Lab on a Chip

View Article Online

PAPER View Journal | View Issue

Node-pore sensing: a robust, high-dynamic range

method for detecting biological species3
Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

Cite this: Lab Chip, 2013, 13, 1302

Karthik R. Balakrishnan,a George Anwar,a Matthew R. Chapman,b
Trongtuong Nguyen,c Anand Kesavarajud and Lydia L. Sohn*a

Resistive-pulse sensing (RPS), which is based on measuring the current pulse produced when a single
particle transits a pore or channel, is an extremely versatile technique used to determine the size and
concentration of cells and viruses and to detect single molecules. A major challenge to RPS is dynamic
range: smaller particles in a heterogeneous sample can go undetected because of low signal-to-noise
ratios (SNRs) and the fact that the pore size must be commensurate with that of the largest particles. Here,
we describe a fundamentally different pore that provides an unprecedented dynamic detection range,
Received 21st November 2012,
Accepted 21st January 2013
from tens of nanometers to several microns in size, without the need for pre-sorting or filtration. Because
of its unique geometry – nodes inserted along the channel – our pore produces distinct electronic
DOI: 10.1039/c3lc41286e
signatures that overcome low SNRs. We demonstrate the power of our device by directly detecting and
www.rsc.org/loc enumerating human immunodeficiency virus (HIV) in human plasma.

Introduction requires the utilization of nanoscale pores such that both

Vparticle/Vpore and SNR significantly increase.
RPS, or the Coulter-counter technique of particle sizing, plays As there is a driving interest to measure single molecules of
an integral role in cell biology and medicine. For example, RPS DNA for sequencing and HIV in human plasma for improving
provides a quick, label-free method to count cells and to disease detection and monitoring, methods for fabricating
determine accurately cell-size distributions1–5 and cellular smaller, more definable pores have grown more sophisticated:
response to stimuli.6,7 Particle sizing with RPS involves from utilizing high-energy nuclear particles to irradiate a
measuring the current across a pore connected to two plastic sheet that is then etched,11 to employing naturally-
reservoirs. When transiting a pore, a particle partially blocks occurring or genetically-modified biological proteins12,13 such
the flow of current, leading to a transient increase in the pore’s as alpha-hemolysin,14–20 to creating hybrid21,22 or solid-state
electrical resistance. The resulting normalized resistance, or pores23–26 using electron-beam lithography and atomic-layer
change in current from baseline, is approximately propor- deposition, to directly employing carbon nanotubes.27–29
tional to the volume ratio of the particle to the pore, i.e. |DI|/I While the resulting pores from each of these different
3 Vparticle/Vpore.8 Commercial Coulter counters can detect |DI|/ methods have been successful in measuring nanometer-sized
I corresponding to Vparticle/Vpore y 1.6 6 1025 when a dynamic objects, as Vparticle/Vpore y 1 in many of these cases, they may
aperture range of 64 000 : 1 by volume is employed (Multisizer suffer from clogging due to larger (contaminating) particles or
4 Coulter Counter). On-chip9 resistive-pulse sensors are even aggregates. Furthermore, if the sample to be measured
more sensitive, with the state-of-the-art capable of measuring consists of a heterogeneous population of particles ranging
|DI|/I as low as Vparticle/Vpore y 4 6 1026. Ultimately, SNR in size from nanometers to several microns, as is the case of
(defined10 as |DI|/IRMS) is the limiting factor to sensitivity, and HIV-infected human plasma, the pore must be commensurate
even the use of sophisticated signal amplification and noise with the larger-sized particles in the sample to prevent
filtering is insufficient for lower Vparticle/Vpore ratios. Detecting clogging. Consequently, significantly smaller-sized particles
nanoscale particles therefore becomes very challenging and in the population, such as HIV, may go undetected, and
alternative methods such as polymerase chain reaction
a
Department of Mechanical Engineering, University of California at Berkeley, 5118 (PCR)30–34 must be employed, adding to time, cost, and
Etcheverry Hall, Mail Stop 1740, Berkeley, CA, 94720-1740, USA. potential loss of sample during preparation.
E-mail: sohn@me.berkeley.edu; Tel: +1 510-642-5434 Here, we address these current limitations and present a
b
Biophysics Graduate Group, University of California, Berkeley, CA, 94720, USA
c
unique, yet simple, method – node-pore sensing (NPS) – for
Engineering Physics, University of California, Berkeley, CA, 94720, USA
d
Department of Bioengineering, University of California, Berkeley, CA, 94720, USA
improving the detection capabilities of resistive-pulse sensors.
3 Electronic supplementary information (ESI) available. See DOI: 10.1039/ NPS, which involves utilizing pores that have a sequence of
c3lc41286e nodes along the channel, produces distinct electronic signa-

1302 | Lab Chip, 2013, 13, 1302–1307 This journal is ß The Royal Society of Chemistry 2013
 View Article Online

Lab on a Chip Paper

Ti/Pt/Au thin film was deposited onto the substrates using an

electron-gun evaporator. Electrodes were gold-etched to expose
Pt.

Preparation of node-pore sensors

Negative-relief masters made of SU-8 photoresist on polished
silicon wafers were used to create the node-pore sensors.
PDMS (Sylgard 184) (10 : 1, pre-polymer : curing agent) was
poured onto the masters after degassing and then cured for 1
h at 80 uC. A slab of PDMS containing the node-pore design
was cut from the masters and entry and exit ports were cored
Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

using 16 G syringe needles. The PDMS was subsequently

plasma bonded (200 mTorr, 80 W) onto the glass substrates
with the pre-patterned electrodes at 80 uC for 30 min.
Fig. 1 Schematic of the node-pore sensor platform consisting of a polydi-
methylsiloxane (PDMS) mold bonded to a glass substrate with platinum (Pt) Colloid experiments
electrodes and gold (Au) contact pads that enable a four-terminal measurement
of the electrical current across the pore. The device consists of two reservoirs, Colloidal samples included polystyrene nanobeads (50 nm and
cored with entry and exit ports, connected by a small channel, i.e. pore. 500 nm from Polysciences, Inc.), polystyrene microspheres
Magnified view: In this particular node-pore sensor design, the pore is (15.45 mm from Bangs Laboratories, Inc.), and latex beads (4.9
segmented into three regions, separated by two nodes of larger cross-sectional
mm and 930 nm from Interfacial Dynamics Corp.). Samples
area than the rest of the pore.
were washed with phosphate buffered saline (1x, pH 7.4)
before being driven through node-pore devices.

tures that can be easily distinguished within the measured

noise and can be translated into an enhanced fast Fourier
transform (FFT) power spectra for rapid detection. We are able
Results
to perform measurements with Vparticle/Vpore as low as 1.2 6 Versatility of design and signal measurements
1029, thus enabling the direct screening of nanoscale particles
To demonstrate the unique electronic signatures that our
with microscale devices without any concern for clogging and node-pore sensors produce, we show how the current pulse
need for signal amplification. generated by a 15.45 mm polystyrene colloid transiting a pore
changes with an increasing number of nodes. Fig. 2a (left) is a
fluorescent image of a conventional pore (25 mm 6 25 mm 6
Methods 800 mm, H 6 W 6 L) and the corresponding current pulse
measured (right). As expected, the pulse has a well-defined
Platform design square shape that is typically recorded in standard RPS
Fig. 1 shows a device created with soft lithography. The device measurements. In contrast, Fig. 2b (left) shows a pore (25
consists of a polydimethylsiloxane (PDMS) mold that is mm 6 20 mm 6 2400 mm, H 6 W 6 L) that has two, equally-
bonded to a glass substrate with pre-defined platinum (Pt) spaced 50 mm-wide nodes and the corresponding pulse
electrodes and gold (Au) contact pads. The mold has two produced (right). Unlike the pulse in Fig. 2b, this pulse is
reservoirs that are connected by a single fluidic channel (the clearly modulated. As the colloid enters the node-pore sensor,
‘‘pore’’). In direct contrast to standard RPS pores, our pore is the current initially drops from the baseline as expected. When
segmented by a series of ‘‘nodes’’. These nodes increase our the colloid enters the first node, however, the current rises,
pore sensitivity three orders beyond current state-of-the-art RPS only to drop again once it exits. This rise and fall in current is
sensors. Because of our use of standard lithography, we have repeated as the colloid enters and exits the next node. Finally,
extensive flexibility and control to include as many nodes, when the colloid exits the pore, the current returns to the
spaced as close or as far apart, as are advantageous for baseline value. The current modulation we observe is a
measurements. We drive particles through the node-pore hallmark of all node-pore sensors. Fig. 2c shows the current
sensor using a non-pulsatile pressure (y2.5 psi), and utilize a pulse recorded when a pore with eleven equally-spaced nodes
four-point measurement with a constant applied AC voltage is employed, and Fig. 2d, when a pore has seven variably-
(0.32 V), as previously published,4,35–38 to measure the current spaced nodes. In each case, the current modulation is clearly
across the pore. Subsequent data analyses, including calculat- identifiable and reflects the number and spacing of the nodes
ing DI/I (Fig. S1, ESI3), low-pass filtering (Fig. S2, ESI3), and in a pore. Thus, flexibility in device fabrication allows for
performing an FFT analysis (Fig. S4, ESI3), are accomplished variation in the number of nodes and their spacing, offering
using custom software written in LabVIEW. design versatility in NPS.
The unique measurement that our pores provide can be
Electrode fabrication understood by analyzing the resistivity changes as a particle
Electrodes were lithographically-patterned onto glass sub- transits through a node-pore sensor. Assuming a non-
strates using Microposit S1813 (Dow) resist. A 75/250/250 Å conducting particle and a pore with radius rparticle and rpore,

This journal is ß The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 1302–1307 | 1303
 View Article Online

Paper Lab on a Chip

Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

Fig. 2 Scale bar, 100 mm. (a–d) Fluorescent images of various pore designs (left) and the current pulses produced when a 15.45 mm colloid transits each pore (right).
(a) A conventional pore without nodes. (b–d) Node-pore sensors with two (b) and eleven (c) equally-spaced nodes, and seven variably-spaced nodes (d). Distinct
current modulation can be detected due to the presence of nodes. (e) NPS theory. (i) Schematic showing a spherical, non-conducting particle (red) with radius, r
particle, centered in a pore with radius, r pore. dx and r correspond to the thickness and radius of a cross-sectional slice where x is along the flow direction axis. (ii)
View of a cross-sectional slice of the entire pore. Red and gray represent the particle cross-section and the cross-sectional area between the pore and the particle,
respectively. (iii) Cross-sectional slice of a particle within a rectangular pore of height H and width W. (iv) Cross-sectional slice of a particle within a rectangular node
region of height H and width Wnode . W.

respectively, the resistance, dR, of a cross-sectional slice of sectional area, the measured change in resistance caused by
thickness dx is1 the particle transiting the pore is constant (eqn (2)). If the
cross-sectional area of the pore changes, then the resistance
rfluid dx
dR~ (1) will also change as the particle transits the pore. Since the
DA
resistance depends on r of each cross-sectional slice of width
where rfluid is the fluid resistivity and DA is the difference dx (at any given time), we can specifically tailor the shape of
between the pore cross-sectional area, Apore, and the particle the pore to provide a desired resistance value (and thus
cross-sectional area, Aparticle (Fig. 2e, i and ii). We can express current) measurement.
the total increase in resistance measured across the pore due
Detection capabilities for heterogeneous populations
to the presence of a particle as
ð rparticle ð rparticle Because the unique electronic signatures produced by node-
rfluid dx rfluid dx pore sensors are easily identifiable, SNR becomes far less of an
R~ { (2)
{rparticle DAðxÞ {rparticle Apore ðxÞ issue than with a standard pore. Even with SNRs % 1, signals
within noise can be extracted using simple low-pass filtering
where x is along the flow direction axis. The above analysis and identified by their unique signature (Fig. S2, ESI3). By
assumes a cylindrical pore; however, the planar lithography utilizing a single, microscale pore (8 mm 6 10 mm 6 500 mm,
process we employ results in rectangular cross-sections with H 6 W 6 L) with 4 nodes that are equidistant apart (Fig. 3a),
width, W, and height, H. Thus, the cross-sectional area we can measure colloids ranging from 50 to 4900 nm in size
difference of a node-pore slice is (Fig. 3b–e). While the current pulse produced by the passage of
the smaller colloids is difficult to distinguish from the
DArectangular(x) = WH 2 r(x)2 (3) baseline current, the unique modulation of the current pulse
in the low-pass filtered signal makes the signature readily
where r(x) is the particle slice radius (Fig. 2e, iii). In a node identifiable for detection (Fig. 3e). By designing pores with
region of the pore, the increased cross-sectional area differ- patterned node arrangements and by screening samples at a
ence of a pore slice is specific flow rate, the current-modulated pulse can be
predicted and identified. Utilizing this particular strategy, we
DAnode(x) = WnodeH 2 r(x)2 (4) are able to screen successfully 50 nm colloids with the large
micron-sized pore shown in Fig. 3a. Furthermore, Vparticle/Vpore
where Wnode is the width of the node region (Fig. 2e, iv). Thus, y 1.2 6 1029, which to our knowledge, is the lowest ever
when a particle transits a node region with width Wnode . W, recorded by an RPS-based system.
DA is larger than that in the constant cross-section region, and
in turn, the change in resistance, DR, drops while the particle
is within the node. Ultimately, if a pore has a constant cross-

1304 | Lab Chip, 2013, 13, 1302–1307 This journal is ß The Royal Society of Chemistry 2013
 View Article Online

Lab on a Chip Paper

high frequency (fhigh) and low frequency (flow) peaks corre-

sponding to the electronic signatures produced by the 100 mm
node-spacing and 500 mm node-spacing segments of the pore,
respectively. Node/pore combinations thus translate spatial
frequencies directly into measurable temporal frequencies
that can be directly detected and analyzed. Therefore, a single
pore with a specific number of nodes optimized to enhance
the corresponding FFT power spectrum has improved detec-
tion capabilities. This unique FFT analysis allows for NPS to
utilize FFT detection algorithms for data extraction, which
could ultimately improve data processing times.
Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

Viral detection
Using the same microscale four-node pore shown in Fig. 3a,
we detected HIV particles, whose size range from 100 to 200
Fig. 3 Detection capabilities of a four-node device (8 mm 6 10 mm 6 500 mm, H
nm in diameter39,40 depending on viral maturity. In three
6 W 6 L) with varying sized colloids. Scale bars, 100 mm. (a) Fluorescent image
of the device. (b–d) Representative current pulse of a 4900 nm (b), 930 nm (c), experiments, we screened 50 nL of human plasma containing
500 nm (d), and 50 nm (e) colloid traversing the node-pore sensor after 100 000 copies per mL of a replication-incompetent strain of
applying a low-pass filter (Fig. S2, ESI3). Four distinct peaks (red circles) within HIV (8E5 LAV Subtype B, SeraCare Life Sciences). As a control,
each pulse correspond to the colloid traveling through the individual nodes. we screened human plasma with no viral content. Fig. 5a
shows a representative current versus time trace when we
measured the HIV-positive human plasma sample. As shown,
Fast Fourier transform data processing there are two distinct modulated current pulses with strikingly
different magnitudes. Reference colloids of known size were
The node spacing in a pore provides a unique ability to utilize
measured with the same four-node device (Fig. S2, ESI3) and
its spatial geometry, translating measured signals into an FFT
their normalized current-pulse magnitudes (Fig. S1, ESI3), i.e.
power spectrum for quick detection analysis (see ESI3). We
|DI|/I, were used to determine the size of the particles in
demonstrate this by employing a specifically designed pore (18
transit (see ESI3). The larger of the two pulses shown in Fig. 5a
mm 6 20 mm 6 2400 mm, H 6 W 6 L) that has two 50 mm-
is a particle .1 mm and corresponds to fibrin in human
wide nodes that are 100 mm apart from one another, a 400 mm-
plasma. The smaller pulse, present only in the HIV-positive
long node ‘‘spacer’’, and two additional 50 mm-wide nodes, 500
human plasma sample and not in the control, corresponds to
mm apart (Fig. 4a). The current pulse produced when a 15.45
an HIV particle of size 140 nm. The particle size distribution
mm colloid transits the pore (Fig. 4b) has two distinguishable
for all our experiments is shown in Fig. 5b–e. We find that the
signatures reflecting the different geometric spacing of the
number of viral particles detected per volume sampled is in
nodes. The associated FFT spectrum (Fig. 4c) displays both
agreement with the expected number given the concentration
of HIV in each sample screened. Thus, our microscale node-
pore sensor is highly accurate in measuring viral particles
without the need for sample preparation or the use of
nanoscale fabrication. Furthermore, its large cross-sectional
area makes it much less susceptible to clogging as compared
to much smaller RPS-based viral sensors.41–43 This makes NPS
appropriate for determining particle concentrations in hetero-
geneous samples.

Conclusions
Unlike commercial Coulter counters, the node-pore sensor we
have developed enables one to perform sensing with unpre-
cedented dynamic range, flexibility, and sensitivity, i.e. Vparticle/
Vpore. Because the node-pore sensor has a large dynamic range,
Fig. 4 Scale bar, 100 mm. (a) An image of an 18 mm 6 20 mm 6 2400 mm (H 6 we can measure nanometer-sized particles without extraordin-
W 6 L) pore with two 50 mm-wide nodes that are 100 mm apart from one ary effort. Optimizing aperture size or using multiple apertures
another, a 400 mm-long node ‘‘spacer’’, and two additional 50 mm-wide nodes, for detection is not necessary. In comparison with other on-
500 mm apart. (b) The current pulse produced as a 15.45 mm colloid travels
chip RPS based sensors, including those we have previously
through the device. The high (shaded in red) and low (shaded in green)
frequency, fhigh and flow, correspond to the sequence of nodes in the device. (c) developed,4,35–37,44 the node-pore sensor distinguishes itself in
The fast Fourier transform frequency spectrum of the total signal after ten a number of ways. First, high signal amplification is not
duplications of the data (Fig. S4, ESI3). needed. Second, complicated and expensive nanofabrication

This journal is ß The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 1302–1307 | 1305
 View Article Online

Paper Lab on a Chip

Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

Fig. 5 (a) Representative current vs. time trace of a sample of human plasma spiked with HIV at 100 000 copies per mL. A large-sized particle, e.g. fibrin (blue box), is
easily detected. After applying a low-pass filter (cutoff y1500 Hz), a nanoscale HIV particle (magnified red box) is readily apparent. (b) Size distribution of particles
detected after screening 50 nL of human plasma without HIV. (c–d) Particle-size distribution of three 50 nL samples of human plasma with HIV. Particles with size
ranging from 100–200 nm are HIV. Particle concentrations can be determined from these distributions. All particle sizes are determined by the magnitude of the
normalized current, |DI|/I, as described in the ESI.3

procedures, such as atomic-layer deposition, electron-beam could be utilized in a number of diverse medical, environ-
lithography, and incorporation of naturally occurring pores, mental, and industrial applications, including screening
are not needed to measure at the nanoscale. With regard to bacterial and other viral populations, assessing water quality
processing volume, we have employed flow rates . 10 mL and determining contaminants, and monitoring protein
min21 without loss of signal detection or resolution. aggregation in pharmaceutical processes.
Throughput of our node-pore is limited by the fact that having In summary, we have created an extremely robust, flexible,
multiple particles traversing the pore simultaneously presents and powerful screening platform. Future work will focus on
signal identification complications; however, multiplexing modifying the node-pore sensor to probe several particle-pore
strategies such as integrating parallel-sensing arrays44,45 could interactions4,46 in a single device. This would enable one to
be utilized to boost processing efficiency, resulting in an take advantage of transit-time measurements corresponding
extraordinarily versatile high-throughput sensing platform. to different segments of the pore, thereby adding a new
While viral detection using RPS has been accomplished dimension to particle characterization. Ultimately, NPS
with nanometer-scaled pores,41–43 the ability to detect viral advances biological detection at both the micro- and nanos-
particles directly in human plasma has been a major challenge cale and could be a valuable tool for improving detection
due to the inability of nanopores to screen fibrin and other capabilities in broader applications.
aggregate components in blood plasma. Because node-pore
sensors are applicable for heterogeneously-sized populations,
they could be utilized at the point-of-care, determining viral Author contributions
particle counts to monitor disease progression and response to
therapy. This would make rapid, low-cost HIV testing K.R.B., G.A., M.R.C., and L.L.S. developed the node-pore
accessible in areas with limited medical facilities. Beyond sensor platform concept. K.B. and L.L.S. designed the
HIV detection, the node-pore sensor, with its ease in experiments and wrote the manuscript. K.B., A.K., and T.N.
manufacturing, dynamic range, and accuracy in detection, fabricated the node-pore sensors, and K.B. performed all the

1306 | Lab Chip, 2013, 13, 1302–1307 This journal is ß The Royal Society of Chemistry 2013
 View Article Online

Lab on a Chip Paper

experiments. G.A. developed the fast Fourier transform 21 A. R. Hall, A. Scott, D. Rotem, K. K. Mehta, H. Bayley and
analysis tools. C. Dekker, Nat. Nanotechnol., 2010, 5, 874–877.
22 P. Xie, Q. Xiong, Y. Fang, Q. Qing and C. M. Lieber, Nat.
Nanotechnol., 2012, 7, 119–125.
23 J. Li, D. Stein, C. McMullan, D. Branton, M. J. Aziz and J.
Acknowledgements A. Golovchenko, Nature, 2001, 412, 166–169.
We thank J. Uzgiris and S. Pandit for helpful discussions and 24 M. Wanunu, J. Sutin and A. Meller, Nano Lett., 2009, 9,
assistance with the viral-detection experiments, J. Hack for 3498–3502.
data analysis assistance, and E. Jabart, M. Chambers, and V. 25 R. S. Wei, D. Pedone, A. Zurner, M. Doblinger and U. Rant,
Tuminelli for helpful discussions. This research was funded by Small, 2010, 6, 1406–1414.
the Berkeley-Siemens Center of Knowledge Interchange and 26 M. Wanunu, T. Dadosh, V. Ray, J. Jin, L. McReynolds and
M. Drndic, Nat. Nanotechnol., 2010, 5, 807–814.
Published on 22 January 2013. Downloaded by UBC Library on 12/5/2024 11:15:34 PM.

Siemens Corporate Technology. K. B. is supported by a

27 L. Sun and R. M. Crooks, J. Am. Chem. Soc., 2000, 122,
National Defense Science and Engineering Graduate
12340–12345.
Fellowship (NDSEG).
28 A. T. Rodriguez, M. Chen, Z. Chen, C. J. Brinker and H.
Y. Fan, J. Am. Chem. Soc., 2006, 128, 9276–9277.
29 H. Bayley and C. R. Martin, Chem. Rev., 2000, 100,
Notes and references 2575–2594.
30 S. Kwok, D. H. Mack, K. B. Mullis, B. Poiesz, G. Ehrlich,
1 E. C. Gregg and K. D. Steidley, Biophys. J., 1965, 5, 393–405.
D. Blair, A. Friedman-Kien and J. J. Sninsky, J. Virol., 1987,
2 H. E. Kubitschek, Nature, 1958, 182, 234–235.
61, 1690–1694.
3 K. Craine and A. D. Waft, J. Clin. Pathol., 1967, 20, 913–915.
31 U. V. Comandini, A. Sonnerborg, A. Vahlne and Z. Yun, J.
4 A. Carbonaro, S. K. Mohanty, H. Huang, L. A. Godley and L.
Virol. Methods, 1997, 69, 171–180.
L. Sohn, Lab Chip, 2008, 8, 1478–1485.
32 P. Simmonds, P. Balfe, J. F. Peutherer, C. A. Ludlam, J.
5 J. Hirsch and E. Gallian, J. Lipid Res., 1968, 9, 110–119.
O. Bishop and A. J. L. Brown, J. Virol., 1990, 64, 864–872.
6 A. K. Bryan, A. Engler, A. Gulati and S. R. Manalis, PLoS
33 J. Bell and L. Ratner, AIDS Res. Hum. Retroviruses, 1989, 5,
One, 2012, 7, e29866.
87–95.
7 Y. F. Wu, J. D. Benson and M. Almasri, Biomed.
Microdevices, 2012, 14, 739–750. 34 C. Hart, G. Schochetman, T. Spira, A. Lifson, J. Moore,
8 D. C. Golibersuch, Biophys. J., 1973, 13, 265–280. J. Galphin, J. Sninsky and C. Y. Ou, Lancet, 1988, 2,
9 X. Wu, Y. Kang, Y. N. Wang, D. Xu, D. Li and D. Li, 596–599.
Electrophoresis, 2008, 29, 2754–2759. 35 O. A. Saleh and L. L. Sohn, Rev. Sci. Instrum., 2001, 72,
10 R. M. M. Smeets, U. F. Keyser, N. H. Dekker and C. Dekker, 4449–4451.
Proc. Natl. Acad. Sci. U. S. A., 2007, 105, 417–421. 36 O. A. Saleh and L. L. Sohn, Nano Lett., 2003, 3, 37–38.
11 R. W. Deblois and C. P. Bean, Rev. Sci. Instrum., 1970, 41, 37 O. A. Saleh and L. L. Sohn, Proc. Natl. Acad. Sci. U. S. A.,
909. 2003, 100, 820–824.
12 S. M. Bezrukov, I. Vodyanoy and V. A. Parsegian, Nature, 38 K. Balakrishnan and L. L. Sohn, Laboratory Methods in Cell
1994, 370, 279–281. Biology: Biochemistry and Cell Culture Academic Press, 2012,
13 J. Clarke, H. C. Wu, L. Jayasinghe, A. Patel, S. Reid and vol. 112, ch. 21, pp. 369–387.
H. Bayley, Nat. Nanotechnol., 2009, 4, 265–270. 39 J. A. Briggs, T. Wilk, R. Welker, H. G. Krausslich and S.
14 M. Rincon-Restrepo, E. Milthallova, H. Bayley and D. Fuller, EMBO J., 2003, 22, 1707–1715.
G. Maglia, Nano Lett., 2011, 11, 746–750. 40 M. Gentile, T. Adrian, A. Scheidler, M. Ewald, F. Dianzani,
15 J. J. Kasianowicz, E. Brandin, D. Branton and D. G. Pauli and H. R. Gelderblom, J. Virol. Methods, 1994, 48,
W. Deamer, Proc. Natl. Acad. Sci. U. S. A., 1996, 93, 43–52.
13770–13773. 41 R. W. DeBlois and R. K. Wesley, J. Virol., 1977, 23, 227–233.
16 L. Q. Gu, S. Cheley and H. Bayley, Science, 2001, 291, 42 K. Zhou, L. Li, Z. Tan, A. Zlotnick and S. C. Jacobson, J. Am.
636–640. Chem. Soc., 2011, 133, 1618–1621.
17 H. Bayley and P. S. Cremer, Nature, 2001, 413, 226–230. 43 Z. D. Harms, K. B. Mogensen, P. S. Nunes, K. Zhou, B.
18 H. Y. Wang, Y. L. Ying, Y. Li, H. B. Kraatz and Y. T. Long, W. Hildenbrand, I. Mitra, Z. Tan, A. Zlotnick, J. P. Kutter
Anal. Chem., 2011, 83, 1746–1752. and S. C. Jacobson, Anal. Chem., 2011, 83, 9573–9578.
19 Y. L. Ying, D. W. Li, Y. Li, J. S. Lee and Y. T. Long, Chem. 44 A. Carbonaro and L. L. Sohn, Lab Chip, 2005, 5, 1155–1160.
Commun., 2011, 47, 5690–5692. 45 J. Zhe, A. Jagtiani, P. Dutta, J. Hu and J. Carletta, J.
20 Y. Wang, D. Zheng, Q. Tan, M. X. Wang and L. Q. Gu, Nat. Micromech. Microeng., 2007, 17, 304–313.
Nanotechnol., 2011, 6, 668–674. 46 M. Wanunu and A. Meller, Nano Lett., 2007, 7, 1580–1585.

This journal is ß The Royal Society of Chemistry 2013 Lab Chip, 2013, 13, 1302–1307 | 1307