Nanosensor and Breath Analyzer for Ammonia
Detection in Exhaled Human Breath
Perena Gouma, Krithika Kalyanasundaram, Xiao Yun, Student Member, IEEE, Milutin Stana´ cevi´ c, Member, IEEE,
and Lisheng Wang
Abstract—The detection and monitoring of gases in exhaled
human breath up to date has been limited by the lack of
appropriate materials and technologies which could rapidly and
selectively identify the presence and monitor the concentration
of trace levels of specific analytes-biomarkers. We present a
metal oxide-based nanosensor that is highly specific to ammonia
gas in breath-simulating environments at low part-per-billion
concentrations. The design of a handheld breath analyzer for gas
detection in exhaled human breath is described. Semiconducting
ceramics are presented as suitable sensor materials for easy and
affordable non-invasive diagnostics.
Index Terms—Chemical sensing, Nanosensors, Breath ana-
lyzer, Metal oxide-based sensors, Ammonia detection
clinical practice, human breath analysis methodologies that
exploit the non-invasive nature of such approach are still
under-developed. Since the time of the ancient physician
Hippocrates, exhaled breath was recognized as a non-invasive
tool to diagnose diseases. Breath testing devices first appeared
in 1784 when Lavoisier detected CO in exhaled breath of
guinea pigs . Since then, colorimetric assays and gas
chromatography columns have been used to analyze VOCs
(volatile organic compounds) in human breath in quantities
varying from millimolar (10−3M) to picomolar (10−12M)
concentrations . The latter gas sensitivity limit was achieved
by Linus Pauling’s gas-chromatography-based breath analysis
device in 1971 . Among the 400 compounds of which
the human breath consists, only 30 have been identified and
most of them are potential indicators (markers) of more than
one type of diseases –: NO which has been widely
studied as a bio-marker for oxidative stress ; exhaled CO
also a marker for cardiovascular diseases, diabetes, nephritis,
bilirubin production; exhaled hydrocarbons of low molecular
LTHOUGH analysis of body fluids (blood, sputum,
urine) for disease diagnosis and monitoring is routine
Manuscript received November 15, 2008; revised January 20, 2009. This
work is supported by NSF NIRT award and the UHMC Nanomedicine
Initiative of SUNYSB on the ”Development of a novel breath analysis
P. Gouma is with the Department of Materials Science and En-
gineering, Stony Brook University, Stony Brook, NY 11794-2275 (e-
K. Kalyanasundaram is with Sensitron, Division of RSM Electron Power,
Inc., Deer Park, NY 11729-4681.
X. Yun and M. Stana´ cevi´ c are with the Department of Electrical and
Computer Engineering, Stony Brook University, Stony Brook, NY 11794-
L. Wang is with the Department of Electrical and Computer Engineering,
The University of British Columbia, Vancouver, BC Canada V6T 1Z4.
mass, such as ethane, n-pentane; ethylene, isoprene (hydrocar-
bon affected by diet which is a marker for blood cholesterol
levels) ; acetone; formaldehyde; ethanol; hydrogen sulfide,
carbonyl sulfides, and ammonia/amines. For example, mea-
surements of exhaled ammonia may differentiate between viral
and bacterial infections in lung diseases to justify the use of
antibiotics , or it may be used to indirectly measure urea
levels if monitoring renal diseases .
Identifying these signaling metabolites (disease markers)
and measuring them in trace concentrations is not a trivial
problem. The low concentrations (ppb) of analyte molecules
present a major challenge, along with the specificity to a given
analyte. On the other hand, the benefit of developing this tech-
nology is tremendous. This paper focuses on the synthesis of
a nanosensor that can selectively measure ppb concentrations
of ammonia gas in breath-simulating environments. In the
present work, a breath testing prototype device has also been
developed based on an NH3-selective MoO3nanosensor and
it is described in detail. This breath analysis tool operates on
the principle of resistive chemosensing, thus it is unique, novel,
inexpensive and it rapidly detects ammonia. The selectivity of
the nanosensor enables use of simple circuitry for detection
and enables design of portable ’cut-off’ prototype for pre-
screening patients with specific diseases. Due to the selectivity
of the sensor, pattern recognition algorithms employed with
general sensor arrays become obsolete and the conductance
of the sensor is directly proportional to specific analyte con-
centration. Thus to sense the concentration of analyte, the
resistance of the sensor is converted to a voltage signal. For
a binary response, the percentage increase is compared to a
predefined threshold value. The threshold value is set through
calibration measurements and the specific LED indicates if the
concentration of the analyte has exceeded the threshold value.
The prototype device that we have developed is hand-held,
offering portability, and it is easy to use.
The paper is organized as follows. Section II reviews the
experimental techniques that have been used in the synthesis,
characterization, and design of the sensor and the sensing
device. Section III describes the experimental results obtained
using the nanosensor as well as the operation of the breath
analyzer. Concluding arguments are given in Section IV.
II. EXPERIMENTAL PROCEDURE
A sol-gel synthesis method  was employed to produce
3D networks of MoO3 nanoparticles through an alkoxide
reaction between molybdenum isopropoxide and 1-butanol.
of the sensing substrate on the TO8 substrate
(a) Schematic of EOS 835 and the sensing chamber; (b) Schematic
Fig. 2.Schematic of the gas sensing setup
The prepared sol was spin-coated onto sensing substrates
(3mmx3mm alumina plated with interdigitated Pt electrodes
produced by our group at Cornell’s Nanofabrication Facility)
producing thin films of MoO3. The amorphous films were
then calcined at 500oC for 8 hr to form the α-MoO3poly-
The sensor was tested using a gas-flow bench . The
gases used were UHP (ultra high purity) nitrogen (Praxair),
UHP oxygen (Praxair), 10 ppm ammonia in N2 (NORLAB
gases). NH3 concentration was varied by varying its flow
rates in connection with N2 and O2 flow rates. The gases
were controlled through 1479 MKS Mass flow controllers
whose channels are connected to a Type 247-MKS 4-channel
readout which reads the flow rate of the gases directly in sccm
(standard cubic centimeter per minute). The combined flow
rate of the gases was maintained at 1000 sccm. A schematic
of the sensor on a TO8 substrate, the sensing chamber and the
gas sensing set up are given in Figure 1 and Figure 2.
The effect of humidity was evaluated using a controlled
humidity chamber, in which relative humidity can vary be-
tween 0 and 100% in the presence of the gas under study.
Humidity is produced by bubbling N2 gas stream through a
standard water bubbler at room temperature. The percentage of
humidity was controlled by varying the ratio of dry to wet N2.
A commercially available CO2filter (NaOH premixed with
Vermiculite in a 10:1 ratio – Decarbite absorption tube, PW
Perkins and Co) was used in these studies. Decarbite reacts
only with highly acidic gases such as CO2, H2S.
Furthermore, a portable breath analyzer was developed by
attaching a heater to the backside of the sensor’s substrate and
then embedding them into a circuit board. A breath collection
tube is attached to the prototype.
III. RESULTS AND DISCUSSION
The sensing element used in this work to detect ammonia
was a nanocrystalline, sol-gel processed thin film of a semi-
conducting ceramic, MoO3. This metal oxide, which is also a
known catalyst, has been stabilized in it’s α-phase polymorph
with an orthorhombic crystal structure. The α-MoO3 phase
has a layered structure with (010) basal plane that is built up
of double chains of edge-sharing [MoO6] octahedra connected
through vertices. Upon reduction in catalytic reaction with
gases this phase forms the Mo18O52structure instead of the
ReO3 type Mo8O23 shear structure . It was discussed
in our earlier publications ,  that the α-phase is
selective to ammonia and highly sensitive to amines (which are
moderate bases) and the sensing mechanism is consistent with
the reduction of MoO3and the formation of ordered phases,
which suggests a reaction-based sensing process. This was
true for sputtered films , as well as for sol-gel processed
ones . In all of our previous studies on MoO3 sensors,
no matter what the processing method used , , the
α-phase exhibits a trend of selective response to ammonia in
the presence of interfering gases. Furthermore, this observed
selectivity included gases that are typically encountered in the
human breath, including NO2, NO, C3H6, and H2; there was
a slight cross-sensitivity reported for sol-gel processed films
to CO (for tested concentrations between 50-500ppm) [figure
6 in ].
In this work, the focus is to demonstrate that the α-MoO3
-based sensor may detect ammonia at concentrations relevant
to the normal levels found in exhaled human breath. Selected
ion flow tube (SIFT) measurements of ammonia in the breath
carried out  on a sample of normal controls versus a
population of stable patients treated for end-stage renal failure
show that the former group exhibit ammonia levels of 425 ppb-
1800 ppb, mean of 960 ppb. The diseased population had
ammonia levels on breath ranging from 820 ppb-14700 ppb,
with a mean of 4880 ppb. The lowest gas concentration we
were able to test in our laboratory was 50 ppb, so this was the
lowest limit of our measurements. Ambient levels of CO2are
around 300 ppm and a typical level in breath is much higher,
at approximately 5 – 6%. At 500oC, CO2is expected to break
down to CO and O2. Given the known cross-sensitivity of the
sol-gel sensor to CO, a filter is used to remove CO2altogether.
The sensor is semiconducting at elevated temperatures only
and the α-phase is stabilized above 500oC. As shown in
Figure 3, when the sensor operated at 500oC was exposed to
various concentrations of NH3gas in a background mixture
of N2and O2(80%/20%) simulating ambient air, NH3was
detected easily down to 50 ppb. When the sensor was exposed
Fig. 3. Sensing response of MoO3sensor to NH3(50ppb and 100ppb).
to various concentrations of NH3and CO2, (using the filter to
scavenge CO2) in the presence of N2and O2, the presence of
CO2did not affect NH3sensing, for concentrations ranging
between 0.5 and 10 ppm. Such results are clearly shown in
Figure 4(a)−(c). Preliminary data also showed that up to 25%
humidity in the gas stream does not have any effect on the
performance of the sensor (data not shown here). The sensor
is reversible; response and recovery times are extremely fast
(milliseconds to seconds) and the expected lifetime is over 1
year at the operating temperature.
A. Binary Prototype
The goal of the designed sensor prototype is to provide
measurement and quick ’cut-off’ response display. For the
quick initial pre-screen procedure, the qualitative measurement
of the gas concentration is not necessary. LED display is
implemented to indicate the binary measurement results. Thus,
when the output resistance is higher than a predefined higher
threshold, a red LED is lit up; when the output resistance is
lower than a predefined lower threshold, a green LED is lit
up; when the output resistance stays in the middle, both LEDs
A schematic of designed prototype is shown in Figure 5. The
output of the sensor is connected with a fixed resistor serially.
With the help of the voltage divider, the output resistance
is converted to a voltage signal, which is proportional to
the output signal. The voltage signal, indicated as Vtest in
the figure, is connected to a comparator and compared to a
threshold voltage, Vth, which is created by another voltage
divider. A potentiometer is used to create the threshold voltage,
so that it is very easy to adjust the threshold. The comparator
compares the two inputs and generates a output signal that
either turns on or turns off the following LED.
For simplicity, Figure 5 only shows the comparison with
higher threshold. If Vtest is higher than Vth, the output of
comparator is high, thus turns on the red LED. The same
schematic is applied for the comparison with lower threshold,
where the only difference is Vtest is now connected to the
negative terminal of comparator, and Vth, which is the lower
(a) CO2 and NH3 without filter; (b) CO2 with filter; and (c) NH3 with
Sensing responses of the MoO3 sensor to 1ppm concentration of
threshold here, is connected to the positive terminal. In addi-
tion, connections for the on-chip heater is also provided in the
Figure 6 shows a photograph of the designed binary sensor
prototype. The MoO3sensor is isolated from the environment
by a chamber. A specially designed channel allows exhaled
human breath flow or controlled gas flow to go through the
chamber and interact with the sensor (not shown here). The
mouthpiece receives the analyte input. The sensor determines
the concentration of NH3. Based on predetermined threshold,
the result is displayed either as a red (positive for high
H+ H+ H+
R R R
Fig. 5. Schematic diagram of the designed prototype.
Photograph of designed binary portable prototype for detection of
ammonia level indicating diseased state) or green (negative for
disease) light signal. The prototype is powered with an external
power adaptor, and regulated with an on-board adjustable
voltage regulator. The adjustable voltage regulator is chosen
to enable testing of the sensor with different biasing voltage.
The comparators are implemented with commercially available
ICs, LM339 from National Semiconductor. Due to the high
current, up to 450mA, required from the heater, a heat sink is
used to help the regulator work properly.
Our prototype system was developed to validate the feasi-
bility of breath analyzer and selectively to sense the range of
concentrations of NH3 that could be present in the human
breath. The designed prototype will be used in future medical
trials that would determine the appropriate threshold value.
This paper reports on a ceramic nanosensor for the selec-
tive detection of ppb levels of ammonia in breath-simulating
environments and on the design of a simple breathalyzer
prototype device utilizing such chemiresistive sensors for
fast, convenient, inexpensive, and non-invasive monitoring of
exhaled gases in human breath for disease diagnosis. The ce-
ramic (metal oxide) science and technology reported here may
revolutionize nanomedicine applications, such as inexpensive
and easy to use non-invasive diagnostic tests.
 M. Phillips, “Detection of Volatile Organic Compounds in Breath,”
in Disease Markers in Exhaled Breath, N. Marczin, S.A. Kharitonov,
M.H. Yacoub, and P.J. Barnes, Eds., New York: Marcel Decker, 2002.
 V.E. Arterbery, W.A. Pryor, L. Jiang, S.S. Sehnert, W.M. Foster, R.A.
Abrams, J.R. Williams, M.D. Wharam, Jr. and T.H. Risby, “Breath Ethane
Generation during Clinical Total Body Irradiation as a Marker of Oxygen-
Free-Radical-Mediated Lipid Peroxidation: A Case Study,” Free Radical
Biol. Med., vol. 17 (6), pp. 569-576, 1994.
 S.A. Kharitonov and P.J. Barnes, “Exhaled Markers of Pulmonary Dis-
ease: State of the Art,” Am. J. Respir. Cell Mol. Biol., vol. 163 (9), pp.
 A.D. Smith, J.O. Cowan, S. Filsell, C. McLachlan, G. Monti-Sheehan, P.
Jackson, and D.R. Taylor, “Diagnosing Asthma: Comparisons between
Exhaled Nitric Oxide Measurements and Conventional Tests,” Am. J.
Respir. Cell Mol. Biol., vol. 169 (4), pp. 473-478, 2004.
 S.M. Studer, J.B. Orens, I. Rosas, J.A. Krishman, K.A. Cope, S. Yang,
J.V. Conte, P.B. Becker, and T.H. Risby, “Patterns and Significance of
Exhaled-Breath Biomarkers in Lung Transplant Recipients with Acute
Allograft Rejection,” J. Heart Lung Transplant, vol. 20 (11), pp. 1158-
 T.H. Risby and S.S. Sehnert, “Clinical Application of Breath Biomarkers
of Oxidative Stress Status,” Free Radical Biol. Med., vol. 27 (11-12), pp.
 T. Karl, P. Prazeller, D. Mayr, A. Jordan, J. Rieder, R. Fall, and W.
Lindinger, “Human Breath Isoprene and Its Relation to Blood Cholesterol
Levels: New Measurements and Modeling,” J. Appl. Physiol., vol. 91 (2),
pp. 762-770, 2001.
 S.A. Kharitonov and P.J. Barnes, “Biomarkers of Some Pulmonary
Diseases in Exhaled Breath,” Biomarkers, vol. 7 (1), pp. 1-32, 2002.
 K. Sawicka, P. Gouma and S. Simon, “Electrospun Biocomposite
Nanofibers for Urea Biosensing,” Sens. Actuators, B, vol. 108 (1-2), pp.
 A.K. Prasad, P.I. Gouma, D. J. Kubinksi, J.H. Visser, R.E. Soltis, and
P.J. Schmitz, “Reactively Sputtered MoO3Films for Ammonia Sensing,”
Thin Solid Films, vol. 436 (1), pp. 46-51, 2003.
 T. Ressler, J. Wienold J, R.E. Jentoft and T. Nelsius, “Bulk Structural
Investigation of the Reduction of MoO3with Propene and the Oxidation
of MoO2with Oxygen,” J. Catal., vol. 210 (1), pp. 67-83, 2002.
 A.K. Prasad, D. Kubinski, and P. I. Gouma, “Comparison of Sol-Gel
and Ion-Beam Deposited MoO3 Thin Film Gas Sensors for Selective
Ammonia Detection,” Sens. Actuators, B, vol. 9, pp. 25-30, 2003.
 A.K. Prasad, “Study of gas specificity in MoO3/WO3thin film sensors
and their arrays,” Ph.D. Thesis, SUNY Stony Brook, NY, 2005.
 K. Kalyanasundaram, “Biomarker Sensing using Nanostructured Metal
Oxide Sensors,” Ph.D. Thesis, SUNY Stony Brook, NY, 2007.
 S. Davies, P. Spanel and D. Smith, “Quantitative Analysis of Ammonia
on the breath of patients in end-stage renal failure,” Kidney International,
vol. 52, pp. 223-228, 1997.
Perena Gouma is an Associate Professor in the
Department of Materials Science and Engineering at
the State University of New York-Stony Brook and
the Director of the Center for Nanomaterials and
Sensor Development. She is a Fulbright Fellow and
a US delegate for the National Academies’ GDEST
Workshop on the Future of Sensors and Sensor
Systems. She has published over 90 peer-reviewed
articles, she has been granted several patents on bio-
chemical sensor technologies. She is an Associate
Editor of the Journal of the American Ceramic
Society, and serves on the editorial boards of several other journals on
sensors and nanomedicine. Her research interests include nanostructured and
nanocomposite materials for selective chemosensors, electronic noses, and
breathanalyzers for non-invasive diagnostics.
5 Download full-text
Krithika Kalyanasundaram received her Bache-
lors in Engineering in Metallurgy from National
Institute of Technology, Tiruchirapally (formerly Re-
gional Engineering College), India in 2003 and PhD
in Materials Science and Engineering from SUNY,
Stony Brook, in 2007. She currently works as a
Sr. Materials Engineer for Sensitron Semiconductor,
Deer Park, NY. Her research interests include nano-
materials, structure property correlations in metal
oxide semiconductors, gas sensors, selective gas
sensing, breath analysis, and electron microscopy.
Xiao Yun (S’06) received the B.S. degree in Elec-
trical Engineering from the Beijing University of
Aeronautics and Astronautics, Beijing in 2005. He
is currently a PhD candidate of Electrical and Com-
puter Engineering at Stony Brook University, NY.
His research interests include low-power low-noise
mixed-signal VLSI circuits for sensor applications.
Milutin Stana´ cevi´ c (S’00-M’05) received the B.S.
degree in Electrical Engineering from the University
of Belgrade, Serbia in 1999, and the Ph.D. degree
in Electrical and Computer Engineering from Johns
Hopkins University, Baltimore, MD, in 2005. He is
an Assistant Professor of Electrical and Computer
Engineering at Stony Brook University, NY. His re-
search interests include mixed-signal VLSI circuits,
systems, and algorithms for parallel multi-channel
sensory information processing with emphasis on
real-time acoustic source localization and separation,
and micropower implantable biomedical instrumentation and telemetry.
Lisheng Wang received his B. Eng. degree in
Materials Science and Engineering from Tsinghua
University, China in 2003 and the M. Eng. from
the same institution in 2005. He received his Ph.D.
degree in Materials Science and Engineering from
Stony Brook University, NY USA in 2008. He
is currently a postdoctoral fellow in University of
British Columbia, Canada. His research interests
include nanostructured metal oxides, polymers and
their composites, design of gas/VOC sensitive de-
vices based on these materials for environmental
monitoring, human disease diagnosis and energy resource detection.