ARTICLE IN PRESS
Journal of Electrostatics 65 (2007) 511–520 www.elsevier.com/locate/elstat
Numerical design of a 3-D microsystem for bioparticle dielectrophoresis: The Pyramidal Microdevice P. Pham, I. Texier, A.-S. Larrea, R. Blanc, F. Revol-Cavalier, H. Grateau, F. Perraut CEA-LETI, Dir. de la Recherche Technologique, 17 rue des Martyrs, 38054 Grenoble cedex 9, France Received 21 April 2006; received in revised form 10 October 2006; accepted 24 November 2006 Available online 22 December 2006
Abstract A device for the detection of bioparticles (latex beads, cells, proteins, DNA,y) present in a small volume of a liquid sample is described. Dielectrophoresis (DEP) is used to enhance the transport of bioparticles towards the reactive surface of the detector. A Pyramidal Microdevice (PM) is designed using numerical simulation. Unlike conventional coplanar electrode microchips, this original 3D configuration is able to collect particles on a surface for both positive and negative DEP. Numerical simulations show the location of the collection zones in the PM according to the DEP mode applied. PM performance is evaluated with the calculation of the collection kinetics. The influence of the voltage is studied. Experiments on PM with latex microparticles confirm calculations. Their replacement by DNA molecules attests to the feasibility of a PM collecting such extremely small particles using DEP. r 2006 Elsevier B.V. All rights reserved. Keywords: Dielectrophoresis; Finite element; AC electrokinetics; Latex particle; DNA
1. Introduction Scientific activity in the field of in vitro diagnostics for medical and industrial applications has developed considerably over the past few years through the development of autonomous devices for detecting biological and chemical agents. The goal is to install them in public areas where the risk of exposure to pathogens that have been identified as potential biological warfare agents is high. It is necessary to be able to detect and diagnose diseases at locations a long way from laboratories and hospitals. Although these monitoring systems must be portable and consume little energy, their miniaturization provides the possibility of greater sensitivity and faster detection. The ongoing revolution of silicon micromachining techniques, which enable the production of quite complex microstructures on silicon wafers, is the starting point for the development of biochip devices. As one human hair is about 100 mm in diameter, channels and electrodes with the same order of dimensions can be patterned for handling Corresponding author. Tel.: +33 438783612/5787.
E-mail address:
[email protected] (P. Pham). 0304-3886/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.elstat.2006.11.008
liquid samples. Building sophisticated, multifunctional and automated microfluidic systems is still in the research and development stage. It is a multidisciplinary activity and experts in biochemistry, physics, optics, numerical simulation and microelectronics engineering are working together to make microdevices a reality. More specifically, our project focuses on biomolecule detection (protein, DNA,y). Most biochip technologies utilize molecular probes tethered on a surface for compound analysis. Although Brownian motion is significant because of their small size, biomolecule transport from the bulk to the reactive surface is the main limiting factor for fast detection. The use of dielectrophoresis (DEP) to enhance biomolecule transport towards a reactive surface is a challenge because handling the motion of specific particles inside small volumes of liquid is an important stage in effective surface detection. DEP is an electrical phenomenon (polarization) that allows particles to be trapped or discriminated by applying nonuniform electrical fields [1,2]. One main advantage of DEP is that AC electric fields can be used. Owing to isolated electrodes, electrochemical reactions on electrodes are avoided and high electric fields can be delivered (over
ARTICLE IN PRESS 512
P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
106 V/m). With a few volts and electrodes spaced less than 100 mm apart, dielectrophoretic forces are effective even for extremely small particles such as DNA [3]. When particles move towards regions of high gradient electric fields, DEP is said to be positive. When moving away from high-gradient fields, DEP is said to be negative. These two DEP modes depend on two parameters: the polarizability of the particle with respect to the medium (buffer) in which it is suspended and the field frequency. Without changing the buffer composition, frequency is the only parameter for controlling the DEP mode. However, for highly conductive buffer the negative DEP mode is dominating regardless of the applied frequency. Actually, most biochips for DEP use coplanar microelectrodes for delivering the electric field inside the liquid with which it is in contact. Such systems capture particles only for the positive DEP: for the negative DEP mode, particles are repelled toward the buffer by the electrodes and collection on a surface is not possible. For treating a wide range of different kinds of particles and buffer compositions, a powerful microdevice should concentrate particles on a surface for both the negative and the positive DEP modes. This paper presents the design of a generic 3-D microdevice (the Pyramidal Microdevice—PM) for DEP on biomolecule samples. The lack of knowledge concerning their dielectric properties with respect to their aqueous buffer has constrained us to simulate this new configuration with latex particles. Experiments with both DNA samples and latex particles are briefly presented to show its feasibility. In the first part, the mathematical models used for numerical simulations are presented. DEP force distribution is calculated from the point dipole model and the AC complex electrokinetics equation. The advection–diffusion equation is used to determine the distribution of the transient concentration of particles submitted to Brownian motion and the DEP force applied. Working frequencies are chosen to be sufficiently high so that electrohydrodynamics (EHD) effects can be neglected. Equation parameters are listed and justified. Equations relating to associated boundary conditions are solved with the Comsol MutltiphysicsTM Finite Element code. In the second part, the numerical design of the PM is introduced. DEP collection zones are identified owing to the DEP mode. DEP collection rates are evaluated owing to the applied voltage. The influence of indium titanium oxide deposit on the cover glass is studied. Simulations are compared with experiments with latex microparticles. In the final part of the paper, experiments with DNA in the PM show the feasibility of this microdevice for such molecules. Positive DEP collection is observed with rather fast kinetics. 2. Mathematical models DEP is the electromechanical response of real dielectric matter due to polarization [1–4]. Real dielectrics assume that
matter has finite conductivity in addition to its dielectric permittivity. In microsystems, DEP is a means for actuating liquids [5] as well as moving particles and molecules. Usually liquid DEP is the name used when liquids are concerned and we will focus here only on particles and molecules suspended in an aqueous liquid (buffer). The DEP force acting on them depends both on the electric field nonuniformity and the particle polarization with respect to the buffer. Particle polarization is represented by the point dipole model which makes the assumption that the dipole does not modify the external applied electric field. This assumption is central because it makes it possible not to take into account the presence of the particle in the liquid sample for the electric field calculation. It is generally filled when particle dimensions are negligible compared with the electrode gap. Previous studies have shown that the point dipole model fails in microsystems when particles are located in the vicinity of the electrodes [6], however it is sufficient for a designing work. Although determination of the DEP force field provides useful input for designing because we obtain information about DEP orientation, in experiments DEP phenomenon is evaluated by the optical detection of particle fluorescence. For comparing simulations to experiments, the concentration of particles must be calculated. EHD is the liquid motion resulting from externally applied electric fields [7]. The liquid motion is due to Joule heating (electrothermal flow due to the permittivity and conductivity gradients) and/or the presence of an electrical double layer on electrodes (AC electroosmosis flow) [8,9]. When using AC electric fields, EHD can compete with DEP. In this study, EHD effects are considered negligible. This assumption is discussed in Section 3.4. 2.1. AC complex electrokinetics equation for the electric field Electric field is governed by the Maxwell’s equations. In the case of real dielectrics, several assumptions are retained to recast Maxwell’s equations into the AC complex electrokinetics equation: the point dipole model, the quasi-static assumption (magnetic field due to the time varying electric field is negligible [7]) and the use of fictive complex numbers to simplify the treatment of time derivatives. The AC complex electrokinetics equation is given by =:ð s = V Þ ¼ 0,
(1)
where V is the phasor of a fictive complex potential defined by V ðtÞ ¼ V eiot ,
(2) 2
t is the time, i the complex number such as i ¼ 1 and o is the angular frequency of the AC voltage applied to the electrode ðo ¼ 2pf Þ. Electric potential V is the real part of V , real(V). In (1), s is the complex conductivity such that s ¼ s þ io,
(3)
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
where s and e are, respectively, the conductivity and the permittivity of the media bounded by the electrodes (the liquid sample and possibly the electrodes coating for insulation). From (1) the complex electric field is deduced from V by E ¼ = V .
(4)
The choice of the calculation domain is guided by the knowledge of the boundary conditions associated with (1). Due to the high conductivity of gold, electrodes are considered isopotential and their thickness is neglected. On gold surfaces, the applied AC voltage is in phase opposition between two adjacent electrodes (Dirichlet condition) V ¼ V 0 ,
(5)
and other boundaries are insulated (Neumann condition) E n ¼ E :n ¼ 0,
(6)
where n is the outer normal vector of the boundary. Note that (1) can be simplified into the Laplace equation ðD V ¼ 0Þ if only one medium is present between electrodes (no electrode coating for example) and its complex conductivity is constant. 2.2. DEP force expression The time average DEP force exerted by an electric field E ¼ realðEÞ on a spherical particle of volume Vp is given by [4] FDEP ¼ hFDEP ðtÞi ¼ 34V p b realðF CM Þ=ðE :E Þ,
(7)
where E is the complex conjugate of E, eb the permittivity of the buffer where the particle is suspended and FCM the Clausius–Mossotti (CM) factor such that F CM ¼
p b . p þ 2b
(8)
ek is the complex permittivity for the buffer ðk ¼ bÞ and the particle ðk ¼ pÞ defined by sk k ¼ k i , o
513
in (9). Thus, in (9) for k ¼ p, O’Konski [10] proposes 2ks ¼ s þ ss , (10) a where s is the bulk conductivity of the particle, ss its surface conductivity and a its radius. ks has to be determined experimentally.
sp ¼ s þ
2.3. Time particle concentration distribution equation Particle concentration C under the joint influence of DEP forces and particle Brownian motion, is given by the transient advection–diffusion equation [11] qC þ r:½D rC þ CuFDEP ¼ 0, (11) qt where FDEP is given by (7) and D is the diffusion coefficient of particles (m2/s) such that D ¼ ukB T, T is the temperature (K), kB the Boltzmann’s constant (1.38 1023 J/K) and u the particle mobility (s/Kg) such that u¼
1 . 6pmb a
(12)
mb is the buffer dynamic viscosity (Kg/m/s). The assumptions retained for setting (11) are described in [7]. Expression (12) is deduced from the Stokes force Fd which represents the drag effect undergone by a single sphere of velocity vp in a low Reynolds viscous flow [7] Fd ¼ 6pmb avp .
(13)
For liquids, the Reynolds number is always very low in microsystems because of the small dimensions. As DEP forces concentrate particles on surfaces, we have to keep in mind that (12) and (13) can become unsuited as particle concentration C increases. Boundary conditions associated with (11) are either Dirichlet conditions when particle concentration is known, C ¼ C0,
(14)
or Neumann conditions for other boundaries for which impermeability to particles or symmetry is expressed, rC:n ¼ 0.
(15)
(9)
where ek is the dielectric permittivity such that k ¼ 0 rk , erk being the dielectric constant and e0 the permittivity of vacuum (8.85 1012 F/m). Relations (8) and (9) were established for real dielectrics theory and fail for aqueous suspensions of latex particles and macromolecules such as DNA. Water is a high polar media, and even when pure this liquid is capable of self-ionization and generation of a double ionic layer on the particle surface. Strong DEP forces arise which are not readily accounted for by simply considering relations (8) and (9) [10]. For spherical latex particles, electrical interactions at the particle surface, between the particle dipole and ions from the buffer, provide a surface contribution ks to particle conductivity
2.4. Models parameters Parameters which appear in previous equations relate to particle and buffer properties. In aqueous buffers, the dielectric properties of the particle depend strongly on interfacial phenomena arising at its surface [1,10]. Green and Morgan [12] measured the effective conductivity of fluorescent 557 nm diameter latex spheres commercialized by Molecular Probes (MP557). MP557 were suspended in a potassium phosphate (PP) aqueous buffer of several molarities to provide variable medium conductivity. They found a value of 1 nS for ks in a 10 mM PP buffer with a conductivity of 0.17 S/m. According to (10) and assuming
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
514 Table 1 Simulation MP557/buffer parameters MP557
Aqueous buffer DI
s (S/m) ss (S/m) er (adim) a (m) mb (Kg/m/s) rb (Kg/m3) T (K)
0 0.072 2.55 278.5 109 — —
Histidine 50 mM 4
5.5 10 — 78.5 — 103 103 300
8 103 — 78.5 — 103 103 300
that the bulk conductivity s is zero for latex, this leads to a surface conductivity ss of 7.18 103 S/m for MP557. In our experiments, MP557 (Ref. F-8813) are suspended in two aqueous buffers: deionized water (DI) and histidine 50 mM, the conductivities of which were measured with a MeterLab CDM210 conductimeter (see Table 1). Although those two buffers are much less conductive than Green’s 10 mM PP buffer, we use Green’s measurement (ss ¼ 7.18 103 S/m) for MP557 in all simulations. Table 1 summarizes the particles/buffer properties used in our calculations. 2.5. CM curves CM curves plot the variation of the real part of the CM factor (8) according to frequency (Matlab calculations). These curves describe the theoretical frequency variation of the DEP force. They are very useful for visualizing the frequency ranges where the DEP mode is positive (real(FCM)40) and negative (real(FCM)o0). Fig. 1 displays CM curve for MP557 suspended in different aqueous buffers with variable conductivities. CM curves for MP557 in aqueous buffer show that increasing the buffer conductivity enhances the negative DEP mode. For high buffer conductivity, whatever frequency is used the positive DEP mode is not possible. 3. PM numerical design Our goal here is to design a generic microdevice which is able to collect particles on a substrate independently of the applied frequency or the buffer composition, i.e. the DEP mode. The numerical studies gave sufficiently good results for the component to be manufactured. Here we will not describe all the numerical stages in designing the component but only the final step where experiments and numerical results were compared. The first prototype of the PM is a microchannel 1 cm long, 200 mm wide and 100 mm deep. Steam oxidation of a 4 in silicon wafer leads to a 1 mm silicon oxide layer. The PM channel is made using deep silicon chemical etching techniques. Two layers, one of titanium (100 nm) and the other of gold (400 nm) are
Fig. 1. Clausius–Mossotti curves for MP557 in several aqueous buffers: deionized water (DI), histidine 50 mM (Hist) and PP 10 mM used by Green [12]. DEP is positive for frequencies below 2.28 MHz (crossover frequency). For buffer conductivity greater than 80 104 S/m, only negative DEP is possible.
deposited on the entire channel surface. Further chemical etching at the bottom of the channel removed the metal deposit to isolate the two tilted electrodes (tilt angle of 54.71). At the bottom, the metal deposit could not be completely removed and the electrodes extend slightly as displayed in Fig. 2 right. The entire structure is then insulated with a 200 nm silicon nitride (Si3N4) layer deposited by a chemical vapor deposition technique. 3.1. Influence of the silicon nitride coating on gold electrodes The influence of insulating Si3N4 and silicon dioxide (SiO2) layers is discussed extensively elsewhere from the numerical and the experimental studies on interdigitated electrode microdevices (Pham et al., in preparation). Briefly, 200 nm thickness of Si3N4 is a compromise between no insulation, which restricts the use of the PM to small voltages (oa few volts) because of the nondesired electrochemical process (bubbles formation on electrodes), and a too thick insulating layer, which screens the electric field and therefore reduces the magnitude of the DEP force inside the aqueous buffer, especially at low frequencies. The Si3N4 dielectric constant er is 7 and its conductivity is assumed to be zero. Fig. 3 shows an example of the Si3N4 influence on the electric field decrease when moving away from the electrode surface (path A of 10 mm length) for various frequencies: These calculations confirm that the electric field is all the more screened by the Si3N4 layer that the frequency is low. For frequencies higher than 100 kHz, the electric field distribution is not frequency dependant anymore and is slightly attenuated (compare the curves with the ‘no Si3N4’
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
515
electrical contacts channel
54.7° 100 µm 1 cm 200 µm
Fig. 2. Pyramidal Microdevice prototype no. 1: picture seen from above (left), channel cross-section dimensions and tilted electrodes (black lines) (right). The two electrodes are polarized with opposite AC voltage.
Fig. 3. Decrease of the calculated electric field intensity from the electrode surface (path A, left) for various frequencies between 500 Hz and 100 kHz (right): the Si3N4 layer thickness on the gold electrode is 200 nm, the applied voltage is 8 V p/p, the electrode gap and the electrode width are 100 mm, the buffer is deionized water (DI, see Table 1).
curve). These results allow us to disregard the presence of the Si3N4 layer for the DEP force calculations in the PM because working frequencies are higher than 100 kHz (4200 kHz). 3.2. DEP force distribution and DEP collection of MP557 2-D numerical simulations are performed on the half cross-section of the PM (see Fig. 2 right). The free surface between the fluid and the air is considered as an isolated surface when solving (1). Fig. 4 displays the DEP force distribution for the positive DEP (left) and the negative DEP (right). We can see that DEP force is oriented toward the electrodes for both DEP modes: at the bottom on electrode edge (zone A) for positive DEP, at the electrode corner (zone C) and at the top on electrode edge (zone B) for negative DEP. Note that the electrode extension at the bottom strongly influences the DEP force field
distribution. To evaluate PM efficiency for collecting particles without experimenting, it was necessary to simulate the distribution of MP557 concentration over time by solving (11). Maximum concentration zones are the collection zones A, B and C displayed on Fig. 4. The spatial integration of concentration C over these regions (squares of size 3 mm) over time gives the change in collection rates. Figs. 5 upper and lower show the influence of the applied voltage on the collection rate. The collection rate is higher and faster as the voltage increases. For positive DEP, 100% of the MP557 can be collected with 40–15 V, but the rate (in zone A) is much slower for 15 V (48 min) than for 40 V (o2 min). In the following simulations and experiments the voltage indicated is the peak value V0. For negative DEP a small proportion of the particles are collected on the electrode corner (zone C: less than 12%) while the majority are collected on the top electrode edge (zone B).
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
516
Fig. 4. DEP force distribution in the half geometry of PM prototype no. 1: for positive DEP (left), and negative DEP (right). Vector length is not proportional to the magnitude of the force.
100 40V
90
20V
collection rate %
80
15V
70 60
10V
50 40
5V
30 20 10 0 0
1
2
3
4
100
12
90
6
40V 20V 15V
10
80
40V
70 60
collection rate %
collection rate %
5
time (mn)
20V
50 40 15V
30
10V
8 6 4
5V
20
2
10
10V
0
0 0
1
2
3 time (mn)
4
5
6
0
1
2
3
4
5
time (mn)
Fig. 5. (Upper) DEP collection rate of MP557 (DI buffer) in zone A as a function of applied voltage for positive DEP. (Lower) DEP collection rate of MP557 (DI buffer) in zones B (left) and C (right) as a function of applied voltage for negative DEP.
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
517
Fig. 6. MP557 (DI buffer) fluorescence in the PM no. 1 bottom seen from above (9 V): for positive DEP (700 kHz, left) and for negative DEP (7 MHz, right).
microscope lens ITO cover glass
54.7°
100 µm
100 µm
Fig. 7. Pyramidal Microdevice prototype no. 2: picture (left), channel cross-section dimensions and tilted electrodes (black lines) (right).
These results were confirmed by experiments. The following pictures show the fluorescence seen from above the PM channel: the fluorescent lines correspond to the two collection zones A (Fig. 6 left) and C (Fig. 6 right), as planned by calculations. In the first version of PM, it was difficult to observe the negative DEP mode with this opened channel. It was not possible to cover the channel because of the too close electrical contacts. Particles could not accumulate at the top on electrode edge (zone B) because of the free surface movements due to the ambient air convection. A second prototype was built with wider electrical contacts and a cover glass was used to close the channel. It is possible to deposit a transparent indium–tin oxide (ITO) layer on the inside surface of the cover glass. ITO is a conducting material. The ITO cover glass is insulated from the gold electrode by the Si3N4 layer (200 nm) and by an UV adhesive layer which bounds the ITO cover glass to the wafer (Fig. 7). The channel is now 100 mm wide and all the metal has been removed from the bottom. Without ITO, the cover glass is considered as an isolated boundary, this case being equivalent to the uncovered PM no. 1 (Figs. 2–5). With the ITO layer, the cover glass is considered isopotential ðV ¼ 0Þ when solving (1). Fig. 8 shows that an ITO cover glass presents an advantage over a cover glass without ITO due its conductive properties. For negative DEP, the accumulation zone is moved from the top on electrode edge (zone B on Fig. 4 right) to the middle
when coating the cover glass with ITO (zone F on Fig. 8 right). For positive DEP zone A (Fig. 4 left) is replaced by zones D and E (Fig. 8 left): For negative DEP the collection zone at the electrode edge (zone C on Fig. 4 right) disappears because the electrode extension on the channel bottom has been removed in the PM no. 2. This explains why collection rates for the negative DEP in the PM prototype no. 2 (see Fig. 9) are slightly better than the ones obtained with the PM no. 1 (Fig. 5b left). Experiments show a good agreement with these calculations. With MP557 in DI buffer, negative DEP collection has been observed in the middle of the ITO cover glass. The existence of a collecting surface for both the positive and the negative DEP mode is the main advantage of PM. As seen from Section 2.5, the negative DEP mode for latex particles MP557 is enhanced for high buffer conductivity and when frequency is increased. Such conditions minimize the electrochemical reaction outbreaks which usually strongly affect the electrode life. 3.3. DEP collection of DNA Simulations with DNA molecules are not possible because of the lack of identified models for their dielectric properties. Few publications reported the possible manipulation of DNA by DEP [3,13] but the conditions for negative and positive DEP are not clearly identified. However, our experiments show that positive DEP
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
518
collection zone F -4
x10 1.15 1.1 1.05 1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 -0.05 -0.1 -0.15
1.15 1.1 1.05 1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 -0.05 -0.1
E
ITO glass cover
collection zones
D -0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
ITO glass cover
-0.1
0
0.1
0.2
0.3
0.4
x10-4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3 -4
x10
Fig. 8. DEP force distribution in the half geometry of PM prototype no. 2 with an ITO cover glass: for positive DEP (left) and negative DEP (right). Vector length is not proportional to the magnitude of the force.
3.4. EHD fluid motion
Fig. 9. MP557 (DI buffer) negative DEP collection rates in zone F according to the applied voltage in the half geometry of PM prototype no. 2.
collection is possible in the PM for such molecules. The PM used for experiments is the PM no. 2 with a cover glass without ITO coating. The collecting zone for the positive DEP is located on the electrode edges on the bottom of the channel as expected by calculations (zone A, Fig. 4 left): These experiments show that the PM is able to collect DNA molecules. In deionized water, only the positive DEP mode was observed. Collection kinetics are rather fast (5 min).
When using AC electric fields in aqueous buffers, EHD fluid motion can compete with DEP [7–9]. Green et al. [14] reported that for aqueous KCl buffers, frequencies above 1 MHz and for coplanar electrodes (gold–titanium), that EHD flows are due to AC electroosmosis. They also studied the influence of several parameters such as voltage, electrode gap and buffer conductivities, on the fluid velocity which reaches a maximum at frequencies always lower than 10 kHz [14]. The authors conclude that this maximum does not depend on voltage but rather on the electrical buffer conductivity. As the conductivity increases, the maximum fluid velocity is moved towards higher frequencies, around 10 kHz for a conductivity of 8.4 102 S/m [14]. More generally, at frequencies around 10 kHz, the DEP regime is usually positive (see Fig. 1) and it is easy to determine if the EHD fluid motion is dominant because particles accumulate at locations different from those expected under positive DEP [8]. Frequencies used in the PM are much greater than 100 kHz: for latex particles, 700 kHz is used for positive DEP, 7–5 MHz for negative DEP. Positive DEP for DNA molecules is observed at a frequency of 200 kHz. For MP557 and DNA molecules, particle accumulation was observed at locations expected under positive DEP (on electrode edges, see Fig. 4 left) which confirms that EHD liquid motion is not dominant under these conditions. Another argument for this assumption is that AC electroosmosis is the result of the force exerted on the electrical double layer
ARTICLE IN PRESS P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
519
Fig. 10. MP557 fluorescence (DI buffer) in the PM no. 2 channel seen from above. Focusing is done on the ITO cover glass. Left: no voltage is applied, fluorescence is uniformly distributed. Right: for negative DEP after 10 min (7.5 V, 5 MHz), a line of fluorescence appears in the middle of the cover. Fluorescence outside the channel is due to water leaks between the cover and the chip.
Fig. 11. Cybr Green II-labeled l phage DNA in deionized water in the PM no. 2 seen from above. The cover glass has no ITO coating. Focusing is done on the bottom of the channel. Left: voltage is 25 V, 200 kHz, fluorescence after 5 min. Right: voltage is 25 V, 900 kHz, fluorescence after 15 min.
by the tangent component of the electric field above the electrode surface [15]. In the PM, as electrodes are placed face to face, the tangential component of the electric field is minimized compared to a coplanar electrode device. A full justification of this discussion needs to be reported in a dedicated paper which is under preparation (Figs 10 and 11). 4. Conclusions Numerical simulation reduces time for development, saves real experiments and leads to better constructions of prototypes—in short: it reduces costs. The Pyramidal Microdevice (PM) has been designed for collecting bioparticles independently of the DEP mode. Experiments show that the PM appears to be a promising device for collecting bioparticles by both negative and positive DEP. The collection zones of latex particle MP557 in deionized water for the two DEP modes and for the two PM prototypes (nos. 1 and 2) were observed as predicted. Collection of l phage DNA in deionized water was done by positive DEP with the PM no. 2. Experimentally, further developments should be focused on the quantification of the bioparticle fluorescence on the collection surface. Then, the kinetics of the particle accumulation could be directly compared to the computed collection rates. Concerning the modeling work, EHD
flows could be studied [15] although frequencies and/or voltages used in the PM prevent them from being dominant. Models exist for latex particle or biological cell polarizability but they do not account for the surface conductance contribution which results from the complex interaction between the particle with its ion atmosphere and which has to be measured [16]. Very recently, Bakewell studied the DEP attraction of 12 kbp plasmid DNA with an interdigitated electrode microdevice [17]. In this work, the conductivity of those DNA/buffer samples is measured to be 5 mS/m. Experimental results are compared to calculations where the DNA effective polarization is estimated from both dielectric spectroscopy measurements and theory. Under 20 MHz, only the positive DEP regime is identified which confirm our observations in the PM. Such a work could be reproduced in the PM, firstly to evaluate more precisely its capability for collecting DNA molecules and secondly to help for the comprehension of the differences between theory and measurements reported in [17].
Acknowledgment The authors would like to acknowledge the French De´le´gation Ge´ne´rale pour l’Armement for financial support (Contract no. 03.010010 00 470 91 50).
ARTICLE IN PRESS 520
P. Pham et al. / Journal of Electrostatics 65 (2007) 511–520
References [1] H.A. Pohl, Dielectrophoresis, The Behavior of Neutral Matter in Nonuniform Fields, Cambridge University Press, Cambridge, ISBN 0 521 21657 5, 1978. [2] G. Fuhr, R. Hagedorn, T. Muller, B. Wagner, W. Benecke, Linear motion of dielectric particles and living cells in microfabricated structures induced by traveling electric fields, in: IEEE Proceedings of Micro Electro Mechanical Systems, MEMS ‘91, An Investigation of Micro Structures, Sensors, Actuators, Machines and Robots, 1991, pp. 259–264. [3] M. Whasizu, O. Kurosawa, T. Nishizaka, T. Shinohara, Molecular dielectrophoresis of biopolymers, IEEE Trans. Ind. Appl. 30 (1994) 835–843. [4] T.B. Jones, Electromechanics of Particles, Cambridge University Press, Cambridge, ISBN 0-521-43196-4, 1995. [5] T.B. Jones, K.-L. Wang, Frequency-dependent electromechanic of aqueous liquids: electrowetting and dielectrophoresis, Langmuir 20 (2004) 2813–2818. [6] A.M. Benselama, P. Pham, P. Atten, Calcul de la force die´lectrophore´tique dans les microsyste`mes biologiques: comparaison du mode`le dipolaire avec le mode`le du tenseur de Maxwell, J. Electrostat. 64 (2006) 437–444. [7] A. Castellanos, A. Ramos, A. Gonza´lez, N.G. Green, H. Morgan, Electrohydrodynamics and dielectrophoresis in microsystems: scaling laws, J. Phys. D 36 (2003) 2584–2597. [8] N.G. Green, A. Ramos, H. Morgan, AC electrokinetics: a survey of sub-micrometer particle dynamics, J. Phys. D 33 (2000) 632–641.
[9] A. Ramos, H. Morgan, N.G. Green, A. Castellanos, AC electrokinetics: a review of forces in microelectrodes structures, J. Phys. D 31 (1998) 2338–2353. [10] C.T. O’Konski, Electrical properties of macromolecules. V. Theory of ionic polarization in polyelectrolytes, J. Phys. Chem. 64 (1970) 605–619. [11] D.J. Bakewell, H. Morgan, Measuring the frequency dependant polarizability of colloidal particles from dielectrophoretic collection data, IEEE Trans. Dielectr. Electr. Insul. 8 (3) (2001) 566–571. [12] N.G. Green, H. Morgan, Dielectrophoresis investigations of submicrometer latex spheres, J. Phys. D 30 (1997) 2626–2633. [13] C.L. Asbury, G. van den Engh, Trapping of DNA in nonuniform oscillating electric fields, Biophys. J. 74 (1998) 1024–1030. [14] N.G. Green, A. Ramos, A. Gonzales, H. Morgan, A. Castellanos, Fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes. I. Experimental results, Phys. Rev. E 61 (2000) 4011–4018. [15] A. Gonzales, A. Ramos, N.G. Green, A. Castellanos, H. Morgan, AC fluid flow induced by nonuniform ac electric fields in electrolytes on microelectrodes. II. A linear double layer analysis, Phys. Rev. E 61 (2000) 4019–4028. [16] W.M. Arnold, U. Zimmermann, Electro-rotation: development of a technique for dielectric measurements on individual cells and particles, J. Electrostat. 21 (1988) 151–191. [17] D.J. Bakewell, H. Morgan, Dielectrophoresis of DNA: time- and frequency-dependant collections on microelectrodes, IEEE Trans. Nanobiosci. 5 (2006) 1–8.
