1
                        NSS/1
                        TIM/1
                        GCS
The solution was injected into the GC and after the
solvent was eluted completely, GCS was executed.   The
                         16

-------
interferometer began to scan after the threshold
voltage was exceeded.  Thus, each single scan spectrum
was stored sequentially on file.  The value of TIM
was equal to 1 plus the value of PKN containing the
most intense spectrum in the file.  The TIM values
obtained were tested for each GC/IR mode.

Profile of the Eluted Peak in the IR Cell

Data collection identical with the GCM=EV mode for TIM
determination was used to define the time dependence
of the concentration of the eluted substance in the IR
cell.  The intensity of an absorption band in the
spectrum of the compound was plotted against time.  This
time corresponded to the value of the PKN parameter
associated with each spectrum.  With NSS=1, consecutive
PKN values were separated by 1 second intervals.

Trap Leakage Rate

Trap leakage was measured using both SS3 and SSI.  A
sixteen scan background spectrum at 8 cm   resolution
was stored in file 1 in SSI.  The parameters in SSI were
set to collect 16 scan spectra at 8 cm   resolution and
to plot the frequency region containing a strong absorp-
tion band of the compound in the absorbance mode.

With the instrument in SS3, the solution of the compound
was injected into the GC and the compound was trapped
automatically.  The toggle switch was manually turned
on to keep the trap closed.  The data collection was

-------
aborted and the system was transferred to SSI.  The SCN
command was executed to get spectra of the trapped
material at one minute intervals.  The intensity of the
absorption band was plotted as a function of time.  A
simpler method used later in this work employed only SS3
in the EV mode.  NSS was set equal to 10 so that each
spectrum represented a time period of 10 seconds.  The
solution of the compound was injected into the GC and
after the solvent was completely eluted the GCS command
was executed.  Trapping of the eluate was done manually
by means of the toggle switch, which operated the trap
valve.  The trap valve was closed after an appropriate
delay time was allowed for by counting the required
number of scans from the instant scanning was initiated
by the signal from the FID.  Data collection was
allowed to continue until all 80 files were filled.  A
plot was made of the absorbance at a particular character-
istic group frequency of the test compound vs. time.

Reconstruction of the Gas Chf oroatogram from IR
Absorbance Data

Spectral data collection was essentially the same as the
one used for TIM determination.  GCM=EV, NSS=1, TIM=1,
were the parameters used.  GCS was executed and the
solution of the diesters was injected into the GC.  The
GC conditions used were

     Temperature of injection port and manifold, 250°C
     Temperature program, 85° to 200°C @12°/min
     Temperature of transfer line, 270°C
                          18

-------
     Temperature of IR cell, 265°C
     Column, 15.24m SCOT SE30
     Carrier gas flow, 9.4 ml/min of He measured at
       the IR cell outlet
     Splitter, 14:1

Data collection was allowed to proceed until all 80
files were filled.  The previous GCS command was
deleted.  The plot parameters were set to display C=O
stretching frequency region of the spectrum in the
absorbance mode and an instruction tape to compute and
plot the spectrum was fed into the punch tape reader.
After all the 80 spectra were plotted, another set of
data was collected, computed, and plotted.  The collec-
tion of the succeeding sets of data was initiated at the
point in the chromatogram at which the previous set was
terminated.  The solution was injected into the GC for
each set of 80 spectra collected.  In this manner a
spectrum for every one second interval in the chromato-
gram was obtained.  Plotting the intensity of the
absorption band vs. time gave the IR reconstructed
chromatogram.

Determination of Sensitivity

The sensitivity for three modes of spectral analysis,
EV, FL, TR were tested using separate solutions of
biphenyl, naphthalene, tetradecane, cyclohexanone, and
dimethyl adipate.  The absorbance between 3100 and 2900
cm   for the first three compounds and that between
1800 and 1700 cm   for the last two were measured.
                         19

-------
Multiple runs were made for each compound.  The quantity
of material injected was plotted vs. the GC peak area
for each compound.  The areas were measured with a
planimeter.  The corresponding plot of absorbance vs.
GC peak area was also made.  The slope of the latter was
divided by the slope of the former.  The result was then
expressed as vg per 10~  absorbance unit.  This absorb-
ance unit corresponded to an absorption of oa. 0.2% of
the incident energy at the frequency of the absorption
band being monitored.
                         20

-------
                       SECTION V
                RESULTS AND DISCUSSION

NOISE AND THE OPTIMUM REGION FOR SPECTRAL ANALYSIS

A single scan non-ratioed background spectrum from the
GC/IR cell is shown in Fig. 2.  The spectrum intensity
is maximum at 2036 cm*" .  The cut-off frequencies are
at 450 and 3900 cm" .  Below 750 cm"  and above 3500
  — 1
cm  , the energy is less than 10%  and 20%, respectively,
of the energy in the spectrum at 2036 cm" .  Figures
3a, b and c show the region between 2020 and 2070 cm
with the abscissa expanded 8 times.  In this figure, a
is plotted without ordinate expansion; b is plotted at
SOX ordinate expansion and c is a  superposition of three
separate single-scan spectra.  In  c, the noise is seen
as a variation in the intensity of each corresponding
peak in the superposed spectra.  On the average this
amounts to 20% of the intensity at SOX ordinate expan-
sion.  Thus the signal-to-noise ratio at 2036 cm   is
oa. 250.  This represents the maximum signal-to-noise
ratio that can be expected from a  single scan non-
ratioed spectrum.

The relative noise level between 500 and 3800 cm   is
shown in Fig. 4.  Trace a  is the  ratio of a pair of
si gle-scan background spectra; b, c, and d  are those
of 16-, 256- and 1024-scans spectra.  All are plotted
in the transmission mode without ordinate scale
expansion.  In a, the noise level  below 800 cm   is
approximately twice the noise level above 3500 cm   and
                         21

-------
      C/D
                                            4.5
                        Cm
Figure 2.   Non-ratioed single  scan background spectrum
                          22

-------
20.7   2Q2
                      20.7   2Q2

                     Cm "I  X I 0 ~2
20.7-  2Q2
Figure 3. (a)
          (b)


          (c)
        The region of maximum intensity of the
        spectrum in Fig. 2 plotted at 8X
        abscissa expansion without ordinate
        expansion

        50X Ordinate expansion of the portion
        of the spectrum shown in  (a)

        A superposition of three single scan
        spectra in the same region under the
        same scale expansion as in (b)
                        23

-------
     38
   30         20
         CrrT1  X  I 0 -2
Figure 4.
Transmission baseline using equal number of
scans for background and sample between 500
and 3800  cm~l
                         24

-------
almost 20 times that of the noise  level between 860 and
3500 cam  ,  This result is in accordance with the energy
distribution in the spectrum in Fig. 2,

The effect of signal averaging on  noise can also be seen
in Fig. 4, especially  at the region below  800 cm  .  In
this region the ratio  of the peak-to-peak  noise level
in traces a, b, c, and d are approximately 20:4:2^1.
The relative noise level in the spectrum between 800 and
3500 cm"  for these same traces is also shown in Fig. 5.
Trace a is plotted without ordinate expansion, while
traces b, c, and d are plotted with 10X ordinate expan-
sion.  The relative noise levels in this region for the
four traces are 10:4:1.5:1.  The noise level changes by
no more than a factor  of 10 throughout this spectral
range; this region, therefore, is  the  optimum range for
GC/IR analysis and spectral sensitivity tests.

The advantage of using a reference background spectrum
containing a large number of scans is  depicted in Fig.
6.  Curves a, b, and c are the transmission base lines
for single, 16- and 256- scans spectra ratioed to a
1024-scan background spectrum respectively. . Curve d is
exactly the same as d  in Fig. 4.   Improvement in the
noise level can be readily seen when corresponding
traces are compared in Figures  4 and 6.  The noise level
in Fig. 6a, b, and c is oa. 3/4 to 1/2 that of the
corresponding traces in Fig.  4.  This  gain is due to the
high signal-to-noise ratio in the  reference spectrum,
which theoretically should be 32 times that of a single-
scan spectrum.   The noise in curves a and b in Fig. 6
                          25

-------
                             10
Figure 5.
The noise level on the transmission
baselines in Fig. 4 between 800 and 3800
cm-1.  (a) no ordinate expansion, (b),  (c)
and  (d)  LD:€ ordinate expansion.
                         26

-------
    b
                                         •4
                           I
     38       30         20         10    5
                 Cm"1  X   !0~2
Figure  6.
Transmission baselines  using a 1024-scan
background spectrum.  NSS for a/ b, c,and d
are 1,  16, 256,and 1024 respectively.
                        27

-------
is due mostly to the poor signal-to-noise ratio in the
1- and 16-scan spectra.

One disadvantage of using a large number of scans for
the background spectrum is that the background intensity
may change during the period that a reference and sample
spectrum-are collected.  This change in background
intensity, which may be due to a change in the purged
condition in the spectrometer or column bleed, results
in poor background compensation in the ratiped spectrum.
It is easily noticeable in d of Pig. 5, where the water
absorption between 1300 and 1800 cm"  was not completely
compensated for.  As noted previously, the maximum
signal-to-noise ratio in the single beam spectrum is 250
at 2036 cm~ .  If the energy at this frequency were
reduced 250 times, the signal-to-noise ratio would
be unity.  In the ratioed spectrum this would corres-
pond to an absorption of 0.4% of incident energy or
1.7x10   absorbance units.  To obtain a common measure
of the sensitivity of the GC/IR system to different types
of compounds, the sensitivity .is arbitrarily expressed
as the quantity of the compound in yg or number of moles
that will yield an absorbance of 1x10   at the selected
characteristic group frequency.  This unit of absorbance
corresponds to an absorption of 0.23% of the incident
energy and is therefore only 56% of the absorption for
a unity signal-to-noise ratio at 2036 cm  .  Furthermore,
the noise level in the region between 800 and 3500 cm
varies by less than a factor of 10.  The unit of sensi-
tivity adopted in this work may consequently be no more
than 20X greater than the true value .of the sensitivity
of the system for a specific compound on the basis of
signal-to-noise ratio of unity.
                           28

-------
In Fourier transform spectroscopy the criterion for
detectability based on a signal-to-noise ratio of unity
is valid only when the absorption bandwidth is much
larger than the noise bandwidth in the spectrum.  When
the absorption band is weak and its halfwidth is
narrower than the resolution setting used for spectral
measurement, it is difficult to distinguish the noise
from the signal for a signal-to-noise ratio of 1.  This
is because the noise bandwidth will always be greater
than the absorption bandwidth.  A better criterion for
detectability in this case would probably be a signal-
to-noise ratio of 2.

GC/IR ANALYSIS

Optimization of GC/IR analysis depends essentially on
finding the proper conditions for obtaining the most
intense absorption spectrum for a given quantity of
eluate.  Two general requirements are  (1) adequate purg-
ing of the spectrometer to remove as much as possible the
CO2 and water vapor from the optical path, and  (2) main-
taining a sufficiently high temperature throughout the
path transversed by the eluate to prevent condensation.
Purging with N2 boiled off from a liquid nitrogen tank
at a rate of 2.5 i/min is sufficient for this purpose,
although the absorption intensity due to water decreases
continuously throughout an 8 hour period at this purge
rate.  This change, however, is relatively slow.  One
hour after purging is initiated, adequate compensation
for H-0 absorption in the ratioed spectrum is obtained
                          29

-------
when the sample and background spectra are collected
within 15 to 30 minutes.of each other.  Collection of a
background spectrum just before a GC/IR run proves in
most cases sufficient to insure adequate background
compensation.  Heating the injection port, manifold,
transfer lines, and IR cell to at least 50° above the
GC column temperature is adequate to prevent condensation,

With these conditions established, specific factors that
may affect the intensity of the spectrum were investi-
gated.  Figure 7 is a time delay calibration curve for
the SCOT capillary column.  The time delay decreases
monotonically with increasing carrier gas flow.  The
time delay calibration curve, which in this case was
obtained with GCM=EV, may be used directly for GCM=FL
but not for GCM=TR.  Observations during the course of
these experiments show that the optimum value of TIM for
the trap mode is 1 to 3 units less than that for the
GCM=FL.  .For sharp GC peaks (halfwidth less than 10 sec.)
this difference is between 1 to 2, and for broader peaks
it may be as much as 3.

From Fig. 7 one may obtain also the upper limit of
carrier gas flow rate for which automatic trapping of
the eluate in the IR cell is practicable in the present
configuration of the system.  For example, a TIM=1 for
GCM=TR will correspond to a TIM=2 or 3 for GCM=FL.
These correspond tp flow rates between 17 to 24 ml/min.
Since the minimum value of TIM for all GC/IR modes is 1
(TIM=0 is not allowed)  the practical upper limit of the
flow rate for trapping the eluate is oa.  24 ml/min.
                          30

-------
to
H
             H-


             I
             CD

             -J
        
   (D
CD
M
(U
           (D
              rt
              H-
              O
              O

           Hi h
3    cn
CD  I"J £r*

CD  rt 5
h  (D H-

   DJ VQ
           f
                            15
                       OL
                            10
                                                                                     I
                                                                     10              15

                                                             CARRIER  GAS  FLOW, mI/min
                                                                                            20

-------
During the transit of the eluted compounds between the
column and the GC/IR cell, separation of the components
may deteriorate because significant change in the dis-
tribution of the eluate in the carrier gas occurs or
the transit time of the eluate through the GC/IR cell is
very large compared with the peak to peak separation of
the eluates in the chromatogram.  Fig. 8 shows the GC/IR
"resolution".  Curves a and b are the respective plots
of the transit time of the eluted peak through the IR
cell and the width of the GC peak us. flow rate.  The
transit time is the width of the peak at half the maximum
absorbance during transit of the eluate.  Such a plot is
shown in Fig. 9a.  The corresponding GC peak is shown in
Fig. 9b.  It is evident from Fig. 8 that the* peak to peak
resolution of the GC/IR system improves with increasing
carrier gas flow rate.  At a flow rate of 3 ml/min the
transit time is twice the GC peak width.  This is reduced
to a factor of 1.6 at 16 ml/min.  We concluded that GC
peaks resolved by twice their peak width in the chromato-
gram should also be resolved in the GC/IR system.  There
is no visible significant difference between the general
shapes of the two curves a and b in Fig. 9.  We
further concluded that the GC resolution is not signifi-
cantly degraded during the transit of the eluate to the
IR cell.

The relation between GC/IR absorption intensity (curve b)
and GC peak width (curve a) is shown in Fig. 10.  A
threefold decrease in the GC peak width results in less
than 6% increase in the IR absorption intensity over a
                           32

-------
    20 —
    15
  o
  0)
  0}
              i—a
o GCIR PEAK  WIDTH

• GC PEAK WIDTH
                  I
       I
I
                  5          10          15

              CARRIER GAS FLOW  RATE, ml/min
Figure 8.  Plot  of (a)  GC peak transit  time  in the
                        IR cell and

                    (b)  GC peak width vs.  flow rate
                          33

-------
      en

      o


      x

      <.
       . 30 -
      UJ
      0
       m
       a:
      •o
       u»
       m
      UJ

      Q_
      Of
      o
                \-b
                                               -I 50
               •  GCIR PEAK


                 GC PEAK
                                                 40
  w


30 I


  >-

  <
  cc
  i—

  oo
                                       ui
                                       Q.
                                       O

                                       O
                                                 10
                     5          10


                        TIME, sec
                             15
Figure  9.
(a)  Concentration profile of  the eluate in

    the IR cell  and


(b)  Mass profile of the same  eluate from

    the chromatograph.  Carrier gas flow

    rate: 17 ml/min.
                            34

-------
four fold increase in the flow rate for the present
system.  Since capillary columns deteriorate rather
rapidly at high flow rates and since no substantial gain
in GC/IR sensitivity is obtained under this condition,
the flow rate optimum for GC resolution should also be
optimum for GC/IR analysis.

Signal averaging is best performed by trapping the
eluate, i.e./ by using GCM=TR mode.  The trap leakage
limits the length of time for effective signal averag-
ing.  Pig. 11 shows the decrease of absorbance with time
for a trapped GC eluate.  The absorbance remains practi-
cally constant for 100 seconds after the trap closes and
then decreases rather sharply for a period of 100 seconds,
Thereafter, it decreases slowly at an approximate rate
of 2.7xlO~  absorbance units/sec.  The best period for
signal averaging, therefore, is within 100 seconds after
the compound is trapped.  About 100 scans can be taken
during this time and an improvement of 10 times the
signal-to-noise ratio of a single scan spectrum may be
expected.

Fig. 12 shows the effect of the number of scans on the
signal-to-noise ratio- at the C=O stretching region for
a trapped cyclohexanone eluate.  The signal-to-noise
ratio increases linearly with the square root of the
number of scans for NSS 100.  It drops sharply for
100 NSS£25£, and somewhat slowly for N3S<256.  This
                           35

-------
          10 —
        o
        V
        m
Q.


O
                   •  GC PEAK WIDTH

                   o  QC4R ABSORBANCE
                      J_
                                     _L
                    .5          10-15

                CARR4ER GAS FLOW RATE, ml/min
                                                    60
O


X
                                                       o
                                                      .2

                                                       CD
                                                     •••••Qe
                                                       O
                                                       CO
                                                       CD
                                                    40 <
                                                       o
                                                       o
                                                    20
figure 10.   Plot of  (a)  GC peak width vs.  flow  rate and

                       (b)  Absorbance
                             36

-------
(jj
         H-
         iQ
         C
         h
         0)
       •C O
       iQ ft
       ^
         O
       CJ Hi
       SB
       (D O
         ®
       (D
       H «i
       C Co
       JU .
       rt
       (D rt
         H-

         I

         H)
         O
         H
                            40  -
O


X
    30
no
ce
o   20
to
CD
    10
                                                    200
                                                 400
600
                                                             TIME, sec
          (D
          Pi

-------
LJ
CO
                p.
                s
                h
                (D
                t\j
                »
         rt
         H
       §
 PJ  3  (P
•O  O
*O  rt W
 (D  H-H-
 cn    PJ
    O  H
•C  Hi  I
iQ     rt
    ft O

    g"rJ
 O     O
 H-« W  H-
         0> £U
         X'-h'H
         0) 0)  (U
         5     rt
         O n  H-
         300
         CD O
         fD    H
         H O 0
         Ci Ml ri-
               ft
               ro
            CO QJ
            CO
               ft)
            Hi CO
            o
rt a

-------
behavior correlates very well with the curve in Fig.
11.  Signal averaging within the period of time
when the density of the trapped eluate in the IR cell
remains constant results in a gain in signal-to-noise
ratio that varies directly with the square root of  the
number of scans.  No further gain in S/N is realized by
signal averaging over longer periods.  In fact, if  the
time for signal averaging includes the period when  the
density of the trapped eluate is decreasing because of
leakage from the trap, the signal to noise ratio of the
resulting spectrum decreases.  This is to be expected
since the signal intensity is directly proportional to
the absorbance, and, as previously mentioned, the noise
is independent of the signal intensity, while signal
averaging is apparently made over non-weighted single
scan spectra.  The maximum signal-to-noise ratio in
this particular case is 34 (See Table 1) and, on the
basis of S/N=1, the absolute detection limit of the
instrument for cyclohexanone is 0.2 yg or 1.6x10-9  moles.
The previous discussion on the adapted generalized
expression for the detection limits (yg/10" A) does not
apply to this particular case since the value of the
detection limit for cyclohexanone is based directly on
the optimum signal-to-noise ratio at the C=0 stretching
frequency for the compound.

The detection limits for other compounds are shown  in
Table 2.  Figs. 13a-e and 14a-e are the least-squares-
fitted- plots of concentration of solutions and of  the
absorbance at the selected group frequencies vs. GC
                          39

-------
Table 1.  SIGNAL TO NOISE RATIO AT THE C=O STRETCHING
          FREQUENCY FOR TRAPPED SAMPLES OF CYCLOHEXANONE
NSS (=NSR)
4
16
100
256
400
1024
—3
Absorbance (10 A)
22
19
19
12
11
5
S/N
5.3
13.0
34.0
13.7
11.5
4.6
                          40

-------
Table 2.  DETECTION LIMITS FOR A NUMBER OF ORGANIC COMPOUNDS
Compounds
Biphenyl
Cy c 1 oh ex anon e
Dimethyl
Adi pate
Naphthalene
Tetradecane
dm/ds
wg/Unit Area
22.5
10.7
17.0
22.3
22.9
dA/ds Detection Limits 	 '"
10""3A/Unit Area yg/lo"3A(Nano-moles/10~3A)
31
41
70
33
16
0.7
0.3
0.2
0.7
1.4
(4.5)
(3.1)
(1.1)
(5.5)
(7.0)

-------
to
       H-


       1
       (D
     — O
     H- O
     >0 H>
     fa^
     (D O
     9 0
     »< a
     H O
       (D
       §
       0)
       8
       T3
       (D
       R-
                         30
                      c*
                      3
Z
O
z
LJ
O
Z
O
O
   20
          BIPHENYL
                               1
                                          GO PEAK AREA (ARBITRARY UNITS)
                                                         10

-------
U)
         (D
       tr M
       *-* o
         rt-

       30
       ,£i **
       *o
       P'O
(U O
H (D
ro P
3 <1-
(D h
  (U
  ft
  H-
  O
  3
         Co
         8
         ro
                          30
3
\
cn



z
o

t~
<

t—
z
LU
O
z
o
o
                   20
                            NAPHTHALENE
                                              GC PEAK  AREA  (ARBITRARY UNITS)
         0
         fa

-------
  "3
  H-

  •a
  to
  *
o
ft) O
§•§
o o
  ft
  n
  0>
  tt
  H-
  O
  3


  §
(p
  (D


  $•
  H
  (D
                 30





               3
              \
               Ot


                f*


              o 20
              ^^
              t—
              <
              LU
              o
              o
              o
                  0
                        TETRADECANE
                                             5                         10

                                    GC PEAK  AREA (ARBITRARY UN ITS)

-------
Crt
         H-

        •8

         fl>

         H
         U)
       3 Ml
       CD
         SO
         o
      »<. 0
       H O
         (D
       0) 3
         ft
         H.
         O
         CO
         O
         O
         (D
                          30 -
3
\

3


Z

E  20
i—

o:
i—
z
UJ
o

o
o
                                 DIMETHYL ADIPATE
                                                _L
                         5                   10

                       GC PEAK AREA (ARBITRARY UNITS!
         (D

-------
          hrj
          H-
it*

-------
   H-
  iQ

   PS
   H
   (D
fir M
— o
 .- rt
w
H-O
^d'l-h
D*
(D P»
» cr
^: w
H O

   g-
   PI
   p
   o
   (D

   «
   CO
   O


  T3
   (D
   fD
   PJ
                        30 -
                    ro
                     o
 .: 20 -
o
m
ce
o
CO
CO
    0
                                B1PHENYL
                                                                                  10
                                          GC  PEAK  AREA  (ARBITRARY  UNITS)

-------
00
       H-

       |

       
-------
*>.
VD
           H-


          1
           (D

           H
         O
         (D O
         rt Mi
(D cn
o o
tu i-t
3 tr
CD P)
  3
  O
  (D
           0)
           8
           (D
           R)
           PJ
           ^
           (D
                                50 -
                    x

                    
-------
Ul
o
         ttj
(D


H
rf*
        §•0
         Hi
       0>

       Bfr
      »< w
       HO

       ft 8*
      wo
      01 (D

      ro 's
         0)
        0>
        CU
        h
        (B
        P
                          150
                       x

                       <
              ui
              o
              m
              oc
              o
              w
              CO
                 00
                 50
                                 DIMETHYL  ADIPATE
                                       5                  10

                                   GC PEAK AREA  (ARBITRARY UNITS)
                                                                                     15

-------
(Ji
H
      *-* O
        ft

      5 0
      0 Hi
o w
x o
s&
§s
(D O
  (D
        Co
        (D
        O
        (D
        (U
                             150 -
CO
 O
                             100
 CO
 a:
 o
 w
 m
                             50
                                   CYCLOHEXANONE
                                                       I
                                              GC PEAK
                            10          15

                           AREA (ARBITRARY  UNITS)
                                                                             20
                                                                                   25

-------
peak area, respectively.  The values in Table II were
calculated from slopes o£ the corresponding curves in
these figures.  It is interesting to compare the detec-
tion limit for cyclohexanone obtained with that described
in the previous paragraph.  The difference is less than
a factor of 2.  Furthermore, the detection limit for
dimethyl adipate is comparable with that for cyclo-
hexanone.  On this basis it is reasonable to conclude
that the unit adopted for the detection limit for the
instrument is valid.

The detection limits for cyclohexanone for the
different GC/IR modes, also shown in Table 3, are prac-
tically identical.  Programmed temperature conditions
are almost a necessity to achieve both speed and reso-
lution in the GC analysis of a multicomponent sample.
However, the carrier gas flow often decreases as the
temperature increases during the run.  As shown in Fig.
7, the time delay parameter is sensitive to changes in
carrier gas flow rate.  If GC/IR analysis with GCM=FL is
used, the synchrony between spectral scanning and the
elution of the components into the IR cell deteriorate?
progressively with each succeeding GC peak.  If the
change in carrier gas flow rate does not produce a change
in the delay time by more than 1/2 of the transit time
of the eluate through the IR cell, it should be possible
to record a spectrum of each eluate in the chromatogram.
The spectra, however, will not have the optimum intensity
and signal-to-noise ratio.  With greater change in the
flow rates, it is possible to miss the spectrum of the
eluates after the first few peaks in the chromatogram.

                         52

-------
Table 3.  DETECTION LIMITS FOR CYCLOHEXANONE AT THE
          C=0 STRETCHING FREQUENCY FOR THE DIFFERENT
          GC/IR MODES
GCM
NSS
NSR
      Detection Limits
[yg/10~3A (Nano-moles/lO^A)
EV
FL
TR
TR
 16
100
 16
 16
 16
100
        0.2   (2.0)
        0.2   (2.0)
        0.3   (3.1)
        0.3   (3.1)
                           53

-------
The Every Scan mode is potentially most suited to
programmed temperature GC/IR analysis.  A spectrum may
be taken and stored as often as every second during the
GC run.  Recording of the spectrum of every GC peak is
virtually assured under normal GC conditions.  Fig. 15a
is a reconstruction of a chromatogram from the IR absorb-
ance data obtained with the Every Scan Mode.  The corres-
ponding chromatogram is shown in Fig. 15b.  The single
scan spectra, one for each eluate, plotted in the absorb-
ance mode are shown in Fig. 16a-g.  These results show
the effectiveness of the chromatogram reproduction from
the GC/IR absorbance data and the quality of the spectra
obtained without the benefit of signal averaging.

Both software and hardware modifications will be needed
in order to use the Every Scan Mode more effectively.
The signal-to-noise ratio in a single scan spectrum is
much too low for recording usable spectra of compounds
in the yg range.  Weighted co-adding of the interfero-
grams taken during the transit of the eluate through the
IR cell will be needed to optimize the signal-to-noise
ratio of the spectrum.  In the current system this can
only be done manually.  Note that in Fig. 15a, the
number of spectra or interferograms needed to reconstruct
a chromatogram from a single GC run far exceeds the
present storage capacity of the disk.  Additional data
storage capacity is required for an uninterrupted data
collection from one or more GC runs.  The quantity of
eluate for GC/IR analysis can be optimized if the GC
splitter is removed and the full GC effluent is relayed
to the IR cell.  In this configuration a means of
                          54,

-------
                   8
                   H
                   Cn
ui
0>

O

H-

P-

P>
H

O
ft O
PI ft
                                o:
                                O
                                CO
                                CO
                                           30      60      90
                                                                  120     150     180     210

                                                                              TIME, sec
                                                                                      240    270
                                                                                                     300
                                                                                                             330
                                                                                                                    360
                                                                                                                           390

-------
          SOLVENT
Figure 16.  Single scan spectra of the eluates in
         Fig. 15
         (a) Chloroform (solvent peak)
                  56

-------
            •5
            (D
Ul
         ^^ H W
         cr ui H-
         *-  3
       3 H-
       •a 3
       C (D

       'i&
 ^ O

3 0

rt pi

0> (D
       H-tJ
       3 0)
       (D pi
       ft *•
O -"
X
p) PI
H 3
       (Dft

         (D
       0>
       9!
       «
      (D

      (0
      O

      s

      en
      O
      rt
      1-5
      PJ

      O
      H>

      rt
      3*
rt
(D
CO

H-
3
                          PEAK  #   I
                                                        PEAK  #  2

-------
           0>
U1

CX)
3 D
P> H-
H (D
O rt
ft M
(D O
  X
^. pi
TJ H
(D D»
ft) ft
*• n>
      ^ ro
        u>
        PU
        a
        H-
           CO
           (D

           CO
           O
cn
>d
CD
o
rt
      O
      H>

      rt
      tr
      CD

      0)
      rt
      (D
      in

      H-
      3
                       PEAK  #  3
                                                   PEAK   #  4

-------
           (D
VO
       PI H- H-
      P. D   M
      H- H- I-1 (D
      3 (D Ui
      (D rt   W
      ft 3*   O
      Pi
       CD
       rt
     fD
       0)
       (D
       =**=
       Ul
W

(D
O
rt
     *T3 I—1
     BJ 3
     rt (a   »-(
     (D h"1  fu
       o
     >^ 3   O
rt

(D

(D
M
C

rt
(D
(Q
                        PEAK   #   5
                                                       PEAK  #  6
       Pi

-------
synchronizing GC/IR data collection with the GC
operation is required without the signal from the FID.
Both hardware and software to carry out this improvement
had been ordered at this writing.
                          60

-------
                   SECTION VI

                   REFERENCES


Lephardt, John Oscar, "Application of Fourier
Transform Spectroscopy for Gas Chromatography
Effluent Analysis and Structural Studies," 1972,
Ph.D. Thesis, University of Maryland, College Park,
Maryland.

Low, M. J. D., "Infrared Fourier Transform
Spectroscopy in Flavor Analysis IV. Spectra of Gas
Chromatography Fractions," J. Agr^ Food Chem. 19,
(6): 1124-1127 (1971).

Lephardt, J. 0. and Bernard J. Bulkin, "On-The-Fly
Gas Chromatography-Infrared Spectrometry Using a
Cholesteric Liquid Crystal-Effluent Interface,"
Analytical Chemistry, 45_, 706-710  (1973).

Fellgett, P. B., "A Propos De La Theorie Du
Spectrometre Interferentiel Multiplex," J. Phys.
Radium, 19, 187-236, 237-240  (1958).

Griffiths, Peter R., "Trading Rules in Infrared
Fourier Transform Spectroscopy," Analytical
Chemistry, 4£, .(11): 1909-1913  (1972).
                       61

-------
 SELECTED WATER
 RESOURCES ABSTRACTS
 INPUT TRANSACTION FORM
                                   w
    ; **'  INFRARED FOURIER TRANSFORM SPECTROMETRY
        OF GAS CHROMATOGRAPHY EFFLUENTS
                                                         5, Report Date
                                                         8.
Azarraga, Leo V., McCall,  Ann C.
Southeast Environmental Research Laboratory
Athens,  Georgia   30601
                                                           Report Ho.-
                                   '   /';;:;,•,•:: ffa.

                                    16ADN-26
                                                         ;, Type . f Report and
                                                           Period Covered
?*5: SpflB*sw'i5 °IS*a- tt/on United Stated 'Environmental Protection Agency
Environmental Protection Agency report number,
 EPA-66Q2-73-03&  Januar  197A
      An evaluation was  made of the performance of a computerized Fourier
 transform infrared spectrometer for the  on-line measurement of the
 infrared spectra of  GC  effluents.  An optimum condition  for GCIR analysis
 was  described.  Detection limits of a few nanomoles were obtained for
 common organic compounds.  The system requires between 10 and 100
 nanomoles of organic substances for qualitative identification.
 (Azarraga-SERL)
 i?3. Descriptors   *Spectroscopy, *Gas Chromatography, *Analytical Techniques,
              *0rganic  Compounds, Interferometry, Infrared Radiation
  irb. identifies   *GCIR/  *Fourier Transform Infrared Spectroscopy,
              *On-The-Fly Vapor Phase  Infrared Spectroscopy,
              Infrared Spectra, Computer Controlled
  17t. COVVRRFit'J & Group  05A
  ,'*-. Avail ability
JS, Security Class.
                     R.'l.
                      ???.  Set 
-------