Anal. Chem.
1997, 69, 587-596
Solid-Phase Microextraction for the Analysis of
Human Breath
Christoph Grote† and Janusz Pawliszyn*
The Guelph-Waterloo Center for Graduate Work in Chemistry, Department of Chemistry, University of Waterloo,
Waterloo, Ontario, Canada N2L 3G1
Solid-phase microextraction (SPME) has been applied to
the quantitative determination of ethanol, acetone, and
isoprene in human breath. The method involves extrac-
tion and preconcentration with a fused silica fiber coated
with a polymeric stationary phase, desorption at 200 °C,
and assay by gas chromatography/mass spectrometry.
Three different fiber coatings have been evaluated with
regard to sensitivity, linear range, precision, and detection
limits. Typical RSD values in the range 2%-6% could be
obtained, depending on the fiber coating and the com-
pound investigated. The calibration curves for the com-
pounds are reproducible and linear over the concentration
ranges found in human breath samples. The method is
capable of detecting concentrations of acetone and iso-
prene reported for healthy subjects. The influence of
temperature and humidity on the extraction process has
been studied in detail. A linear relationship between log
K versus 1/T allows the calibration of the method for any
given temperature. The device is portable, economical,
and easy to use in patient sampling.
In recent years there has been increased interest in the
determination of breath compounds in medicine and clinical
toxicology. More than 100 compounds have been identified in
normal human breath by gas chromatography and mass spec-
trometry.1 These voliatile organic compounds (VOCs) are pro-
duced by metabolic processes and partition from the blood stream
via the alveolar pulmonary membrane into the alveolar air. This
implies that the concentration measured in breath is related to
the concentration in blood. Breath analysis can be used as a
diagnostic tool because increased or decreased concentrations of
some compounds have been associated with various diseases or
altered metabolism.2 The major advantage of breath analysis is
that it is a noninvasive procedure. It is well accepted by patients
because it does not incur the inconvenience of the collection of
blood or urine samples. Although many diseases have not yet
been examined in the context of using breath analysis as a
diagnostic tool, the usefulness has already been demonstrated by
the investigation of various clinical conditions. Volatile hydro-
carbons, especially ethane and pentane, have been used as
markers of lipid peroxidation, the oxidative degradation of
† Present address: Department of Analytical Chemistry, Fraunhofer Institute
of Toxicology and Aerosol Research, Nikolai-Fuchs-Strasse 1, D-30625 Hannover,
Germany.
(1) Krotoszynski, B. K.; Gabriel, G.; O’Neill, H. J. Chromatogr. Sci. 1977, 15,
239-244.
(2) Manolis, A. Clin Chem. 1983, 29 (1), 5-15.
S0003-2700(96)00749-4 CCC: $14.00 © 1997 American Chemical Society
polyunsaturated fatty acids.3 Several organic compounds can be
associated, for example, with lung cancer,4 liver disease,5,6 and
myocardial infarction.7 Increased concentrations (>50 nmol/L)
of acetone have been detected in the breath of patients suffering
from diabetes, a chronic disorder in which either the pancreas is
unable to produce insulin or the body is unable to effectively use
insulin.8-10 Moreover, breath acetone can be used to monitor
patients on diets. A weight reduction of about 225 g/week is
indicated by an acetone concentration of 500 nmol/L. Eating
meals with a high carbohydrate content significantly decreases
this concentration.10 Breath acetone analysis can, therefore, be
used as a motivational tool during weight-loss programs. Inhala-
tion toxicology is another area of interest. Exposure of workers
to environmental pollutants can be monitored by breath analysis.11
Routine breath analysis is applied for the determination of ethanol
because its concentration in breath is directly related to the
concentration in blood. A breath-to-blood ratio of 1:2100 is used
when breath alcohol measurement devices are calibrated to reflect
blood alcohol concentrations.12 Values displayed by breathalyzers
show the blood concentration of alcohol that would be assumed
by the breath alcohol concentration.
Breath compounds are present in nanomolar or even lower
quantities.1 To improve the sensitivity and precision of the
determination of these compounds, the sample has to be concen-
trated before assay with gas chromatography/mass spectrometry
(GC/MS). The three main methods currently utilized for pre-
concentration are chemical interaction, adsorptive binding, and
cold trapping. For example, cold trapping utilizes solid adsorbents
cooled to temperatures of liquid nitrogen or solid carbon
dioxide.13-15 Trapping systems operated at ambient temperatures
(3) Van Gossum, A.; Decuyper, J. Eur. Respir. J. 1989, 2, 287-291.
(4) Gordon, S. M.; Szidon, J. P.; Krotoszynski, B. K.; Gibbons, R. D.; O’Neill,
H. J. Clin. Chem. 1985, 31 (8), 1278-1282.
(5) Kaji, H.; Hisamura, M.; Saito, N.; Murao, M. Clin. Chim. Acta 1978, 85,
279-284.
(6) Chen, S.; Mahadevan, V.; Zieve, L. J. Lab. Clin. Med. 1970, 75, 622-627.
(7) Weitz, Z. W.; Birnbaum, A. J.; Sobotka, P. A.; Zarling, E. J.; Skosey, J. L.
Lancet 1991, 337, 933-935.
(8) Levey, S.; Balchum, O. J.; Medrano, V.; Jung, R. J. Lab. Clin. Med. 1964,
63, 574-584.
(9) Rooth, G.; Ostenson, S. The Lancet 1966, ii, 1102-1105.
(10) Crofford, O. B.; Mallard, R. E.; Winton, R. E. Trans. Am. Clin. Climatol.
Assoc. 1977, 88, 128-139.
(11) Raymer, J. H.; Pellizzari, E. D.; Thomas, K. W.; Cooper, S. D. J. Exposure
Anal. Environ. Epidemiol. 1991a, 1 (4), 439-451.
(12) Van Berkom, L. C. Alcohol Drugs Driving 1991, 7, 229-234.
(13) Van Gossum, A.; Shariff, R.; Lemoyne, M.; Kurian, R.; Jeejeebhoy, K. N.
Am. J. Clin. Nutr. 1988, 48, 1394-1399.
(14) Roberts, R. J.; Rendak, J.; Butcher, J. Dev. Pharmacol. Ther. 1983, 6, 170-
178.
Analytical Chemistry, Vol. 69, No. 4, February 15, 1997 587
Figure 1. SPME device modified for breath analysis.
have also been used.16 Several published reviews discuss the
currently used sampling techniques.2,17 These methods require
complex devices, and each method has its particular problems.
The methods based on adsorptive binding or cold trapping are
susceptible to contamination and require an additional water trap
to remove the large amount of water vapor present in human
breath. Otherwise this water would saturate the traps, and after
thermal desorption it could damage the GC column. Gas sampling
bags, often used to transport a definite volume of breath to a place
where the concentration and analysis are carried out, can be the
source of contamination and may lead to selective adsorption at
the inner wall. Current sampling devices used to preconcentrate
breath compounds at the time of sample collection normally
consist of a complex arrangement of tubing and valves which often
lead to a buildup of contaminants and loss of analytes by leakage.
The difficulty in quantitative delivery of the sample to a gas
chromatograph is another problem of the collection methods used
currently.
A technique is needed that does not concentrate H2O or CO2
and that can be applied directly to the mouth to preconcentrate
the compounds so that possible contamination due to tubing
material is avoided.
Solid-phase microextraction (SPME), developed in recent years
by Pawliszyn et al.,18-20 is a rapid, inexpensive, and efficient
technique for sampling solid, liquid, or gaseous samples. It
requires no solvents or complicated apparatus and provides linear
results over a wide range of analyte concentrations. Analysis of
the extracts is performed using GC, GC/MS, or HPLC.18-20 In
the SPME technique, a fused silica fiber coated with a polymeric
stationary phase is contained in a specially designed syringe whose
needle protects the fiber when septa are pierced. The fiber is
directly exposed to a liquid or gaseous sample to extract and
concentrate the analytes. After the absorption equilibration is
attained, the fiber is withdrawn into the needle and introduced
into an injector of a gas chromatograph, where the extracted
compounds are thermally desorbed and analyzed. SPME can be
performed manually or by an autosampler. The method is
economical, because one single fiber can be used repeatedly.
Unlike in conventional methods for analysis of gaseous samples,
modified equipment like a complex valve injection system, a
thermal desorption device, or a cooling trap is not required by
SPME. The SPME fiber is easily cleaned by desorbing any
(15) Snider, M. T.; Balke, P. O.; Oerter, K. E.; Francalancia, N. A.; Pasko, K. A.;
Robbins, M. E.; Gerhard, G. S.; Richard, R. B. Life Chem. Rep. 1985, 3,
168-173.
(16) Lawrence, G. D.; Cohen, G. Anal. Biochem. 1982, 122, 283-290.
(17) Schaeffer, H. J. J. High Resolut. Chromatogr. 1989, 12, 69-81.
(18) Chen, J.; Pawliszyn, J. Anal. Chem. 1995, 76, 2530-2533.
(19) Zhang, Z.; Yang, M. J.; Pawliszyn, J. Anal. Chem. 1994, 66, 844A-853A.
(20) Boyd-Boland, A. A.; Pawliszyn, J. Anal. Chem. 1996, 68, 1521-1529.
588 Analytical Chemistry, Vol. 69, No. 4, February 15, 1997
contaminants in a hot GC injector. The SPME technique coupled
with GC has been used to analyze volatile and semivolatile air
compounds.21 The fiber can be exposed directly to the air or to
a sample collected in a gas sampling bulb. This paper reports on
the first application of solid-phase microextraction for the analysis
of human breath. The validation of the method was based on
ethanol, acetone, and the nonpolar compound isoprene, which was
found to be the main endogenous hydrocarbon of human breath.1
EXPERIMENTAL SECTION
Chemicals and Materials. Chemicals. Acetone (99.9%+,
HPLC grade) was purchased from Sigma-Aldrich (Mississauga,
ON, Canada) and isoprene (99%, inhibited with 100 ppm p-tert-
butylcatechol) from Aldrich (Milwaukee, WI). Undernatured
ethanol (99.9%+, H2O < 0.1%) was obtained from a local pharmacy.
Carbon disulfide was purchased from BDH Inc. (Toronto, ON,
Canada). Dimethyldichlorosilane (Supelco, Canada) was used to
deactivate the glassware. For the Varian GC/MS system, helium
was purchased from Praxair (Waterloo, ON, Canada).
Materials. Gas sampling bulbs (0.5 and 1.0 L) were used for
standard gas preparation (Supelco). Prior to the first use, all bulbs
were silanized to deactivate the interior glass surface using a 10%
(v/v) solution of dimethyldichlorosilane in toluene. The solution
was kept in the bulbs for 12 h. After the bulbs were rinsed with
toluene and methanol several times, they were dried at 100 °C.
After use, the gas bulbs were cleaned with deionized water,
followed by rinsing with methanol and drying at 100 °C. The bulbs
were then purged with purified nitrogen for 30 min to remove
any contaminants.
The SPME fibers and the holder were purchased from Supelco.
Prior to their first use, new fibers have been conditioned under a
helium stream to clean the fiber from any contaminants present
in the coating. Conditioning was performed in a normal GC
injection port following the instructions of the manufacturer.
SPME Device Modified for Breath Analysis. The SPME
fiber can be directly exposed into the mouth of a subject. To
avoid the subject’s tongue coming into contact with the fiber
during sampling, the fiber has to be protected. This is achieved
simply by using a small piece of inert tubing which fits onto the
depth guide of the SPME device. The tubing also serves as a
mouthpiece, which can be easily replaced for hygienic reasons.
One little aperture in the tubing allows the subject to replace the
dead volume in the tubing and to exhale during the sampling
procedure. Figure 1 shows the modified SPME device. The
breath sampling procedure is described below.
Preparation of Standards. Preparation of Standards for the
Instrument Calibration. The standards used to calibrate the mass
(21) Chai, M.; Pawliszyn, J. Environ. Sci. Technol. 1995, 29, 693-701.
spectrometer were prepared in carbon disulfide. Three stock
solutions containing 100 µL of acetone, 200 µL of ethanol, and 20
µL of isoprene, respectively, were prepared in 50.0 mL volumetric
flasks. Eight standards were then prepared by taking appropriate
aliquots of the stock solutions and diluting them with carbon
disulfide in 25.0 mL volumetric flasks. The solutions were mixed
in an ultrasonic bath for 15 min and refrigerated at 4 °C. For
analysis, 0.5 µL aliquots of the solutions were injected into the
instrument. Manual injections were performed using a 10 µL
syringe (Hamilton, Reno, NV).
Preparation of Gaseous Standards. The stock solutions of
acetone and ethanol were prepared in 50.0 mL volumetric flasks
using deionized water as a solvent (NANOpure, ultrapure water
system, Barnstead/Thermolyne, Dubuque, IA). The solutions
were mixed in an ultrasonic bath for 15 min. The temperature of
human breath is about 35 °C. Moreover, it is almost saturated
with water vapor. It has been shown that both temperature and
humidity affect the extraction of analytes on the fiber.21 These
results require the setup of a calibration curve with gas standards
kept at 35 °C and spiked with the appropriate amount of water to
give a saturated atmosphere.
A water bath big enough to accommodate four gas sampling
bulbs at the same time was used, which made the procedure quite
practical. A thermostat with a circulator was installed to keep
the temperature at 35 °C. A second thermostat was used to heat
water flushing through a copper coil at the bottom of the water
bath to ensure a more accurate temperature control. The
temperature of this additional thermostat was controlled by a
contact thermometer. Manual adjustment was not necessary, and
the temperature could be kept constant in a small range of
approximately (0.2 °C. The temperature was measured using a
commercial digital oral thermometer. Its accuracy is reported to
be (0.1 °C in the range between 33 and 43 °C.
The amount of water which is necessary to give a saturated
atmosphere was calculated by the ideal gas law using the vapor
pressure of water at 35 °C. The 1.0 L bulbs had to be spiked
with 39.7 µL of water to give a saturated atmosphere. This water
evaporates to about 55 mL of vapor, which causes a significant
pressure increase in the gas bulb. Therefore, the bulb was first
spiked with 34 µL of pure water. After the water had been
evaporated at 35 °C, the pressure was decreased to atmospheric
pressure by quickly turning the stopcock of the bulb. Then 5 µL
of water containing the appropriate amount of ethanol and acetone
was spiked into the gas sampling bulb through a half-hole septum
using a microsyringe (Hamilton). The total amount of water
injected gave a relative humidity of approximately 100%. Standards
with higher concentrations of ethanol were prepared by spiking
neat ethanol into the bulb using a 1.0 µL syringe with no dead
volume (SGE, Australia). Since isoprene is not miscible with
water, the procedure described above could not be used for this
analyte. Therefore, isoprene was spiked into the bulb using a
gas-tight syringe (Hamilton). The appropriate aliquots were taken
from gaseous stock solutions obtained by spiking a 500 mL gas
sampling bulb with 10 and 20 µL of pure isoprene, respectively.
The gas sampling bulbs were kept at 35 °C for at least 30 min
prior to analysis to guarantee a homogeneous mixture. The
isoprene stock solutions were kept in the dark at room temper-
ature. Using this procedure, no visible condensation occurred
on the inner wall of the sampling bulb. Condensation must be
avoided because the analytes would partition into the aqueous
phase.
Selection of SPME Coating. Four different types of coatings,
a 100 µm polydimethylsiloxane (PDMS) fiber, an 85 µm polyacry-
late fiber, a 65 µm polydimethylsiloxane/divinylbenzene fiber, and
a 65 µm Carbowax/divinylbenzene fiber, were assessed by
comparing the extraction time profile and sensitivity of SPME
analysis with these fibers. The PDMS fibers with thinner films
were not investigated because, although they equilibrate faster,
they have a lower capacity which is not useful for analyzing highly
volatile trace level compounds.
Analytical Instrument Parameters. Instrument Calibration.
A Varian 3400 GC coupled with a Varian Saturn ion trap mass
spectrometer (Varian Chromatography Systems, Walnut Creek,
CA) was used to carry out the experimental work. Data acquisi-
tion was achieved using the Varian Saturn software. A 30 m SPB-5
column with 1.0 µm film thickness was connected to a 30 m SPB-5
column with 0.25 µm film thickness using a glass seal connector
(Supelco). Removing this thick film column was easily possible
without shutting down the mass spectrometer.
Carbon disulfide was chosen as a solvent because it elutes after
the target compounds. The best chromatographic results were
obtained by using a precolumn and starting the column temper-
ature at 45 °C, slightly below the boiling point of the solvent.
During injection, the septum-equipped programmable injector
(SPI) was kept at the same low temperature of 45 °C and then
ramped to 200 °C with a rate of 200 °C/min. The acquisition was
stopped before the large amount of solvent entered the ion trap
to protect the electron multiplier. The column temperature was
then ramped to 120 °C with a rate of 30 °C/min to remove all the
solvent. Since the peaks were narrow, the highest possible scan
rate was applied (0.14 s/scan).
Different characteristic masses for the three analytes have been
checked for the quantification. At higher concentrations of
ethanol, the mass spectrum changes due to secondary reactions
occurring in the ion trap. A steady decrease in detector response
factor with increasing concentration was observed when only m/z
) 45 was used for the quantification of ethanol. In contrast, the
response increased when m/z ) 47 was used. Due to reactions
in the trap, the fragment M - 1 (m/z ) 45) turned into M + 1
(m/z ) 47). Therefore, the sum of m/z ) 45 and m/z ) 47 was
used to quantify ethanol. As for acetone, both m/z ) 43 (base
peak) and m/z ) 58 (molecular ion) can be used for quantification.
Although the fragment m/z ) 58 is more significant for acetone
than the common fragment m/z ) 43, the latter one was used
during the study, because of the much higher intensity of the
peak. The quantitation of isoprene gave good results when m/z
) 53, m/z ) 67, or the sum of both characteristic masses was
taken.
Instrument Parameters for the Analysis of Gaseous Standards
and Breath Samples. Several column temperatures have been
investigated. Since the time required for the thermal desorption
of the volatile analytes in the hot injector port is relatively short
(<15 s), an initial temperature of 15 °C was sufficient to focus
the analytes at the beginning of the column. This cryofocusing
was achieved using CO2 as the cooling agent. All tests were
carried out with standard gas mixtures. Examples include
investigation of different fiber coatings, measurement of the
extraction and desorption time profiles, and determination of the
reliability of the method. Unless otherwise stated, all the breath
Analytical Chemistry, Vol. 69, No. 4, February 15, 1997 589
Table 1. Instrument Parameters
calibration
GC
injector SPI, inlet sleeve, 54 mm × 4.6 mm, 0.75 mm i.d.
column (1) precolumn, 0.8 m deactivated fused silica
(2) 30 m SPB-5, 0.25 mm i.d., 1.0 µm film
(3) 30 m SPB-5, 0.25 mm i.d., 0.25 µm film
carrier gas helium
head pressure 24 psi
linear velocity 34 cm/s
column temp 45 °C, 5 min, 10 °C/min, 75 °C, 30 °C/min, 120 °C
injector temp 45 °C, 200 °C/min, 200 °C
MS
detector ion trap
ionization mode EI
transfer line temp 220 °C
manifold temp 220 °C
scan rate 0.14 s/scan
mass range 35-150 amu
auto ion control on
samples were analyzed using the method described in Table 1.
Using this method, the overall run time was about 10 min. Total
ion current was monitored for all samples; however, quantitation
was performed by selecting characteristic masses. This provided
better sensitivity and a linear calibration curve for standard
injections. Table 1 summarizes the optimized instrumental
parameters for both calibration and analysis of breath samples.
RESULTS AND DISCUSSION
Extraction Time Profiles. The first step in the development
of an SPME method is the determination of the time needed for
the analyte to reach equilibrium between the sample and the fiber.
The extraction time profiles of the four differently coated fibers
were established by plotting the detector response versus the
extraction time. The equilibration time is reached when a further
increase of the extraction time does not result in a significant
increase in the detector response. Figure 2 shows the extraction
time profiles for 100 µm PDMS, 85 µm PA, 65 µm Carbowax/
DVB, and 65 µm PDMS/DVB fibers. All times were investigated
by analyzing duplicates.
The 100 µm PDMS fiber reached equilibrium after about 30 s
for all three compounds investigated in this study. Acetone and
isoprene equilibrated even faster when the Carbowax/DVB fiber
was used. Only 10 s was necessary for these compounds, whereas
ethanol required about 1 min to reach equilibrium. The new
PDMS/DVB fiber also showed fast equilibration times. Isoprene
equilibrated in about 15 s, acetone in 30 s, and ethanol in 1 min.
The more polar the compound, the longer it took to reach
equilibrium. As can be seen in Figure 2b), the equilibration times
for the PA fiber for all three compounds were much longer. This
fiber probably requires a longer equilibration time because it is a
more crystalline polymer than the PDMS coating. Therefore,
diffusion into the polymer does take longer for PA than for
PDMS.23
Since the breath sampling device will be applied directly into
the subject’s mouth, a short equilibration time is required.
Therefore, the appropriate fibers are the PDMS, the Carbowax/
DVB, and the PDMS/DVB fibers. Since equilibrium does not
(22) Supelco chromatography products catalog, 1996, Supelco Inc., Canada.
(23) Louch, D.; Motlagh, S.; Pawliszyn, J. Anal. Chem. 1992, 64, 1187-1199.
590 Analytical Chemistry, Vol. 69, No. 4, February 15, 1997
breath samples
SPI, inlet sleeve, 54 mm × 4.6 mm, 0.75 mm i.d.
(1) precolumn, 0.8 m deactivated fused silica
(2) 30 m SPB-5, 0.25 mm i.d., 1.0 µm film
(3) 30 m SPB-5, 0.25 mm i.d., 0.25 µm film
helium
24 psi
36 cm/s
15 °C, 0.25 min, 5 °C/min, 45 °C, 20 °C/min, 100 °C
200 °C
ion trap
EI
220 °C
220 °C
0.14 s/scan
35-150 amu
on
have to be reached as long as the same extraction time is applied
for both the breath and the standards, a very short sampling time
is possible. This is especially true for the Carbowax/DVB fiber,
for which a sampling time of only 10 s is possible. Although the
equilibrium for ethanol is not reached after 10 s, the decrease in
sensitivity can be tolerated, considering that ethanol reaches much
higher concentrations in human breath than acetone and isoprene.
Unless otherwise stated, the analytes were thermally desorbed
from the fiber for 15 s at 200 °C throughout the study. The
determination of the appropriate desorption time is important
because all of the analytes should be desorbed from the fiber to
reach the highest possible sensitivity and to avoid carryover. Less
than 1% carryover was found even when standards with high
concentrations of ethanol (3.5 µmol/L) were extracted. Carryover
was determined by analyzing the fiber blank directly after the
initial injection and expressed as percentage of the initial peak
area. Fibers thermally desorbed for 30 s at 250 °C did not show
any detectable carryover. Therefore, each fiber was predesorbed
in an extra injector port for 30 s at 250 °C prior to each extraction.
This procedure also ensured that no analytes were extracted from
the ambient air due to passive sampling.
Selection of SPME Coating. The choice of an appropriate
coating is essential for the SPME method. Depending on the
molecular weight and the polarity of the analytes to be extracted,
the sensitivity of each fiber is different. Four types of coatings
were investigated. A standard sample of ethanol, acetone, and
isoprene with a relative humidity of 99% was analyzed four times
with each fiber. The extraction time was 1 min for the PDMS,
Carbowax/DVB, and PDMS/DVB fibers, and 4 min for the PA
fiber. The desorption time was always 15 s except for the PA
fiber, which was desorbed for 20 s. The temperature of the water
bath was kept at 35 ( 0.1 °C. Figure 3a shows the absolute
amount of each analyte extracted by each type of fiber obtained
by comparing the detector response with the calibration curves
generated with liquid standards. For a better comparison, the
amount of each analyte extracted by the PDMS/DVB fiber was
set at 100%. The results are illustrated in Figure 3b. As can be
seen from Figure 3b, the PDMS/DVB fiber, recommended for
polar volatiles, exhibited a higher sensitivity for the analytes of
interest in comparison to the other fibers. This was especially

Figure 2. Extraction time profiles. Concentrations of analytes in
standard: ethanol ([), 136 nmol/L; acetone (9), 109 nmol/L; and
isoprene (2), 100 nmol/L. Relative humidity, 98%; extraction tem-
perature, 35 °C; desorption temperature, 200 °C; desorption time,
15 s. The lines only connect the data points to show the general
trend. They do not convey mathematical relation.
true for the less polar compounds, acetone and isoprene, inves-
tigated in this study. Ethanol can best be extracted with fiber
coatings recommended for polar analytes. Compared to the
nonpolar PDMS coating, the sensitivity could be increased by a
factor of 3 when the PA fiber was used, by a factor of ap-
proximately 4 when the Carbowax/DVB fiber was used, and by a
factor of 5 when the PDMS/DVB fiber was used. Whereas
PDMS, Carbowax/DVB, and PA basically have the same sensitiv-
ity for acetone, the amount extracted was 3 times higher when
the PDMS/DVB fiber was used. For isoprene, the new PDMS/
DVB fiber exhibited about 3 times higher sensitivity compared
to Carbowax/DVB, 5 times higher compared to PDMS, and almost
15 times higher compared to PA. In addition to the poor
sensitivity for isoprene, the PA fiber also required a much longer
extraction time to reach equilibrium. Therefore, the PA fiber was
not further investigated in this study. If only the sensitivity is
considered, the PDMS/DVB fiber would be the fiber of choice,
but characteristics like precision and linear range are also
important and will be discussed later in this study.
Linear Range. The linearity of the method was investigated
by determining calibration curves over the concentration ranges
of interest. The concentration of ethanol was about 3 orders of
magnitude higher than the concentrations of acetone and isoprene.
Therefore, ethanol was spiked into the gas sampling bulbs as a
neat liquid using a 1.0 µL syringe (SGE). Acetone and isoprene
were added as described earlier. Duplicate samples of gas
standards were extracted with three different fiber types and
analyzed using GC/MS. The line of best fit for the relationship
between the average peak area and the concentration of analyte
in the sample was determined by linear regression. The line was
obtained by integrating the selected m/z chromatograms. The
results are presented in Table 2, together with the results obtained
for precision and detection limits.
Precision. Eight replicate measurements were performed by
extracting standards with the same concentration. Each standard
was prepared fresh, and four different bulbs were used. The
extraction was performed 30 min after the analytes had been
spiked into the bulb. The temperature was kept at 35 °C, and
the relative humidity was set at 98% prior to spiking of target
analytes. The precision was evaluated by calculating the mean,
standard deviation, and relative standard deviation (%RSD) of the
observed values. Since only one extraction was performed out
of each gas sampling bulb, the calculated precision include not
only the precision of the analytical method itself (SPME fiber plus
GC/MS system) but also the repeatability of the preparation of
the gaseous standards. The results are shown in Table 2. Relative
standard deviations of retention times for the target compounds
were lower than 0.2% RSD (n ) 8), indicating good repeatability,
which is crucial for a reliable chromatographic method.
Accuracy. Quantification of Acetone and Isoprene. There are
no standard reference materials available for human breath.
Reference methods were also not applicable for acetone and
isoprene because they require additional custom-made equipment.
Therefore, the accuracy of the method for acetone and isoprene
was investigated by creating a calibration curve in the range of
interest for both acetone and isoprene. The analysis was
performed using the 65 µm PDMS/DVB fiber, because of its
highest sensitivity toward isoprene compared to the other fibers
investigated.
For acetone, a five-point calibration curve was set up in the
range 6-56 nmol/L (y ) 44447x + 21546, r2 ) 0.994). A new
standard was then prepared following the same procedure and
containing 20.44 nmol/L acetone. Triplicate analysis was per-
formed. Using the above calibration curve, a concentration of 21
( 1.2 nmol/L at the 95% confidence level was found (2.3% RSD).
For isoprene, a five-point calibration curve was set up in the range
1-12 nmol/L (y ) 3608x - 960). A standard containing 3.00
nmol/L isoprene was then analyzed three times and quantified
with the calibration curve to be 2.9 ( 0.15 nmol/L at the 95%
confidence level (2.1% RSD). The studies reported to date were
performed using always the same fiber. If several fibers have to
Analytical Chemistry, Vol. 69, No. 4, February 15, 1997 591
Figure 3. (a) Amounts extracted by different SPME fiber coatings. (b) Sensitivity relative to PDMS/DVB. Concentrations of analytes in the
standard: ethanol, 136 nmol/L; acetone, 109 nmol/L; and isoprene, 100 nmol/L. Relative humidity, 99%; extraction temperature, 35 °C; extraction
time, 4 min for PA, 1 min for PDMS, Carbowax/DVB, and PDMS/DVB; desorption temperature, 200 °C Desorption time, 15 s for PDMS, Carbowax/
DVB, and PDMS/DVB, 20 s for PA.

Table 2. Linear Range, Precision, and Limits of

Detectiona the fiber at equilibrium, Vf is the volume of the fiber coating, and
c, is the concentration remaining in the sample at equilibrium.23
SPME fiber
The problem in determining the K value is that cs is the
100 µm 65 µm 65 µm
compound PDMS Carbowax/DVB PDMS/DVB concentration at equilibrium (after the extraction took place) and
not the initial concentration of the sample, cs°. In certain cases,
Linear Range (r2 > 0.99)
ethanol 0-34 µmol/L 0-25 µmol/L 0-8 µmol/L cs in eq 1 can be replaced by the initial concentration cs°. This
acetone 0-545 nmol/L 0-136 nmol/L 0-545 nmol/L simplification is possible when the sample volume Vs is at least 2
isoprene 0-100 nmol/L 0-100 nmol/L 0-100 nmol/L orders of magnitude larger than the product KVf.24 This assump-
Precision, %RSD (n ) 8) tion is true when 1.0 L gas sampling bulbs are used, considering
ethanol 5.7 8.3 1.7
acetone 2.2 4.2 4.8
a constant fiber volume of the PDMS fiber of Vf ) 6.91 × 10-4
isoprene 3.9 12.8 3.2 cm3 and relatively small K values (less than 1000) for the highly
Limits of Detection (nmol/L) volatile compounds of interest. The determination of K values is
ethanol 20 4.6 5.8 then possible by using the equation
acetone 5.0 5.2 1.8
isoprene 2.1 1.1 0.3 K ) nf/cs°Vf (2)
aConditions: relative humidity, 98%; extraction temperature, 35 °C;
extraction time, 1 min; desorption time, 15 s; desorption temperature,
200 °C. This equation can only be applied for the 100 µm PDMS fiber
because this coating can be treated as a “normal” liquid. We
believe that, due to presence of solid polymer particules (DVB)
be used, the fiber length has to be taken into account as well, in the cross-linked structure of the Carbowax/DVB and PDMS/
because a variability in the length of (1 mm would affect the DVB, these fibers extract the analytes primarily by adsorption
results by (10%. rather than by absorption. Therefore, the equilibrium theory on
Effect of Temperature on the Extraction. The temperature effect which eq 2 is based cannot be applied. For these fibers, a different
on the extraction process was studied by analyzing a standard K value can be defined, taking into account the surface area of
with 99% relative humidity. Since the effect of temperature on the fibers. Since this surface area is not known, the K values
the extraction of volatile compounds has already been investi- have been determined only for the 100 µm PDMS fiber.
gated,21,24 only three temperatures (26, 35, and 45 °C) have been To determine the K values, a standard containing 136 nmol/L
studied. At each temperature, four measurements were performed ethanol, 109 nmol/L acetone, and 100 nmol/L isoprene was
with each type of fiber. Since the instrument was calibrated prior prepared. Three different temperatures were investigated. The
to the analysis, this study was also used to determine the partition relative humidity was set at 99% in all cases. Four replicate
coefficients (K values) at different temperatures. extractions were performed at each temperature. The average
The partition coefficient K between the fiber coating and the detector responses were converted into the absolute amount of
gaseous sample matrix can be expressed as analytes extracted by the fiber, nf, using a calibration curve
generated prior to the study. The K values for ethanol, acetone,
K ) nf/csVf (1) and isoprene were 88, 101, and 59 at 26 °C, 53, 72, and 45 at 35
°C, and 35, 43, and 40 at 45 °C, respectively, for the 100 µm PDMS
fiber. The relatively low K values are due to the low boiling points
where nf is the amount of analyte extracted from a gas sample by
and high vapor pressures of the analytes. Several extractions can
(24) Martos, P. A.; Pawliszyn, J. Anal. Chem., in press. be performed using one bulb without a significant depletion in

592 Analytical Chemistry, Vol. 69, No. 4, February 15, 1997

Table 3. Effect of Temperature on the Extraction
Process of the 100 µm PDMS Fibera

log K ) a(1/T) + b
compound a b r2
ethanol 2014 -4.8 0.9935
acetone 1866 -4.2 0.9887
isoprene 852 -1.1 0.9425
a Concentrations of analytes in the standard: ethanol, 136 nmol/L;
acetone, 109 nmol/L; and isoprene, 100 nmol/L. Relative humidity,
99%; extraction temperatures investigated, 26, 35, and 44 °C; extraction
time, 1 min; desorption time, 15 s.

the concentration because of the small amount of analyte extracted

during one sampling process. As can be seen, the K values
strongly depend on the temperature. Once the relationship
between K and T is determined, the plot of log K vs 1/T gives a
straight line.24 Using the equation of that line allows the
determination of K at any given temperature. It has to be
mentioned that the K values are calculated for a relative humidity
of 99%. Table 3 shows the data obtained when log K is plotted
versus 1/T for the 100 µm PDMS fiber. As predicted by the
theory, a straight line was obtained.
Since Carbowax/DVB and PDMS/DVB coatings cannot be
treated as “normal” liquids, a partition coefficient has to be defined
which takes into account the adsorption process. It is assumed
for the linear portion of the adsorption isotherm that the amount
extracted by the fiber is proportional to the concentration in the
sample. Therefore, a similar relationship between K or the mass
of analyte extracted and the temperature should be obtained as
well. Since the volume and the surface of these fibers are not
known, a new K value was not defined. The log of the mass versus
1/T was plotted instead, giving straight lines similar to those
shown in Table 3.
Changing the temperature had a significant effect on the
amount of analyte extracted by the fiber. The positive slope in
the figures illustrates that the amount of analyte extracted by the
fiber increases with decreasing temperature, indicating that the
extraction is an exothermic process. The study demonstrates that
the temperature had to be carefully monitored. However, the
temperature of the standards does not necessarily have to be
monitored at the same temperature as human breath, as long as Figure 4. Effect of relative humidity on the extraction. Relative
humidities investigated: 12.6%, 90.7%, 100%. Extraction time, 1 min;
the saturated humidity is carefully adjusted. The equations of
extraction temperature, 35 °C; desorption time, 15 s; desorption
the straight lines obtained as described above allow the calculation temperature, 200 °C; ethanol, 136 nmol/L, quantitation ions m/z )
of K at any given temperature. Knowing K and the amount of 45 + m/z ) 47; acetone, 109 nmol/L, quantitation ion m/z ) 43;
analyte extracted by the fiber, the concentration of the sample isoprene, 100 nmol/L, quantitation ions m/z ) 53 + m/z ) 67.
can be calculated.
Effect of Relative Humidity on the Extraction. The effect of the bars were defined as mean (2SD and calculated individually for
relative humidity on the amount of analyte extracted by each fiber triplicate measurements of each point. The results are illustrated
type has been investigated for three different humidities (12.6%, in Figure 4. It can be seen from Figure 4a that the extraction
90.7%, and 100%). Each standard was prepared freshly. The 5 efficiency of the 100 µm PDMS fiber is not significantly affected
µL of water containing the appropriate amount of acetone and by the amount of water present in the bulb. Considering the error
ethanol which was always spiked into the gas sampling bulbs in triplicate measurements, there is no significant difference in
determined the lower limit of the humidity range. Not more than the amount of ethanol and acetone extracted with varying relative
three extractions were performed out of the same bulb. Since humidity. The amount of isoprene extracted increases with
the extraction efficiency largely depends on the temperature, the increasing humidity. About 21% less is extracted at a relative
water bath was carefully kept at 35 °C. To exclude an instrumental humidity of 12.6% compared to a relative humidity of 100%. The
drift in response, the analysis of the standard with a relative effect observed for isoprene follows a trend opposite to what would
humidity of 90.7% was repeated after the analysis of the standard be expected. One reason for this phenomenon might be a slightly
with 100% relative humidity. No difference was observed. Error higher concentration of isoprene in the gas phase at saturated

Analytical Chemistry, Vol. 69, No. 4, February 15, 1997 593

humidity. Although no visible condensation of water was observed
on the inner glass wall of the sampling bulbs, there is a certain
amount of water adsorbed at the active sites of the glass. Due to
its polarity, water has a higher affinity to these sites than isoprene,
which might also be adsorbed to some degree. At higher relative
humidity, more water is present, which can replace some of the
adsorbed isoprene, resulting in a higher concentration in the gas
phase. It is not the purpose of this study to carefully investigate
the mechanism by which humidity affects the extraction process.
It should rather demonstrate that the standards have to be
prepared at 100% relative humidity because this also affects the
accuracy of this analytical method.
The importance of humidity is even more obvious when the
Carbowax/DVB and PDMS/DVB fibers were investigated. The
hypothesis that these fibers extract the analytes by adsorption
rather than by absorption is consistent with the observations. The
affinity to the coating material now plays a major role, and
replacement of analytes by compounds with higher affinity is very
likely to occur. Whereas for all three compounds the difference
in extraction efficiency between a relative humidity of 90.7% and
100% is not significant, the response differs significantly at a
relative humidity of 12.6%. When the relative humidity is
increased from approximately 12% to 100%, the amount of acetone
extracted by the fiber drops by 13% and that of isoprene by 33%.
This can be explained by replacement of these compounds by
water. Ethanol shows the opposite trend. When the relative
humidity increases from about 12% to 100%, the response increases
by a factor of 2. Hence, the more water that is present causes
more ethanol to be extracted. This is likely caused by a change
of the properties of the sorbent material, depending on the amount
of sorbed water at different humidities. A hypothesis that
adequately explains the data may be that water will be included
in the porous structure of the polymeric Carbowax/DVB coating.
This water, now available as a kind of stationary phase, will lead
to a higher extraction efficiency for polar ethanol molecules.
These molecules have high affinity to water due to possible
formation of hydrogen bonds. Whereas the nonpolar isoprene
molecule has no affinity to water, the polar acetone provides less
hydrogen bonding interaction. Thus, acetone and isoprene do
not show the same effect.
The extraction efficiency for the 65 µm PDMS/DVB fiber
strongly depends on the relative humidity as well (Figure 4c).
The amount of analytes extracted by the fiber coating decreases
by about 50% for all the analytes of interest when the relative
humidity increases from 12% to 100%. This result indicates that
water replaces some of the analytes from the fiber due to the
higher affinity to the coating.
All findings obtained for the three types of fibers investigated
in this study clearly show that it is crucial to set up the standards
at 100% relative humidity to obtain accurate results with the SPME
method.
Effect of Other Matrix Compounds on the Extraction. Extraction
efficiency for trace level compounds might be affected by other
compounds present in much higher concentrations in the sample.
In breath this can occur when the subject has consumed alcoholic
beverages. Even after one drink, the concentration of ethanol is
much higher than the concentrations of the other compounds of
interest. To investigate the effect of high ethanol concentrations
on the extraction of acetone and isoprene, the following experi-
ment has been performed. Standards with the same concentra-
594 Analytical Chemistry, Vol. 69, No. 4, February 15, 1997
tions as used for the calibration curves were prepared. The
relative humidity was set at 98%. Thirty minutes after acetone
and isoprene had been spiked into the bulb, two extractions were
performed using the 100 µm PDMS fiber. Ethanol was then
spiked into the same bulb, and 30 min later the bulbs were
extracted again in duplicate.
Using this procedure, six different concentrations were ana-
lyzed. It was found that high ethanol concentrations (i.e., 17.0
µmol/L ethanol compared to 136 nmol/L acetone and 20.0 nmol/L
isoprene) do not significantly affect the extraction of acetone and
isoprene when the 100 µm PDMS fiber is used. This allows
reliable quantification of acetone and isoprene, even if large
amounts of ethanol are present in the breath.
A short study has also been performed to check the effect of
high ethanol concentrations on the extraction of acetone and
isoprene when the Carbowax/DVB and the PDMS/DVB fibers
were used. Two standards containing 136 nmol/L ethanol, 109
nmol/L acetone, and 100 nmol/L isoprene were prepared with a
relative humidity of 98%. The extraction was performed at a
temperature of 35 °C. Each standard was analyzed three times
with the same type of fiber. Then, 0.2 µL of pure ethanol was
spiked into the bulb, giving a new ethanol concentration of 3.55
µmol/L, which is approximately 26 times greater than before. This
standard which still has the same acetone and isoprene concentra-
tion was then analyzed again three times. All six extractions in
total were performed out of the same bulb. It was found that the
effect of high ethanol concentrations on the extraction of acetone
and isoprene depends on the type of fiber used for extraction.
The Carbowax/DVB fiber indicates no difference in sensitivity to
acetone and isoprene when the ethanol concentration is signifi-
cantly increased. In this case, the concentrations of the com-
pounds are probably too small, so that the active sites of the fiber
are not completely occupied. The results obtained for the 65 µm
PDMS/DVB fiber indicate that ethanol replaces acetone and
isoprene from the fiber coating. A decrease in sensitivity by about
10% is observed, demonstrating that, whenever high ethanol
concentrations are present in human breath, the uptake of acetone
and isoprene by the PDMS/DVB coating will be affected.
Limit of Detection. The sensitivity of the SPME technique
was considered in terms of limit of detection (LOD). The limits
were estimated on the basis of the signal-to-noise ratios obtained
with standards containing the compounds of interest at low
concentration levels. Since the sensitivity of the PDMS/DVB fiber
is much higher compared to those of the PDMS and Carbowax/
DVB fibers, two standards with different concentrations were
prepared in such a way that the signal-to-noise ratios for the
analytes were less than 20 when the characteristic masses were
monitored. The LOD was defined as that concentration of an
analyte which produced a signal 3 times greater than the baseline
noise with GC/MS detection. The average signal-to-noise ratio
of four measurements was used to calculate the LOD. The fiber
blanks always contained coeluting impurities at the same position
as acetone with the same characteristic fragment m/z ) 43.
Therefore, eight blank samples at a relative humidity of 98% were
analyzed, and the average peak are of these measurements was
considered as noise. The signal-to-noise was then calculated
manually, and the limit of detection was determined in the same
way as described above. The results are shown in Table 2. The
results indicate that all three types of fiber coatings are suitable
for detecting low-level concentrations of ethanol, acetone, and
isoprene in human breath. Jones et al. reported a range of
isoprene concentrations found in human breath of 1.60-10.33
nmol/L.25 A mean acetone concentration in normal subjects of
23.2 nmol/L was reported by Phillips and Greenberg.26 This
indicates that all fibers are capable of detecting concentrations of
acetone and isoprene found in the breath of healthy subjects.
Stability of Standards. To ensure that the developed method
will be practical for everyday use, the gaseous standards should
be stable for at least 1 work day. Once prepared in the morning,
the standards could then be used during the day to quantify the
compounds of interest. To study the stability of the standards,
four gas sampling bulbs were spiked with the analytes in the same
way as described earlier (98% relative humidity). The temperature
of the water bath was carefully kept at 35 °C. The standards were
extracted using the 100 µm PDMS fiber. The extraction time was
1 min and the desorption time 15 s. The first extraction was
performed 30 min after the analytes were spiked. The other times
investigated were 1, 2, 4, 8, 24, and 48 h. During the study, seven
extractions in total were performed out of each bulb. After the
end of the study (48 h), a control sample with the same
concentration was prepared and analyzed 30 min after the analytes
had been spiked to check for any trend in instrument response
during the study. The mean value for four replicate measure-
ments was calculated for each time period investigated. The
relative standard deviation was then calculated for these mean
values (n ) 7) to evaluate the stability of the standards during 48
h. The RSD values were 4.4% for ethanol, 3.7% for acetone, and
3.8% for isoprene. Assuming an inherent random error and
comparing the results with those for the control sample, it was
found that the standards are stable for at least 2 days. Thus,
several calibration standards could be prepared at one time and
stored in the water bath for use throughout the working day.
Storage Time. An analysis normally cannot be carried out
immediately after the collection of the sample. Therefore, it is
essential to know how long a sample can be stored on the fiber.
The storage time for selected VOCs in air has previously been
investigated.21 The best results were obtained by sealing the
needle of the SPME device with a septum and keeping the
sampled fiber in a box filled with dry ice. Martos improved the
storage by retracting the fiber approximately 1.5 cm back into
the needle.24 Therefore, the fiber was removed from the holder,
retracted into the needle, and embedded in a Thermogreen LB-2
septum (Supelco) immediately after extraction. The fiber was kept
in a box with dry ice for 8 h. It has been shown that the standards
are stable during 1 working day. Before and after analyzing the
fibers which were stored on dry ice for 8 h, two control samples
were run to monitor any instrumental drift. Since one single fiber
of eacy type was used to study the storage conditions, the whole
procedure was repeated on consecutive days to ge three values
for each fiber type. The standards were prepared fresh daily. The
results have been statistically evaluated by comparing the mean
values of the control samples which had been analyzed prior to
the stored sample (n ) 6) with the mean values of the stored
samples (n ) 3) by applying Student’s t-test at the 95% confidence
level.27 Prior to the t-test, an F-test was applied to verify that the
standard deviations did not differ significantly.
(25) Jones, A. W.; Lagesson, V.; Tagesson, C. J. Chromatogr. B 1995, 672, 1-6.
(26) Phillips, M.; Greenberg, J. J. Chromatogr. B 1987, 422, 235-238.
(27) Miller, J. C.; Miller, J. N. Statistics for Analytical Chemistry, 2nd ed.; Ellis
Horwood Limited: Chichester, England, 1988.
At the 95% confidence level, the data revealed no significant
loss of analytes, even after a storage time of 8 h for the Carbowax/
DVB and the PDMS/DVB fibers. The mean values obtained for
the PDMS fiber differed significantly for acetone (increase by 10%
after storage of 8 h), whereas for ethanol and isoprene no
significant difference was observed. This phenomenon was not
further investigated.
Breath Analysis. Breath Sampling. As mentioned before, the
amount of analyte extracted by the fiber is independent of the
sample volume if the sample volume is large compared to the
fiber volume (Vs . KVf). That is true when the fiber is exposed
directly into the mouth. Moreover, the analyte uptake of the fiber
does not depend on the flow rate for analytes with K values lower
than 5000, which has been shown by Martos and Pawliszyn.24 The
whole breath normally contains about two-thirds alveolar breath.
The other third is dead space air of the respiratory tract (mouth,
nose, pharynx, trachea, and bronchi), with lower concentrations
of breath compounds.28 To obtain reproducible samples, it is
important to use alveolar air for breath analysis. The following
procedure should gaurantee that the fiber is exposed mainly to
the end exhaled breath with the highest concentration of com-
pounds of interest. Using the modified SPME device described
above, the subject has to inhale through the nose and hold the
breath for about 5 s to attain equilibration between the blood
stream and the alveolar air. Then the subject exhales ap-
proximately the first third of the breath quickly through the mouth
into the specially designed mouthpiece. The end part of the
breath should be exhaled as slowly as possible until the total
extraction time of 30 s is reached. Following this procedure was
quite convenient and easy as stated by the subject involved in
this study.
Quantitation of Acetone and Isoprene. Three breath samples
from a healthy subject were obtained by the method described
above (65 µm PDMS/DVB fiber) and assayed for acetone and
isoprene immediately after the extraction (RSD ) 1.2% for acetone,
5.7% for isoprene). Background samples from the room air and
fiber blanks assayed prior to the collection of the breath samples
did not show any peaks eluting at the same time as acetone and
isoprene.
The concentrations of acetone and isoprene in the breath were
determined to be 38 ( 1.1 nmol/L for acetone and 10 ( 1.4
nmol/L for isoprene at the 95% confidence level. These values
are within the range found for healthy people.25,26 Figure 5 shows
the total ion chromatogram of a breath sample taken 1 h after
the consumption of approximately 0.5 L of beer and stored for 3
h on dry ice prior to analysis. The peaks, which were identified
by comparing the retention times and mass spectra with reference
materials, correspond to air, CO2 (1), acetaldehyde (2), ethanol
(3), acetone (4), isoprene (5), and carbon disulfide (6).
CONCLUSIONS
The present work has shown that solid-phase microextraction
is applicable for the collection of breath samples. SPME provides
an effective method for the quantitative analysis of ethanol,
acetone, and isoprene. Comparing the SPME method to currently
used sampling techniques which are usually cumbersome and
costly shows numerous advantages. SPME is a rapid method
because the sampling process takes less than 1 min. Hence,
donating a breath sample does not require much effort from the
(28) Periago, J. F.; Prado, C.; Ibarra, I. J. Chromatogr. A 1993, 657, 147-153.
Analytical Chemistry, Vol. 69, No. 4, February 15, 1997 595

Figure 5. GC/MS chromatogram of a breath sample. Breath
sample taken after the consumption of alcohol, 65 µm Carbowax/
DVB fiber. Peak assignment: air, CO2 (1), acetaldehyde (2), ethanol
(3), acetone (4), isoprene (5), and carbon disulfide (6). Extraction
time, 30 s; desorption time, 15 s; desorption temperature, 200 °C.
patient. A complete analysis is performed in less than 10 min,
which allows a much higher sample throughput compared to other
methods described so far. The technique is sufficiently sensitive
to detect 5.8 nmol/L ethanol, 1.8 nmol/L acetone, and 0.3 nmol/L
isoprene, respectively. The response was linear over the concen-
tration ranges of interest, and the data obtained for accuracy and
precision are satisfactory. However, further evaluation of this new
sampling approach against well-understood blood measurements
is necessary before it can be relied upon for clinical measure-
ments.
The device is easy to handle, and because of its simplicity,
the buildup of contaminants is unlikely. The mouthpiece can be
596 Analytical Chemistry, Vol. 69, No. 4, February 15, 1997
further improved by adding two one-way valves. This would not
only simplify the sampling procedure, but it could also be used
to provide the subject with a pure air supply for inhalation to
reduce atmospheric contamination. The technique is economical,
and since the device is portable, it is suitable for patient sampling.
Samples can be stored on the fiber for 8 h without any significant
loss of analytes. This storage time can certainly be increased and
should be investigated in further studies. Breath samples can,
therefore, be easily obtained at the patient’s room without the
requirement to carry large equipment. The SPME device is
simple enough for the potential use in a physician’s consulting
room.
To date, the method is limited to compounds with relatively
high concentrations in human breath, but as more coatings
become available, the sensitivity and selectivity of the method
could be improved. The initial results of this study justify further
research in this new use of SPME for application to other
compounds of interest in human breath.
ACKNOWLEDGMENT
Financial support from the Natural Sciences and Engineering
Research Council of Canada, Supelco, and Varian is greatly
appreciated. This work was presented in part at the 21st
International Symposium on Chromatography, September 15-
20, 1996, Stuttgart, Germany.
Received for review July 25, 1996. Accepted November
18, 1996.X
AC960749L
X
Abstract published in Advance ACS Abstracts, January 1, 1997.