Folded Floating Gate CMOS Biosensor for Nano Charge Sensing

2906 IEEE SENSORS JOURNAL, VOL. 11, NO. 11, NOVEMBER 2011
Authorized licensed use limited to: Iowa State University. Downloaded on April 11,2022 at 19:20:18 UTC from IEEE Xplore. Restrictions apply.
Folded Floating-Gate CMOS Biosensor for the
Detection of Charged Biochemical Molecules
Baozhen Chen, Archana Parashar, and Santosh Pandey
Abstract—A folded floating-gate CMOS biosensor is realized
for the detection of charged biochemical molecules. The biosensor
comprises a field-effect transistor with a floating-gate, a con-
trol-gate, and a sensing area. Charged biochemical molecules
placed on the sensing area induce a voltage on the floating-gate
and a relative shift in the threshold characteristics. Compared to
conventional devices, the floating-gate here is folded to span the
entire device. This allows us to place the sensing area above the
field-effect transistor (FET), thus reducing the total device area.
The fabricated devices show good sensitivity to the polarity and
concentration of charged poly amino acids. Possible applications
of the device include electronic detection of spatial and temporal
charge migration such as in biological processes.
Index Terms—Bio-FET, CMOS sensor, electronic charge
sensing, floating gate transistor.
I. INTRODUCTION
N RECENT YEARS, there has been significant advances in
the area of noninvasive, portable sensors that would couple
chemical, and biological entities with semiconductor devices
and integrated circuits [1]–[5]. These include electronic sensors
for the detection of chemicals, DNA hybridization, enzyme
reactions, action potentials, and protein complexes. To address
the issues of postprocessing, label-free detection, and bulky
instrumentation in traditional electrochemical sensors [2]–[6], a
CMOS-compatible extended floating-gate field-effect transistor
(EGFET) device have been demonstrated for the electronic
detection of electrolytes [7] and DNA hybridization [8], [9]. In
these EGFETs, the floating-gate is capacitively coupled to the
sensing area (where biochemical species are adsorbed) in one
direction and to the control-gate (where the readout voltage is
applied) in the other direction [7]–[9]. The threshold voltage
is a function of the applied control-gate voltage and the charge
on the sensing area. Fluctuations in the amount of biochemical
charge are electronically detected by a corresponding shift in
the threshold voltage. However, in the present EGFETs, the
floating-gate topology is bifurcated and spans in two directions
[7], [8], which increases the overall device area. In addition,
the sensing area of the device is large (40 40 to
100 100 ) [8] and is placed far from the FET in order
Manuscript received December 21, 2010; revised April 22, 2011; accepted
April 25, 2011. Date of publication May 02, 2011; date of current version Oc-
tober 26, 2011. This work is supported by a NSF-CMMI-1000808 grant from
the National Science Foundation. The associate editor coordinating the review
of this paper and approving it for publication was Dr. Dwight Woolard.
The authors are with the Department of Electrical and Computer Engineering,
Iowa State University, Ames, IA 50014 USA (e-mail: chen.baozhen@gmail.
com; archana@iastate.edu; pandey@iastate.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JSEN.2011.2149514
Fig. 1. (a) Structure of the proposed bio-FET showing its floating-gate, control-
gate, and sensing area. (b) Equivalent capacitive model for the device.
to isolate the test species from the transistor [7]–[9]. As a
solution, nonplanar EGFETs have been proposed to shrink the
sensing area by combining microelectromechanical systems
(MEMS) micromachining and CMOS fabrication steps [4].
Here, we present an extended floating-gate biosensor with a
folded floating-gate topology (Fig. 1). The device comprises a
FET with a control-gate, a floating-gate, and a sensing area. The
basic working principle is similar to previous EGFET designs
[7]–[9]. However, the floating-gate here is placed above the con-
trol-gate and is folded to cover the entire device (Fig. 1). This
allows us to utilize the top device area as the sensing area which
is now directly above the FET. Thus, the sensing and detection
is done at the same location on the chip. The device is fully
CMOS-compatible and its charge sensitivity is comparable to
other EGFET devices [7]–[9].
II. METHODOLOGY
The induced voltage on the floating-gate of the biosensor
device shown in Fig. 1 is written as [7]–[9]
(1)
(2)
where represents capacitance between control-gate and
floating-gate, is the capacitance between sensing area and
floating-gate, is the capacitance between floating-gate and
the bulk, is the total capacitance of the device, is the
applied voltage on the control-gate, is the static charge on
1530-437X/$26.00 © 2011 IEEE
I

CHEN et al.: FOLDED FLOATING-GATE CMOS BIOSENSOR FOR THE DETECTION OF CHARGED BIOCHEMICAL MOLECULES 2907
floating-gate, and is the total induced charge on sensing area.
The other device parameters used in (1) and (2) are the oxide
capacitance , floating-gate to source capacitance ,
floating-gate to drain capacitance , and the FET depletion
capacitance . The effective threshold voltage is related
to the original threshold voltage (i.e., with no solution on
the sensing area) and can be modulated by charges on top of
the sensing area as shown in (3) and (4). Assuming there is
minimal change in the applied voltages ( and ) and the
amount of static charge , the effective change in threshold
voltage is proportional to the amount of fluctuation in the
surface charge on the sensing area
(3)
(4)
In our experiments, is taken as the reference charge on
the floating-gate with deionized (DI) water on the sensing area.
When the charged biomolecules are brought close to the sur-
face of the sensing area, they induce countercharges (i.e., )
in the floating-gate [1]. This alters the induced voltage on the
floating-gate and shifts the effective threshold voltage. Posi-
tively charged molecules increase and thus a smaller is
required to turn on the device. This produces a lower threshold
voltage ( ) and a negative shift in the current-voltage (I–V)
curve compared to that for DI water. Alternatively, negatively
charged molecules lower the floating-gate voltage and re-
quire a higher control-gate voltage to turn on the device.
This is measured by a higher threshold voltage and a positive
shift in the I–V curve compared to that for DI water.
The biosensor devices are designed in standard CMOS AMI
Semiconductor (AMIS) 1.5 library. After a series of design
optimization steps, the sensing area is conservatively chosen as
14.4 28.4 . The smallest possible device area can be
further scaled as , where is the minimum FET gate
length or half of the technology node ( in our case).
The ratio is 3:1. The thicknesses of the gate oxide and
interpoly oxide are approximately 31 and 60 nm, respectively.
A 1 thick spacer separates the floating-gate and the
charged molecules on the sensing area. Next, a 3-D model is
built in Synopsys Sentaurus Structure Device Editor adopting
the design topology and dimensions [Fig. 2(a)]. In each sim-
ulation, a fixed amount of electronic charge is placed on the
sensing area [Fig. 2(b)]. The simulation results indicate that
the floating-gate voltage increases linearly with the con-
trol-gate voltage . Furthermore, for a fixed , increasing
the charge density linearly increases the amount of voltage in-
duced on the floating-gate similar to [9]. From our simulations,
1 of charge on the sensing area produces
a 0.95 V change in the floating-gate voltage . As shown in
(4), decreasing the sensing area does not necessarily improve
the device sensitivity as both and are reduced. For
comparison purposes, the reported EGFETs have a sensitivity
of 5.44 V/C (i.e., one electron charge per produces 14
voltage shift) [9].
Fig. 2. (a) Three-dimensional device structure created using Sentaurus Struc-
ture Device Editor. (b) A graph of the floating-gate voltage versus the applied
control-gate voltage at three fixed charge densities on the sensing area (SA). The
inset shows the variation of the floating-gate voltage with charge density on the
sensing area (SA) at three fixed control-gate voltages.
The layout is fabricated in CMOS AMIS 1.5 process and
packaged in 40-pin dual in-line package (DIP). Each chip com-
prises individual devices of varying areas (14.4 28.4
to 100 100 ). The packaged chips are tested using a
HP4145 semiconductor parameter analyzer. Unlike conven-
tional ChemFETs [2], no postprocessing or additional coating
is needed. The sensing area is physically isolated from the
floating-gate by overlay glass (1 thick) deposited during
the last step of CMOS fabrication. A polydimethylsiloxane
(PDMS) layer with a 1.5 1.5 mm opening is placed on the
chip for protection of the contact pads and bonding wires. A
certain volume (0.8–1.2 ) of a previously prepared solution
of biochemical molecules is dropped on the fabricated chip. A
custom LabView program is used to apply the voltage on
the control-gate. The series (5–10) of I–V curves are recorded
and compared with that for DI water (used as reference). To
refresh the device or to test another biomolecular concentration,
the chip’s surface is first rinsed (3–5 times) and then tested
with DI water to obtain the reference I–V curve. The new
solution is subsequently dropped on the chip surface and the
I–V characteristics are recorded.
The following charged biochemicals are used for testing the
devices: poly-L-lysine (0.01%), poly-L-histidine (0.01%), poly
glutamic acid (0.5%), poly aspartic acid (PAA) (0.5%), poly

Fig. 3. Transconductance versus control-gate voltage are plotted for poly
amino acids and compared with the DI water reference curve.
styrenesulfonic acid sodium salt (PSASS), and poly diallyl
dimethyl ammonium chloride (PDDAC).
III. RESULTS AND DISCUSSION
Fig. 3 plots the measured transconductance ( ) versus con-
trol-gate voltage for different charged poly amino acids
on the sensing area of the biosensor device. The drain voltage
is fixed at 0.4 V and the control-gate voltage is swept from 0
to 4 V. Each curve is an average of 10–12 experimental runs.
The curve with DI water is similar to the measured response
under dry conditions. Compared to the DI water reference, the
curves for poly-L-histidine and poly-L-lysine are shifted to
the left. This is because both poly amino acids, being positively
charged, produce an increase in the floating-gate voltage and a
decrease in the threshold voltage. Negatively charged poly as-
partic acid and poly glutamic acid shift the curves to
right of DI water reference.
The threshold voltages are extracted from Fig. 3 using the
transconductance method [7]. For the case of DI water refer-
ence, the threshold voltage is 1.5 V. A differential threshold
voltage ( ) is calculated as the difference
between the threshold voltages of a device under test and that for
DI water. For positively charged poly-L-histidine and poly-L-
lysine, the values are and , respectively. The
values for negatively charged poly aspartic acid and poly
glutamic acid are 0.35 V and 1.2 V respectively. Both poly-L-
histidine and poly-L-lysine have a much lower (0.01% w/v) con-
centration which explains their smaller magnitude. Poly
aspartic acid and poly glutamic acid are at a higher (0.5% w/v)
concentration and hence produce a larger magnitude. Each
I–V measurement is done 5–8 times and repeated (5–8 times)
again after refreshing the device using the above mentioned
steps (Section II). The values are consistent which in-
dicates reliability in detecting biochemical charges using the
bio-FETs.
We also measure the subthreshold characteristics of the
biosensor devices. The drain voltage is fixed at 2.5 V and the
control-gate voltage is swept from 0 V to 2 V. The subthreshold
I–V curves are recorded for different concentrations of PAA
solution (Fig. 4). The devices are refreshed after each run of
Fig. 4. Subthreshold current versus drain voltage is plotted for different PAA
concentrations and compared with the DI water reference curve.
Fig. 5. Relative shift in the logarithmic drain current (difference in a device’s
drain current with poly amino acid and with DI water) versus percentage con-
centration of four poly amino acids (PAA, PDDAC, PSASS).
5–8 trials and reused. The subthreshold I–V curves show a
consistent shift to the left with increasing PAA concentration.
A higher concentration of PAA molecules increases the sur-
face potential and lowers the threshold voltage . The sub-
threshold slope is unaltered, indicating that the charge distri-
bution is undisturbed during individual experiments [7]. These
results are also expected from our simulations.
To show concentration-dependent change in the threshold
voltage, two other poly amino acids (PSASS and PDDAC) are
tested (Fig. 5). Similar to the PAA case in Fig. 4, subthreshold
- curves are measured for different concentrations of the poly
amino acids. The relative shifts in the logarithmic subthreshold
I–V curve from that for DI water are plotted in Fig. 5. Error
bars are included for each data point. In all the three cases,
we observe a small current shift at lower ( w/v) charge
concentrations. As the charge concentration increases, the
current shift increases and eventually approaches saturation at
around 5% w/v concentration. The saturation point corresponds
to the scenario where all the surface sites are occupied with
charged molecules. The standard deviation in each data point
is small ( ) which indicates consistency and reliability of
the measurements.
The folded floating-gate device topology can be useful for
detecting spatial charge variations within a biological specimen.
Since the sensing area is directly above the FET, each device can

CHEN et al.: FOLDED FLOATING-GATE CMOS BIOSENSOR FOR THE DETECTION OF CHARGED BIOCHEMICAL MOLECULES 2909
Fig. 6. Snapshot of the fabricated chip showing single and an array of biosensor
devices. Cell lines placed on the chip adhered to the surface and are growing into
neurons (translucent globules).
be viewed as a pixel for charge sensing. Several devices in an
array can then be viewed as an electronic pixel array that can be
read in parallel. It is worth mentioning here that such an array
implementation would be challenging with earlier EGFETs as
their floating-gate is bifurcated and the large sensing areas are
placed far away from the FETs.
To show initial feasibility of the biosensor devices for VLSI
applications and biological detection, we fabricated an 8 8
array of the proposed devices (14.4 28.4 ) (Fig. 6).
Bit and word lines are included for parallel readout of multiple
devices in the array [6]. The control-gates of a single row are
connected to form a bit line while drain electrodes of a single
column are connected to form a word line. Each of the 8 bit
and word lines is addressable by an external analog-digital con-
verter and a custom LabView program. A device with simulta-
neously activated bit and word lines is automatically read and
its I–V characteristics are recorded. We observe that the indi-
vidual biosensor devices in the array behave electrically similar
to the isolated devices (Figs. 3–5). Fabrication in a smaller tech-
nology node (possibly 0.18 or 40 nm) can further scale the
array dimensions.
As a “proof-of-concept” for on-chip cell growth and viability,
neuronal cell lines (suspended in standard cell culture media)
are dropped on the fabricated chip which then adhered to the
chip surface and grew over time (Fig. 6). The on-chip cell cul-
ture is monitored for three days. The snapshot in Fig. 6 shows
successful adherence and growth of neurons on the chip surface
three days after placing the cell lines. The measured I–V char-
acteristics of the biosensor devices in the array are stable all
through the three days of neuronal growth. This indicates good
physical and chemical isolation of the floating-gate and min-
imal penetration of ions in the culture media through the overlay
glass.
Floating-gate MOS devices are known to trap residual
charges during the CMOS fabrication process, which alter the
effective threshold voltage of these devices [10]. Techniques
such as UV erasing, Fowler–Nordheim tunneling, and hot
electron injection are employed to remove this charge from
the floating-gate [10]. Studies have shown that the amount of
trapped charge is almost the same in all floating-gate transistors
fabricated on the same die and in the same fabrication run [11].
In our experiments, we also did not observe any noticeable
changes in threshold voltage among the different devices fab-
ricated on the same die. This may be because our devices are
placed close to one another on the same die and fabricated at
the same time. Furthermore, since we are measuring the differ-
ential threshold voltage (4), the effects of residual charge on
our experimental results are expected to be minimal. However,
if required, this initial charge can be eliminated by any of the
abovementioned techniques used for other floating-gate MOS
devices.
IV. CONCLUSION
In conclusion, we presented a folded floating-gate topology
for an EGFET where the sensing area is placed directly above
the FET. This allows us to shrink the lateral device dimensions
compared to the conventional bifurcated floating-gate topology.
Our tests suggest good physical and chemical isolation of the
floating-gate from the sensing area. The transconductance
and subthreshold characteristics provide a reliable means of
detecting charge fluctuations on the sensing area of individual
devices. The biosensor device is fully CMOS-compatible
and is shown feasible for VLSI applications with parallel
readout. Biocompatibility and successful growth of neuronal
cells on the chip surface has been demonstrated. Such an
electronic biosensor can enable possible real-time, noninvasive
observation of biologically relevant processes which involve
charge migration, or localization including cell growth and
regeneration.
REFERENCES
[1] M. Shinwar, M. Deen, and D. Landheer, “Study of the electrolyte-insu-
lator-semiconductor field-effect transistor (EISFET) with applications
in biosensor design,” Microelectron. Reliab., vol. 47, pp. 2025–2057,
2007.
[2] M. J. Schoning and A. Poghossian, “BioFEDs (Field-effect devices):
State-of-the-art and new directions,” Electroanalysis, vol. 19, pp.
1893–1900, 2006.
[3] L. Chen, M. Topinka, B. LeRoy, R. Westervelt, K. Maranowski, and
A. Gossard, “Charge-imaging field-effect transistor,” Appl. Phys. Lett.,
vol. 79, no. 8, 2001.
[4] D.-S. Kim, J.-E. Park, J.-K. Shin, P. Kim, G. Lim, and S. S. Park, “An
extended gate FET-based biosensor integrated with a Si microfluidic
channel for detection of protein complexes,” Sens. Actuators B, vol.
117, pp. 488–494, 2006.
[5] S. Meyburg, M. Goryll, J. Moers, S. Ingebrandt, S. Bocker-Meffert,
H. Luth, and A. Offenhausser, “N-Channel field-effect transistors with
floating gates for extracellular recordings,” Biosens. Bioelectron., vol.
21, pp. 1037–1044, 2006.
[6] J. M. Rabaey, A. Chandrakasan, and B. Nikolic, Digital Integrated Cir-
cuits. Upper Saddle River, NJ: Prentice-Hall, 2003.
[7] N. Shen, Z. Liu, C. Lee, B. Minch, and E. Kan, “Charge-based chem-
ical sensors: A neuromorphic approach with chemoreceptive neuron
MOS transistors,” IEEE Trans. Electron Devices, vol. 50, no. 10, pp.
2171–2178, Oct. 2003.
[8] M. Barbaro, A. Bonfiglio, L. Raffo, A. Alessandrini, P. Facci, and I.
Barakbarak, “A CMOS, fully integrated sensor for electronics detec-
tion of DNA hybridization,” IEEE Electron Device Lett., vol. 27, no. 7,
pp. 595–597, Jul. 2006.
[9] M. Barbaro, A. Bonfiglio, and L. Raffo, “A charge-modulated FET for
detection of biomolecular process: Conception, modeling, and simula-
tion,” IEEE Trans. Electron Devices, vol. 53, no. 1, pp. 158–166, Jan.,
2006.
[10] T. Tokuda, A. Yamamoto, K. Kagawa, M. Nunoshita, and J. Ohta, “A
CMOS image sensor with optical and potential dual imaging function
for on-chip bioscientific applications,” Sens. Actuators A, vol. 125, pp.
273–280, 2006.

[11] E. Rodriguez-Villegas, M. Jimenez, and R. Carvajal, “On dealing with
the charge trapped in floating gate MOS (FGMOS) transistors,” IEEE
Trans. Circuits Syst.-II Express Briefs, vol. 54, no. 2, pp. 156–160,
2007.
Baozhen Chen received the B.S. degree in condensed matter physics from the
University of Science and Technology of China (USTC), Hefei, China, in 2005
and the Ph.D. degree in electrical engineering from Iowa State University, Ames,
in 2010, and worked on the design, fabrication, and testing of CMOS-based
charge sensors. He also worked on microfluidic assays for high-throughput be-
havioral analysis of micro-organisms.
He is currently with a semiconductor company in Rayleigh, NC.
Archana Parashar received the B.Sc. degree in biochemistry from Patna Uni-
versity, Patna, India, in 2009 and the M.S. degree in biochemistry from the
Industrial Toxicology Research Centre (ITRC), India, in 2010. Currently, she
is working towards the Ph.D. degree at the Department of Electrical and Com-
puter Engineering, Iowa State University, Ames.
Her current projects include the design and measurement of CMOS-compat-
ible bioelectronics, behavioral microfluidics for C. elegans, and high-throughput
screening platforms for pathogens.
Santosh Pandey received the B.Tech. degree in electrical engineering from the
Indian Institute of Technology, Kharagpur, in 1999 and the Ph.D. degree in elec-
trical engineering from Lehigh University, Lehigh, PA, 2006. His doctoral dis-
sertation was focused on developing an integrated MEMS/CMOS silicon-based
platform for performing high-throughput cellular electrophysiology.
In 2006, he joined the Department of Electrical Engineering, Iowa State Uni-
versity, as an Assistant Professor. His research interests lie in engineering high-
throughput screening platforms for cell cultures and whole organisms by inte-
grating semiconductors, microfluidics, systems biology, and image-processing
processing tools.

Folded Floating Gate CMOS Biosensor for Nano Charge Sensing

More Related Content

Similar to Folded Floating Gate CMOS Biosensor for Nano Charge Sensing (20)

More from Iowa State University (20)

Recently uploaded (20)

Folded Floating Gate CMOS Biosensor for Nano Charge Sensing