betermination of Partition Coefficients by Automatic
qUilibrium Headsnace Gas Chromatography by Vapor
ase Calibration r
I~.Kolb* / C. Welter / C. Bichler
I~adenseewerk Perkin-Elmer GmbH, 7770 12/berlingen, Germany
l(ey Words
~as Chromatograph
bleads . Y
Pa..... Pace samphng
'tttlon coefficients
Multiple headspace extraction
SU~naary
~qu!librium headspace gas chromatography has been
c~JPlled to the determination of the partition coeffi-
teeLts,of volatile compounds in water-air systems. Only
sp~t~n~ques that are suited to a fully automatic head-
nil_~ Procedure usin~ the oneumatic headspace sam-
~""g-t,~ ,- -
s ~ . "-Camque have~ been considered. Particularly
- formation on the solubility and adsorptivity of hazard-
~pU~l,,e is the technique of "vapor phase calibration -
Cal:L Where an external vapor standard is used to
he,Orate the concentration of the volatile analyte in the
fou~Space, while the concentration in the sample is
.1,~.."'~,tram the difference in the total amount in the vial.
s technique is described in detail for 2-butanone in
~,.,cr. Finally, the water-air partition coefficients of
ie;e ral selected volatile compounds at different
e,. Peratures are listed together with their temperature
"UtlCtlons.
Introduction
~qr~!!!brium headspace GC (E-HSGC)is based on the
Ors~t,!~ of a volatile compound between the liquid
b,, ,,ad Sample and the surrounding gas phase followed
thJetransfer of the volatile compound in the gas phase to
nr,,gas chromatograph for analysis. The partitioning
~vh'~ess is governed by the distribution coefficient,
the ~' re.the case of gas-liquid systems is usually termed
nr^ Parhtion coefficient K. Although this distribution
~'eadess is the most important parameter, both for the
USpace sensitivity and for any quantitative calibra-
. ediCaledto Professor Leslie S. Ettre on the occasion of his 70th
birthday.
(2hrornatographia Vol. 34, No. 5-8, September/October 1992
0009-5893/92/9 0235-06 $ 03.00/0 9 1992 Friedr. Vieweg & Sohn Verlagsgesellschaft mbH
tion technique, the value of K is in general not known
for most practical headspace applications and with few
exceptions is not used for quantitative calibration tech-
niques. Its knowledge, however, would be very useful
for quantitative determination, particularly for compar-
ing headspace samples of differing sample volume. The
sample volume in a headspace vial influences the
headspace sensitivity as does the partition coefficient
and the sensitivity depends strongly on the value of K
and the phase ratio 13,as discussed recently by Ettre and
Kolb [1]. The main application of E-HSGC is in the
quantitative analysis of volatile compounds in liquid or
solid samples, but there are also other interesting ap-
plications for the determination of physico-chemical
data, based on phase partitioning systems. It is of inter-
est, particularly for environmental analysis to obtain in-
ous compounds in water and soil and their release to the
atmosphere. Also for biomedical purposes, especially
industrial hygiene, the partitioning of noxious com-
pounds between blood and air in the lung needs to be in-
vestigated for measuring BEI values (biological ex-
posure index) or for BAT values (biological tolerance
values for working materials). Owing to this increasing
interest in such partition systems, there is a need to
measure a large number of partition coefficients and an
automatic technique is highly desirable for this purpose.
Rohrschneider [2] first used an automated headspace
instrument to measure the partition coefficients of six
reference compounds with different polarities in 80 sol-
vents. These measurements were performed on a rela-
tive basis, a particular system with a known partition
coefficient (n-octane in squalane/air at 25 ~ being
used for reference purposes. This procedure, however,
is not generally applicable, as it depends on the
availability of data for any desired temperature and
their accuracy depend on the reproducibility of the ref-
erence system.
The partition coefficient is the ratio of the concentra-
tion of a volatile compound in two phases and although
it can be derived from quantitative analysis of both
phases, there is a variety of procedures in the literature
which are briefly discussed below.
Of all these techniques the equilibrium headspace
method offers the best prerequisites for automation,
235
and has already been applied with various modifica-
tions. In looking for the best procedure for an
automated technique, none of those briefly discussed
below have been found ideal, i.e., sufficiently simple for
this purpose, hence a new version of the headspace tech-
nique, termed "vapor phase calibration (VPC)", had
therefore been developed.
Jones et al. [3] determined the partition coefficient of
ethyl acetate in water at O ~ from independent
analyses of the total amount of ethyl acetate in the
headspace vial and from the amount in the gas phase9
The gas phase was calibrated against an external liquid
standard injected by a microsyringe. Kieckbusch et al.
[4] also used a microsyringe but for the analysis of the
liquid phase. The combination of headspace injection
by gas syringe and liquid injection by microsyringe
needs very careful calibration between both sampling
devices and is hard to automate 9
Two determinations using different sampling techni-
ques can be avoided if the same instrumentation is used
and if, in the prevailing two-phase system, one of the in-
fluencing parameters is varied. For such an approach
the total amount in the two-phase partition system can
be altered if, after headspace analysis, the equilibrated
gas phase is removed and replaced by fresh inert gas.
After re-establishing equilibrium, the amount of analyte
in the gas phase becomes smaller. The partition coeffi-
cient can be derived from this difference9This technique
was first applied by McAuliffe [5] followed by others
[6-8] but needs a gas-tight syringe to separate mechani-
cally and completely both the liquid and the gas phases9
It is obvious that these techniques, although they have
become quite popular, are difficult to automate 9
The gas phase, however, need not be replaced totally to
shift the equilibrium. It can be removed partially simply
by venting a pressurized headspace vial to atmosphere.
It is thus possible to derive K from the resulting dif-
ferences of the peak areas in the sequential headspace
chromatograms determined in a similar way [9]. This
MHE (multiple headspace extraction) procedure,
described in detail in a recent paper [10], can be carried
out automatically, but again various practical variants
are possible. One of these variants is used here in com-
bination with MHE, where the sample volume and thus
the phase ratio is varied.
It is obvious for the purpose of automation that the
same sampling technique should be used. With head-
space sampling instrumentation only the gas phase is ac-
cessible for analysis of the volatile analyte. Whenever
physico-chemical data are determined, the original
composition of the sample is usually well defined, In the
case of partition coefficient determination, it is the gas
phase composition which is determined analytically.
From the total amount of volatile analyte, exactly
known from the preparation of the original sample, the
concentration in the liquid or solid sample under the
prevailing temperature conditions is derived from the
difference. Quantitative analysis of the gas phase can
very simply be carried out by external standard calibra-
tion. This procedure, which therefore is termed here:
236
"vapor phase calibration t e c h n i q u e - VPC", f u l f i l ~
requirements for automation and is described ano ,,is
cussed in detail below. It should be mentioned that tt~e
. variant
a specxal . of. the general
. prmc~ple wher e 1( Call-e~ ~
derived from the resulting difference in peak atto6
when the sample volume is varied, since with this *"
technique the volume of the sample matrix in the refer'
ence system becomes zero9
"Vapor Phase Calibration (VPC)"
Technique
The VPC technique is based on the nrinciple that t!eo
9 .
concentrataon of the volatile analyte r.]n the g a s P_I~a~'
/f
can be determined by external standard calibrations'^',.
the total amount present in the vial is known, the co~
centration in the sam P le P hase at e q uilibrium is fott]j~
from the difference.
This technique, where the distribution of a volatile c.~ I
pound between a two-phase system in a headspaCe *.'~s
is determined bv usinz a pure vapor as a reference, w~
. . . . . tllu
originally applied and performed with an autOna~...
- -coP"
headspace
. . . .
sampler by". Kolb for
.
the
.
determination u'.,~l
9 l/,j
tnbutlon coefficients m gas-hqmd [11] and gas-soltd [..,
systems and later by Schoene et al " [13] to deterr~, iJI0
solubility coefficients of vapors in both solid and llq.,
polymers. Although these investigations have t,
. . . dtstnbutxon
focused on non-hnear . as found both in coO" .,
centrated solutions and in gas-sohd adsorption sys~ .,,
of partition coefficients. After having screened alte[~.~
tives, it has been found very useful for a fully automat~
procedure. Analytical determinations are p e r t o r w
automatically under constant instrumental cond~t~o,
while the necessary variation of the sample compositl~
is manual when preparing samples.
Fundamentals of VPC
When an amount m ~ of the volatile analyte is injeCte!
into a vial containing the sample matrix with the voltlt~~
Vs, it is distributed between the gas phase and t
sample as described by Ea (1), and the amount ~n ~,
sample (ms) after equilibration results from the o,
ference, Eq. (2):
CS ms VG (1)
CO mo VS
ms = m ~ - mo (~)
The peak area As from a headspace analysis co[
responds to the gas phase concentration C6 with @ ~
calibration factor which includes instrumental Park
meters such as injected volume of headspace gas ~tl0
detector response but no specific headspace paramete~
such as phase ratios or matrix effects. @ can therefo rel~
Chromatographia Vol.34, No. 5-8, September/October 199~
 ~etern~ined simply by external standard calibration with Sample Preparation and Results
Por Sample of the anab,te For this nurt~ose a small
Four vials, each containing 3.0 mL water, were prepared
~OUnt of the analyte, e.g. m~ * is ~njected into an
9-ply vial either a - and 2.0, 3.0, 4.0 and 5.0 laL of M E K added by means of
Whe- 9 s the pure compound or as a solutmn a microsyringe. Identical volumes of M E K were added
tUrore ~t vaporizes totally under the prevailing tempera-
to the empty reference vials.
.calib,^.
.e conditions 9 T h. .~. . .r .o.~. n. l t i n ~~,,~
n e a k area A C is used to
in E ,~te the concentration CG ~ of the analyte vapor as The results of the VPC-technique according to Eq. (9),
q. (3), from which Eq (4) is derived. where the partition coefficient is determined from the
area difference of each pair of samples, are listed in
C~ ~- m~ / V v = @ A c (3) Table I.

rn~ = @ AC V v (4)
~'he - Table |. Water-air partition coefficient K of MEK at 70 *C by VPC,
COncentration ,~f
t0110..... , , , &~ e a n a lyt e in the g as P hase , C G, Eq. (9).
"~ xrom Eq. (5) and the amount, mG, by Eq. (6):
A d d e d volume of M E K [/aLl 2.0 3.0 4.0 5.0
CG= m G / V o = @ A s (5) peak area A c 35202 51357 66903 83657
peak area As 4868 7005 9207 11434
k'IG .~ @ As VG (6)
K 47.29 48.04 47.58 47.92
~'~ am~ ms, of analyte in the sample phase is deter- mean value for K: 47.70 +/- 0.7 % C.V., N = 4
ecl from the difference, Eq. (7):
rns = @ [Ac V v - As VG] (7) conditions: vial volume Vv = 22.3 mL, matrix volume Vs = 3.0 mL,
gas volume Vc~ = 19.3 mL.
Introduction of E^s r2~ and (73 in E (1) gives Eqs. (8)
antlrm . q.~ J , , q"
t~) Where @ is finally eliminated.
Results by Linear Regression
K = ms VG @[AcVv-AsVG]VG
- (8) Whenever differences in peak areas are used, it is good
Vs mG @ A s VG Vs
practice to check whether both areas are still in the
range of headspace linearity; this is a linear function of
K -- A c V v - As VG (9) concentration in the sample and its partial vapor pres-
As Vs sure expressed by a constant K, independent of con-
centration. Having the wide range of partition coeffi-
~,ie~cpC technique has been applied to 2-butanone
cients in mind and the corresponding variation of
be~ K) in water at 70 ~ This model system was chosen solubility and resulting concentration, it is advisable to
res~ use the partition coefficient of roughly 47 is a rep- include such a linearity test e.g., by linear regression.
o;^entative value in the wide range of partition coeffi-
~'Cnts The data in Table I can also be subjected to linear
f0r L from K = 0 to 104, where a reasonable volatility regression. K is derived by replacing AC and As in Eq.
"eadspace applications ends. (9) by the values of the corresponding slopes ac and as
according to the modified Eq. (9a):

EXPerimenta ! ac V v - as V o
K= (9a)
The. as Vs
fie- wa!er-air partition coefficient of M E K at 70 ~ was
terramed by VPC under the following conditions. ac = slope of regression line for vapor standard
as = slope of regression line for aqueous sample

Instrun~ental Conditions The necessary values to derive K from the data in

Table I and Eq. (9a) are the slopes of the regression line
t" e tig.tio. we e.rriod out wit
g"a~'lO.1 Automatic Headspace Sampler and Model 8700
a erk n- mor from the reference 'empty' vial (ac - 16091.10; correla-
tion coefficient r = 0.99990) and that of the sample vial
stY. e~romatograph. G C conditions: 2 m x 1/8" packed (as = 2190.0; r = 0.99974) together with the volume of
(2~Inless steel column, 1 5 % Carbowax 1500 on the sample matrix (Vs = 3.0 mL) and the vial volume
l~i~~ W, NAW, 60-80 mesh at 60 ~ detector: (Vv = 22.3 mL). From these data K = 48.2, which is in
Jec,'" Headspace conditions: thermostat time 60 rain, in- good agreement with the mean value of K = 47.7,
sa;l~n time 0.05 rain, pressurization time 2.0 min, +/- 0.7 % C.V. listed in Table I.
Pie matrix: deionized water.

~,L.YOther amount, not necessarily m* of the analyte can be used. Accuracy of VPC Technique
t].~s makes the VPC technique flexible enough to adjust the some-
-ues big differences in peak areas for the external vapor standard The accuracy of the VPC technique was tested by two
and the Sample in the case of a high K value. independent procedures. The M H E technique was used

{2hr~ Vol. 34, No. 5-8, S e p t e m b e r / O c t o b e r 1992 237

to determine the partition coefficient of MEK at 70 ~
and the result compared with that obtained by VPC. For
a quite different approach the partition coefficient of
cyclohexane in squalane was determined by VPC and
compared with the corresponding literature value.
Linear Regression Calculation by the M H E Technique
The value of K for MEK in water-air at 70 ~ deter-
mined by VPC, was compared with that obtained by the
M H E technique. The fraction of the gas phase removed
from a pressurized headspace vial during the M H E pro-
cedure depends on the pressure difference between the
headspace pressure Ph and the final lower pressure, for
which atmospheric pressure Pa is usually assumed ac-
cording to Eq. (10). If this process of venting the pres-
surized headspace gas is repeated several times the
decrease in the peaks is described by the slope q
(q = An+l/An). The relationship of q to K, the phase
ratio, 13, and the pressure difference between the pres-
sure of the pressurized headspace vial Ph and the end
pressure Pa, is given by Eq. (10):
( K / ~ ) + (Pa/Ph)
q = (K/I~) + 1 00)
K can be derived if the absolute values of both Pa and Ph
are known. Usually atmospheric pressure is assumed for
Pa. Onda et al. [14] have used this absolute procedure to
determine distribution coefficients of flavor compounds
in solid coffee samples. Aqueous samples, however,
have to deal with the problem that the end pressure
after venting the pressurized gas phase in the vial is the
prevailing water vapor pressure rather than atmos-
pheric pressure. Since other parameters such as time of
venting, may also affect the volume of the expanded gas
phase due to the internal water vapor pressure, the cal-
culation based on Pa/Ph was not considered reliable
enough. These problems with internal water vapor pres-
sure can be eliminated if, again, two determinations are
carried out under the same instrumental conditions.
The reference vial must contain an aqueous sample to
avoid any additional corrections for the water vapor at
the prevailing temperature. Two equations are obtained
with the same Pa/Ph, while the phase ratio 1~ is varied.
Two vials with 1.0 mL and 5.0 mL water respectively
were prepared to which the same volume of 3.0 pL
MEK was added. The result of a 4 step M H E procedure
is shown in Figure I together with the analytical and in-
strumental conditions. From the slopes of both sere/log
regression lines the partition coefficient was determined
by Eq. (10) with the same pressure drop (Pa/Ph) for both
measurements. K = 46.5 was obtained, which agrees
fairly well with those from VPC (47.7, +/- 0.7 % C.V and
48.2 by linear regression). This agreement is surprisingly
good, considering the small difference in the slopes, par-
ticularly the very slow decrease in the 5 mL sample with
q = 0.96, where the differences in two successive areas
approaches the statistical variation. This, therefore,
demonstrates that the M H E approach works well in
principle, but is restricted to partition systems with
238
A
3 354:3 I~'~
Flllht
B
Figure 1
Determination of water-air partition coefficient (K = 46.5)otl011~
at 70 *C by a 4-step MHE procedure according to Eq. t j i~'tle5
derived from different slopes q (q = An,l/An) of regressi~
from two vials with different samples volumes but containing~'
total amount (3/aL) of MEK.
A VsA = 1.0 mL, VGA = 21.3 mL, qa = 0.815, 1~a = 21.3
B Vs8 = 5.0 mL, Von = 17.3 mL, q~ = 0.959, [5n = 3.46
o,ds
small K values (K < 100). The slope, however, dep~o
not only on K but on 15 as well and for systems wl~.
high K value, e.g. alcohols-water, a lower VS is reC~'~
mended to decrease q. This can be done without ~"~
loss in headspace sensitivity [1].
In contrast to VPC, where quantitative analysis bY~.~e
ternai standard calibration is carried out using a v~os
vapor standard, the M H E technique applied here ~'~
two aqueous samples with different phase ratios.
Comparison with Literature Values
9 I b~sis
Although partition coefficients are the phys~ca d"
and thus the most important parameter for any heo~
space application, it is surprising to find little publiS~us
data9 The reason might be that most determinatt~
based on headspace measurements have been carrt~
. J {fl~
out usmg a manually-operated, gas-tight syringe ano ,t
measurements have thus mainly been carried out [.
room temperature. Most practical headspace aPpllC.~.
tions, however, are nowadays carried out with attt
mated headspace samplers and at elevated tenaper~"
tures to increase sensitivity. The data are, ther ef~
hardly comparable.
A different technique has been used here to test th.e.~d0
curacy of VPC with proven literature values. Partlti A
coefficients have been used for this purpose deteraaiPT~
by chromatographic measurements from retention dat,.'
Very precise measurements were carried out in the e~tfl)
days of GC by Porter, Deal and Stross t[151 9 us1 g..|t'
squalane column. From their data the partition coefA
-- t0
cient of cyclohexane in squalane has been cornPaf,, s
with that obtained by VPC. The conditions and rest~t~
are listed in Table II.
Chromatographia Vol. 34, No. 5-8. September/October 299~ :
"rahleII. Partition coefficients K of cyclohexane in squalane. 8
Ln K
~ [15]~ headspace gas chromatography* t 7
1~5 KK=ll2 K = 109.6, +/- 0.4 % C.V., N = 3 6
*C = 60.1 K = 61.2, +/- 4.2 % C.V., N = 3
5
Itgas .Phase: helium.
~rad~Pace gas chromatozraphy by vapor phase calibration ac- 4
10~otng !o Eq. (9); system: "2.6mL squalane + 2.0 ~L solution
Stala.~':ycl~ in toluene in 22.3 mL headspace vial, reference
air, uara: 2.0 ~tL eyclohexane solution in empty vial, gas phase: 3

cTyhc0agreement b e t w e e n the partition coefficients for

n: aexane in s q u a l a n e d e t e r m i n e d by different tech- 4'0
3'.z
sb
t
6'0
~
3.o
7'0
1.9
8b
i
3
toc]
3
[ ozra
t~ques is SUrprisingly good, thus showing that the V P C
eehrlique is in fact a reliable m e t h o d . Figure 2
Temperature function of water-air partition coefficient K of ethanol
(1), MEK (2), and diehloromethane (3) from 40 ~ - 80 *C by linear
TenlPerature D e p e n d e n c e of Partition Coefficient regression: y = ax + b with y = In K, x =lfI" (T in K), a = A/R
and b = -B according to Eq. (11).
~he d e p e n d e n c e of K on t h e a b s o l u t e t e m p e r a t u r e , T, is
g~vea [7] by Eq. (11): ethanol(l) MEK (2) CHzCIz(3)
a 5083 4298 2778
laK- A b -9.02 -8.62 -7.14
RT
B (11) r(corr, eoeff.) 0.9992 0.9994 0.9902

f~erttransformation (cf. Figure 2) K can be d e r i v e d for *conditions: sample volume Vs = 3.0 mL, vial volume Vv = 22.3 mL,
ern- m p e r a t u r e b y linear regression, at least in the 3.0/.tL mixture of three compounds added.
sev~erature r a n g e w h e r e K has b e e n d e t e r m i n e d at
ro, er.al t e m p e r a t u r e s . A n e x a m p l e of the linear
~ a h o n s h i p in a s e m i l o g plot is s h o w n in Figure 2 with
t h e V P C t e c h n i q u e with linear r e g r e s s i o n calculations.
~ a r t o l , M E K a n d d i c h l o r o m e t h a n e at five t e m p e r a -
A m i x t u r e o f these c o m p o u n d s was p r e p a r e d and 1.0,
~es in the r a n g e 40 ~ - 80 ~
3.0 and 5.0 p L w e r e injected into 5.0 m L of d e i o n i z e d
water. T h e resulting partition coefficients are listed in
T a b l e III. T h e t e m p e r a t u r e d e p e n d e n c e was deter-
SUPplemen t mined. B o t h coefficients a a n d b f r o m the linear regres-
sion, as s h o w n in Figure 2, are also listed to allow recal-
~v~ ition coefficients of several volatile c o m p o u n d s
culation for o t h e r t e m p e r a t u r e s .
tier" Wtdely differing solubilities in w a t e r h a v e b e e n
errained in t h e t e m p e r a t u r e r a n g e 40 ~ to 80 ~ by

Table III. Partition coefficients of selected compounds in water-air system.

Partition coeff. K regression coeff.

compound 40 *C 60 *C 70 ~ 80 *C b* a*

tetrachloroethene 1.48 1.27 0.78 0.87 -5.13 1736

1,1,1-trichloroethane 1.65 1.47 1.26 1.18 - 2.52 950
o-xylene 2.44 1.31 1.01 0.99 -7.61 2647
cyclohexane 0.07 0.05 0.03 0.02 -12.9 3228
n-hexane 0.14 0.043 0.012 - -28.6 8375
ethyl acetate 62.4 29.3 21.8 17.5 - 7.28 3563
n-butyl acetate 31.4 13.6 9.82 7.58 - 9.27 3971
toluene 2.82 1.77 1.49 1.27 -6.03 2198
isopropyl alcohol 825 286 179 117 - 10.6 5413
methyl isobutyl ketone 54.3 22.8 16.2 11.8 - 9.56 4237
dioxane 1618 642 412 288 - 7.94 4798
n-butanol 647 238 144 98.9 -10.2 5234

*cf. Figure 2.

Qhr~ Vol. 34, No. 5-8, September/October 1992 239

List of Symbols and Abbreviations
Ac Peak area of ana~yte from external vapor standard
As Peak area of analyte from headspace sample
ac slope of regression of analyte from external vapor
standard
as slope of regression of analyte from sample vial
Cs Concentration of analyte in sample
Co Concentration of analyte in gas phase of sample
vial
C o C Vapor concentration of analyte in external vapor
standard
K Partition coefficient
m ~ Total amount of analyte in vial
m o Amount of analyte in gas phase
ms Amount of analyte in sample
Pa Atmospheric pressure
Ph Headspace pressure in pressurized vial (Ph = Pa +
PGC) with POC as excess pressure given by pres-
sure of G C
VG Volume of gas phase (headspace) in vial
Vs Volume of sample in vial
V v Volume of headspace vial
13 Phase ratio ([~ = VGIVs)
@ Calibration factor
References
[1] L.S. Ettre, B. Kolb, Chromatographia 32, 5 (1991).
/2] L. Rohrschneider, Anal. Chem. 45, 1241 (1973)- :,~
[3} W.J. Jones, M.J. Egoville, E.O. Strolle, E.S. DellamOn~'
J. Chromatogr. 455, 45 (1988).
[4] T.G. Kieckbusch, C.J. King, J. Chrornatogr. Sei. 17, 273
(1978).
[5] CA. McAuliffe, Chem Tech, 46-51, Jan. 1971.
[6] A.G. Vitenberg, B.V Ioffe, Z.S. Dimitrova, t.L. Btaae~
J. Chromatogr. 112, 319 (1975). ja~lalr
[7] B . V..Ioffe,
. . A G Vitenberg, Head-Space Analysis an~ ~."~c,',.
Methods ~,nGas Chrot~atography. John Wiley & 5on~,
York, NY, 1984. Chromatogr.~gl,
[8] R. Guitart, A. Puigdemont, M. Arboix, J.
271 (1989). ,det
[9] P Vospisil, B. Kolb, paper presented at the "At~ce'_'0r
Kolloqu[um tiber die Gaschromatographische Damplr~
analyse", l~ad Nauheim, 3.-4. October 1983.
[10] B. Kolb, L.S. Ettre, Chromatographia 32, 505 (1991)'
[11] B. Kolb, J. Chromalogr. 112, 287 (1975). ,-,~,001~'
[12] B. Kolb, in: B. Kolb (Ed,), Applied Headspace Oas~;080/.
tograpy. Heyden & Son Ltd., New York, NY, pp. 1-tl t~',55,
[13~ K. Schoene, J. Steinhanses, A. KOnig, J. Chrornatog"
57 (1988).
[14] N. Onda, A. Shhwhara, 1-1. Ishii, A. Sato, HRc&CC 14,
357 (1991). Sor ~'a,
[~5] P.E. Porter, C.H. Deal, EH. Stross, J. Am Chem. '
2999 (1956).

Received: April 1,1~2

Revised manuscript~ ~
received: May 8, ~9i9~
Accepted: May 15,

240 Chromatographia Vol. 34, No. 5-8, September/October 199~