Harry 1.

Pardue,
Michael F. Burke,
and Jack R. Barnes
Purdue University
Lafayette, Indiana 47907

Quantitative Gas Chromatography

An experiment

conventional experiments involving gas

chromatography have emphasized the usefulness of
this technique as a separation tool for qualitative
identification ( 1 , 2). Little emphasis is placed on obtaining significant quantitative data. However, in
present day analytical chemistry, the making of quantitative measurements on complex mixtures is as important as the qualitative identification of the components present in the mixture. Formal training of
chemistry students in this area is extremely worthwhile.
This report describes an experiment which is designed to illustrate and evaluate several methods of
handling chromatographic data, as well as to demonstrate the desirability of automatic instrumentation.
These methods range from rrra~hicalmeasurement and
manual computation of the data to electronic measurement and computation with direct readout of the data
in digital form. The experiment illustrates the use of
operational amplifiers for measurement and analog
computation and, depending on how the experiment is
presented, it can also illustrate details of circuit design. The experiment also emphasizes the proper use
and understanding of detector response in chromatographic analyses.
The experiment consists of the determination of relative response factors for a multicomponent mixture
and the use of these factors in determining the percent
composition by weight of an unknown mixture.

Quantitofive Gas Chromatography

Detector Response

A thermal conductivity (TC) detector was used in

this experiment since it offers sufficient sensitivity for
student use. These detectors employ either a heated
metal filament or a thermistor (a semiconductor of
fused metal oxides) as the resistance e1ement:for sensing
changes in thermal conductivity. A detector element
and matching reference element are opposed to obtain
a diiei-ential signal. The heated elements are cooled
by the pure carrier gas stream and assume a definite
resistance. As the binary mixture of the eluted solute
and the carrier gas passes over the detector element, the
mixture of diierent thermal conductivity changes the
rate of heat loss, and the change in temperature causes
a change in resistance of the element. This difference
in resistance between detector and reference is a function of the instantaneous concentration of the component
in the gas stream. The reference and sensing resistance
elements are incorporated in a Wheatstone bridge, and
the out-of-balance signal is applied to a recorder. The
response of a TC detector due to temperature change is

a complex function of such things as heat capacity of

the solute, free and forced convection, and radiation
heat losses, as well as thermal conductivity. However,
by proper cell design and assuming the rate of heat loss
to be constant, the TC cell output voltage, I&, is given
by
Eo

= cAT(k./k,

- 1)

(1)

where c is a constant, and k , and k , are the thermal

conductivities, respectively, of the carrier gas and the
mixture.
Thermal conductivity, in general, decreases with
increasing molecular weight. The approximate relationship between molecular weight and thermal conductivity for members of an homologous series is given
by the proportionality
k d h = (MUMI)'/*
(2)
Therefore, the thermal conductivity will decrease
rapidly for the first few members of such a series and the
higher members will be detected with increasing sensitivity. For higher molecular weight molecules, this
difference in response approaches a limiting value and
the cell response closely corresponds to the weight of
solute emerging from the column (3).
For good quantitative work, therefore, it is necessary
to compensate for these differences in response. A
fairly exact relation between the response of a TC cell
and the amount of solute eluted from the column may
be obtained bv the use of a relative resnonse factor.
This approachrequires the determination of a specific
peak area per unit weight for each component of the
mixture. These values are then normalized with respect to a chosen standard reference compound under the
same detector conditions. The resulting relative
response factor may then be used over a reasonable
temperature range to yield highly accurate analyses (4).
Correction for detector response to various molecules
is also necessary with all other types of detectors commonly used for gas chromatography. The use of relative response factors is recommended for all truly
quantitative work.
Measurement o f Peak Areas

With the weight percent of the solute being proportional to the percent area of the recorded peak, it is
now necessary to determine the area beneath each
component peak. Several graphical methods have
been suggested; however, the most accurate is that
described by Cremer and Muller (5), which shows the
area beneath a Gaussian (bell shaped) peak to be best
approximated by multiplying the peak height by the
width a t half height.
Volume 44, Number 7 7, November 1967

695

An electromechanical approach to integration is that

of a ball and disc (6) where the number of rotations of
the disc is displayed by a sweeping recorder pen tracing.
In t,his case, a full swing of the integrator pen is assigned an arbitrary value of 100 and partial swings a
proportionally smaller value. The sum of the pen
sweeps during the recording of a peak represents the
area of a peak and is reported in "Disc Units."
A third and very powerful technique of integration of
peak areas to he discussed is that of electronic in&
gration using an operational amplifier as described
elsewhere (6). In this case, when the output potential of a T C detector is to be integrated, a basic operational amplifier feedback circuit is used (see Fig.
1) with a capacitor in the feedback loop. A resistor is

Figure 1.

Figure 2.

Block diagram of direct reading inlegrotor.

Direct Reading Integrator. A block diagram representmg the functions of the instrument is illustrated
in Figure 2. The signal resulting from a chromatographic peak is connected a t E;. and compensated for
detector response by the sensitivity compensator. The
signal is then integrated with respect to time and the
integrated signal for each peak is stored. After all
the peaks of a chromatogram have been processed in
this manner, the sum of all the stored integrals is
measured and adjusted until a reading of 1.000 is observed on the digital voltmeter (om). Each individual integral is then sequentially measured and the
digital voltmeter presents the data in the form of
weight percent for each component.
The circuit layout illustrated in Figure 3 contains
components which perform the functions described
below.

Basis operational amplifier integrator circuit.

put in the input circuit so that the input current is proportional to the signal voltage. The output voltage is
then given by

DYM.

Since the potential of the charge stored on the capacitor

is equal to Eo,this charge can then be measured and
is proportional to the area under the peak.
These three methods of area measurement are used
and con~paredin the experiment reported.
The Experiment

Reagent grade hydrocarbons were used without further

purification. Mixtures of known composition by
weight are made up in 2 ml vials which are capped with
rubber septums. Care must be taken to insure that
the composition of the sample is not changed by evaporation of the lower boiling components.
The column used is a 5 f t by '/,-in. od aluminum
column packed with 15% Apiezon L on 60/80 mesh
Chromocoil W (Johns-Manville). The column oven
is maintained a t SO%, with the detector and injector
a t 200C. Helium is used as the carrier gas at a flow
rate of 40 ml/miu. The detector filament current is
140 ma.
Chromatograph. A Variau-Aerograph Model 202
(Varian-Aerograph, Walnut Creek, Calif.) gas chromatograph with a thermal conductivity detector is used.
Reco~rler. A Leeds and Northrup Model (Leeds and
Northrup, Philadelphia, Pa.) potentiometer recorder
equipped with a Disc integrator (7) is used to record the
peaks and the areas measured by the ball and disc type
integrator.
696

lournal of Chemical Education

Figure 3.

Circuit diogrom of direct remding integrator.

SWI
SWZ
SWI
SWa

Single-pole-single-throw switch
Four.pole-six-position rotory switch
Four-pole-single-throw rotory switch
Normallv ooen wsh-bullon swibh
RI-R; 10K ten'turn poientiometer
Rs
1K 10% resirtor
1 OOK ten turn potentiometer
RD
C L - 6Mylor, 2/rf. ZOO-w.v.d.c. capacitors (Goodoll)
O.A. Heath Model EUW-IOA operational amplifier (Heath Co., Benton
Harbor, Michigan)
FOL Unit No. 1 of above system in the follower position

Response Compensation. The signal from each peak

must he modified to compensate for differences in relative response of the detector to the different sample
materials. According to the integration Eqn. (3), the
output voltage is proportional to (l/Ri,) and the output
is increased by decreasing the input resistance for a
given voltage-time integral. The effect of dialing the
relative response factor on the input potentiometer is
to increase the output potential for materials of less d e

tector response so that the same weight of each material results in the storage of equal charges on the c a
pacitors. This function is performed by a bank of five
10-turn, 10-kilohm potentiometers (RI,
etc.) used in
the input to the integrator. These potentiometers are
switched, by a set of contacts on the resistor-capacitor
(RC) selector switch (SWJ into the integrator circuit
one a t a time to receive the integration signal arising
from the individual peaks.
Integration. The operational amplifier (O.A.) is one
section of a Heath Model Em-1OA operational amplifier system (Heath Co., Benton Harbor, Mich.)
stabilized with a Phibrick K2-P chopper stabilizer (8)
(Philbrick Researchers, Boston, Mass.). This ampliier is connected in the conventional integrator configuration (Fig. I), with capacitor in the feedback to
accumdate and store the integrated signal. After each
peak, the selector switch is set to a new RC combination
for integration of the signal for the next component.
It should be noted that each potentiometer is always
associated with the same capacitor. The capacitors
must be closely matched in capacitance value and must
be made with very low-leakage dielectric. Those used
in this work are 10% tolerance Mylar, 2-pf 200-wvdc
(Goodall) matched to within +0.5% by smaller parallel
capacitors.
When the INT-READ switch (ST3) is placed in the
READ position, a 1K resistor (Re) is placed in the feedback loop of the amplifier to prevent i t from limiting.
Cmnputation. The integrals are summed by connecting the capacitors in series a t the input of amplifier
No. 1 in the follower configuration. The high input
impedance of the follower prevents drainage of charge
from the capacitors during readout. The output
from the follower is sampled by a ten turn, 100 kilohm
potentiometer. This potentiometer is adjusted to divide the output giving a reading of 1.000 v on the digital
voltmeter representing 100.0%. Leaving the potentiometer a t this setting, each capacitor is switched individually into the input to the follower and the output voltage, representing percent of that component,istabulated.
Digital Readout. A DigiTec Model 201 (United
Systems Corp., Dayton, Ohio) digital voltmeter with a
2.000-volt full-scale range is used to convert the analog data into digital data.
Procedure
Determination of Relative Response Factors

The chromatograph must be allowed sufficient time

to reach thermal equilibrium and the recorder and
integrator should be allowed to warm up for approximately one hour. With the output of the chromatw
graph connected to the recorder, and the ON-OFF switch
of the direct readmg integrator set to OFF, a 0.5 p1
sample of the mixture of known composition is injected in the chromatograph. As the peaks are traced
by the recorder pen, the chromatograph attenuator is
adjusted such that the largest peak approaches full
scale on the recorder. Several attempts may be necessary to obtain the correct setting. Then replicate
chromatograms of the known-composition mixture are
run by making anew injection of 0.5 pl after thelast peak
of the previous injection is eluted.
The areas of the peaks are then measured by taking
peak-height times width at half height and by counting

the number of "Disc Units." Using these areas and

the known weights of the components, response facton
(&/unit area) are calculated and then normalized
with respect to the largest factor.
The input of the direct-reading integrator is connected in parallel with the recorder. The chromatograph attenuator is set on the XI position, the response
compensating potentiometers are all set to 100 and the
INT-READ switch (SW3) is set to INT.
With the ON-OFF
switch (SW2) set to OFF, the integrating ampliier
balance control is set for zero output to the DVM. The
ON-OFF switch is then set to ON, and the detector
bridge zero control is used to obtain zero output to the
DVM. A 0.1 pl sample is injected and the peaks i n t e
grated by switching to an empty capacitor after each
peak when the recorder pen returns to the baseline.
After all peaks have been integrated, the ON-OFF
switch
is set a t OFF, and the INT-READ switch is set to READ.
The RC selector switch (SF2) is moved to the various
capacitors used and the voltages recorded.
The response factor for each component is determined
by dividing the mass of that component in the sample
by the output voltage for that component. Relative
response factors are calculated by dividing each individual factor by the largest. This procedure is
repeated in triplicate. The average numerical value of
each relative response factor is then dialed on the
corresponding potentiometer.
Determination of Compositions of Unknown Mixture

The weight percent composition of an ~ n k n 0 mix~n

ture is determined using the relative response factors
obtained above. Using the same procedures, three
replicate samples of an unknown mixture are chromatographed. The areas of those peaks are measured
by peak height times width at half height and the
"Disc Units." The areas are then corrected by multiplying by the appropriate factor (&/unit area X area)
summed and the percent composition by weight calculated.
For the direct-reading integrator the procedure is the
same as that used in the determination of the response
factors except that after all of the peaks of a given
sample are integrated, the RC selector switch is moved
to the SUM position. The voltage divider (Ro) is then
adjusted to give a readmg of 1 . 0 0 0 ~on the D ~ M . Now
the RC selector switch is rotated to each of the capacitors used, and the value shown, which is the percent
by weight of each component, is noted.
All of the capacitors used are discharged by depressing the RESET switch (SW4) as the RC selector
switch is set to each position.
Results and Discussion

The student is first impressed with the need for using

response factors by comparing the known percent by
weight composition with the area percent determined
by simply summing the areas and calculating their
percentages. With this in mind he then determines
the relative response factors and then uses these to
determine the mixture of unknown composition.
In reporting his results, the student is expected to
discuss the limitations and advantages of the various
methods. Results obtained by the various methods
are compared with respect to repeatability, accuracy,
Volume 44, Number 1 1 , November 1967

697

Table 1.

Relative Response Factors by Each Method

Manual

Disc

Dinct
Reading Int.

Heptane
Octane
Decane

0.92 i 0.02
0.96 f 0.02
1.000

0.93 & 0.02

0.95 i 0.01
1 ,000

0.910 i 0.015
0.949 10.010
1.I100

Table 2.

Percent by Weight of a Mixture Determined

by Each Method
-

Actual

D/,
.

Heptane
Octane
Decane

32.0
33.6
34.4

Direct
Readine Int.

Disc

Manual

32.110.21 31.710.35 31.5i0.72

3 3 . 3 i 0 . 3 5 32.7*0.38 3 3 . 8 i 0 . 4 5
34.6 &0.51 3 5 . 6 i 0 . 4 0 3 4 . 5 1 0.60

relative ease, and speed of obtaining the final result.

Typical numerical results obtained by an undergraduate student performing the experiment are given in
Tables 1 and 2. The experimental work described in
the report required about three hours for completion.
It should be noted that the chromatographic equip
ment, as well as the samples used, are not limited to those
presented in this report. Caution should be used in
selecting a system from two different points of reference. First the components should be selected such
that all can be chromatographed in less than ten
minutes in order that all of the required runs can be
completed in the alloted time. Also highly volatile
samples will change composition by evaporation of the
lighter components over a three-hour period.
The direct reading integrator can be used in suh-

698 / Journul of Chemical Education

sequent periods for studying experiment variables.

These variables can be either chromatographic, such
as the effect of changes in flow rate, temperature, etc.,
on the response factors as well as variations in the
rlcctronics. Sever31 students havt! nsrd this upparutu*
for k~llowincthe kineticsol esterificatio~~
of ethanol with
acetic acid.- This was done by measuring the increase
in ethyl acetate simultaneously with the decrease in the
ethanol and acetic acid concentrations. Clearly, the
range of problems which can be examined profitably
is limited only by the time available. There is much
to be said for illustrating the use of automatic equipment in quantitative analysis as described herein.
Literature Cited
(1)
, , MEMAN.C. E.. AND KISER.R. W.. "Problems and Emeriments'
~&trume& 'Ak,lys&," Charles E. M&~U
Books, Inc., Columbus, Ohio, 1963.
D. T., "Experimenh for In(2) REILLEY,C. N.. AND SAWYER,
strumental Methods," McGraw-Hill Book Co., Inc., New
York, 1961.
(3) D u NOQARE,
S., AND JUVET,JR.,R. S., "Ga~LiquidChre
mrttography," Interscience Publishers (division of John
Wiley & Son, Inc.), New Ymk, 1962.
(4) Rosm, D. M., MD GROR,R. L., Anal. Chem., 29, 1263
(1957).
(5) CREMER,E., AND MUKLER,M., Krochem. A&., 36, 553
(1951).
(6) BARNES,J. R., AND PARDUE,H. L., Anal. Chem., 38, 156
(1966).
(7) Disc Instruments Ine., Sents Anq Calif., Bull. 200.
(8) P~RDOE,
H. L., AND DAHL,W. E.,J. Eleetroanol Chem., 8 ,
264 (1964).
~~

~~

~~~