In-tube solid phase micro-extraction gas chromatography of
volatile compounds in aqueous solution
Boon Chong Dennis Tan,a Philip J. Marriott,*a Hian Kee Leeb and Paul D. Morrisona
a
Royal Melbourne Institute of Technology, Department of Applied Chemistry, GPO Box
2476V, Melbourne, Victoria 3001, Australia. E-mail: [e-mail]
b
National University of Singapore, Department of Chemistry, 10 Kent Ridge Crescent,
Singapore 119260, Singapore
Received 2nd March 1999, Accepted 30th March 1999
This paper describes the use of conventional coated capillary gas chromatography columns for sorption of organic
solutes from aqueous solution, with subsequent gas chromatographic analysis. The essential principles are similar
to those of solid phase extraction (SPE) and solid phase micro-extraction (SPME); this approach may be referred
to as in-tube solid phase micro-extraction (ITSPME). The technique was evaluated using toluene in water as the
initial test solute, and a mixture of BTEX solutes (benzene, toluene, ethylbenzene, xylenes) in Milli-Q water was
used to further characterise ITSPME. A 1 m length of capillary GC column was used for sorption of analytes from
aqueous solution passed through the capillary by using nitrogen pressure. Collection of small fractions of aqueous
solution issuing from the capillary enabled a sorption profile to be generated, with initial fractions depleted in
analyte. A Boltzmann curve could be fitted to the sorption profile data, exhibiting good agreement with
experimental data. For recovery of sorbed toluene, a single 100 mL aliquot of hexane was passed through the
column as a stripping solvent. The back-extraction step was quantitative. Equilibrium extraction of solutes shows
that the total amount of recovered solute is proportional to its initial concentration in the extracted aqueous
solution and allows distribution constants to be readily estimated. For BTEX solutes, K values were similar to
those reported for SPME and literature Kow values. For toluene, log K decreases from 2.47 to 1.48 when the
sorption column temperature increases from 20 to 30 °C; adding salt or reducing the pH of the aqueous solution
increases the degree of extraction of phenols, agreeing with general considerations on solute partitioning
behaviour.
at equilibrium is directly related to its concentration in the
Introduction
sample solution, which can be described in a similar manner to
SPME1,4 by equation (1):
Solvent free sample preparation methods or those employing
less organic solvent are becoming increasingly important1 and
KVsampleCsampleVs
may induce a major change in analytical methodology.2
Ms = (1)
KVs + Vsample
Practical alternatives to existing sample preparation methods
may therefore need to be formulated. Solid phase extraction
where Ms is the mass of an analyte sorbed by the stationary
(SPE) and solid phase micro-extraction (SPME) have emerged
phase, Vs and Vsample are the volumes of the stationary phase and
as efficient, popular alternative extraction techniques.3 A broad
the sample passing through the capillary column, respectively,
array of applications have been reported, such as determining
octanol water partition coefficients,4 detecting BTEX in water5 K is the partition coefficient of the analyte between the
stationary phase and the sample matrix, and Csample is the initial
and analysing pesticide solutions. Gas-phase extraction and a
concentration of the analyte in the sample in mass per unit
variety of headspace applications are available. Whilst these
volume. As in SPME, if Vsample is large (Vsample > KVs), the
show that SPE and SPME are good alternative methods, basic
amount of the analyte extracted8 is:
principles still require evaluation.6 An effective alternative to
these techniques is presented in this paper, in-tube solid phase
Ms = KCsample Vs (2)
micro-extraction (ITSPME). Recently, Pawliszyn and co-
workers have reported the automation of a similar approach Following aqueous sample extraction, the sorbed analyte can be
involving coupling of the extraction column to high-perform- stripped from the stationary phase with minimum amount of
ance liquid chromatography (HPLC).6 By extension, ITSPME organic solvent, and the extract analysed by GC-FID. In
should also have the potential to be coupled to other analytical addition, the aqueous solution can be monitored both prior to
instruments, such as gas chromatography (GC) or capillary extraction and also in the stream which issues from the
electrophoresis (CE). extraction capillary. The problem of analysing aqueous samples
ITSPME utilizes similar principles to SPME. Exhaustive has been addressed.9 Conceptually, ITSPME should preserve
extraction might not occur, with equilibrium partitioning of the the advantages of SPE and SPME, but may offer potential
analyte between the sample solution and the extraction medium benefits regarding quantitation and automation. ITSPME may
(the stationary phase coated to the inner walls of the extracting use conventional capillary GC columns for sorption, with
capillary column). The sorbing phase can be selected according thermally stable, non-extractable bonded phases making the
to the type of analytes to be determined, e.g., a non-polar phase phase more robust than thick film phases on fibres. Conversely,
if the analyte of interest is non-polar.1,7 In ITSPME, analyte thick films are not readily prepared for wall-coated capillary
solution is passed through the capillary at a reasonably slow columns. Thus, ITSPME possesses complementary advantages
flow rate. The amount of analyte sorbed by the stationary phase to SPME, as outlined in this paper.
Analyst, 1999, 124, 651 655 651
analytical results of the collected aliquots could be used to
Experimental
generate a sorption profile. A separate standard was prepared in
water to serve as a reference solution against which the
Reagents
collected aliquots could be compared. The extraction experi-
ment may be conducted with different linear velocities of
All reagents used were of analytical reagent grade. Benzene,
aqueous solution passing through the capillary column, by
toluene, ethylbenzene, o-xylene, m-xylene, p-xylene, m-ni-
means of controlling the head pressure of nitrogen in the vial
trophenol, p-cresol, p-tert-butylphenol, 2,4-dichlorophenol,
using control gauge pressure.
2,4,6-trichlorophenol, acetone, hexane, methanol, potassium
chloride, sodium chloride, concentrated hydrochloric acid and
Back extraction. After a specific volume of solute in water at
2-hexanone were purchased from BDH (Sydney, Australia) or
a fixed concentration was passed through the extraction
Aldrich (Sydney, Australia); Milli-Q water (Millipore, Bedford,
capillary column, the capillary was dried with a nitrogen flow
MA, USA) was used throughout. The following stock standard
and then a minimum volume of organic solvent was passed
solutions were prepared in water and hexane (BTEX analysis)
through the extracting capillary to strip the sorbed solute. A
or methanol (phenols analysis): 100 mg L21 toluene, 100
suitable internal standard was added, and the solution was
mg L21 ethylbenzene, and mixtures of 100 mg L21 of each
analysed by using GC-FID. The result was then compared with
component of BTEX and 100 mg L21 of each component of
the calibration plot from a series of standard solutions in the
phenols. 2-Hexanone (100 mg L21) in hexane solution and 100
same organic solvent to estimate the amount of recovered
mg L21 o-xylene in methanol were also prepared. In order to
solute.
ensure that the chemicals were adequately dissolved in the
solvent, 0.5 mL of acetone was used to dissolve the reagent first,
followed by dilution as required with solvent Milli-Q water or
hexane.5,10 The effect of pH and salt on efficiency of extraction
Results and discussion
in the ITSPME technique was examined. A pH 2 buffer was
prepared with 25 mL of 0.2 m KCl and 6.5 mL of 0.2 m HCl in
Sorption profiles
100 mL of water, and saturated salt solutions were prepared
with NaCl.
Using forward extraction, four sorption profiles of aqueous
toluene passing through the 1 m capillary GC column at flow
rates of 20, 30, 50 and 70 mL min21 were obtained. The 100 mL
Instrumentation
fractions collected were analysed by GC-FID after addition of
the internal standard. Each flow rate was repeated 3 times and
A Shimadzu GC 17A with an autosampler and FID detector
Fig. 2 presents representative results from individual studies.
(Shimadzu Scientific Instruments, Rydalmere, NSW, Australia)
was used for all gas chromatographic analyses. The fused silica
capillary column used for GC was 30 m 3 0.25 mm with 0.25
mm film thickness BPX5 phase (SGE International, Ringwood,
Australia). The conditions for the analysis were as follows.
BTEX analysis: column flow approximately 1.9 mL min21;
linear velocity approximately 35.6 cm s21; column oven at
93 °C (this temperature allowed acceptable analysis time
without causing the toluene peak to overlap with trace acetone
solvent peak); injection port at 250 °C with split injection (split
ratio 1 : 20); FID detector at 320 °C. Phenols analysis: column
flow approximately 1.4 mL min21; linear velocity approx-
imately 30 cm s21. The injector, operated in splitless mode, was
maintained at 200 °C; the FID detector was at 275 °C. The
temperature program was 50 °C for 1 min, ramp to 190 °C at
10 °C min21, final hold time 1 min at 190 °C.
Fig. 1 Experimental set up for the ITSPME technique.
In-tube solid phase micro-extraction (ITSPME). To
perform extraction of BTEX and phenols using ITSPME, 2
different types of capillary GC columns were used. The first was
1 m long, 0.25 mm internal diameter with 3.5 mm thick BP1
(100% methylsiloxane) stationary phase, while the second was
1 m long, 0.32 mm internal diameter with 1 mm thick BP20
(polyethylene glycol) stationary phase (both columns from SGE
International, Ringwood, Australia). Nitrogen gas was used to
provide head pressure to the sample vial to force the aqueous
solution through the capillary (Fig. 1). Most extractions were
carried out at 20 °C and the capillary may be immersed in a
water bath for temperature control. Two forms of extraction
were performed, forward extraction and back extraction, as
described below.
Forward extraction. A volume of solution (in water) was
prepared at the desired concentration from the stock solution. It
was then forced through the capillary by applying nitrogen head
Fig. 2 Experimental sorption profiles for toluene at different flow rates.
pressure. Solution was passed through as a continuous stream,
Lines of best fit based on a Boltzmann distribution are shown for each set
and collected in separate vials in 100 mL volumes or fractions.
of data. (2) 20 mL min21; (!) 30 mL min21; (/) 50mL min21; (&!) 70
A suitable internal standard was added in the fractions. The mL min21.
652 Analyst, 1999, 124, 651 655
Variation in peak areas or area ratios may arise from (i) injection apparently the methyl siloxane stationary phase behaves
volume uncertainty, or (ii) variation in volume of either the similarly to octanol in the octanol water partition experiment.
collected toluene fraction or the added ethylbenzene volume.
The toluene volume has the greatest uncertainty since the 100
mL volumes collected could not be controlled with precision. An Dependence of ITSPME on extraction temperature,
alternative procedure would involve weighing the collection solution pH and salt content
vial. For the 20 mL min21 flow data, the absence of toluene in
fractions up to fraction 4 is noted; fraction 13 and later fractions The back extraction procedure for 20 mg L21 toluene was
have essentially the same level of toluene as that in the reference repeated at least 4 times with the same piece of capillary to
solution, shown as a broken line in Fig. 2, indicating 100% ensure that the extraction was consistent. The extraction was
 breakthrough . As the flow rate, and hence velocity, of the then carried out at a temperature of 30 °C, with a decrease in the
aqueous toluene through the extracting capillary increases, amount of toluene sorbed expected and confirmed (Fig. 3).
extraction is less complete for the early collected fractions, and Increasing the temperature decreases the analyte Cs; in other
traces of toluene could be detected in fraction 1 of the sorption words, there is less affinity for the stationary phase. Log K at
profiles for faster flow rates. The curves also become less steep 30 °C is estimated to be 1.48 and a temperature increase of
than those at slow velocities. Experimental uncertainties meant 10 °C decreases the value of K by a factor of 10. This result is
that data for the extraction profile did not precisely fit a smooth as expected from chromatographic results, where higher
curve; however, a Boltzmann-type curve could be fitted to temperature gives a smaller retention volume in GC and in
experimental data, as seen for the sorption profile data in Fig. 2. HPLC, and so smaller k and K values. Since salt affects solute
The steepness of the curve increased for slower flow rates. solubility, a further study to test ITSPME for extraction of
Integration of the Boltzmann equation can be used to give an phenols from aqueous solution showed that both saturated salt
estimate of the total sorbed solute amount. solution and lowering the solution pH increase the extent of
extraction by up to 10 20 times. For example, from Table 3
data, the peak for 2,4,6-TCP increases by about 3.5-fold, and p-
tert-butylphenol by about 25-fold. Fig. 4 is a representative GC
Determination of partition coefficient, K, for toluene
trace of the extraction of the saturated salt, pH 2 buffered,
between water and BP1 phase
aqueous solution. Since pKa values of phenols are !7, there is
The 1 m BP1 capillary has a stationary phase volume of 2750
nL. Each 100 mL fraction of aqueous solution passing through
Table 2 Distribution coefficient, K, data determined by ITSPME in
the capillary has a maximum amount of 2 mg of toluene that can
comparison with literature log Kow and log K (SPME) (for 100% methyl
be sorbed, with less sorbed when breakthrough occurs. Using
siloxane coating)
results for a 20 mL min21 flow of aqueous toluene, the total
BTEX Log K
amount of toluene sorbed by the capillary is 16.2 mg (Table 1).
compound (ITSPME)a Log Kow Log K (SPME) b
Using K = Cs/Cm (where Cs is the concentration in the
stationary phase and Cm the concentration in the mobile phase)
Benzene 1.77 (7.4) 2.1317, 2.1315 2.3011, 2.1012, 2.308
the K value of toluene between water and stationary phase can
Toluene 2.47 (4.2) 2.6917, 2.6915 2.8811, 2.5312, 2.888
be calculated. At equilibrium (i.e., at 100% breakthrough) Cs =
Ethylbenzene 2.75 (3.9) 2.8413, 3.1515 3.3311, 2.7212, 3.338
p- and m-Xylene 2.81 (4.8) 3.1514 3.3111, 3.318
16.2 mg per 2750 nL = 5801 mg L21 and Cm is the
o-Xylene 2.69 (5.0) 2.7714, 2.7715 3.2611, 2.8212, 3.268
concentration of the original toluene standard, i.e., Cm = 20
a
Experimental value, this work; triplicate determinations; %rsd values in
mg L21. Thus, log K = 2.47, which is in good agreement with
b
parentheses. Values quoted for SPME studies.
the K value determined by equation (2), log Kow and log K
determined from SPME (Table 2)8,11 16 reported for the same
temperature. Recovery of the sorbed toluene by back extraction
should yield 81 mg L21 of toluene (16.2 mg per 100 mL of
hexane, diluted 1 : 1 with IS solution). The 100 mL hexane strip,
with an added IS, was found to contain 66 mg L21 toluene,
indicating a recovery of about 80% toluene; again, the volume
uncertainty can lead to error in the calculated value.
Determination of partition coefficient, K, for BTEX
compounds
K values for BTEX compounds were determined as above, in
triplicate. A 2.5 mL mixture of BTEX (each 20 mg L21) in
Milli-Q water was passed through the capillary and it was
assumed that each solute reached 100% sorption. Results
showed that different analytes had different recoveries, and
hence a different affinity for the stationary phase of the
extraction column. K values are reported and compared to
Fig. 3 Chromatograms of the hexane strip with different temperatures
literature log Kow and log K (SPME) values (for a BP1-like
used for the aqueous extraction of 20 mg L21 of toluene. Curve a, extraction
coating fibre) (Table 2). Good agreement was found, so capillary at 20 °C; curve b, extraction capillary at 30 °C.
Table 1 Estimated amount of toluene sorbed per 100 mL of aqueous solution for the 20 mL min21 sorption profile at 20 °C
Fraction number 1 2 3 4 5 6 7 8 9 10 11 12 13 Total
Toluene sorbed/mg 2 2 2 2 2 1.7 1.6 1.3 0.9 0.5 0.2 0 0 16.2
Analyst, 1999, 124, 651 655 653
Table 3 Comparison of extracted amounts of phenols (10 mg L21 each)
from different aqueous matrices, as area ratio per cent of phenol peak/
xylene internal standard
pH 2 buffer
Milli-Q water pH 2 buffer + salta
p-Cresol 10.1 (4.0)b 11.3 (6.2) 27.0 (3.7)
2,4-DCP 8.2 (9.6) 11.7 (3.4) 55.1 (2.0)
p-tert-butylphenol 21.3 (2.8) 25.2 (9.1) 72.3 (4.3)
2,4,6-TCP 1.9 (6.2) 5.8 (3.5) 47.1 (5.5)
m-Nitrophenol 0.7 (14.3) 1.6 (12.5) 4.5 (17.8)
a b
Refer to Fig. 4 for the chromatogram of this solution. %RSD (n =
3) values in parentheses.
Fig. 5 Chromatograms of the hexane strip for aqueous extractions of
toluene aqueous solution concentrations of (a) 10 mg L21, (b) 20 mg L21
and (c) 40 mg L21. Ethylbenzene is added to each extracted solution as
internal standard at 25 mg L21.
Conclusions
Initial studies have proven ITSPME to be an effective extraction
technique. Due to the availability of different polarity stationary
phases, optimised extractions for target analytes of interest in
routine analysis should be possible. If the extraction column is
directly connected to the analytical column, maximum mass
conservation in transferring the sorbed analyte to the analytical
step can be realised. The first experiments using this have been
Fig. 4 Chromatogram of phenols sorbed from solution saturated with salt
encouraging. A solvent vent or waste line is required, and rather
and buffered to pH 2. BP20 polyethylene glycol capillary used for sorption,
than solvent stripping, thermal desorption of extracted analyte
with phenols back extracted from the capillary with 190 mL methanol. o-
will be explored with refocussing of the analyte at the head of
Xylene (10 mL of 25 mg L21 concentration) was added to collected
the analytical column to minimise band broadening. A simple
methanol as internal standard.
solution to this problem will be to use a recently demonstrated
cryogenic technique to focus the target analyte band on the
only a moderate increase in extracted amounts with reduction in analytical column prior to chromatographic analysis.18 This
pH. preconcentration effect makes ITSPME suitable for sample
preparation for trace analysis.
Results presented here were logical and in agreement with
Linearity of ITSPME extraction of toluene from water expectations. The K values determined by ITSPME are
comparable to those obtained with SPME and literature Kow
Equilibrium extraction conditions are established between the values. It is anticipated that ITSPME will be suitable for routine
aqueous and stationary phase, so concentration and recovered analysis. The extraction capillary column can be re-used; thus
solute should be linearly correlated.11,16 According to chroma- far, there have been no carry-over problems, nor any evidence
tographic theory, the partition ratio, k, is [equation (3)] of performance deterioration in the extracting column. Bonded
phase capillary columns are stable to organic solvent flushing.
CsVs Ns Vs
k = = = K
Large particulate material may require filtration prior to
(3)
CmVm Nm Vm
extraction, and, if necessary, a water rinse can be included
between sorption and back extraction.
where Ns and Nm are the number of moles of toluene in the
stationary phase and mobile phase, respectively, and Vs and Vm
are the volumes of the respective phases (note that Vm/Vs is
normally referred to as the phase ratio). Rearranging the
expression (3), and substituting CmVm for Nm, gives
Acknowledgement
Ns = KVsCm = ACm
The authors wish to thank SGE International, Ringwood,
where A = KVs = constant at a given temperature. Back
Australia for providing the GC capillary columns for perform-
extraction of 10, 20 and 40 mg L21 aqueous solutions of
ing the extractions and Shimadzu Scientific Instruments,
toluene, with added internal standard, was conducted in
Rydalmere, Australia for GC facilities.
triplicate with an %RSD of about 4% for each concentration.
Fig. 5 shows an overlay of one representative GC trace for each
extracted concentration. The calibration graph of concentration
versus area ratio (toluene/IS) had an R2 of 0.999. The gradient
References
of the line allows determination of the partition coefficient if the
volume of the stationary phase is known. The slope A is
1 Z. Zhang, M. J. Yang and J. Pawliszyn, Anal. Chem., 1994, 66,
approximately 0.83 mL, giving an estimated log K value of
844A.
2.48, which is close to the previous estimation of log K for
2 D. Noble, Anal. Chem., 1993, 65, 693A.
toluene between water and BP1 stationary phase. 3 C. L. Arthur and J. Pawliszyn, Anal. Chem., 1990, 62, 2145.
654 Analyst, 1999, 124, 651 655
4 J. R Dean, W. R. Tomlinson, V. Makovskaya, R. Cumming, M. 12 C. L. Arthur, L. M. Killam, S. Motlagh, D. W. Potter and J.
Hetheridge and M. Comber, Anal. Chem., 1996, 68, 130. Pawliszyn, Environ. Sci. Technol., 1992, 26, 979.
5 S. P. Thomas, R. Sri Ranjan, G. R. B. Webster and L. P. Sarna, 13 C. T. Chiou, D. W. Schmedding and M. Manes, Environ. Sci.
Environ. Sci. Technol., 1996, 30, 1521. Technol., 1982, 16, 4.
6 J. Pawliszyn and R. Eisert, Anal. Chem., 1997, 69, 3140. 14 K. Verschuren, Handbook of Environmental Data on Organic
7 R. G. Belardi and J. Pawliszyn, Water Pollut. Res. J. Can., 1989, 24, Chemicals, Van Nostrand Reinhold, New York, USA, 2nd edn.,
179. 1983.
8 C. L. Arthur, D. W. Potter, K. D. Buchholz, S. Motlagh and J. 15 C. T. Chiou, Environ. Sci. Technol., 1985, 19, 57.
Pawliszyn, LC-GC, 1992, 10, 656. 16 L. P. Sarna, G. R. B. Webster, M. R. Friesen-Fischer and R. Sri
9 K. Grob, Split and Splitless Injection in Capillary Gas Chromato- Ranjan, J. Chromatogr. A, 1994, 677, 201.
graphy with some remarks on PTV Injection; Huthig Verlag, 17 T. Fujita, J. Iwasa and C. Hansch, J. Am. Chem. Soc., 1964, 86,
Heidelberg, Germany, 1993. 5175.
10 C. L. Arthur, L. M. Killam, K. D. Buchholz, J. Pawliszyn and J. P. 18 R. M. Kinghorn and P. J. Marriott, Anal. Chem., 1997, 69, 2582.
Berg, Anal. Chem., 1992, 64, 1960.
11 D. W. Potter and J. Pawliszyn, J. Chromatogr., 1992, 625, 247. Paper 9/02567G
Analyst, 1999, 124, 651 655 655