Sensors and Actuators B 117 (2006) 50–57

Two-step preconcentration for analysis of exhaled gas

of human breath with electronic nose
Seong M. Cho a,∗ , Young Jun Kim a , Gwi Suk Heo b , Sang-Man Shin b
a Biosensor Team, Basic Research Laboratory, Electronics and Telecommunications Research Institute (ETRI),
161 Kajeong-dong, Yuseong-Gu, Taejeon 305-350, Republic of Korea
b Gas Metrology Group, Korea Research Institute of Standards and Science, 1 Doryong-dong, Yuseong-Gu,

Taejeon 305-600, Republic of Korea

Received 18 March 2005; received in revised form 28 October 2005; accepted 28 October 2005
Available online 15 December 2005

Abstract
The two-step preconcentrator, which performs concentrating volatile organic compounds (VOCs) and removing water vapor from samples, was
suggested for the analysis of the exhaled gas of human breath. The suggested two-step preconcentrator consisted of two separate adsorbent stages,
and it was operated simply by the serial adsorption/desorption of these two adsorbent stages. By using an additional adsorbent stage, the performance
of water removal was improved four times without loss of the VOCs, compared with that of the conventional single-step preconcentrator. The
suggested two-step preconcentrator is considered to be widely applicable to many fields in which interference by water vapor is critical.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Preconcentration; Electronic nose; Adsorbent; Breath analysis

1. Introduction eral technical barriers to overcome for realization of diagnostic

methods using electronic noses. First of all, the concentrations
Recently, diagnostic methods using electronic noses have of biomarkers in exhaled gases of human breath are very low,
attracted much attention [1–3]. The electronic nose consists of a typically in the range of several ppb to several hundreds of ppb.
series of sensors and the responses of the sensors are collectively These low concentrations make it very hard to be analyzed with
analyzed as a fingerprint of a sample [4]. The sensing mechanism conventional sensors. For this reason, it is almost essential to pre-
of electronic noses is basically different from that of the conven- concentrate VOCs in samples before analysis. The easiest way
tional specific gas sensors in the aspect of its non-specificity. It to preconcentrate VOCs is to use adsorbents, and it is widely
is rather similar to that of human olfaction. Diagnosis with the used in many application fields [10–16].
electronic noses would have a great advantage over the conven- The second problem in the analysis of human breath is the
tional method, because it is a very simple, non-invasive, fast and large amount of water vapor in exhaled gases. Generally, the
painless method. concentration of water vapor in human breath is several tens
According to the previous results by other researchers, of thousands of ppm. The concentration is predominantly high
volatile organic compounds (VOCs) in exhaled gases of human compared with those of VOCs. Because most of gas sensors
breath contain important information about the metabolism more or less respond to water vapor, some proper pretreat-
related with the health state of the person [5–9]. Exhaled gases ment method is required to minimize the effect from water
of patient breath differ from those of healthy persons in the com- vapor. When using adsorbent-type preconcentrators, the con-
position of VOCs, and it is believed that these differences can be centration of water vapor can be substantially lowered, typically
smelled out with electronic noses. However, there are still sev- several thousands of ppm, by adopting hydrophobic adsorbents
[15,16]. Such lowering of water vapor prevents the chromatog-
raphy columns from clogging, enabling the gas chromatography
∗ Corresponding author. Tel.: +82 16 525 5262; fax: +82 42 860 6836. analysis. A flame ionization detector (FID), which is typically
E-mail address: [email protected] (S.M. Cho). used in gas chromatography, is insensitive to water vapor. But,

0925-4005/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2005.10.050
 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57 51

in electronic nose systems, such lowered levels of water vapor second adsorbent. The flow of the sample is controlled by a car-
are still high, compared with the concentration of VOCs, and rier gas (N2 ) and a mass flow controller in the inlet position of
would surely influence the results of analysis. the carrier gas. Fig. 2(c) shows the final desorption process of
In the present study, with these backgrounds in mind, a the second adsorbent and the injection of the desorbed gas to an
two-step preconcentration method was proposed to effectively electronic nose system with a help of a four-port valve.
remove water vapor from samples and selectively concentrate The adsorbent tubes were made with quartz and wound with
VOCs. heating wire for desorption. The desorption performance of the
two-step preconcentrator is mainly determined by the heating
speed of the second stage adsorbent. A thin and slim tube was
2. Experimental used in the second stage for fast heating. The inner and outer
diameters (ID and OD) were 2.7 and 3.78 mm, respectively. The
2.1. Two-step preconcentrator heating rate of the second adsorbent tube was 20 ◦ C/s in the
desorption process. On the contrary, the heating speed of the
Fig. 1 shows the schematic operation principle of the two-step first adsorbent tube is not so critical. The ID and OD of the
preconcentrator. First, exhaled gases containing a large amount first adsorbent tube were 3.83 and 6.3 mm, respectively, and
of water vapor pass through the first stage adsorbent tube. In the the heating rate was 10 ◦ C/s. Desorption temperature for both
first stage, most of VOCs are adsorbed on the adsorbent surface, adsorbents was set at 310 ◦ C. A well-known Carbotrap (Supelco
while a significant amount of water vapor is not trapped because Co., USA) of 80 mg was used as the adsorbents of both stages.
of the hydrophobicity and small water-adsorbing capacity of the The flow rate of sample charging was 50 sccm, and the same
adsorbent. After sufficient adsorption, the first stage adsorbent flow rate was used in the desorption process from the first stage,
is heated for desorption. The desorbed gas sample will have while the flow rate was 6 sccm in the desorption of the second
enriched concentrations of VOCs and a substantially lowered stage. In all cases, the carrier gas used was dry nitrogen.
amount of water vapor. Then the desorbed gas goes into the The desorbed gases from the second adsorbent were analyzed
second adsorbent tube. Most of the VOCs are adsorbed on the with an electronic nose and FID. The FID gives informative
second adsorbent, but most of water vapor is not captured there. reference signals for the VOCs in the desorbed gases because it
In the gas samples desorbed from the second adsorbent upon is insensitive to water vapor.
heating, the VOCs are considerably enriched, while the water
content is much lowered. 2.2. Electronic nose system
Fig. 2(a and c) show the system layout of the two-step pre-
concentrator and its operation sequence. Fig. 2(a) describes the Fig. 3(a–c) is the photographs of the electronic nose system
adsorption process on the first stage adsorbent. A sample gas in used. Fig. 3(a) is the printed circuit board (PCB) for acquisition
a commercial sampling bag of polyethylene (PE; Flek Sampler, of sensor signals, which consist of reference resistances, ampli-
Japan) is injected to the first adsorbent tube through a six-port fiers, filters and so on. Initially, the responses of the sensors are
switching valve. The flow of the sample gas is controlled by a changes in electrical resistance, which are converted to electri-
vacuum pump and a mass flow controller which are located in the cal signals by a simple voltage dividing circuit on the PCB. The
exhaust line. Fig. 2(b) shows the desorption from the first stage applied voltage for the voltage dividing circuit was 10 V, and it
adsorbent and the adsorption on the second adsorbent stage. The corresponded to the field strength of 100 kV/m. These electri-
six-port valve is switched to carry the desorbed sample to the cal signals are digitized and transferred to the computer system

Fig. 1. Schematic diagram of the two-step preconcentrator. MFC is a mass flow controller.
52 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57

Fig. 2. System layout and operation sequence of the two-step preconcentrator: (a) sample charging step, (b) desorption of the first adsorbent stage and (c) desorption
of the second adsorbent stage. In each figure, the sample flow is indicated as bold line.

through a data acquisition card (DAQ card; National Instrument lulose (EC), poly(ethylene-co-vinyl acetate) 40% (PEVA40),
Co., USA). Fig. 3(b and c) is the sensing chamber and the sen- polyvinyl-pyrrolidone (PVP) and polycaprolactone (PCL). The
sor electrodes of the electronic nose. The sensor electrodes were carbon black–polymer composite sols were prepared with chlo-
made on a slide glass in interdigitated shape by Au evaporation. roform media, and were sprayed as thin films on the glass
The spacing of the interdigitated electrodes was 100 ␮m. The substrates mounted with Au electrodes. The limit of detection
sensing chamber was made by O-ring sealing and the volume (LOD) of the carbon black–EC composite sensor was around
was designed as small as possible. The dimension of sensing 1 ppm for benzene vapor. The LOD differs from sensor mate-
chamber was 3 mm × 25 mm × 1 mm. It corresponds to a vol- rials and also differ from analyte species. Generally, the higher
ume of about 75 ␮l. The performance of preconcentration is the vapor pressure of an analyte, the higher the LOD.
influenced by the volume of the sensing chamber. Because the
desorption is completed in a short time, typically less than 5 s, 2.3. Samples
if the volume of the sensing chamber is larger compared with
the volume of the desorbed gas, the desorbed gas will be diluted The correlations between VOCs in human breath and related
in the sensing chamber. Assuming the desorption time to be 3 s, diseases have been studied by many researchers [5–9]. Though
the volume of desorbed gas is approximately 300 ␮l, since the the absolute indicator of any kinds of disease was not found,
desorption flow rate is 6 sccm. If the sample gas flows ideally rough correlations have been reported between certain compo-
in the chamber, then the time of sample passage in the sensing nents in human breath and some diseases. For example, alkanes
chamber is about 4 s. (such as hexane), benzene derivatives, o-toluidine and aniline
The number of channels in the electronic nose was four. have been reported to be related with lung cancer [6], acetone
The sensor materials used were carbon black–polymer com- with diabetes [5,9], n-pentane with lipid peroxidation [9] and
posites [17–21]. The matrix polymers used were ethyl cel- isoprene with oxidative damage of the fluid lining of the lung
 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57 53

Fig. 3. Photographs of the electronic nose system used: (a) PCB which consists of reference resistors, amplifier, filter and so on, (b) sensing chamber and (c) sensor
electrodes and sensing materials deposited on them.

[9], and so on. These VOCs exist at very low concentrations in elapsed time after sample preparation. The carbon black–PVP
the human breath typically in the range of several ppb to several composite is the most sensitive material to water vapor among
hundreds of ppb. the carbon black–polymer composites examined. The nitrogen
The samples used in the present experiments are selected sample was prepared by filling dry nitrogen in a sampling bag,
among the VOCs, which are relatively abundant in the exhaled which had been washed with dry nitrogen 10 times. Although dry
breath of human and are believed to be potential biomarkers nitrogen was initially filled into the sampling bag, the concentra-
for diseases. The selected samples were isoprene, pentane, hex- tion of water vapor increased with time because the bag was per-
ane and heptane. All the analyte samples were prepared by meable to water vapor in the atmosphere to some extent. Filled
needle injection of a pure liquid sample in dry nitrogen. Sam- circles in Fig. 4 are the results for the preconcentrated samples
ples with the concentrations of several hundreds of ppm were for 1 min, and open circles are those of the raw samples, which
prepared first, and then diluted step-wise with dry nitrogen. A were not preconcentrated. In this case, the preconcentration was
benzene–toluene–xylene mixture was selected as a mixture sam- carried out with a conventional single-step preconcentrator, and
ple. Finally, the feasibility of the analysis of human breath was the amount of the Carbotrap adsorbent used was 80 mg. The car-
tested. The experiments were tested only to see the possibility
of discrimination of the signal of the VOCs in human breath
from the overwhelming water background. Further systematic
research on the diagnostic application will be done with the two-
step preconcentrator in future.
To analyze the effects of water vapor, a nitrogen gas sample
containing water vapor was used as a blank. Dry nitrogen was
filled into a sampling bag, and was left for 24 h to equilibrate with
the humidity of room air (60%, 25 ◦ C). This sample contains
only water vapor and no VOCs, so that it can be used as a good
sample of water vapor.

3. Results and discussions

3.1. Estimation of the water capacitance of the Carbotrap

adsorbent

Fig. 4 shows the sensing response of the carbon black–PVP Fig. 4. Sensing responses of the carbon black–PVP composite to water vapors
composite to water vapor in the nitrogen sample as a function of in nitrogen sample as a function of elapsed time after sample preparation.
54 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57

bon black–PVP composite was sufficiently stabilized with dry

nitrogen before measurement, so that the response signal should
be absolutely attributed to the responses to the water vapors in
the samples. When FID was used in the same measurements,
however, there was no sensor response. Therefore, it was clear
that these signals corresponded to the water vapor in the samples.
The increase in the signal strength of open circles (not pre-
concentrated samples) with time means the increase in the
concentration of water vapor in the sampling bag. The most
interesting phenomenon was that the signals of filled circles (pre-
concentrated for 1 min) did not monotonously increase, while
those of open circles increased almost linearly with time. The
saturation reached within 10 min. This is attributed to the satu-
ration of water adsorption sites in the Carbotrap adsorbent. The
water-adsorbing capacity of the adsorbent can be approximately
expressed with the Langmuir adsorption as follows [14]: Fig. 5. Reduction of water vapor by the two-step preconcentration; the dotted
line is the signal from single-step preconcentration and the solid line from two-
mkC
We = (1) step preconcentration.
1 + kC
where We is the adsorption capacity, m asymptotic maximum in the two-step preconcentrator was only ∼23% of that of the
solid-phase concentration (mg/g), k the equilibrium constant single-step. The content of water vapor in the finally desorbed
(1/ppm) and C is the water vapor concentration. In Fig. 4, the sample estimated ∼860 ppm for the two-step preconcentrator,
saturation corresponds to the region where We is independent of while it was ∼3400 ppm for the single-step. This result means
C, that is We ∼ m. It means that the adsorption sites were fully that by using the two-step preconcentrator the water removal
adsorbed with water molecules. When the preconcentration time performance could be improved by four times.
was increased to 3 min, it was observed that the resulting signal The single-step preconcentration was observed to show a
increased only by 20%. This also means the low water capacity hump prior to the main peak of desorption. The hump is thought
and fast saturation of the absorbent. to have come from the desorbed water before heating. The hump
When measured by a reference sample, it was confirmed that can be reduced by purging with dry nitrogen before desorption
the signal amplitude of 1 V corresponded to roughly 2000 ppm [22]. But, no hump was detected when the two-step preconcen-
of water vapor. Therefore, the content of water vapor in the tration was used. The absence of hump was likely caused by
desorbed sample was estimated to be about 3000 ppm, which the reduced amount of the total water adsorbed in the second
remained independent to the preconcentration time. Since the adsorbent.
desorption time is very short, the adsorbed amount of water in
the Carbotrap is very small, less than 1 ␮g. But, such lowered 3.3. Recovery of the VOC in the two-step preconcentration
concentration (3000 ppm) of water vapor is still much higher
than those of the biomarkers in real samples even if the concen- Another important parameter in the two-step preconcentrator
trations were enhanced by the preconcentrator. To obtain more is loss of the VOCs due to an additional adsorbent bed. In prin-
accurate analysis of human breath, more efficient removal of ciple, only water vapor should be selectively removed by the
water vapor is required without affecting the concentrations of two-step preconcentrator, and the VOCs should be preserved
VOCs in the samples. throughout the procedure. Fig. 6 compares the two-step and
single-step preconcentration for a sample of heptane. The con-
3.2. Improved water removal by two-step preconcentration centration of heptane was 17 ppb. FID was used as a detector.
Charging time was 10 min and the corresponding preconcentra-
Fig. 5 shows the effects of the two-step preconcentration on tion factor was around 1000, which was confirmed by measuring
the capability of removing water vapor when compared with a a reference sample of 10 ppm. In Fig. 6, there is no significant
single-step preconcentration. The samples used were nitrogen, difference in the desorption peaks between the two-step and
and the sample bag was left for 24 h to be balanced with the single-step preconcentration. This means that the loss of heptane
humidity of room air (25 ◦ C, 60%) before measurements. The by the additional adsorbent trap was negligible and the recov-
carbon black–PVP composite was used as a sensing material. ery of heptane was almost perfect. This was also confirmed by
As can be seen from Fig. 5, it is clear that the two-step pre- observing the breakthrough signal during the desorption from
concentration substantially reduces the content of water vapor in the first adsorbent. The increase of the FID background by the
the sample. By simply comparing the heights of the desorption desorption from the first adsorbent bed was very small and less
peaks, it can be noticed that the desorption peak height of the than 0.1% of the height of the desorption peak from the second
two-step preconcentrator was only ∼26% of that of the single- adsorbent. These results support the advantage of the two-step
step. More accurate analysis should be made with the areas of preconcentration, which reduces only the water vapor without
the desorption peaks. The calculated area of the water vapor influencing preconcentration of VOCs. The same tests were also
 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57 55

Fig. 6. Comparison of desorption peaks for the two-step and single-step pre- Fig. 7. FID signal for the two-step preconcentrated sample of a benzene–
concentrated samples. The sample was 16.9 ppb heptane and FID was used as a toluene–xylene mixture. The concentration was 20 ppb for each component.
detector.
gas chromatography, but in the case of the electronic nose it can
carried out with isoprene, pentane and n-hexane, and no signif- be used as important information. The time variable techniques
icant difference was observed. have already been suggested by several researchers [23–26].
Fig. 8 shows the results of the two-step preconcentration of
3.4. Desorption of mixture gas the exhaled gas of human breath. A four channel electronic nose
was used as a detector. Sample charging time was 20 min. Before
Fig. 7 shows the results of the two-step preconcentration for the desorption of the first stage adsorbent, dry nitrogen purging
a benzene–toluene–xylene mixture. The detector used was FID. was applied for 5 min to reduce the water vapor in the first adsor-
The concentrations of the constituents were all 20 ppb. It can bent. The solid lines in the figure are the results of the human
be seen that the desorption consists of three peaks, which are breath and the dotted lines are those of the nitrogen blank. As
attributed to the different desorption of benzene, toluene and mentioned before, the signals from the nitrogen blank corre-
xylene, respectively, in the order of the desorption sequence. spond to the responses to water vapor. From Fig. 8 it can be
These phenomena would be a serious problem in the case of confirmed that the signals from the human breath consist of sev-

Fig. 8. Measured signals with a four channel electronic nose for the two-step preconcentrated sample of the exhaled gas of human breath.
56 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57

References
[1] C. Di Natale, A. Macagnano, E. Martinelli, R. Paolesse, G. D’Arcangelo,
C. Roscioni, A. Finazzi-Agro, A. D’Amico, Lung cancer identification
by the analysis of breath by means of an array of non-selective gas
sensors, Biosens. Bioelectron. 18 (2003) 1209–1218.
[2] M. Fleisher, E. Simon, E. Rumpel, H. Ulmer, M. Harbeck, M. Wandel,
C. Fietzek, U. Weimar, H. Meixner, Detection of volatile compounds cor-
related to human diseases through breath analysis with chemical sensors,
Sens. Actuators B 83 (2002) 245–249.
[3] J.W. Gardner, H.W. Shin, E.L. Hines, An electronic nose system to
diagnose illness, Sens. Actuators B 70 (2000) 19–24.
[4] J.W. Gardner, P.N. Bartlett, Electronic Noses: Principles and Applica-
tions, Oxford University Press, Inc., New York, 1999.
[5] A. Manolis, The diagnostic potential of breath analysis, Clin. Chem. 29
(1983).
[6] H.J. O’Neil, et al., A computerized classification technique for screening
for the presence of breath biomarkers in lung cancer, Clin. Chem. 34
Fig. 9. Comparison of the signals from FID and the electronic nose for the (1988) 1613–1618.
two-step preconcentrated sample of the exhaled gas of human breath. [7] M. Phillips, K. Gleeson, J.M.B. Hughes, J. Greenberg, R.N. Cataneo, L.
Baker, Volatile organic compounds in breath as markers of lung cancer:
eral independent peaks. By comparing with the signals from the a cross-sectional study, Lancet 353 (1999) 1930–1933.
[8] M. Phillips, Method for the collection and assay of volatile organic
nitrogen blank, the first peak can be assigned to water vapor in
compounds in breath, Anal. Biochem. 247 (1997) 272–278.
the sample, and other peaks to VOCs. The signals due to VOCs [9] J.K. Schubert, W.P.E. Müller, A. Benzing, K. Geiger, Application of a
are clearly distinguished from those of water vapor except chan- new method for analysis of exhaled gas in critically ill patients, Intensive
nel 3. These clear separations between VOCs and water vapor Care Med. 24 (1998) 415–421.
could be attributed to the effective water removal by the two- [10] J. Nieß, T. Hamache, P.S. Lammers, E. Weber, P. Boeker, A miniaturized
thermal desorption unit for chemical sensing below odor thereshold,
step preconcentration method. In channel 3, a carbon black–PVP
Sens. Actuators B 95 (2003) 1–5.
composite was used as a sensor material. The response to VOCs [11] F. Bender, N. Barié, G. Romoudis, A. Voigt, M. Rapp, Development of
was not observed because of its strong response to water vapor. a preconcentration unit for a SAW sensor micro array and its use for
Thus, the materials too sensitive to water vapor are not appro- indoor air quality monitoring, Sens. Actuators B 93 (2003) 135–141.
priate in the analysis of human breath. [12] C.J. Lu, E.T. Zellers, Multi-adsorbent preconcentration/focusing module
for portable-GC/microsensor-array analysis of complex vapor mixtures,
Fig. 9 compares the results of the electronic nose with those
Analyst 127 (2002) 1061–1068.
of FID. The sample used was the exhaled gas of human breath, [13] C.-J. Lu, E.T. Zellers, A dual-adsorbent preconcentrator for a portable
which had been preconcentrated with the two-step preconcen- indoor-VOC microsensor system, Anal. Chem. 73 (2001) 3449–
trator. The solid line in Fig. 9 corresponds to the FID signal, 3457.
while the dotted line to that of the second channel (carbon [14] W.A. Groves, E.T. Zellers, G.C. Frye, Analyzing organic vapors in
exhaled breath using a surface acoustic wave sensor array with precon-
black–PEVA40 composite) of the electronic nose. Compared
centration: selection and characterization of the preconcentrator adsor-
with the FID signal, the signal from the electronic nose was slow bent, Anal. Chim. Acta 371 (1998) 131–143.
and low in time resolution. However, the observed multi-peak [15] J.M. Sanchez, R.D. Sacks, GC analysis of human breath with a series-
structure of both the FID and the electronic nose were similar coupled column ensemble and a multibed sorption trap, Anal. Chem. 75
to each other, except the first peak in the electronic nose, which (2003) 2231–2236.
[16] D. Helmig, L. Vierling, Water adsorption capacity of the solid adsor-
was originated from the water vapor.
bents Tenax TA, Tenax GR, Carbotrap, Carbotrap C, Carbosieve S III,
and Carboxen 569 and water management techniques for the atmo-
4. Conclusion spheric sampling of volatile organic trace gases, Anal. Chem. 67 (1995)
4380–4386.
[17] M.C. Lonergan, E.J. Severin, B.J. Doleman, S.A. Beaber, R.H. Grubbs,
A two-step preconcentrator was used to analyze the exhaled
N.S. Lewis, Array-based vapor sensing using chemically sensitive carbon
gas of human breath. The two-step procedure was found to suc- black–polymer resistors, Chem. Mater. 8 (1996) 2298–2312.
cessfully remove the water vapor from the sample gas without [18] B.J. Doleman, R.D. Sanner, E.J. Severin, R.H. Grubbs, N.S. Lewis,
deteriorating the preconcentration performance. This two-step Use of compatible polymer blends to fabricate arrays of carbon
preconcentrator may be applied to many practical fields in which black–polymer composite vapor detectors, Anal. Chem. 70 (1998)
2560–2564.
the interference from water vapor is critical. Further efforts are
[19] S.-C. Ha, Y. Yang, Y.S. Kim, S.-H. Kim, Y.J. Kim, S.M. Cho, Envi-
under process to minimize the size of the preconcentrator system ronmental temperature-independent gas sensor array based on polymer
and to reduce the cycle time of the measurement. composite, Sens. Actuators B 108 (2005) 258–264.
[20] S.-C. Ha, Y.S. Kim, Y. Yang, Y.J. Kim, S.-M. Cho, H. Yang, Y.T. Kim,
Acknowledgements Integrated and microheater embedded gas sensor array based on the
polymer composites dispensed in micromachined wells, Sens. Actuators
B 105 (2005) 549–555.
This work was supported in part by the Ministry of Informa- [21] Y.S. Kim, S.-C. Ha, Y. Yang, Y.J. Kim, S.M. Cho, H. Yang, Y.T. Kim,
tion and Communication of Korea and in part by the Ministry of Portable electronic nose system based on the carbon black–polymer com-
Science and Technology of Korea through the NRL program. posite sensor array, Sens. Actuators B 108 (2005) 285–291.
 S.M. Cho et al. / Sensors and Actuators B 117 (2006) 50–57 57

[22] J. Gawlowski, T. Gierczak, E. Pietruszynska, M. Gawrys, J. Niedzielski, Young Jun Kim received his PhD in polymer science from the Univer-
Dry purge for the removal of water from the solid sorbents used to sity of Akron, Akron, Ohio, US, in 1997. He, as a research associate, had
sample volatile organic compounds from the atmospheric air, Analyst been involved with controlled synthesis of block and functional polymers, and
125 (2000) 2112–2117. their applications in surface modifications related with biocompatible surfaces
[23] R.E. Shaffer, S.L. Rose-Pehrsson, R.A. McGill, Multiway analysis of and nanostructured materials at the University Connecticut and Princeton
preconcentrator-sampled surface acoustic wave chemical sensor array University until 2000. Since then at Electronics and Telecommunications
data, Field Anal. Chem. Tech. 2 (1988) 179–192. Research Institute (ETRI) he has been involved with nanoencapsulating
[24] T. Nakamoto, Y. Isaka, T. Ishige, T. Moriizumi, Odor-sensing system memory molecules and synthesizing nanoclusters for memory media and
using preconcentrator with variable temperature, Sens. Actuators B 69 biosensors.
(2000) 58–62.
[25] T. Nakamoto, K. Sukegawa, E. Sumitomo, Higher-order sensing using Gwi Suk Heo received his PhD in chemistry from Texas Tech University,
QCM sensor array and preconcentrator with variable temperature, Proc. Lubbock Texas, US, in 1983. His research topic during the doctorate program
Transducers (2002) 366–371. was synthesis of crown ethers with multifunctional groups and its application
[26] Y.S. Yang, S.-C. Ha, Y.S. Kim, A matehed-profile method for simple of selective metal separation at separation science field including analytical
and robust vapor recognition in electronic nose (E-nose) system, Sens. chemistry. He, as a research associate, had been involved with synthesis
Actuators B 106 (2005) 263–270. of nitroso compounds related with medical diagnosis at the University of
Oregon until 1985. Since then at Korea Research Institute of Standards and
Science (KRISS) he has been involved in organic analytical research field.
Biographies One of his recent interests is development of organic analytical methods for
air pollutants including volatile organic compounds (VOC) at trace level of
Seong M. Cho received the BS, MS and PhD degree in materials engineering ppt using various concentration techniques.
from Pohang University of Science and Technology (POSTECH), Korea, in
1992, 1994 and 2001, respectively. After getting MS degree, he had joined Sang-Man Shin received the BS, MS degree in environmental engineering
Samsung Electronics, Korea, in 1994, where he had worked for the devel- from Chungnam University, Korea, in 2003 and 2005, respectively. During
opment of lithography technology for 256 M and 1 G DRAM until 1996. He his MS program, he has been involved in method development of trace
joined Electronics and Telecommunications Research Institute (ETRI), Korea, VOC analysis at KRISS. Cryogenic concentration technique and adsorption
in 2001 as a senior researcher in the Basic Research laboratory. His current concentration technique were studied to get high precision analytical results
research interests are mainly focused on the sensor technologies such as an to certify reference materials of VOCs, which were developed at KRISS at
infrared image detector, an electronic nose, etc. ppb level.