PAPER www.rsc.org/loc | Lab on a Chip

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online / Journal Homepage / Table of Contents for this issue

Enhancing DNA hybridization kinetics through constriction-baseddielectrophoresis

Nathan Swami,*a Chia-Fu Chou,*bc Venkatraman Ramamurthyc and Vasudha Chaureya

Received 29th May 2009, Accepted 10th August 2009

First published as an Advance Article on the web 8th September 2009

DOI: 10.1039/b910598k

The enhancement of signal sensitivity through the scaling-down of sensors presents mass transport

limitations that can arrest the sensitivity gains obtained as a result of miniaturization. To alleviate these

limitations, we study the application of constriction-based dielectrophoresis methods to enhance

transport through pre-concentration of target DNA in the vicinity of the diffusion layer of the sensor,

on which capture probe DNA molecules were immobilized. We demonstrate that constriction-based

DEP pre-concentration was not impeded by scaling-down of the sensor, as long as the sensor electrode

was composed of nanostructured edges and was coupled to an equally scaled down insulating

constriction within a microfluidic channel to enhance the focusing effects of the constrictions and

edges. Furthermore, as a result of the high focusing fields, pre-concentration of single-stranded target

DNA occurred in the vicinity of the sensor pad within the relatively high ionic strength buffers required for

DNA hybridization, with minimal degradation of capture probe molecules. Finally, constriction-based

DEP resulted in an almost immediate pre-concentration of target DNA in the vicinity of the sensor

electrode diffusion layer, resulting in a ten-fold enhancement of the DNA hybridization kinetics at target

concentration values down to the sensitivity limit of 10 pM for the sensor platform.

I. Introduction

The role of mass transport limitations in the sensitivity of

biosensor assays, especially for miniaturized sensor arrays, has

been recognized in prior work.1–3 For many detection platforms,

such as electrochemical, surface enhanced Raman spectroscopy

(SERS),4 electrical, resonance frequency shifts, and surface

plasmon resonance5 based methods, miniaturization improves

signal transduction and sensitivity to varying degrees. However,

at successively lower DNA target concentration values, fewer

target DNA molecules are available for diffusion towards the

miniaturized sensor surface for hybridization with surface-bound

capture probe DNA molecules. This slows down binding

kinetics, thereby considerably delaying signal onset and satura-

tion (i.e. complete hybridization of all capture probes on the

sensor surface). Hence, while signals barely above the noise can

be measured for various biomolecular assays at picomolar or

lower target DNA concentration values,6 these assays operate

under mass transport limitations of the analyte rather than in

a signal transduction limit. This suggests that, in order to further

improve sensitivity within practical assay time scales, methods to

direct the transport of target biomolecules towards the sensor

surface are necessary. For protein sensors, improved transport

becomes even more crucial for proteins with �100 kDa molec-

ular weight, which diffuse at a rate which is an order of magni-

tude slower. Furthermore, miniaturization imposes penalties on

aDepartment of Electrical Engineering, University of Virginia,Charllottesville, VA, 22904. E-mail: nswami@virginia.edubInstitute of Physics; Genomics Research Center; Research Center forApplied Sciences; Academia Sinica, Taipei, 11529, Taiwan. E-mail:cfchou@phys.sinica.edu.twcBiodesign Institute, Arizona State University, AZ, 85284, USA

3212 | Lab Chip, 2009, 9, 3212–3220

passive transport of biomolecules to the sensor surface for

hemispherical and disk electrodes and presents no significant

advantages for transport to nanowire electrodes.2 Hence, there is

a need for active transport methods where miniaturization aids,

rather than aggravates, transport, so that the sensor can

capitalize on improved signal transduction, as well as improved

transport as a result of miniaturization.

The improvement of mass transport through higher flow rates7

or stirring,8 were reported to enhance hybridization kinetics at

nanomolar or higher target concentration values.9 However, no

more than a 3–4-fold enhancement can be obtained for the

typical flow rates supported in a microfluidic channel,2 since

the enhacement of binding kinetics requires the stirring of the

diffusion layer around the sensor electrode, which is difficult to

achieve for surface-based miniaturized sensor assays. Electroki-

netic methods such as electrophoresis have been demonstrated to

direct transport;10 however, while significant pre-concentration

may be achieved, the requirement of a ‘‘permeation layer’’ and

a non-conducting buffer limits its application towards various

assays. Enhanced transport through magnetic methods,11 as well

as through asymmetric electric-field assisted methods have also

been demonstrated, although these have not been performed

within microfluidic systems at sub nano-molar target concen-

tration values.12 In all cases, methodologies for directed trans-

port of analyte biomolecules to the sensor surface to enable

simultaneous DNA pre-concentration, hybridization and real-

time sensing of the enhancement to DNA hybridization kinetics

are a pressing need.

Dielectrophoresis (DEP) has been widely used for the

manipulation, separation and characterization of cells, DNA,

viruses, and colloid particles (at both micro- and nano-scales) in

various microfluidic platforms,13 since it has numerous unique

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

advantages. First, DEP utilizes the intrinsic dielectric properties

of the material, and can therefore be applied to a wide category

of materials. Hence, applications of DEP, notably in bioparticle

sorting or manipulation in microdevices, do not require labelling

of the bioparticles under study, although labelling could often

assist in improved characterization of the DEP response of

bioparticles or the efficiency of bioparticle separation. Second, it

is relatively straightforward to implement DEP in microdevices

where high-fields and field gradients may be generated by thin-

film deposited metal electrodes separated by a few microns. The

conventional implementation of metal electrodes-based DEP

(MDEP) for force generation has been extended recently by

using dielectric constrictions or insulating obstacles to generate

localized DEP force fields, called electrodeless or insulator-based

dielectrophoresis (EDEP or iDEP, repectively),14,15 and by the

addition of electrically floating metal electrodes within these

constrictions, to further enhance the electric field gradient at

electrode edges.16 In this paper, these so called ‘‘constriction-

based’’ DEP methods are applied to trap bio-molecules, not at

the drive electrodes, where the voltage may damage them, but at

specially designed constricted insulator and metal edge points,

where electric field gradients and force fields are highest, even

though these points are electrically floating.

From a modelling perspective, the constriction-DEP configu-

ration simplifies the interpretation of the DEP phenomena due to

the symmetry of the system (3D symmetry of the force field

rather than 2D planar symmetry in MDEP). Consequently the

physical picture of the force field is less obscure, since in

comparison to MDEP, electrode polarization and electrochem-

istry occurring at metal electrodes do not complicate the picture.

Our prior work has shown the great promise of constriction-

based DEP technology applied to various functionalities of

a micro-total analysis system for molecular diagnostics, such as

cell trapping and separation, cell lysing, and DNA pre-concen-

tration for detection in a microarray platform.14,17,18 However, in

prior work, these operations were carried out within different

chambers on the chip. Simultaneous DNA pre-concentration

and sensing is a major challenge due to the highly focused field

and current conditions of constriction-based DEP, which can

effect sensor performance due to DNA capture probe degrada-

tion on the sensor electrode at the constriction, Joule heating

associated with the focused current, and the effects of convective

fluid flow induced by the temperature gradients from this local

heating effect. On the other hand, the higher force field due to

constriction-based DEP could enable the trapping of low

dielectric strength biomolecules (such as ss-DNA), even under

the relatively high salt conditions required for DNA hybridiza-

tion, in spite of the lower biomolecular polarizability under

such conditions. Hence, to study the relative enhancement of

DNA-hybridization kinetics due to constriction-based DEP

pre-concentration of single-stranded target DNA molecules in

relatively high salt buffers, the simultaneous pre-concentration,

binding and sensing of these molecules is examined herein.

In this paper, we examine the application of constriction-based

DEP methods towards the directed pre-concentration of target

DNA molecules within a micro-constricted fluidic channel onto

a microelectrode sensor with nanostructured edges, on which

DNA capture probes were immobilized. Electric field simulations

were used to guide the development of device designs by

This journal is ª The Royal Society of Chemistry 2009

understanding the trade-offs in scaling of the sensor electrode

with respect to the constriction size, so that miniaturization was

optimized for the directed transport of biomolecules, as well as

for improved signal transduction. These simulations were used to

develop conditions and designs for the trapping of target DNA

molecules due to constriction-based DEP, by imaging the pre-

concentration of fluorescently-labelled DNA target molecules in

solution. Finally, to measure the enhancement of DNA hybrid-

ization under constriction-based DEP, we used a two-potential

electrochemical probe19 to simultaneously probe, in real-time,

the signal from capture probe DNA molecules immobilized on

the sensor surface and target DNA molecules hybridized to the

capture probes. This platform allowed us to optimize electric-

field conditions, so that pre-concentration of target molecules

was enhanced on one hand, while capture probe degradation due

to the highly focussed fields was minimized on the other hand.

This enabled an accurate measurement of the effect of DNA pre-

concentration on hybridization kinetics. To enable the develop-

ment of a versatile biomolecule trapping methodology that

may be applied to various detection platforms, we aim to develop

constriction-based DEP methods that circumvent the use of

a ‘‘permeation layer’’ to protect against capture-probe degrada-

tion (as required within electrophoresis based assays), and

that can simultaneously carry out DNA preconcentration,

hybridization and sensing in the same high-conductivity elec-

trochemical buffer, without any wash steps. As a result, it is

anticipated that the results from the modelling and experiments

described herein will aid in the development of design rules for

effective implementation of constriction-based DEP methodo-

logies within various biosensor platforms.

II. Materials and methods

2.1. Modelling methodology

The classical DEP theory states that the dielectrophoretic force

arises from the interaction of the induced dipole of a polarizable

object and an external non-uniform electric field (dc or ac).20 The

DEP force is given by:

FDEP ¼ aðuÞE vE

vy(1)

a(u) is the polarizability of the object at the angular frequency u,

and y is the direction of the applied external field, E. For

a spherical object of radius a, the dielectrophoretic force may be

solved analytically in the form:14

FDEP ¼ 2pa33m Re

3*

p � 3*m

3*p þ 23*

m

!VE2 (2)

3*p and 3*

m are the complex permittivity of the dielectric object and

the medium, and E may be replaced by Erms, namely the root-

mean-square of the external field, in an alternating field. In

general, the DEP force results in the transportation of the

polarizable object towards (positive DEP) or away from (nega-

tive DEP) the high field gradient region, depending on the

contrast of the dielectric response between the material and the

immersed medium to the applied field. In an aqueous solution,

positive or negative DEP occurs when the dielectric response of

Lab Chip, 2009, 9, 3212–3220 | 3213

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

the object is higher (3*p � 3*

m > 0) or lower (3*p � 3*

m < 0) than the

solvent (in this case, water), respectively.

To elucidate the device-level computation of the dielectric

response, we focussed on the interplay of multiphysical param-

eters of a given constriction and sensor design with a set of

experimental inputs. The Laplace equation was solved simulta-

neously with the current continuity equation, which, in the

frequency domain, is V$[s + ju3]Vf. If s [ u3, this reduces to

DC case: V$sVf. Accordingly, the potential, field, current and

field gradient distributions were computed to simulate measur-

able physical parameters that were compared to our experi-

mental observations. Based on this, we aimed to develop a set of

design rules for constriction-DEP applications in microfluidic

systems.

We implemented electric, thermal, and fluidic modules from a

multiphysics computational fluidics dynamics software package

CFD-ACE+ (ESI R&D U.S., Huntsville, AL) using the device

layout in Fig. 1. The layout consists of a single insulator

constriction in a unit cell of 500 � 500 � 5 (xyz) mm. The device

structure of the constriction is formed by insulating plastic

constriction capped by PDMS (poly-dimethyl-siloxane)-coated

cover slide, on a substrate with metal electrodes insulated by

Si3N4 on a substrate of SiO2/Si. The gap of the constriction

and size of the sensor electrode were gradually varied to study

the resulting trends of field focusing factors due to the variation

of these sizes. The electric potential was applied across the

constriction at both ends of the unit cell in the y direction. The

buffer condition was set at 10 mM NaCl in an aqueous solution

with 1 S cm�1 conductivity.

2.2. Experimental methodology

The device structure for the DEP and detection experiments is

shown schematically in Fig. 1. Pairs of platinum electrode arrays

were patterned by photolithographic lift-off methods (10 mm

long across the fluidic channel width and to 100 nm thickness),

and separated by 500 mm to serve as the ‘‘driving electrodes’’ in

Fig. 1, onto which the DEP voltage was applied. Gold electrodes

were vapor deposited at the center of the Pt electrodes to serve as

Fig. 1 Schematic of constriction-based DEP device, showing Pt driving

electrodes and a Au sensor electrode with exposed edges patterned on an

SiO2/Si substrate, with an insulator (PDMS or oxide) defined microfluidic

channel that is constricted at the sensor electrode. DNA capture probes

were immobilized on the sensor electrode, and the target DNA pre-

concentration was characterized by fluorescence probes and DNA

hybridization was measured by electrochemical methods in separate runs.

3214 | Lab Chip, 2009, 9, 3212–3220

the sensor electrodes for DNA immobilization, and these were

varied in size from 1 to 50 mm as an experimental and model

parameter. A high quality silicon nitride layer was deposited by

PECVD methods to insulate the interconnecting lines of Pt and

Au from the electrolyte, and dry etching was used to only expose

the active areas for Pt and Au electrodes. Special care was taken

in designing etch masks to expose the Au electrode edges of

�50 nm thickness to enable bending of the electric field lines. The

syntheses of DNA capture probes, signaling probes, target

mimics, and ferrocene attachment to DNA for signaling at

different formal potentials have been described in detail in prior

publications.21–23 The deposition procedure for immobilization

of DNA capture probe monolayers involves the incubation of

30 nL drops of the deposition solution consisting of DNA

capture probe and other components of the self-assembled

monolayer in a high-salt buffer for 30 min, as described previ-

ously.19 To enable deposition of different DNA capture probes

on neighboring sensor electrodes to serve as the ‘‘positive’’ and

‘‘negative’’ controls, plasma treatments involving oxygen,

hydrogen, and silane were optimized to enable a contact angle of

35–40� on the insulating silicon nitride layer, adjacent to the

clean and hydrophilic gold sensor electrode surface. After

capture probe DNA spotting, the substrates were placed in

a high-humidity chamber for 30 min, and no significant evapo-

ration or accumulation was observed. The substrates were then

rinsed in de-ionized water, dried under a stream of nitrogen and

stored in nitrogen-filled foil-lined polythene bags until use.

PDMS spin-casting on a silicon master was used to fabricate 100,

200 and 500 mm microfluidic channels, with constrictions of

variable sizes. The channel depth was estimated to be 5 � 1 mm.

After a short oxygen plasma treatment cycle, the PDMS layer

was bonded to the device substrate, so that the constriction was

partially on top of the etched silicon nitride opening on the Au

sensor electrode. The variability of the PDMS constriction sizes

were estimated to be within �1 mm, however, upon bonding to

the substrate, an additional �1 mm variation was observed in

some cases. The mis-alignment of the constriction from the

center of the sensor electrode was estimated at �1 mm.

Following DNA immobilization on Au sensor pads and device

assembly with the microfluidic channel with constrictions, AC

fields of 10–200 Vpp cm�1 at 100–1000 Hz frequency were

applied to the Pt electrodes. Initial experiments to observe pre-

concentration at the constriction points using fluorescence

measurements were performed on dye-intercalated (YOYO-1)

T5 double stranded DNA (38 kbp) at 0.1 mg mL�1 concentration

in 0.5 � TBE buffer with varying degrees of salt concentration

(1 to 50 mM). After optimizing pre-concentration conditions,

we used asymmetric PCR techniques (1 : 10 ratio of

forward : reverse primers) to obtain fluorescein-labeled single-

stranded model DNA targets (�200 bases) with a complemen-

tary sequence of 20 bases matching the capture probe DNA

sequence. The experiments on simultaneous pre-concentration,

hybridization and electrochemical sensing, were carried out in

a buffer containing target DNA of the appropriate concentration

(varied from 100 nM to 10 pM), with 50 mM sodium chloride,

and added components of 1 mM mercaptohexanol (to keep the

DNA monolayer intact), 10% v/v fetal calf serum, and 10 mM

trizma buffer to adjust pH to 6.5. Based on prior work, we

optimized the capture probe concentration on the Au sensor

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

surface at �2 � 1013 molecules cm�2, where hybridization

efficiency was highest, for a large range of target concentration

values.19

III. Results

Field focusing due to dielectric constrictions

To optimize the force fields during DEP pre-concentration, the

electric potential and field distribution around a single dielectric

constriction (i-DEP configuration), without gold sensor elec-

trode edges, was modeled in Fig. 2(a). This shows that the electric

field (the derivative of the potential) was highly focused at the

constriction. To ensure a high field-focusing factor we set the

channel width constriction from 500 to 1 mm, within a few

microns of channel length.

Field focusing due to dielectric and metal-edge constrictions

Fig. 2(b) shows the field distribution around a single constriction

trap, in the presence of an additional metal electrode edge

(Au sensor electrode) to enhance bending of the fields, and the

normalized DEP force: V(E2)/2. In absence of the metal edge, the

focusing field was peaked only at the tips of the constriction trap

and decreased towards the centre of the trap due to attenuation

of the field. As a result of the metal-edge in addition to the

constriction trap, the focusing field was peaked not only at the

constriction tips, but also towards the edges of the sensor elec-

trode; thereby enhancing the extent of focusing.

Constriction size versus sensor size

It is apparent from Fig. 2(a) that a narrower insulator constric-

tion is preferred since this would enhance the force fields as per

eqn (2). However, an understanding of the relationships and

spatial extents arising from the focusing effects of the insulator

constriction versus the sensor electrode edge is necessary to

develop designs that optimally exploit both methodologies for

local enhancement of the force fields. For this purpose, we

coupled Au sensor electrodes of four different sizes (1, 5, 10 and

20 mm in diameter) to a 1 mm insulator constriction to simulate

focusing fields of the constricted DEP device at the plane of the

Fig. 2 Electric field distribution around a single i-DEP trap: (a) without

Au sensor edge; and (b) with Au sensor edge. In (a), the field focusing

occurs due to the dielectric constriction of the trap, while in (b) the sensor

edge and the tip of the constriction trap are responsible for the field

focusing.

This journal is ª The Royal Society of Chemistry 2009

sensor electrode edge and mid-way through the microfluidic

channel. As shown in Fig. 3(a) for profiles at the plane of the

sensor electrode edge, the focusing fields (electric field and field

gradient) for the larger sized Au sensor electrode edges were

greatly attenuated as was apparent from the double peaks cor-

responding to the respective electrode edges for each electrode

size. As the electrode size was decreased from 20 to 1 mm, the

DEP force increased about 100 times. Additionally, as the Au

sensor electrode size was decreased, the two focusing field peaks

due to the edges become closer, resulting in a force field that was

more focused towards the center of the constriction. Hence, for

the 1 mm sensor electrode, force fields due to the insulator

constrictions and metal edges merged, but the broad peak with

shoulders correspond to the continued presence of the metal-

edges on the field. Fig. 3(b) shows the electric field profile above

the plane of the electrodes, at a height corresponding to the

Fig. 3 (a): Electric field profile by finite volume analysis at the plane

of the sensor electrode, for a 1 mm dielectric constriction coupled to four

different Au sensor electrode sizes (Au pad on the channel floor was

centered at 250 mm between a pair of Pt driving electrodes, 500 mm apart

in a channel of 500 mm width and 5 mm height). For the 1 mm sensor

electrode, shoulders corresponding to the continued presence of the

metal-edges on the field were apparent. (b): Electric field profile by finite

volume analysis at the mid-way plane of the microfluidic channel, for

a 1 mm dielectric constriction coupled to four different Au sensor elec-

trode sizes (Au pad on the channel floor was centered at 250 mm between

a pair of Pt driving electrodes, 500 mm apart in a channel of 500 mm width

and 5 mm height). The inset shows an expanded view of the E-field at the

1 mm Au sensor electrode, to show shoulders (indicated by arrows)

corresponding to the sensor edge points.

Lab Chip, 2009, 9, 3212–3220 | 3215

Fig. 4 (a) Bright field image of a 1 mm constriction centered on a 50 mm

Au sensor electrode that is dry-etched to expose Au edges through a 1 mm

hole in the Si3N4 dielectric. (b) Fluorescence image of (a), showing the

initiation of pre-concentration of fluorescently tagged DNA within a few

seconds of E-field; (c) fluorescence image of (a), showing pre-concen-

tration of fluorescently-tagged DNA under field (100 V cm�1, 500 Hz,

2 min); (d)–(f) are similar to (c), with Au electrode sizes of 5 mm (d), 20 mm

(e), 40 mm (f). Note that in all these cases the preconcentration of double-

stranded DNA was carried out in the absence of salt in the buffers.

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

midpoint of the channel. Since the 1 mm constriction extended

through the entire thickness of the microfluidic channel, its

focusing field was observed at the mid-way plane of the micro-

fluidic channel, whereas the focusing field of the sensor electrode

edges was localized to its surface plane only. Hence, for a 1 mm

constriction, while the focusing fields from the 5 mm and 1 mm

sensor electrode edges were somewhat comparable at the plane of

the sensor (Fig. 3(a)), these focusing fields were far higher for the

1 mm sensor electrode edge when measured mid-way through the

microfluidic channel (Fig. 3(b)). It may be noted, however, that

the peak width for the focusing effect from the 1 mm constriction

coupled to the 1 mm sensor electrode was broadened at the floor

of the channel (Fig. 3(a)) due to the greater role of sensor edges

and was sharper midway through the channel (Fig. 3(b)) due to

the dominant role of the constriction, even though small shoul-

ders were also apparent in the latter case (inset of Fig. 3(b)). This

suggests that, for the purposes of field focusing at the floor of

the channel, such as would be necessary for a surface bound

DNA assay at the sensor electrode, the optimal combination of

focusing fields from the insulator constriction and metal edges

can be obtained by coupling a 1 mm constriction to a 5 mm

electrode edge. On the other hand, to extend the focusing field

through all planes of the microfluidic channel, the 1 mm

constriction should be coupled to a 1 mm electrode edge. In other

words, for an extremely narrow insulator constriction (1 mm), an

equally small Au sensor electrode size (1 mm) resulted in the

greatest field enhancement, but this occurred in a non-uniform

manner at the center of the sensor electrode. Alternatively, if

highly uniform field focusing at the plane of the sensor electrode

is sought, then the sensor electrode size should be marginally

larger than the constriction. Sensor electrode sizes greater than

5 mm did not result in optimal utilization of the respective field

enhancements. In summary, for optimal field focusing, the sensor

electrode size must be equal to or marginally greater than the size

of the constriction it was coupled to. Hence, miniaturization

of both sensor and constriction sizes aids biomolecule trapping

from microfluidic channels, thereby alleviating transport

limitations.

Experimental verification of pre-concentration

To verify results on the simulated focusing fields, we used

fluorescently labeled double-stranded DNA to follow DNA pre-

concentration for varying electrode sizes (1–40 mm) coupled to

a fixed constriction size (1 mm). DEP conditions of 100 V cm�1,

500 Hz, were applied within the microfluidic channel for 2 min.

From Fig. 4, it is apparent that the strongest degree of pre-

concentration was obtained for the 1 mm constriction coupled to

a 1 mm Au sensor electrode, followed by a significant, yet lesser

level of pre-concentration on a 5 mm Au sensor electrode. Upon

coupling of the same 1 mm constriction to larger sensor elec-

trodes, the focusing effect at the constriction becomes succes-

sively more faint, and much of the pre-concentration was

confined to the electrode edges. Note that the fluorescence signal

from electrode edges under the insulator was not due to DNA

pre-concentration, but due to refraction of light at the metal/

insulator bump. These results reiterate that for optimal combi-

nation of the focusing fields from the insulator constriction and

sensor electrode edge, the sizes of the respective structures must

3216 | Lab Chip, 2009, 9, 3212–3220

be almost equally small, otherwise the enhacement due to the

constriction is not fully realized. In these and other experiments

to characterize the degree of pre-concentration, the fluorescence

intensity of DEP-concentrated target molecules (after back-

ground subtraction for intensity from buffer, substrate, and

electronic noise) was compared that from non pre-concentrated

target molecules at equilibrium.

DNA pre-concentration in high-ionic strength buffers

While the pre-concentration data in Fig. 4 were acquired in the

absence of salt in the buffer, biomolecular sensors are usually

based on DNA hybridization assays, and these require a rela-

tively high salt concentration in the buffer (�50 mM NaCl).

Hence, in the next set of experiments we aimed to optimize the

appropriate AC field and frequency conditions for DNA pre-

concentration under varying salt concentration of buffer. The

pre-concentration of double-stranded DNA as a function of

constriction size and for different salt concentration values in the

buffer was studied in Fig. 5(a) and (b). While a good degree of

DNA pre-concentration in relatively short times (�1 min) was

obtained in buffers free of salt (Fig. 4) with 100 V cm�1 fields of

500 Hz frequency, higher field strengths were required for pre-

concentration from 2 mM and 50 mM salt containing buffers. In

the latter case, a field of �200 V cm�1 was required for

a comparable degree of pre-concentration within a minute.

However, signals from the immobilized DNA capture probes

(obtained by the two-potential electrochemical characterization

of capture probe) at the constriction were degraded due to the

highly focused fields at 200 V cm�1. Hence, we identified field

conditions of 140–150 V cm�1 at 500 Hz, where a reasonable level

of pre-concentration could be obtained within �5 min, with

minimal capture probe degradation. Upon application of these

field conditions for up to 60 min, the pre-concentration degree

This journal is ª The Royal Society of Chemistry 2009

Fig. 6 Pre-concentration of 1 nM single-stranded target DNA from high

ionic strength buffers (50 mM NaCl) using fields of 150 V cm�1 at 500 Hz,

and enhanced using a 1 mm constriction coupled to a 5 mm sensor

electrode edge, measured after 30 min for indicated channel widths.

Fig. 5 Fluorescence signals of preconcentration of double-stranded

DNA at various PDMS constricted microfluidic channels for 2 and

50 mM NaCl salt concentration values in buffer. The 500 and 200 mm

channels were constricted to 10, 5 and 1 mm, respectively, and images

were acquired using AC fields of 140 V cm�1 at 500 Hz, for 5 min.

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

was substantial. However, for higher pre-concentration times

�15–20% capture probe degradation was observed, with even

higher degradation at the higher ionic strength buffers, possibly

due to electrolysis. The results in Fig. 5 show significant pre-

concentration within 5 min from buffers with 2 mM NaCl for all

the three constriction sizes. The fluorescence signals were peaked

at the tips of the constriction for the 10 mm constriction, and the

peaks begin to merge for the smaller 1 and 5 mm constrictions.

Furthermore, no clear differences between the two channel sizes

(500 and 200 mm) were apparent. With 50 mM NaCl in the buffer

which is closer to the condition for DNA hybridization, the

utility of the narrower constriction from the larger channel for

net enhancement of DNA trapping per unit area of sensor was

apparent. In this case, the 1 mm constrictions from the 500 mm

channel showed the highest fluorescence intensities over all the

other cases. It should be noted that our observations of lower

fluorescence during DEP pre-concentration from high salt

buffers could arise from either a lower DNA polarizability in

conducting buffers or due to salt concentration dependent

lowering of the fluorescence intensity.

Fig. 7 Enhancement of hybridization kinetics due to constriction DEP

pre-concentration (DEP) applied for 30 and 60 min to various concen-

trations of target DNA, as compared to hybridization under passive

diffusion (No DEP). Inset has the same axes plotted over a smaller range

to show the enhancement down at 10 pM target.

Pre-concentration of single-stranded target DNA under high

ionic strength conditions

To further simulate the conditions of DNA hybridization assays, we

studied the DEP pre-concentration of single-stranded target DNA

from within high ionic strength buffers. Asymmetric PCR was used

to synthesize fluorescein-labeled single-stranded model DNA

targets of 200 bases. Electric field strengths of�150–200 V cm�1 at

500 Hz were required for the onset of pre-concentration. Using

fields of 150 V cm�1 at 500 Hz that were tested previously for

minimal capture probe degradation, a substantial degree of

pre-concentration was obtained after 30 min of the E-field. It is

apparent from Fig. 6, that the pre-concentration from 500 mm

channels constricted to a 1mm trap coupled to a 5 mm sensor

electrode edge was greater than that obtained using similar

conditions from 100 or 200 mm channels. Based on these fluo-

rescence results, to screen for device designs and field conditions

under which a considerable degree of DNA pre-concentration

could be obtained from high-salt buffers without significant

capture probe degradation, subsequent constriction-based DEP

pre-concentration experiments were carried out in 50 mM NaCl

buffers, by applying fields of 150 V cm�1 at 500 Hz, that were

This journal is ª The Royal Society of Chemistry 2009

focused from a 500 mm channel to a 1 mm constriction that was

coupled to a 5 mm sensor electrode (the 5 mm sensor electrode was

used instead of a 1 mm electrode to allow for easier alignment of

the sensor electrode to be within the center of constriction and

thereby obtain an array of working sensor electrodes).

Enhancement of DNA hybridization kinetics

The two potential electrochemical platform was applied to probe

the enhancement of DNA hybridization kinetics due to DEP pre-

concentration, since signals from immobilized DNA capture

probe and hybridized DNA target molecules could be quanti-

tatively detected simultaneously, in real-time.19 This allowed us

to screen out the pre-concentration conditions (electric field,

frequency, time, type of buffer, etc.) that may cause degradation

of the capture probe. Furthermore, the signals arise primarily

from hybridized target DNA molecules rather than from

pre-concentrated unbound or non-specifically bound target

molecules. Signals from this sensor platform displayed a three-

orders-of-magnitude dynamic range. Since capture probe

concentration on the Au sensor surface was �1013 molecules

cm�2, where hybridization efficiency was highest, the signal at the

saturation point was obtained from �1013 bound target mole-

cules cm�2, and signal at the detection limit was obtained from

�1010 bound target molecules cm�2. Hence, for a 5 mm sensor

pad, a signal was distinguishable from as few as �100 bound

target molecules on the surface. Fig. 7 shows the signals from

Lab Chip, 2009, 9, 3212–3220 | 3217

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

DNA hybridization under passive diffusion conditions (labelled

‘‘No DEP’’) for various target concentration values from 100 nM

down to 10 pM, in comparison to that obtained for DNA

hybridization under active DEP pre-concentration for 30 and

60 min. Considering the case of hybridization under passive

diffusion, while the signal from 100 nM target DNA saturates

within 30 min and that from 10 nM target DNA saturates within

an hour, target DNA concentration values of 1 nM and lower

run into diffusion problems that considerably delay signal satu-

ration. The signal from 1 nM target saturates after 4 h; that from

100 pM target reaches close to saturation after 10–12 h; and for

the case of 10 pM, the signals do not saturate in the time frame of

the experiment (the monolayer degrades after 16 h in the buffer

solution). Degradation of the signals from their saturation value

was observed upon repeat scanning due to capture probe

degradation. The target DNA concentration range for the DEP

pre-concentration experiments was chosen to be in the sub-

nanomolar range (10 pM to 1 nM), since the concentration

within this range was low enough to slow diffusion but high

enough to possibly reach signal saturation with DEP pre-

concentration. Hence, the time necessary to obtain signal satu-

ration could be used to judge the effect of DEP pre-concentration

on DNA hybridization kinetics. The constriction-based DEP

pre-concetration was applied with microfluidic channels con-

stricted from 500 mm to 1 mm on 20 mm and 5 mm sensor pads.

The DEP pre-concentration was applied for 30 and 60 min only,

since prolonged DEP and/or repeated electrochemical scans

caused some degradation of capture probe signals which posed

problems in signal comparison and quantification. Considering

the case of 1 nM target, it is apparent that, as a result of DEP

preconcentration, the hybridization kinetics were improved to

raise signal values to an equivalent level as the respective values

for 10 nM target DNA without DEP pre-concentration (only

passive diffusion). Similarly, signals from 100 pM target under

DEP pre-concentration were raised to an equivalent level as the

respective values from 1 nM target DNA without DEP pre-

concentration (only passive diffusion). Furthermore, from

a comparison of DEP preconcentration of 1 nM target on 20 mm

versus 5 mm sensor pads, it is apparent that the signals were

higher on the latter as would be expected from the pre-concen-

tration results within Fig. 3 and 4. These further re-affirm the

observation that a smaller sensor size was optimal for DEP pre-

concentration as long as the insulator constriction size was

considerably smaller. A final observation of interest from Fig. 7

was that while signals from DEP pre-concentration on 100 pM

target after 30 and 60 min were equivalent to the expected signal

value at 1 nM without DEP preconcentration, and similarly

(from inset of Fig. 7), the signal from 10 pM target pre-

concentrated under DEP for 30 min seems comparable to that

obtained from 100 pM without DEP pre-concentration, the same

signals for 10 pM target after 60 min of DEP pre-concentration

were significantly lower than the expected signal value at 100 pM

without DEP preconcentration. We attribute this difference to

the depletion of target DNA molecules within the microfluidic

channel at the longer pre-concentration times for the lower

target concentration values (10 pM or lower). A flow cell DEP

preconcentrator would alleviate this problem. In summary, the

results indicate that the role of DEP preconcentration is to

almost immediately increase the target DNA concentration by at

3218 | Lab Chip, 2009, 9, 3212–3220

least ten-fold in the vicinity of the sensor pad’s diffusion layer,

and the kinetics of the ten-fold pre-concentrated target were

followed.

IV. Discussion

Versatility of constriction-based DEP pre-concentration method

From the E-field simulation results, it is apparent that target

DNA pre-concentration through constriction-based DEP was

not impeded by scaling down the sensor. In fact, scaling down

sensor sizes for the purposes of improved signal transduction was

not detrimental to constriction-DEP based analyte transport as

long as the sensor electrode was composed of nanostructured

edges and was coupled to an equally scaled down insulator

constriction in a microfluidic channel to enhance the focusing

effects of the constrictions and edges. As a result of these higher

force fields due to constriction-based DEP, the trapping of low

dielectric strength biomolecules (such as ss-DNA), even under

the relatively high salt conditions required for DNA hybridiza-

tion was demonstrated in spite of the lower biomolecular

polarizability under such conditions. Furthermore, since target

DNA trapping does not take place at the driving electrodes,

where the voltage may damage them, but at specially designed

constricted insulator and metal edge points where electric field

gradients and force fields were highest, the degradation of

capture-probe DNA immobilized on sensor electrodes at the

constriction can be minimized since these electrodes are electri-

cally floating during DEP. This eliminates the need for a perme-

ation layer above the sensor electrode to arrest capture probe

degradation. As a result of these factors, constriction based DEP

can be applied towards analyte pre-concentration within a wide

variety of sensor platforms. Local heating and convective flow

due to the ensuing temperature gradients during constriction-

based DEP do not arrest target DNA pre-concentration. These

characteristics attest to the versatility of the constriction-based

DEP pre-concentration method.

Applying DEP pre-concentration to enhance DNA hybridization

kinetics

While target DNA pre-concentration under electric fields has

been demonstrated previously within various studies, its appli-

cation in conjunction with DNA hybridization and sensing has

proved to be a significant challenge. First, the optimal ionic

strengths of electrolytes for DNA pre-concentration do not

match those required for DNA hybridization and sensing.

Second, the force fields causing target DNA pre-concentration

can cause capture probe DNA degradation on the sensor. Third,

the three-dimensional configuration of the target DNA mole-

cules pre-concentrated under electric field may not be suitable for

hybridization with immobilized DNA capture probe molecules.

In this study, we get around the first limitation by enhancing the

force fields for DNA pre-concentration through the use of

constrictions and edges, so that a significant degree of pre-

concentration may be obtained within the high ionic strength

buffers required for DNA hybridization and sensing. Further-

more, the constriction-based DEP format was optimized to

screen out the pre-concentration conditions that cause capture

probe degradation to address the second limitation. Finally,

This journal is ª The Royal Society of Chemistry 2009

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

through the use of AC fields for pre-concentration at constriction

points rather than at the driving electrodes, we anticipate that the

three-dimensional configuration of pre-concentrated target

DNA molecules were not too constrained to participate in DNA

hybridization. However, it must be noted that to optimize DEP

pre-concentration to address each of these three limitations, the

degree of pre-concentration was somewhat reduced from its

highest level (�1000-fold) to enable the simultaneous testing of

target DNA pre-concentration, hybridization and sensing.

Degree of enhancement to DNA hybridization kinetics

In order to quantify the degree of enhancement to DNA

hybridization kinetics through DEP pre-concentration, we used

the two-potential electrochemical sensing platform. At the

outset, we realize that while nanowire and cantilever sensor

platforms may display higher signal sensitivity, this particular

electrochemical sensor platform was chosen due to its nearly

three-orders-of-magnitude dynamic signal range, a large target

concentration range where signal saturation could be obtained

(100 nM down to 100 pM), and the ability to probe the degra-

dation of capture probe DNA on the sensor surface as a function

of the constriction-based DEP field in real-time. These features

enabled a quantitative real-time assessment of the enhancement

to hybridization kinetics due to DEP pre-concentration.

Furthermore, while constriction based DEP for enhanced

hybridization kinetics may be observed for target concentration

down to the sub-picomolar range, we focus this kinetics study

in the 10 pM to 1 nM range, to enable a direct quantitative

comparison to the case of passive diffusion, since the concen-

tration within this range was low enough to slow diffusion but

high enough to eventually reach signal saturation with DEP-pre-

concentration. The enhancement to DNA hybridization kinetics

observed due to DEP pre-concentration in comparison to passive

diffusion occurred due to an almost immediate ten-fold target

DNA pre-concentration in the vicinity of the sensor pad’s

diffusion layer. It should be noted that while much of the prior

work on enhancement of DNA hybridization kinetics using

a similar sensor platform was reported at 10 nM,8,9 we report

a ten-fold enhancement to hybridization kinetics for target

concentration values down to the sensitivity limit of 10 pM for

this sensor platform. In principle, this enhancement should

continue down to even lower target concentration values, but the

sensitivity limitations of our platform prevented us from

reporting it here. Furthermore, since the sensor covers only the

floor of this constricted volume, we report only a ten-fold

enhancement in Fig. 7, whereas Fig. 6 shows a far greater pre-

concentration factor since it integrates signals from the entire

constricted volume. Hence, sensor methodologies based on the

entire constricted volume would potentially display greater

sensitivity enhancements due to constriction-DEP based

enhancement of hybridization kinetics. Even considering just the

floor of the constricted volume, it is likely that only a fraction of

the pre-concentrated target molecules are bound to the sensor

surface due to losses arising from capture probe degradation and

three-dimensional conformation of arriving target molecules

under the AC field, resulting in only a limited measurement of the

enhancement of DNA hybridization kinetics. Nevertheless,

based on the enhanced hybridization kinetics observed, in spite

This journal is ª The Royal Society of Chemistry 2009

of limitations of the sensor platform, we envision that constric-

tion-based DEP methods can enhance sensitivities within

a variety of sensor platforms through alleviating transport limi-

tations that arise due to sensor miniaturization.

Conclusions

In this study, we have optimized the application of constriction-

based DEP methodologies to alleviate transport limitations

arising from sensor miniaturization and to enhance DNA

hybridization kinetics. The chief conclusions are:

(1) Constriction-based DEP alleviates transport limitations

without being impeded by scaling down of the sensor, as long as

the sensor electrode was composed of nanostructured edges and

was coupled to an equally scaled down insulator constriction

within a microfluidic channel to enhance the focusing effects of

the constrictions and edges.

(2) Due to the high focusing fields obtained as a result of

constriction-based DEP, the pre-concentration of low polariz-

ability biomolecules such as single-stranded target DNA was

demonstrated within the relatively high ionic strength buffers

required for DNA hybridization, with minimal capture probe

degradation.

(3) Through optimization of the constriction-based DEP device

for simultaneous DNA pre-concentration, hybridization and

sensing, a ten-fold enhancement of the DNA hybridization kinetics

was observed for target concentration values down to the sensi-

tivity limit of 10 pM for this sensor platform, due to an almost

immediate ten-fold enhancement of target DNA pre-concentration

in the vicinity of the sensor electrode’s diffusion layer.

Acknowledgements

N. Swami acknowledges support for this work from NSF CBET

Award 0403963 and NSF EFRI 0736002. C. F. Chou acknowl-

edges support from AZ Prop 301, Academia Sinica and NSC 96-

2112-M-001-024-MY3, Taiwan, ROC.

References

1 J. Bishop, A. Chagovetz and S. Blair, Nanotechnology, 2006, 17,2442–2448.

2 P. R. Nair and M. A. Alam, Appl. Phys. Lett., 2006, 88, 233120,233120: 1–3.

3 P. E. Sheehan and L. J. Whitman, Nano Lett., 2005, 5, 803–807.4 A. G. Brolo, E. Arctander, R. Gordon, B. Leathem and

K. L. Kavanagh, Nano Lett., 2004, 4, 2015–2018.5 T. Rindzevicius, Y. Alaverdyan, A. Dahlin, F. Hook,

D. S. Sutherland and M. Kall, Nano Lett., 2005, 5, 2335–2339.6 Y. W. C. Cao, R. C. Jin and C. A. Mirkin, Science, 2002, 297, 1536–

1540.7 C. Phillips, M. Jakusch, H. Steiner, B. Mizaikoff and A. G. Fedorov,

Anal. Chem., 2003, 75, 1106–1115.8 R. H. Liu, R. Lenigk, R. L. Druyor-Sanchez, J. N. Yang and

P. Grodzinski, Anal. Chem., 2003, 75, 1911–1917.9 R. Lenigk, R. H. Liu, M. Athavale, Z. J. Chen, D. Ganser,

J. N. Yang, C. Rauch, Y. J. Liu, B. Chan, H. N. Yu, M. Ray,R. Marrero and P. Grodzinski, Anal. Biochem., 2002, 311, 40–49.

10 M. J. Heller, A. H. Forster and E. Tu, Electrophoresis, 2000, 21,157–164.

11 R. L. Edelstein, C. R. Tamanaha, P. E. Sheehan, M. M. Miller,D. R. Baselt, L. J. Whitman and R. J. Colton, Biosens. Bioelectron.,2000, 14, 805–813.

12 F. Fixe and et al, Nanotechnology, 2005, 16, 2061–2071.13 R. Pethig and G. H. Markx, Trends Biotechnol., 1997, 15, 426–432.

Lab Chip, 2009, 9, 3212–3220 | 3219

Publ

ishe

d on

08

Sept

embe

r 20

09. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 2

2/06

/201

4 14

:26:

35.

View Article Online

14 C. F. Chou, J. O. Tegenfeldt, O. Bakajin, S. S. Chan, E. C. Cox,N. Darnton, T. Duke and R. H. Austin, Biophys. J., 2002, 83,2170–2179.

15 E. B. Cummings and A. K. Singh, Anal. Chem., 2003, 75, 4724–4731.16 M. Washizu, Electrostatics, 2004, 178, 89–94.17 C. F. Chou and F. Zenhausern, IEEE Engineering in Medicine and

Biology, 2003, 22, 62–67.18 J. Gu, R. Gupta, C.-F. Chou, Q. Wei and F. Zenhausern, Lab Chip,

2007, 7, 1198–1201.

3220 | Lab Chip, 2009, 9, 3212–3220

19 N. Swami, C. F. Chou and R. Terbrueggen, Langmuir, 2005, 21,1937–1941.

20 H. A. Pohl, Dielectrophoresis: The Behavior of Neutral Matter inNonuniform Electric Fields. Cambridge, UK: Cambridge UniversityPress, 1978.

21 C. Bamdad, Biophys. J., 1998, 75, 1997–2003.22 C. J. Yu and et al, J. Am. Chem. Soc., 2001, 123, 11155–11161.23 R. M. Umek and et al, Journal of Molecular Diagnostics, 2001, 3,

74–84.

This journal is ª The Royal Society of Chemistry 2009