A Portable Impedance Biosensor Instrument for
Rapid Detection of Avian Influenza Virus
Jianhan LinI, Jacob LumII, Ronghui WangI, Steve
TungIII, Billy Hargis IV, and Yanbin LiI,II,IV
I
Department of Biological and Agricultural Engineering
II
Cell and Molecular Biology Program
III
Department of Mechanical Engineering
IV
Department of Poultry Science
University of Arkansas
Fayetteville, AR, USA
{jianhan, jdlum, rwang, chstung, bhargis,
yanbinli}@uark.edu
Abstract — The objectives of this study were to improve our
previous portable impedance biosensor instrument with more
rapid, reliable and quantitative features for detection of avian
influenza (AI) virus and to evaluate its performance using H5N1
and H5N2 virus. A high-intensity and high-gradient magnetic
field based separator was designed and fabricated for rapid
separation of AI virus. KPL measuring solution was used as
background medium to obtain higher stability. Twin BNC
connection between the electronic circuit and the biochip was
employed to prevent radio frequency interference. The
amplitude limiting average filtering method was used to
eliminate false signals and a linear regression model was built
for the embedded control software. A data acquisition software
was developed for communication and data processing. This
impedance biosensor instrument could work stand-alone or be
connected with a laptop via USB interface. The experimental
results showed that the impedance biosensor could identify
H5N1 virus with a detection limit of 103 EID50/mL in 30 min. A
prototype of the improved impedance biosensor was fabricated
for evaluation with cloacal swab samples from AI H5N2 virus
infected chickens. Compared with viral isolation, this biosensor
instrument had a false negative rate of 10%, whereas realtime
RT-PCR showed a false negative/positive rate of 20%.
I. INTRODUCTION
According to World Health Organization statistics on July
2010, avian influenza A/H5N1 was reported in more than 60
countries for animal cases and in 15 countries for human cases
with 501 people infected and 297 deaths since 2003 [1]. Avian
influenza has caused an increasing threat to animal and human
health worldwide and a key to effectively control the spread of
avian influenza is early detection followed by eradication of
infected animals, quarantine within a two-mile radius to
prevent movement of people and animals and vaccination of
This research was supported by USDA/NRI grant (project #2008-35204-
18662)
978-1-4244-8168-2/10/$26.00 ©2010 IEEE
Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
Huaguang Lu
Animal Diagnostic Lab,
Penn State University
University Park, PA, USA
hxl15@psu.edu
Luc BerghmanI, II
I
Department of Poultry Science
II
Department of Veterinary Pathobiology
Texas A&M University,
College Station, TX, USA
berghman@poultry.tamu.edu
animals outside the quarantine zone. However, conventional
methods used to detect avian influenza, such as viral isolation,
RT-PCR, ELISA and strips, are either time-consuming, or
performed only under laboratory conditions, or not sensitive
enough. Therefore, simple, rapid, robust and reliable methods,
suitable for point-of-care diagnostics and in-field detection,
are urgently needed.
Impedance biosensor is a promising technique for simple,
rapid and in-field detection that measures the electrical
impedance of an interface under a steady AC condition with a
constant DC bias. Interdigitated array (IDA) microelectrode is
often used for the development of impedance biosensor to
enhance the impedance change at the interface, shorten the
detection time, minimize the interference caused by non-
specific reaction, and improve the signal/noise ratio [2-4]. In
short, impedance biosensor has the advantages over the
conventional methods, such as label-free, simplicity,
rapidness, low-cost and in-field test [5].
A research prototype of impedance biosensor was
developed for detection of H5N1 virus [6]. However, this
prototype instrument was designed to focus on in-field
screening of AI H5N1 virus with limited ability to provide a
quantitative result. Besides, this instrument had limited
reliability at 90% confidence level in impedance
measurement. In this study, we improved this portable
impedance biosensor instrument with more rapid, reliable and
quantitative features for detection of avian influenza virus,
including design of a strong magnetic separator for rapid
separation of target virus from complex background and
improvement of our previous impedance detector to obtain a
higher stability in impedance measurement, and evaluated its
performance using H5N1 and H5N2 virus.
1558 IEEE SENSORS 2010 Conference
 II. DESIGN AND FABRICATION OF THE IMPEDNACE
BIOSENOSR INSTUMENT Tube
N S N S
A. Systme Overview A B C
This impedance biosensor instrument consisted of a S N S N
newly-designed magnetic separator for separation of target S N S N
virus, a microfluidic biochip with a gold IDA microelectrode, 2.54cm
N S N S
and an improved impedance detector (Fig. 1). First, the
magnetic separator was used to separate H5 subtype viruses 2.54cm 1.27cm
from a sample using anti-H5 antibody-nanobead complexes
and to concentrate it in a small volume of measuring solution. (a) Layout of the magnetic separator and its magnetic intensity
The prepared sample was injected into the biochip and then 1.40
the N1 subtype viruses were captured by the anti-N1 A C

Intensity (Tesla)
antibodies immobilized on the microelectrode’s surface. 1.30
Finally, the impedance of the biochip with H5N1 viruses was 1.20
measured and analyzed using the impedance detector and the
B
data acquisition software. 1.10

1.00
0.00 0.20 0.40 0.60 0.80 1.00
Length (cm)

(b) (b) Magnetic intensity vs. length

(d) Figure 2. Simulation of the magnetic separator. The intensity and the
gradient of the magnetic field at a tube were >1 T and 0.68 T/cm,
respectively.

To verify this, eight NdFeB Grade N52 magnets from K&J

Magnetics, Inc (Cat #: BX0X0X0-N52, Jamison, PA, USA)
were used to fabricate the magnetic separator and the
(a) magnetic intensity was measured by the 7030 guassmeter from
F.W. Bell (Orlando, FL, USA) with 1.32 T at both sides and
(c) 1.01 T at the center.

C. Use of the Microfluidic Biochip

Figure 1. The impedance biosensor instument. (a) impedance detector, (b)
magnetic separator, (c) microfluidic biochip, (d) a laptop with data
acquisition software
B. Design of the Magnetic Separator
Based on the superposition principle of magnetic field,
Finite Element Method Magnetics software (FEMM 4.2) was
used for magnetic field simulation [7]. The material NdFeB 52
MGOe permanent magnet in FEMM’s library was employed
to simulate cubic NdFeB Grade N52 magnets. As shown in
Fig. 2a, eight magnets were laid out in four columns. Two
magnets in the same column were placed with their same
polarity (south or north) together. The space between the
adjacent columns was 1.27 cm for placing a 1.5 mL sterile
tube. According to the simulation result, the magnetic
intensity in the tube ranged from 1.39 T at both sides (points
A and C) to 1.04 T at the center (point B) (Fig. 2b). The
average magnetic field gradient was 0.68 T/cm.
The microfluidic biochip was designed and fabricated by
Dr. Tung’s group at High Density Electronics Center at
University of Arkansas, Fayetteville, AR, USA. The biochip
featured a 25-pair gold IDA microelectrode with 10 µm finger
width and 10 µm inter-finger gap embedded in a PDMS
microchamber with 1723 µm length, 500 µm width and 40 µm
height. The detailed procedure of design and fabrication of a
similar biochip has been reported in our previous publication
[8].
To improve the stability in impedance measurement using
the biochip, KPL wash solution concentrate from KPL, Inc
(Cat #: 50-63-00, Gaithersburg, MD, USA) was diluted
1:200,000 with deionized water from Millipore (Milli-Q,
18.2 MΩ cm, Bedford, MA, USA) as measuring solution to
replace Mannitol solution (0.1M, Sigma---Aldrich, St. Louis,
MI,USA) in our previous research. The stability was
investigated by measuring the impedance of the same chip
with measuring solution at 10 kHz for eight times and
calculating the standard deviation of the impedance values.
The experimental results shown in Fig. 3 demonstrated that
KPL solution had a better stability with a standard deviation of
37 Ω than Mannitol solution with a standard deviation of 13
kΩ.

1559

Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
 6.0E+04 Since the phase angle of most biochips in impedance
measurement using KPL measuring solution was around -35o,
5.0E+04
solid R//C circuits with the phase angle from -34.5o to -35.5o
Impedance (kΩ)

4.0E+04
Mannitol were used to simulate the biochip for modeling. The 886
3.0E+04 LCR/ESR meter from BK Precision Corp (Yorba Linda, CA,
KPL
2.0E+04 USA) was employed for impedance calibration. As shown in
1.0E+04 Fig. 5, a linear relationship between the impedance magnitude
0.0E+00 of the solid circuit (|ZMC|) measured by the meter and the
1 2 3 4 5 6 7 8 inverse of the magnitude of the impedance (|ZIC|) obtained
Test Time from the converter was found and expressed by equation (2)

Figure 3. Impedance measurement using the same chip in Mannitol and

KPL solution ZMC =12119/ ZIC - 2.18 (2)

D. Improvement of the Impedance Detector 100

Based on our previous work [6], the impedance detector |Z MC | = 12119/ |ZIC |- 2.18
80
was developed using AT89C51 microcontroller (Atmel, San R2 = 0.99

|ZMC| (kΩ)
60
Jose, CA, USA) and AD5934 impedance converter (Analog
Devices, Inc, Norwood, MA, USA) (Fig. 4). Twin BNC 40
connector purchased from Digikey Corp (Cat #: ARF1038- 20
ND, Thief River Falls, MN, USA) was used to connect the
0
impedance detector and the microfluidic biochip to prevent
radio frequency interference. LCD display and keypad were 0 2 4 6 8
1000/|ZIC|
used for stand-alone operations, while USB communication
interface was employed for PC-terminated use. The memory Figure 5. The impedance magnitude of the solid circuits vs. the impedance
was employed to store the experimental data and setting magnitude of the converter
parameters.
The embedded control software of the impedance detector
LCD Display Impedance Microfluidic was developed using C language. The flowchart was shown in
Converter Biochip
Fig. 6. First, KPL measuring solution was injected to verify
Microcontroller

whether the microfluidic biochip was working properly. A

Keypad Memory control and then a sample were injected into the biochip for
impedance measurement. The impedance values of the control
USB (Zcntl) and the sample (Zsmpl) were saved and the impedance
Power Supply
Communication Computer change (ΔZ) between the control and the sample was
Module calculated by subtracting the impedance magnitude of the
control from that of the sample and returning its absolute
Figure 4. Schematic of the impedance detector value. The testing result was determined by the impedance
change according to the following criteria:
In the impedance converter, the measuring frequency, the
feedback resistor and the excitation voltage were set at 10 • If ΔZ is greater than or equal to the upper limit, the
kHz, 10 kΩ and 198 mVp-p, respectively. After each result is invalid;
impedance measurement at this frequency, a real data (R) and
an imaginary data (X) were obtained from the converter. The • If ΔZ is less than the upper limit and greater than or
impedance magnitude of the converter (|ZIC|) was calculated equal to the threshold, the result is positive;
using the following equation: • If ΔZ is less than the threshold, the result is negative.

Z IC = R 2 +X 2 (1)

Since the impedance of the biochip was observed to be

decreasing and tended to be stablized in 2 min after the
injection of the control/sample was stopped. Therefore, the
impedance measurement was performed 2 min after the
injection finished and the amplitude limiting average filtering
method was employed to eliminate false signals by collecting
five impedance data with the interval of 1 sec from the
impedance converter followed by eliminating the data that
were out of the acceptable range and averaging the rest.

1560

Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
 samples (Fig. 7). The collected data were automatically saved
Start in a .txt file with the information, such as date, sample
number, control impedance, sample impedance, threshold and
Yes result.
Inject KPL solution Test a biochip?

No
Wait for 2 min
Inject a control

Measure KPL impedance

Wait for 2 min

No Yes
ZKPL is valid? Measure control impedance

No
Change the biochip ZCntl is valid?
Figure 7. The Data Acquisition Software
Yes Yes
No Send Zcntl to a computer
Test times>3? III. EVALUATION OF THE IMPEDANCE BIOSENSOR
INSTRUMENT
Wash the biochip
A. Materials
Inject a sample Phosphate buffered saline (PBS, 10 mM, pH 7.4) and
Protein A were purchased from Sigma–Aldrich (St. Louis,
Wait for 2 min
MO, USA). Bovine serum albumin (BSA) from EM Science
(Gibbstown, NJ, USA) was prepared in PBS (1.0%, w/v) as
blocking solution. 150 nm streptavidin coated magnetic
Measure sample impedance nanobead was purchased from R&D Systems, Inc (Cat #:
MAG999, Minneapolis, MN, USA). EZ-Link Sulfo-NHS-
No Biotinylation Kit was purchased from PIERCE (Rockford, IL,
Zsmpl is valid?
USA). Avian influenza A/H5N1 virus (Scotland 59) was
Yes provided by USDA-APHIS National Veterinary Services
Send Zsmpl to the computer Laboratory (NVSL, Ames, IA, USA). Avian influenza H5N2
swab samples and monoclonal anti-H5 antibody were
ΔZ=|Zsmpl – Zcntl| prepared by Animal Diagnostic Lab at Penn State University,
PA, USA.

No
The KDS100 syringe pump from KD Scientific Inc
ΔZ≥Upper
limit? (Holliston, MA, USA) was used for injection of measuring
solution, controls and samples.
No ΔZ Yes
≥Threshold B. Experimental
The test on inactivated H5N1 viruses with the
Yes
concentrations from 102 to 105 EID50/mL was conducted in
Result: “Negative” Result: “Positive” Result: “Invalid” Biosensors and Bioinstrumentation Lab at University of
Arkansas and the test on chicken cloacal swab samples with
Wash the biochip Change the biochip Change the biochip H5N2 viruses was conducted in Animal Diagnostic Lab at
Penn State University. One biochip was used only one time
for a test and anti-H5 antibodies were used for both tests.
No
End Next test?
1) Magnetic separation
Yes Prior to magnetic separation, the anti-H5 antibody was
Yes No biotinylated using the EZ-Link Sulfo-NHS-Biotinylation Kit.
Skip control test?
The inactivated H5N1 from NVSL and the H5N2 swab
samples were prepared and stored at -80oC prior to use.
Figure 6. Flowchart of the embedded control software for the impedance
detector First, wash 25 µL of streptavidin coated nanobeads with
250 µL of PBS using the magnetic separator to capture the
E. Development of the Data Acquisition Software beads for 5 min and re-suspended in 100 µL of PBS.
Secondly, add 50 µL of biotinylated anti-H5 antibody and mix
A data acquisition software was developed by using
and incubate them for 30 min. Thirdly, capture the beads for 5
Microsoft Visual Basic software to collect and analyze the
min to remove redundant antibodies and re-suspend the beads
experimental data, including the impedance of controls and
1561

Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
 in 200 µL of a sample and incubate for 30 min at 37oC. D. Detection of H5N2 in Cloacal Swabs
Finally, capture the beads for 5 min to remove waste and The cloacal swab samples from chickens infected with AI
concentrate them in 150 µL of measuring solution for H5N2 virus were tested in parallel using the impedance
impedance measurement. biosensor instrument, real-time RT-PCR and viral isolation.
2) Impedance measurement Viral isolation was used as standard method.
Prior to use, rinse a biochip with deionized water and The threshold of the impedance biosensor instrument was
stabilize it using KPL measuring solution. Rinse the biochip set at 8 kΩ, which was determined by the low detection limit
with deionized water for 5 min and inject measuring solution (6 kΩ) plus the standard deviation (1.4 kΩ). The results shown
for 5 min followed by incubation for 2 min. Repeat the in Table 1 demonstrated that this biosensor instrument,
injection and incubation step for five times and measure the compared with viral isolation, had a false negative rate of 10%
impedance to verify the stabilization of the biochip. Then, in detection of AI H5N2 virus in chicken cloacal swab
Protein A was injected and absorbed onto the gold surface of samples whereas realtime RT-PCR showed a false
the IDA microelectrode. Finally, BSA was used to block negative/positive rate of 20%.
uncoated surface sites and prevent non-specific binding. The
detailed procedure of surface modification was described in
our previous publication [9]. TABLE I. THE RESULT OF THE TESTS ON CLOACAL SWAB SAMPLES
FROM AI H5N2 INOCULATED CHICKENS IN PARALLEL USING IMPEDANCE
BIOSENOSR, REALTIME RT-PCR AND VIRAL ISOLATION
First, inject 30 µL of measuring solution at a flowrate of 1
mL/h for 2 min using the syringe pump and wait for 2 min. Impedance Biosensor
Then, measure the impedance of the biochip using the Sample (Threshold: 8 kΩ) Realtime Viral
impedance detector to verify whether the biochip was working No. Control Sample RT-PCR Isolation
Result
properly. After washing with 150 µL of deionized water for 5 (kΩ) (kΩ)
1 24 31 - - -
min, inject 30 µL of a control for 2 min, wait for 2 min and 2 24 29 - - -
measure the impedance of the control. Wash the biochip with 3 27 34 - +/- -
150 µL of deionized water for 5 min and then inject 30 µL of 4 27 30 - + -
the sample for 2 min and wait for 2 min. Finally, measure the 5 27 19 + - +
impedance of the sample and determine the target virus by 6 30 25 - - -
calculating the impedance change between the sample and the 7 30 25 - - -
control. 8 30 19 + - +
9 34 39 - + +
C. Detection of Inactivated H5N1Virus 10 34 43 + + +
A linear relationship between the impedance change and
logarithmic value of the virus concentration was found in a
range of H5N1 concentrations from 102 to 105 EID50/mL (Fig. IV. CONCLUSIONS
8). The linear regression model was described by equation (3): The previous impedance biosensor instrument was
improved for detection of avian influenza virus with higher
ΔZ =5.05*lg N − 3.25 (3) instrument stability and separation efficiency. Compared with
where, N is the concentration of AI H5N1 virus in EID50/mL. viral isolation, the impedance biosensor instrument had a false
negative rate of 10% in detection of AI H5N2 in cloacal swab
samples, whereas realtime RT-PCR showed a false
20 negative/positive rate of 20%. The detection time for one test
Impedance Change ∆Z (kΩ)

was shortened to 30 min using 150 nm magnetic nanobeads

15 and the lower detection limit was reduced to 103 EID50/mL.
ΔZ = 5.05 * lg N - 3.25
R2 = 0.98
10 ACKNOWLEDGMENTS
The authors thank Lee C. Schrader at Department of
5 Biological and Agricultural Engineering, University of
Arkansas for helping fabricate the magnetic separator.
0
2 3 4 5
10 10 10 10 REFERENCES
Concentration of AI virus N (EID50/mL)
[1] WHO,
Figure 8. Impedance change between the sample and the control vs. the http://www.who.int/csr/disease/avian_influenza/country/cases_table_20
10_07_22/en/index.html, July 22, 2010.
concentration of AI H5N1 virus.
[2] S.M. Radke and E. C. Alocilja, “A high density microelectrode array
biosensor for detection of E. coli O157:H7,” Biosens. Bioelectron. 20,
The detection limit of this biosensor instrument was 1662-1667, 2005.
determined based on signal/noise ratio > 3. The standard
[3] J. H. Thomas, S.K. Kim, P. J. Hesketh, H. B. Halsall, and W. R.
deviation of five controls using the same biochip was 0.96 kΩ. Heineman, “Microbead-based electrochemical immunoassay with
Therefore, the detection limit of the biosensor instrument was interdigitated array electrodes.” Anal. Biochem. 328, 113-122, 2004.
103 EID50/mL.

1562

Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
 [4] M.S. Harrington and L.B. Anderson, “Analytical strategies using
interdigitated filar microelectrodes,” Analy. Chem. 62, 546-550, 1990.
[5] J. S. Daniels and N. Pourmand, “ Label-free Impedance Biosensors:
Opportunities and Challenges”, Electroanaly. 19(12), 1239-1257, 2007.
[6] Y. Li, B. Hargis, S. Tung, L. Berghman, W. Bottje, R. Wang, et al.,
“Methods and systems for detection of contaminants”, International
Patent, Publication Number: WO 2008/028124 A1, 6 March, 2008.
[7] D. Meeker, http://www.femm.info, Nov 2, 2009.
1563
Authorized licensed use limited to: Instituto Federal Fluminense (IFFLUMINENSE). Downloaded on September 05,2021 at 23:01:22 UTC from IEEE Xplore. Restrictions apply.
[8] M. Varshney, Y. Li, B. Srinivasan and S. Tung, “A label-free,
microfluidics and interdigitated array microelectrode-based impedance
biosensor in combination with nanoparticles immunoseparation for
detection of Escherichia coli O157:H7 in food samples,” Sensors and
Actuators, 128, 99-107, 2007.
[9] R. Wang, Y. Wang, K. Lassiter, Y. Li, B. Hargis, S. Tung, et al.,
“Interdigitated array microelectrode based impedance immunosensor
for detection of avian influenza virus H5N1”, Talanta, 79, 159–164,
2009.