origin from the point corresponding to

that flow. Hence, there can be a real

increase in analysis speed, resulting in
more plates per unit time. In gas
chromatography,
most
ordinary
columns employ particles for which
d, is between 0.1 and 0.2 mm. For most
vapors in hydrogen, D o is between 10
and 40 sq. mm. per second. Thus,
D, becomes comparable with D, when
u is between 20 and 50 cm. per second.
For vapors in nitrogen, D, is about fourfold less, and thus D, is comparable
with D owhen u is between 5 and 15 cm.
per second. These speeds are a little,
but not much, greater than are currently
used.
The quantity K forms a good parameter by which to judge the quality of
packing of a column. I n our previous
paper (14), this quantity was tabulated
under the heading C,D,. It describes
the extent to which unevenness of flow
contributes to the plate height. Values
for good ordinary packed columns are
and it is
about 20 to 30 sq. em. x
seen that two of the columns described
here have values in this region. Figure
1 shows that the minimum plate height
for heptane on column I is less than
0.35 mm. in nitrogen and only just
greater than 0.4 mm. in hydrogen;

these figures are characteristic of a good

column. Nevertheless, in this column,
reference to Figure 3 shows that 30%
of this plate height is caused by H*.
This indicates that the techniques by
which columns are packed do not produce an evenly permeable bed.
The extra relaxation provided by
anastomosis at fast flows merely reduces the term H*, which results almost
entirely from imperfect column packing.
It may well be argued that rather than
alleviate bad packing by the use of
fast flows, it would be better to improve
the packing. Nevertheless, the fact
remains that no one has yet found a way
of packing a powder into a tube in a
completely homogeneous manner, and
until this is done, operation of columns
a t fast flows may well prove desirable.
LITERATURE CITED

(1) Bernard, R. A., Wilhelm, R. H.,

Chem. Eng. Progr. 46, 233 (1950).
( 2 ) Giddings, J. C., ANAL. CHEM. 34,
1186 (1962).
(3) Ibid., 35, 439 (1963).
(4) Ibid., 36, 1170 (1964).
(5) Giddings, J. C., J. Chromatog. 5, 46
(1961).
(6) Golay, M. J. E., Nature 182, 1146
(1958).
( 7 ) Halasz, I., Horvath, C., ANAL.
CHEM.35, 499 (1963).

(8) Harper, J. M., Hammond, E. G.,

Ibid., 37, 486 (1965).
(9) Hirschfelder, G. O., Curtiss, C. F.,
Bird, R. B., hfolecular Theory of
Gases and Liquids, Wiley, New York,
19.54
---.
(10) Huyten, F. H., van Beersum, w.,
Rijnders, G. W. A., Gas Chromatography 1960, Proceedings of 3rd
International Symposium, Edinburgh,
R. P. W. Scott, ed., Butterworths,
T,nnrlon.
1960.
~~,
..~.
.1) Kieselbach, R., ANAL. CHEM. 35,

1342 (1963).

. 2 ) Lapidus, L., Separation Processes

in Practice, R. F. Chapman, ed.,

Reinhold, New York, 1961.

.3) Littleyod, A. B., Gas Chromatography, p. 169, Academic Press,
New York, 1962.
.4) Littlewood, A. B., Proceedings of
5th International Symposium on Gas
Chromatography, Brighton, September
1964, A. Goldup, ed., Institute of
Petroleum, London, 1965.
(15) Present, R. D., Kinetic Theory of
Gases, McGraw-Hill. New York. 1958.
(16) Sternberg, J. C., Poulson, R. E.,
ANAL.CHEM.36. 1494 (1964).
~~.--,.
(17) Van Deemter; J. J., Zuiderweg, F. J.,
Klinkenberg, A., Chem. Eng. Sci. 5,
271 (1956).
RECEIVEDfor review June 7, 1965.
Accepted September 27, 1965. Troisihme
Symposium de Chromatographie en Phase
Gazeuse, organis6 par le Groupement
pour lAyancement des MBthodes Spectrographiques, B 1Ecole Polytechnique,
Paris, May 19, 1965.

Effects of Temperature on Carrier-Gas Flow Rates in

Packed and Capillary Columns
GARY HORLICK and W. E. HARRIS
Department of Chemistry, University o f Alberta, Edmonton, Alberta, Canada

H. W. HABGOOD
Research Council o f Alberta, Edmonfon, Alberta, Canada
The temperature dependence of
carrier-gas flow rate under conditions of constant inlet and outlet
pressures was measured for packed
and capillary columns of metal and
of glass using helium, hydrogen, and
nitrogen as carrier gases. With all
three gases over the temperature
range 25 to 200 C. the flow rate
as measured at a constant temperature may be taken as proportional
to ( l / T ) 1 . 7 for packed metal columns
and for capillary columns of either
metal or glass, and as proportional
to ( 1 /T)1.8 for packed glass columns.
The flow rates expressed at column
temperature will be proportional to
(1 / T ) O . 7 and ( 1 /T)O.*, respectivey. The
temperature coefficients for packed
columns are about 0.1 higher than
would be predicted theoretically.

the temperature of a column

for gas chromatography is increased, the carrier-gas flow rate will
decrease unless the flow rate is mainHEN

tained by increasing the column inlet

pressure. The principal effects of increased temperature are to increase gas
viscosity, to increase gas volumn, and
to change slightly the premeability of
the column. The overall effect of
temperature on carrier-gas flow rates is
of interest to the theory of programmed
temperature
gas
chromatography
(PTGC) and of concern to those doing
experimental work both in PTGC and
in isothermal chromatography a t different temperatures. Although a few
observations have been reported (2, ?),
this aspect of gas chromatography
seems to have been largely neglected.
When a column is operated so that the
volumetric flow rate is constant a t the
column outlet as measured a t a constant
temperature, the changes in gas volume
and viscosity with increasing temperature require an increased pressure drop
across the column. As a result, the
pressure-gradient correction factor decreases with increasing temperature.
I n contrast, when a column is operated
under conditions of constant inlet and

outlet pressures,
the
decreasing
volumetric flow rate as measured a t the
column outlet and a t a constant temperature results in increasing peak areas
(using a thermal conductivity detector).
In general, the time taken for analysis
and the efficiency of a separation can be
influenced by the changes in flow resulting from temperature changes. I n
this study the effect of temperature on
flow rate is investigated for several
gases of interest in chromatography.
THEORY

The flow of a gas through a packed

chromatographic column may be described by Darcys law (1, 5 ) ; for gas
flow in a porous bed

F_, -- - BdP
A
t dz

(1)

where F , is the volumetric flow rate a t

Present address, Department of Chemistry, University of Illinois, Urbana, Ill.
VOL. 38, NO. I, JANUARY 1966

Table 1.

Estimate of Thermal Expansion Contribution, Act, to the Temperature

Dependence of Flow Rate

One may then write for the flow rate

Metal column
(co per or
stainPess steel)
Linear coeff. of expansion
of column
Linear coeff. of expansion
of packing (estd.)
Temp. range

18

A a

AB

A c ~(packed column)
ACY(capillary column)
c

FOo 0: ( l / T ) *
where a is (1 N
ACY)

3.3

The preceding discussion has been

for packed columns; for capillary
columns similar expressions may be
derived. For conditions of practical
chromatographic interest the flow rate
in a capillary column is governed
(1, 6 ) by Poiseuille's equation

25'to 175' C.
0.05%
0.1%
-0.470
-1.2y0
+ O . 03
-0.004

assumed to be 0.40 in a packed column.

Touloukian ( 4 ) , are 0.664 for helium,

0.678 for hydrogen, and 0.718 for
nitrogen.
Insertion of the expression for viscosity from Equation 4 in Equation
3 yields

position z along the bed, A is the crosssection area of the bed, B is the:permeability of the bed, 7 is the gas viscosity,
and dp/dz is the pressure gradient a t
position z. For an ideal gas the flow
rate and pressure a t position z are
related to the flow rate and pressure a t
the column outlet, F o and p a , according

F -?!!!
"-LK

to
Fzps = Fopo

(2)

AB

--

[LlN
T

F,O

(5)

PdP

KTN

where F,O is the outlet flow rate corrected to 0' C..

For operation of the column a t constant inlet and outlet pressures, it is
seen from Equation 6 that the outlet
flow rate expressed a t a standard
temperature will vary with temperature
according to the following relation :

(4)
The quantity (ABIL) will change only
slightly with temperature. Assume
that its variation may be expressed in
the form

The values of N for three gases of chromatographic interest, calculated for the
temperature range 0' to 200' C. from
results summarized by Hilsenrath and

Table II.

Column material
Solid support
Length, m.
Nominal inside diameter, mm.
Mesh size of sup ort
Approx. liquid pgase
loading
Temp. range, ' C.
Inlet pressure, cm. Hg
a Fisher Scientific Co.
b Burrell Corp.
c Johns Manville co.

Borosilicate

Column Paka

Firebrickb

glass

Borosilicate
glass

Chromosorb Wc

2.5

2.5
30-60
20% Di-n-decyl

5
3C-60
2OY0 Carbowax
1500
25-175
15

5
6C-80
2075 Silicone
oll200
25-200
23

(2-175
48

ANALYTICAL CHEMISTRY

then Equation 9 should be applicable to

capillary columns as well as to packed
columns with ACYbeing the temperature
coefficient of the appropriate geometrical factor in the flow-rate expression.
For both packed and capillary
columns the coefficient ACY represents
the effects of thermal expansion of
column materials on flow rates. The
changes in R, A , and L with temperature may be calculated directly from
the linear expansion coefficients of t,he
column materials. According to the
Kozeny equation, B is proportional to
the cube of E , the void fraction in the bed
(1, 5 ) . The change in B may be estimated as three times the change in e,
which in turn may be calculated from
the differential expansion of the packing

Characteristics of Columns Studied

Copper

phthalate

If for capillary columns ACYis specified

by the equation

7Po

where L is the length of the column and

pi is the inlet pressure.
The change in viscosity of a gas with
temperature may be expressed as a
function of absolute temperature to the
power N :

which may be compared with Equation

3 for packed columns. By arguments
similar to those relating to packed
columns, the temperature dependence of
flow rate for constant pressure operation
is given as

(pv)

If the flow rate is expressed a t a standard

temperature--e.g., 0' C.-we obtain

Combination of Equations 1 and 2,

with subsequent integration, yields an
expression for the outlet flow rate :

(9)

+ +

x 10-6/0 c.
8 x lo-+/' C.

C.

8X
C.
25" t o 175' C.
0.3%
0.6%
0.75%
2.2%
-0.06
-0.02

AL. AR
AA'

Borosilicate
glass column

Stainless
steel
Capillary

Borosilicate
glass
Capillary

15.2

0.9

0.25

0.5

None'

None'

25-213
152

25-200
3.1

'

1.4

Table 111.

1.2

Temperature Coefficients of Flow Rate

a for Helium

Column

1.0

Co8%%n-Pak
Borosilicatefirebrick
BorosilicateChromosorb
Steel capillary
Borosilicate
capillary

0.8

Observed Predicted

Foo

a ( 1 /T)Ol

OL for Nitrogen
Observed Predicted Observed Predicted

CY

for Hydrogen

1.67

1.60

1.69

1.62

1.72

1.66

1.81

1.69

1.85

1.71

1.85

1.75

1.81
1.63

1.69
1.64

1.80

1.71
1.66

1.86

1.75
1.70

1.66

1.66

...
...

1.67

...
...

1.71

0.6

0.4

1000/T
Figure 1. Relation between flow rate
and temperature for a 2.5-m. X
2.5-mm. copper column packed with
20% didecyl phthalate on Column Pak
Inlet pressure 47.8 cm. Flow rate expressed at
column outlet pressure and a t room temperature
(25 C.)

and of the column. Perhaps the relative

expansion of the stationary liquid film
should also be allowed for, but for
packed columns a t least, the liquid is
largely on the interior surfaces and
should not have any significant effect
on the permeability. This assumption
is supported by the results of Reisch,
Robison, and Wheelock (6). Table I
summarizes the calculations for the
thermal expansion factors, A a , for
several columns.
With the values of N quoted above
and of A a from Table I, the predicted
value of the coefficient a in Equation 9
may be compared with that determined
experimentally.

measured with a thermometer checked

against calibrated ASTM standard
thermometers. Flow rates were measured with a soap-bubble flow meter
a t room temperature, the column and
flow meter being separated by a 30- X
1/4-inch cooling coil immersed in water
at room temperature. (This cooling
coil was found to be an unnecessary
refinement. Subsequent measurements
showed that after passing through
only a few centimeters of rubber tubing
the gas had come to within a few degrees
of room temperature. It was also
found that the gas a t the column inlet
attained bath temperature within the
first 2 cm. of packing.) The pressure
a t the column inlet was measured with
a mercury manometer, and for each
column the pressure was maintained a t
a constant value by means of an adjustment valve on the gas line and a
fixed mark on the manometer.
For each column-gas combination
the observed flow rates were plotted
on log-log paper against the reciprocal
of the absolute temperature.
RESULTS

Figure 1 shows the data obtained for

the copper-Column Pak column. The
exponent CY is the slope of the log-log
plot.
The values of a obtained for all
columns and gases are compared in
Table I11 with the predicted values.
The absolute standard error for a coefficient as evaluated from the experimental data and reported in Table I11
is 0.02--e.g., the temperature coefficient
for a packed copper column with
helium carrier gas is 1.67 f 0.02.

EXPERIMENTAL

Measurements of carrier-gas flow

rates a t different temperatures between
25 and 200 C. were made for three
packed columns using helium, hydrogen,
and nitrogen carrier gases successively
in each and for two capillary columns
with helium carrier gas. The column
characteristics are listed in Table 11.
The columns were mounted in a wellstirred oil bath, and temperatures were

DISCUSSION

The experimental values of the

temperature coefficient a for packed
columns, given in Table 111,tend to lie
significantly higher than the predicted
values by about 0.07 for the copper
column and 0.12 for the borosilicate
glass columns.
This difference is
almost certainly outside the limits of

uncertainty of determining a from the

measured flow rates, and in addition
the fact that the deviation is in the same
direction for ,all systems probably
indicates a real effect. We have no
suggestions to offer a t this time as to the
source of this small discrepancy for the
packed columns. For the capillary
columns the observed and predicted
values are in agreement. The average
differences in the observed coefficients,
a , for the three gases agree with the
differences in the coefficients, N (Equation 4),for the temperature dependence
of viscosity.
For practical chromatography it is
probably adequate to take a value of
1.7 for a with packed metal columns
and capillary columns and of 1.8 with
packed glass columns. These values
refer to flow rates expressed a t a
standard temperature. Flow rates
expressed a t column temperature will
vary with the reciprocal of the absolute
temperature to the 0.7 power and the
0.8 power. The first of these confirms
the coefficient 0.7 suggested by Golay,
Ettre, and Norem (3) for capillary
columns.
LITERATURE CITED

(1) Carman, P.

C., Flow of Gases

Through Porous Media, Chap. 1,
Academic Press, New York, 1956.
(2) Ettre, L. S., Kabot, F. J., ANAL.
CHEM.34, 1431 (1962).
(3) Golay, 31.J. E., Ettre, L. S., %em,
S. D., Gas Chromatography, M.
van Swaay, ed., p. 139, Butterworths,
London, 1962.
(4) Hilsenrath, J., Touloukian, Y. S.,
Trans. A.S.M.E. 76, 967 (1954).
(5) Purnell, H., Gas Chromatography,
Chap. 5, Wiley, New Yofk, 1962.
(6) Reisch, J. C., Robison, C. H.,

Wheelock, T. D., Gas Chrornatography, N. Brenner, J. E. Callen, and

M. D. Weiss. eds.. D. 91. Academic
Press, New York, 1962.

(7) Rowan, R., Jr., ANAL. CHEM. 34,

1043 (1962).

RECEIVEDfor review July 30, 1965.

Accepted October 18, 1965.

VOL. 38, NO. 1, JANUARY 1 9 6 6