Home
Search
Collections
Journals
About
Contact us
My IOPscience
Label-free DNA detection with a nanogap embedded complementary metal oxide
semiconductor
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2011 Nanotechnology 22 135502
(http://iopscience.iop.org/0957-4484/22/13/135502)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 143.248.174.71
The article was downloaded on 23/02/2011 at 04:10
Please note that terms and conditions apply.
IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 22 (2011) 135502 (5pp)
doi:10.1088/0957-4484/22/13/135502
Label-free DNA detection with a nanogap
embedded complementary metal oxide
semiconductor
Chang-Hoon Kim1 , Cheulhee Jung2 , Kyung-Bok Lee3 ,
Hyun Gyu Park2 and Yang-Kyu Choi1,4
1
Department of Electrical Engineering, College of Information Science and Technology
KAIST, 335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea
2
Department of Chemical and Biomolecular Engineering (BK 21 Program), KAIST,
335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea
3
Division of Life Science, Korea Basic Science Institute, 113 Gwahangno, Yuseong-gu,
Daejeon 305-333, Republic of Korea
E-mail: ykchoi@ee.kaist.ac.kr
Received 1 October 2010, in final form 28 January 2011
Published 22 February 2011
Online at stacks.iop.org/Nano/22/135502
Abstract
A nanogap embedded complementary metal oxide semiconductor (NeCMOS) is demonstrated
as a proof-of-concept for label-free detection of DNA sequence. When a partially carved
nanogap between a gate and a silicon channel is filled with charged biomolecules, the gate
dielectric constant and charges are changed. When the gate oxide thickness reduces, the
threshold voltage is significantly affected by a change of the charges, whereas it is scarcely
influenced by a change of the dielectric constant. In the case of DNA, those two factors act on
the threshold voltage oppositely in an n-channel NeCMOS but collaboratively in a p-channel
NeCMOS because of the negative charges of DNA. Hence, a p-channel NeCMOS with a thin
gate oxide is more attractive for DNA detection because it enhances the shift of threshold
voltage; that is, it improves the sensitivity of DNA detection. In addition, the shift of threshold
voltage according to the nanogap length is also investigated and the longer nanogap shows more
shift of the threshold voltage.
event through keeping track of changes occurred in the
physical and chemical properties of the recognition layer. An
ion-sensitive field-effect transistor (ISFET) was proposed for
chip-based DNA detection without a labeling process [15]. The
ISFET sensed charged biomolecules by monitoring changes in
the drain current ( ID ) or threshold voltage (VT ) on an opened
silicon (Si) channel. Such changes originated from the charged
biomolecules near the interface of an ionic solution and the
Si-channel. A solid-state gate in a conventional MOSFET
was replaced with a liquid gate; that is, the ionic solution
in the ISFET. When non-charged or neutralized biomolecules
are introduced onto the ISFET, it cannot work properly as
a high-sensitivity sensor. In spite of that disadvantage, the
ISFET has attracted attention by virtue of its partial CMOS
compatibility and its mature and established technology for
signal transfer. Recently, a Si nanowire FET-based biosensor
has been reported [16]. Its operational principle is the
1. Introduction
Investigation of DNA hybridization events is important in
the fields of genomics, diagnostics, pharmacology, and gene
expression. For a study of gene expression, polymerase chain
reaction (PCR) [1, 2], which is the most famous technique for
DNA detection, has been widely used for the amplification
of DNA. Thereafter many DNA detection methods such as
fluorescence [3–5], electrochemistry [6–8], enzyme [9], and
nanoparticle [10–14]-based techniques have been developed.
However, those techniques need a labeling step.
The
incorporation of a labeling step into DNA detection assay has
shortcomings of limited labeling efficiency, complex multistep analysis, and contamination involved in the preparation of
samples. Many researchers have made an attempt to develop a
label-free technique, which directly detects the hybridization
4 Author to whom any correspondence should be addressed.
0957-4484/11/135502+05$33.00
1
© 2011 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 22 (2011) 135502
C-H Kim et al
phosphorous, respectively. The active channel was defined
by photolithography and Si mesa etching. Afterward, a
gate oxide of 50 nm was grown thermally. The gate oxide
supports a gate poly-Si and also acts as a sacrificial layer for
nanogap formation. The edge of the gate oxide was etched
to make nanogaps after poly-Si gate patterning. Sequentially,
a 250 nm layer of undoped poly-Si was deposited by low
pressure chemical vapor deposition (LPCVD) and patterned by
photolithography and poly-Si etching for the gate electrode.
The undoped poly-Si and a source/drain (S/D) region were
implanted by 5 × 1015 cm−2 and 50 keV of arsenic and
3 × 1013 cm−2 and 30 keV of phosphorous for the n-channel
NeCMOS. Similarly, 4 × 1015 cm−2 and 20 keV of BF2
and 2 × 1013 cm−2 and 30 keV of boron were used for the
p-channel NeCMOS, respectively. Rapid thermal annealing
(RTA) at 1000 ◦ C was then applied for 10 s. The nanogap
length in the source to drain direction was controlled by
varying the duration of the lateral wet etching with 1:6 diluted
buffered oxide etchant (BOE). The time was split into 3, 6, and
10 min periods. The NeCMOS was subsequently reoxidized
at 700 ◦ C for 30 min to regrow a gate oxide with a thickness
of 4 nm in the nanogap region. Figure 1 shows a crosssectional transmission electron microscopy (TEM) image of
the NeCMOS. The length of the nanogap was 250 nm, 430 nm,
and 735 nm for 3 min, 6 min, and 10 min of lateral wet etching,
respectively.
same as the ISFET. When external charged biomolecules are
immobilized on the nanowire, electrical characteristics such
as drain current ( ID ), transconductance (gm ), and threshold
voltage (VT ) are accordingly changed. But, it can work
only for charged biomolecules. To resolve that problem, the
dielectric modulated field-effect transistor (DMFET), which
was proposed to detect non-charged biomolecules with use of
an n-channel [17], is very similar to a conventional MOSFET
except that it has caved nanogaps filled with air at the edge
of the gate electrode. When biomolecules are introduced
into the nanogaps, the dielectric constant is increased from
one (air) to a specific number larger than one (biomolecules).
Consequently, ID and VT can be changed by the increased
gate dielectric constant in the gate dielectric from the initial
characteristics of the air gap state. This type of biosensor has
been used to detect the binding of biotin–streptavidin [17] and
antigen–antibody in the case of avian influenza [18], which
are weakly charged, thus the effect of the gate dielectric
constant was primarily investigated without consideration of a
charge effect arising from the charged biomolecules. However,
the charge effect becomes increasingly important for highly
charged biomolecules such as DNA. And it is uncertain which
is more important between the effect of the dielectric constant
and the charge for detection of highly charged biomolecules.
Herein we fabricated both an n-channel NeCMOS that uses
inverted electrons and a p-channel NeCMOS that uses inverted
holes so that we could investigate the charge polarity effect of
the charged biomolecules and quantitatively compare the effect
of the dielectric constant and the charge strength. It should be
noted that the charge effect tends to be dominant when the gate
oxide thickness becomes thinner because those charges more
strongly influence the channel potential. Thus, a very thin
gate oxide thermally grown after the nanogap formation was
employed. Highly negatively charged DNA and neutralized
peptide nucleic acid (PNA) are used as the target biomolecules.
The increase of dielectric constant makes VT be shifted toward
the negative side in the n-channel ID –VG plot whereas it pushes
VT to move toward the positive side in the p-channel one. In
contrast, the negatively charged DNA makes VT be shifted
toward the positive side in both the n-channel and p-channel
ID –VG plots. From these plots, it can be deduced that the effect
of the dielectric constant and the charge act on VT oppositely
in the n-channel but in the same direction in the p-channel
device. Consequently, the p-channel is highly preferred for the
detection of DNA hybridization.
2.2. Reagents and DNA hybridization
To achieve the covalent attachment of amine-terminated probe
DNA (pDNA) onto the nanogap sample, we used a method
of common reactive intermediates, pentafluorophenyl (PFP)
esters. We introduced the PFP ester groups onto the surface of
the nanogap and reacted the activated surface with the amine
group of the pDNA, because the PFP ester groups were easily
coupled with amines, leading to the formation of amide bonds.
The PFP ester-terminated self-assembly monolayer (SAM)
was prepared from amine-terminated SAM by following the
two-step surface reactions. For the SAM formation, the OHgroup was first created by the application of a hot piranha
solution (H2 SO4 :H2 O2 = 1:1; safety note: piranha solutions
can react violently with organic materials) to the oxide surface
of the nanogap for 10 min. The formed OH-group was
sequentially treated with 4% of aminopropyl-trimethoxy silane
(APTMS) in toluene for 3 min and washed with toluene and
EtOH. The reaction of the APTMS and the OH-group was
enhanced by a baking process at 110 ◦ C for 5 min on a hotplate.
The terminal amine groups were immersed in 0.1 M of succinic
anhydride in dimethylformamide (DMF) for 6 h to produce a
carboxylic acid (COOH)-group at one end of the SAM. The
surface of the COOH-terminated nanogap was activated by
immersing the nanogap into an ethanol solution of 1-ethyl-3(dimethylamino) propylcarbodiimide (EDC) (0.1 M) and PFP
(0.2 M) for 30 min. The PFP-activated nanogap samples were
then rinsed with ethanol, dried under a stream of nitrogen,
and used immediately thereafter. Next, the synthesized pDNA
was pretreated with 100 mM dithiothreitol, purified using an
NAP™ 5 column, and dissolved in DI water. Then, the
2. Experiment
2.1. Device fabrication
The n- and p-channel NeCMOSs were both fabricated on an
SOI wafer. The wafer had a buried oxide layer with a thickness
of 380 nm, a silicon body with an initial thickness of 230 nm,
and a boron doping concentration of 5 × 1015 cm−3 . The body
was thinned to 100 nm by iterative oxidation and wet etching.
Channel doping was achieved by boron and phosphorous
implantation for the n- and p-channels, respectively. The
dose and energy used for implantation were 2 × 1012 cm−2
and 30 keV for boron and 2 × 1012 cm−2 and 40 keV for
2
Nanotechnology 22 (2011) 135502
C-H Kim et al
Figure 1. Cross-sectional images of TEM according to the gate oxide etching time: (a) 3 min, (b) 6 min, and (c) 10 min etching, respectively.
Figure 2. (a) Schematic of fabricated device and (b) threshold voltage change due to the immobilized charge density, according to the
regrown gate oxide thickness.
PFP-activated nanogap samples were immersed into 10 μM
aqueous pDNA solution for 2 h. After washing with water and
ethanol, the nanogap samples were moved into 1 mM aqueous
mercaptohexanol solution for 30 min. Finally, the fabricated
nanogap samples were washed with water and ethanol, and
dried under nitrogen atmosphere. 10 μM solutions of the
target DNA (tDNA), non-target DNA (ntDNA), and target
PNA (tPNA) were applied and allowed to hybridize at room
temperature for 2 h, followed by washing with water and
ethanol. To avoid non-specific binding by the NH2 -group of
tPNA, the NH2 -terminal of PNA was capped with an acetyl
group. Washing with EtOH between each step was carried out
to eliminate any water residue in the nanogap. All reagents
were dissolved in DI water so that the positive ions did not
encapsulate the negatively charged DNA. The DNA sequence
used was artificially synthesized based on the BRCA 1 gene,
which is related to breast and ovarian cancers. It is composed
of a specific 12 mer base-pair, which is a small fraction of the
total BRCA 1 gene. This small number of base-pairs is not
enough to detect the BRCA 1 gene stably, but it is enough
to demonstrate the proof-of-concept [19]. The experimental
reagents were as follows:
probe DNA (pDNA): amine-‘5-CTTCTTCATATT-3’,
target DNA (tDNA): ‘5-AATATGAAGAAG-3’,
non-target DNA (ntDNA): ‘5-GATGGTCAGGTA-3’,
target PNA (tPNA): acetyl-N-AATATGAAGAAG-C.
3. Results and discussion
In this work, NeCMOS with a regrown thin gate oxide
was fabricated to investigate the charge effect. Figure 2(a)
shows a schematic of the fabricated device. One of the
purposes of the thick gate oxide (Tox ) in the center region
was a sacrificial layer for the subsequent nanogap formation.
Another purpose is to support the poly-crystalline silicon
(poly-Si) gate mechanically. After the nanogap formation,
another thin gate oxide (Trox ) was thermally regrown in the
nanogaps to avoid direct current flow, i.e., electrical short
between the gate and the channel via conductive biomolecules.
Trox corresponds to the distance between the biomolecules and
the Si-channel. By the reduction of Trox , the charge effect
on the shift of VT is intentionally amplified. The VT shift by
the charges was investigated for various Trox by the use of
a commercial semiconductor simulation tool (SIVACO) [20].
Figure 2(b) shows the VT shift of the n-channel NeCMOS
by the immobilized charge density for various Trox . For this
simulation, Tox was fixed at 50 nm and Trox was varied from 5
to 45 nm. The charge density was split to 1 × 1011 , 5 × 1011 , and
1 × 1012 cm−2 . In figure 2(b), the VT shift was increased as the
gate oxide thickness was reduced. Thus the thin gate oxide of
5 nm thickness unveils the charge effect, which was concealed
by the dielectric constant effect in previous works [17, 18].
Two types of devices: n- and p-channel NeCMOSs were
3
Nanotechnology 22 (2011) 135502
C-H Kim et al
Figure 3. Transfer characteristics ( ID –VG ) of (a) an n-channel and (b) a p-channel NeCMOS for a targeted work group and various control
groups. Statistical variation of the VT shifts for tDNA, ntDNA, and tPNA hybridizations in (c) the n-channel and (d) the p-channel NeCMOS.
Each data set is collected from four devices.
negative charges. In a control group, as shown in figure 3, the
ntDNA did not show any significant VT shift.
The NeCMOS had a larger VT shift than the ISFET
because of the material state of the gate; that is, the NeCMOS
had a solid gate whereas the ISFET had a liquid gate. In the
ISFET, the channel is exposed to a solution that contains the
target biomolecules and various ions. Because counter ions in
the solution make an ionic double layer, the negative charges
of the DNA are screened by this ionic double layer [22].
Thus, the sensitivity can be decreased. In the NeCMOS, the
gate was made of solid-state poly-Si and the gate voltage was
applied through the contact pads; hence, only a small fraction
of the channel was exposed to air. The characteristics of the
NeCMOS were therefore measured in a dry atmosphere and
the ionic double layer effect was eliminated. Consequently, the
VT shift is a few 100 mV in the NeCMOS but a few 10 mV in
the ISFET.
Figures 3(c) and (d) show the statistical distribution of
the VT shift for the tDNA detection in the n- and p-channel
devices, respectively. After the tDNA hybridization, there is a
VT shift of +0.19 V in the n-channel device and +0.84 V in
the p-channel device. This outcome clearly confirms that the
p-channel device is attractive in terms of sensitivity because
of the synergy of the dielectric constant and the charge effect.
For the tPNA hybridization, there is a VT shift of −0.56 V
in the n-channel device and +0.1 V in the p-channel device.
The dielectric constant-induced VT shift is known to depend
on the body doping concentration [23, 24]; hence, it is difficult
to make a fair comparison between the separately designed
n- and p-channel devices. However, the charge-induced VT
fabricated to verify the charge polarity effect. Figures 3(a)
and (b) show the representative transfer ID –VG characteristics
from the n- and p-channel NeCMOSs for each bio-treatment
step. The target molecules are a charged DNA and an
uncharged PNA sequence. In the n-channel device, the VT shift
was +0.36 V by tDNA and −0.55 V by tPNA. The negative
VT shift by tPNA was clearly due to the tPNA hybridization,
which induced increment of the dielectric constant in the gate
dielectric layer. To understand the positive VT shift caused
by the tDNA hybridization, we need to consider the two
previously mentioned factors: the dielectric constant and the
charge effect. The tDNA to induce increment of the dielectric
constant in the gate dielectric layer caused VT to shift toward
the negative side. However, the negative charges of DNA,
which play the role of the negative fixed charges in the gate
dielectric layer, cause VT to shift toward the positive side [21].
As a result, two competing factors have an opposite effect
on the direction of the VT shift in tDNA hybridization. This
difference reveals that the detection sensitivity was lower in an
n-channel device than in a p-channel device. In figure 3(a),
VT was shifted to the positive side. Hence, the charge effect
is the dominant factor in the VT shift in tDNA hybridization.
In the p-channel device, the increased dielectric constant and
negative charges of the DNA together made VT be shifted
toward the positive side. Figure 3(b) shows the tDNA detection
in the p-channel device. The VT shift was +0.96 V after tDNA
hybridization and +0.12 V after tPNA hybridization. For tPNA
detection, VT was less shifted toward the positive side by only
the increment of the dielectric constant. For tDNA detection,
on the other hand, VT was more shifted toward the positive
side by both the increment of the dielectric constant and the
4
Nanotechnology 22 (2011) 135502
C-H Kim et al
the Nano R&D program through the National Research
Foundation of Korea funded by the Ministry of Education,
Science, and Technology (2010-0018931). This work was
partially supported by the Basic Science Research Program
through the National Research Foundation of Korea (NRF)
funded by the Ministry of Education, Science and Technology
(No. 2010-0000838).
References
[1] Saiki R K, Scharf S, Faloona F, Mullis K B, Horn G T,
Erlich H A and Arnheim N 1985 Science 230 1350
[2] Saiki R K, Gelfand D H, Stoffel S, Scharf S J, Higuchi R,
Horn G T, Mullis K B and Erlich H A 1998 Science 239 487
[3] Bars M, Cloarec J P, Bessueille F, Souteyrand E, Martin J R
and Chauvet J P 2000 J. Fluorescence 10 247
[4] Marton M J et al 1998 Nat. Med. 4 1293
[5] Lu E, Peng X, Song F and Fan J 2005 Bioorg. Med. Chem. Lett.
12 255
[6] Huang X J, Choi Y K, Im H S, Yarimaga O, Yoon E S and
Kim H S 2006 Sensors 6 756
[7] Palecek E, Fojita M, Tomschik M and Wang J 1998 Biosens.
Bioelectron. 13 621
[8] Marrazza G, Chianella I and Mascini M 1999 Biosens.
Bioelectron. 14 43
[9] Defillipo K A and Grayeski M L 1991 Anal. Chim. Acta
249 155
[10] Miller M M, Sheehan P E, Edelstein R L, Tamanaha C R,
Zhong L, Bounnak S, Whitman L J and Coltaon R J 2001
J. Magn. Magn. Mater. 255 138
[11] Tansil N C and Gao Z 2006 Nano Today 1 28
[12] Hansen J A, Mukhopadhyay R, Hansen J O and Gothelf K V
2006 J. Am. Chem. Soc. 128 3860
[13] Rochelet-Dequaire M, Limoges B and Brossier P 2006 Analyst
131 923
[14] Wark A W, Lee H J, Qavi A J and Corn R M 2007 Anal. Chem.
79 6697
[15] Bergveld P 2003 Sensors Actuators B 288 1
[16] Zhang G J, Chua J H, Chee R E, Agarwal A, Wong S M,
Buddgaraju K D and Balasubramanian N 2008 Biosens.
Bioelectron. 23 1701
[17] Im H S, Huang X J, Gu B S and Choi Y K 2007 Nat.
Nanotechnol. 2 430
[18] Gu B S, Park T J, Ahn J H, Huang X J, Lee S Y and
Choi Y K 2009 Small 5 2407
[19] Futreal P A et al 1994 Science 266 120
[20] Atlas 2008 Two Dimensional Device Simulation Program
(SILVACO)
[21] Bendriaa F, Le-Bihan F, Salaun A C, Mohammed-Brahim T,
Tlili C, Jaffrezic N and Korri-Youssoufi H 2005 Sensors 3
412
[22] Song K S, Zhang G J, Nakamura Y, Furukawa K, Hiraki T,
Yang J H, Funatsu T, Ohdomari I and Kawarada H 2006
Phys. Rev. E 74 041919
[23] Kim A, Ah C S, Yu H Y, Yang J H, Baek I B, Ahn C G,
Park C W and Jun M S 2007 Appl. Phys. Lett. 91 103901
[24] Nair P R and Alam M A 2007 IEEE Trans. Electron Devices
54 12
Figure 4. The VT shift dependenc on the nanogap length.
shift is insensitive to the doping concentration. From VT =
VT (tDNA)−VT (tPNA) in figures 3(c) and (d), the VT shift was
0.75 V in the n-channel and 0.74 V in the p-channel; that is, the
shift is almost the same regardless of the channel polarity. This
behavior confirms that the same amount of tDNA is hybridized
to pDNA and that the result is statistically reliable. Note,
as expected, that a negligible VT shift was observed in both
the n- and p-channel NeCMOSs in the case of the ntDNA
hybridization, which is another control group.
Figure 4 shows that the VT shift is dependent on the length
of the nanogap, which is determined by the lateral wet etching
time when the buffered oxide etchant (BOE) is used to remove
the oxide. In this experiment, only p-channel devices are used
because of their improved sensitivity. As predicted, the longer
the nanogap, the larger the VT shift. This behavior occurs
because the dielectric constant and the charge both intensively
affect the channel potential at the interface of the gate oxide
and the channel.
4. Conclusions
A DNA sequence was detected with n-channel and p-channel
nanogap embedded MOSFETs to demonstrate the proof-ofconcept of label-free DNA detection and verify the effect of the
dielectric constant and the charge. The two factors affect the
VT differently in the n-channel device and synergically in the pchannel device. As a result, the VT shift was four times greater
in the p-channel device than in the n-channel device. The
nanogap length dependence was also investigated, and a longer
nanogap is preferred for improved sensitivity. It is planned that
the limit of detection by use of the p-channel device with long
nanogap length will be investigated in further works.
Acknowledgments
This work was supported by the National Research and
Development Program (NRDP, 2010-0002108) for the development of biomedical function monitoring biosensors and
5