1
Instituto Tecnológico y de Estudios Superiores de Monterrey
Campus Monterrey
School of Engineering and Sciences
Geometry optimization and prediction of voltage requirements for particle
trapping in EK-driven insulator-based microfluidics
A thesis presented by
Jesus Martin de los Santos Ramirez
Submitted to the
School of Engineering and Sciences
in partial fulfillment of the requirements for the degree of Master of Science
In
Nanotechnology
Monterrey Nuevo León, November 2020
4
Dedication
This work is dedicated to my wife, Yessica, for all her support along this journey and Eren our future son. This work is also dedicated to my parents, Martin and Ludivina.
5
Acknowledgments
I would like to thank my advisor, Dr. Victor Hugo Perez Gonzalez, for his advice and guidance in the development of this project as well as to the committee members, Dr.
Roberto Carlos Gallo Villanueva, and Dr. Mario Moises Alvarez for their comments and recommendations for the improvement of this thesis work, and finally to Cinthia Janet Ramirez Murillo, who has contributed in the literature collection and review.
I also want to thank to Tecnológico de Monterrey for its support on tuition and to Consejo Nacional de Ciencia y Tecnologia (CONACyT), CVU 896178, for the support in living expenses during the master program.
6
7
Optimization and prediction of voltage requirements in EK-driven insulator-based microfluidics particle trapping
by
Jesus Martin de los Santos Ramirez Abstract
Electrokinetically-driven microfluidics devices have shown to be of great interest for researchers especially when applied in biomedical engineering, medical diagnosis, and biological research. A growing field for this technology is the development of Point- of-Care devices, due to their capability to offer portable, fully integrated, easy to use, and low-cost diagnostic platforms. This work presents an analysis of current electrokinetically-driven (i.e., dielectrophoresis, electrophoresis, electroosmosis) devices from the point of view of voltage requirements, especially for insulator-based devices which are characterized by high (hundreds to thousands of volts) input voltages. Moreover, this work addresses the high voltage problem by presenting a geometry optimization methodology, based on recent advances in the area, to reduce and predict the voltage requirements in insulator-based devices.
8
List of Figures
2.1 Graphic representation of DEP, EP and EO principle………...16
3.1 Examples of electrode-based microdevices………25
3.2 Insulator-based devices examples………...38
3.3 Examples of hybrid microdevices……….…………43
4.1 The three design that were used in the simulations……….….47
4.2 Material domains for the geometry designs………50
4.3 Domain and boundary conditions for the electric currents node.………52
4.4 Meshing resulting from two different meshing configuration……….………53
4.5 Flat channel and its material and voltage conditions at the moment of the simulation……….54
4.6 Circular geometry, the material and voltage conditions in the boundaries when performing the simulation as well as the variables: gap and diameter for the circular geometry………55
4.7 Triangular geometry, the material and voltage conditions in the boundaries when performing the simulation as well as the variables: gap and triangle base for the triangular geometry……….…………56
4.8 Plasma treatment used to produce a permanent bonding between the glass slides (left) and the PDMS device (right)……….………....57
4.9 Micrographs of two microfluidic devices and experimental setup……….59
5.1 Example of surface plot and amplification factor plot for a circular geometry, gap=20 µm and diameter=700 µm……….………...61
5.2 3D graphic representation of the electric field amplification factor for the circular geometry as a function of gap and diameter.……….62
5.3 3D graphic representation of the electric field amplification factor for the triangular geometry as a function of gap and base…...……….64
9
List of Tables
3.1 Stimulation voltage requirements in current electrode-based devices……….26-30 3.2 Stimulation voltage requirements in current insulator-based devices………...39-41 3.3 Stimulation voltage requirements in current hybrid devices………...44-45 4.1 Variable and constant parameters used in the model...………48 5.1 Trapping voltage prediction, example 1………...65 5.2 Trapping voltage prediction, example 2………...65 5.3 Comparison of expected and experimental results based on [16]. Trapping
voltage for 1 µm polystyrene beads in a flat channel was 942.3 V………...66
10
Content
Abstract ... 7
List of Figures ... 8
List of Tables ... 9
Chapter 1 ... 12
Introduction ... 12
1.1 Motivation ... 12
1.2 Problem Statement and Context ... 13
1.3 Solution overview ... 13
Chapter 2 ... 15
Theoretical background ... 15
2.1 Electrophoresis and electroosmosis ... 15
2.2 Dielectrophoresis ... 17
2.3 Non-linear electrokinetics ... 17
Chapter 3 ... 19
State of the art ... 19
3.1 Voltage requirements in eDEP devices ... 19
3.1.1 2D electrodes ... 19
3.1.2 3D electrodes ... 21
3.1.3 Optically induced (virtual) electrodes ... 22
3.1.4 Liquid metal electrodes ... 23
3.2 Voltage reduction strategies in insulator-based devices ... 31
3.2.1 iEK ... 31
3.2.2 cDEP ... 34
3.2.3 Other variations (c-iDEP curvature induced, r-DEP, g-iDEP) ... 35
3.3 Hybrid systems ... 42
Chapter 4 ... 46
Methodology ... 46
4.1 Design ... 46
4.2 Simulation ... 47
4.2.1 Geometry construction ... 48
4.2.1.1 Geometry node... 48
4.2.1.2 Materials node ... 49
4.2.1.3 Electric Currents node ... 50
4.2.1.4 Meshing node ... 53
4.2.2 Study ... 54
4.2.2.1 Flat geometry ... 54
11
4.2.3 Circular geometry ... 54
4.2.4 Triangular geometry ... 55
4.3 First experimental stage ... 56
4.3.1 Fabrication ... 56
4.3.2 Experimental setup and experimental procedure ... 57
Chapter 5 ... 60
Results and discussion ... 60
5.1 Electric field amplification ... 60
5.1.1 Amplification factor ... 60
5.2 Trapping voltage prediction ... 64
5.2.1 Preliminary results from the first experimental stage ... 65
Chapter 6 ... 67
Conclusion and future work ... 67
6.1 Conclusion ... 67
6.2 Future work ... 69
Appendix A ... 70
Appendix B ... 70
Abbreviations ... 75
Bibliography ... 77
Published papers ... 89
12
Chapter 1 Introduction 1.1 Motivation
Microfluidics can be defined as the science and engineering of manipulating fluidic systems at the microscale [1]. Thanks to the small size of these systems, it is possible to use different physical phenomena (e.g. electrokinetics) for the manipulation of fluids and particles of interest.
In the last few decades microfluidics has drawn the attention of researchers in especial when applied in electrokinetically-driven (EK) microfluidic devices. These devices use different electrokinetic phenomena (e.g. dielectrophoresis [DEP], electrophoresis [EP], electroosmosis [EO]) to manipulate fluids and particles. One of the most attractive applications for this technology is its implementation in the development of point-of- care (POC) devices. These devices achieve manipulation of fluids and particles in the microscale, therefore, the amount of sample and reactants needed is minimal and this can be translated into cheaper and faster tests. Besides, because of their small size, portability and ease of use are also added to the lists of advantages.
Generally speaking, electrokinetics refers to the effects of an electric field on a material (fluids and particles in suspension) movement [2]. The main electrokinetic phenomena used are DEP, EP, and EO. Both EP and EO manifest in the presence of an electric field while DEP in the presence of spatially non-uniform electric fields.
The effects of EP and EO act on charged particles and ions in the fluid, respectively.
In both cases the responsible for EP an EO is an electric double layer (EDL). The main difference is the reference framework, in EP the fluid is considered motionless while in EO the particles are considered motionless [3].
However, DEP acts only on polarizable particles. For example, when a polarizable sphere is placed in a non-uniform electric field an effective electric dipole is induced which aligns with the electric field. Due to the non-uniformity of the electric field, the magnitude of the field is greater on one side of the sphere than the other which produces a net force on the sphere that will cause a displacement [4]. Figure 2.1 shows a schematic of the DEP, EP, and EO principles.
Today, it is possible to use computational tools in the analysis and development of microfluidic devices. The advantage of using computational tools in different steps of
13
a design is that it allows for predictions on the behavior of systems prior to an experimental stage. Besides making predictions, these tools have also been used to optimize the performance of the microfluidic devices [5–8]. With these, it is easier to know which variants of the device are more likely to work and, at the same time, they serve as bases for even further improvement. Some example of these tools is software as MATLAB, COMSOL Multiphysics, and Wolfram Mathematica.
1.2 Problem Statement and Context
Although it has been mentioned that the electric fields are responsible for EK phenomena, it is also true that the electric field will require a voltage source to be generated. Here is where one of the most essential characteristics of design comes to light and it is the voltage requirement, especially in insulator-based EK (iEK) devices.
It has been reported that iEK devices can easily require hundreds to even thousands of volts to manipulate particles [9]. Hence, it is of utmost importance to devise strategies to reduce input voltage requirements. Recently, a portable power supply with an output in the range of 200 V has been developed [10]. However, a voltage requirement of thousands of volts still prevents the miniaturization of the voltage source [11]. Moreover, it has been reported that high voltages in microfluidic channels filled with conductive solutions generate significant Joule heating [12, 13]. This can cause temperatures to rise above the optimal levels for cell viability and generate electrothermal flow (ETF). Although there have been efforts to effectively use ETF for particle/cell manipulation [13], it remains generally viewed as a negative effect for cell trapping. Also, the application of high voltages, independently of the heating effects they produce, may damage the biological sample, decreasing cell viability [14]. Thus, high input-voltages represent an obstacle towards creating fully integrated, stand- alone platforms that can be used outside of a laboratory.
1.3 Solution overview
It is not until recently that the nature of the phenomena and behaviors present in iEK devices are starting to be understood on a more fundamental level. In 2018, Perez- Gonzalez et al. [15] noticed that when simplifying the design of an iEK device it was possible to reduce the voltage needed for particle trapping. Later, a work done by Cardenas-Benitez et al. [16] proposes a new theory and shows evidence supporting
14
this new iEK theory. Their works support the idea of focusing on the geometry of the devices to enhance the distribution and increase the magnitude of the electric fields without the need of increasing the input voltage. That theory has been already tested and has shown consistent results [17, 18].
Starting from these bases, it was developed a methodology that consists of modeling an iEK microfluidic device and evaluate the enhancement in the distribution and magnitude of the electric field when the geometry of the device changes under determined conditions keeping the voltage constant.
15
Chapter 2
Theoretical background
In this section, the principles of electrokinetics will be presented. In specific the principles of electrophoresis, electroosmosis, and dielectrophoresis. Moreover, non- linear electrokinetics will be presented, a recent model to describe the behavior of microsystems when in the presence of high electric fields.
2.1 Electrophoresis and electroosmosis
As mentioned in the introduction section, both electrophoresis and electroosmosis are electrokinetics phenomena that show when an electric double layer forms in the interface of a fluid and a solid and an electric field is present, electrophoresis acting on particles suspended in a media while electroosmosis acting on the fluids through the dissolved ions [3]. Figure 2.1C shows and schematic for the representation of both phenomena.
The difference is the boundary conditions taken into account for each phenomenon.
Electrophoresis considers the fluid motionless and the particles to be movable, which results in the following equations describing the electrophoretic motion:
𝐮EP = μEP𝐄 (1)
μEP =εmζp
η (2) Where uEP is the electrophoretic velocity, µEP is the electrophoretic mobility, E is the electric field, εm is the medium permittivity, ζp is the particle zeta potential, and η is the medium viscosity.
Now, for electroosmosis, the conditions are that the fluid is movable, and the walls of the channel are motionless, giving:
𝐮EO = μEO𝐄 (3)
μEO = −εmζw
η (4)
16
Where uEO is the electroosmotic velocity, µEO is the electroosmotic mobility, E is the electric field, εm is the medium permittivity, ζw is the channel walls zeta potential, and η is the medium viscosity.
Finally, adding both velocities up the drag for on the particle (spherical) can be obtained:
𝐮p = 𝐮EP+ 𝐮EO (5)
𝐅p= 6πη𝑎𝐮𝐩 (6)
Where up is the particle velocity, Fp is the drag force on the particle (spherical), and 𝑎 is the particle radius.
Figure 2.1 Graphic representation of DEP, EP, and EO principle. A Alignment of an effective electric dipole moment in the presence of a non-uniform electric field. B Electric field distortion in an electrode-based device and its effect on a particle that is more polarizable than its suspending fluid medium (positive DEP – pDEP). C Effects of EP and EO on a negatively charged particle. EP induces a movement on the particle towards the positive terminal. EO induces electroosmotic flow (EOF) on the medium’s ions (dragging the particle with it) towards the negative terminal.
17
2.2 Dielectrophoresis
Dielectrophoresis is the movement of a polarizable particle when they are in the presence of a spatially non-uniform electric field. When the formation of a dipole occurs under these circumstances, the dipole will align with the electric field. Then, as the electric field has a different distribution, this will cause a net force on the dipole [4].
Figure 2.1A and B represent this phenomenon acting on a spherical particle. The dielectrophoretic force acting on a spherical particle is:
𝐅DEP= 2π𝑎3εmRe[K]∇(𝐄 ∙ 𝐄) (7) Where ∇ is the Del operator, and Re[K] is the real part of the Clausius-Mossotti factor.
K is defined as:
K =εεp∗−εm∗
p∗+2εm∗ (8) Where ε*=ε-jσ/ω is the complex permittivity and the subscript p and m are the particle and medium, respectively, j is the imaginary unit, σ is the conductivity, and ω is the angular frequency of the applied voltage signal. Re[K] can be approximated as:
Re[K] =σσp−σm
p+2σm (9) For low frequencies and
Re[K] =εεp−εm
p+2εm (10) For high frequencies.
2.3 Non-linear electrokinetics
Non-linear electrokinetics is a new approach that tries to explain the behavior observed in DC-iEK which in most cases is not in agreement with the commonly accepted formulations that are used to describe iEK systems. It is necessary to mention that non-linear electrokinetic is an area that is in constant development and there is no ultimate model capable to accurately describe iEK systems. The formulations presented here are those derived in the work by Cardenas-Benitez et al.
[16], which at the same time is based on the work by Schnitzer and Yariv [19], the SY model.
18
The model focuses on trying to explain the electrophoretic and electroosmotic behavior of a microfluidic system when is exposes to high electric fields. In [16] are presenter two variants of the model. Here, the model is not going to be deeply analyzed, however, the formulation for the two variants are shown
Variant 1:
𝐮EP = μEP(1)𝐄 + μEP(3)𝐄𝟑 (11) μEP(1) = −εmφT
η (ζ0+Du∙ln(16)
1+2Du ) (12) μEP(3) = −𝑎2εm
ηφT f(Du, ζ0, α, ὰ) (13) Variant 2:
𝐮EP = εmφT2
η𝑎 (ζ0𝐄 + Du𝓤𝟏) (14) As can be seen from the subscripts, both variants are proposals for describing electrophoretic velocity. In the first variant, it depends on μEP(1) and μEP(3) which are first and third-order electrophoretic mobilities, respectively. Both mobilities, at the same time, are nontrivial functions depending on parameters as: Dukhin number, Du;
thermal voltage, φT; dimensionless zeta potential ζ0; and dimensionless coefficients α and ὰ. Variant 2 introduces a function, 𝓤𝟏, which varies with power 3/2 of the electric field. Luckily, for the electroosmotic component of the model, equations (3) and (4) are still in use without modification.
Something noticeable is that using variant 1 in equation (5), and equating it to cero, it is possible to obtain an expression for the electric field of electrokinetic equilibrium condition (EEEC), which is an estimation of the electric field required for particle trapping.
EEEC= √−μEP
(1)+μEO
μEP(3) (15)
19
Chapter 3
State of the art
In this chapter, it is shown a recompilation of previous work done in the area of electrokinetics apply to microfluidics. The focus of this chapter will be on electrode- based dielectrophoresis (eDEP), insulator-based electrokinetics, and hybrid systems which are combinations of eDEP with iEK and another microfluidic manipulation technique and what the voltage requirements for those systems are and how these voltage requirements have changed.
3.1 Voltage requirements in eDEP devices
Electrode based dielectrophoresis devices are commonly known to operate at what would be considered as low voltages, making this their principal advantage.
Additionally, eDEP devices are capable of producing strong electric fields and, consequently, significant non-uniformities on which the dielectrophoretic force depends on. These features are mainly due to the short distances that exist between electrodes. Therefore, since the electric stimulation requirements are low in eDEP based systems, just a few works have aimed at further reducing the input voltage.
Nonetheless, eDEP brings disadvantages that include electrode fouling, possible sample contamination, and high fabrication costs [20–22]. Figure 3.1 shows examples of eDEP devices.
3.1.1 2D electrodes
In eDEP devices, flat electrodes are fabricated on the floor or ceiling of the channel.
This kind of electrode design is known as 2D electrodes and it is the most widely used approach to DEP. Figure 2.1B-D shows examples of 2D electrodes.
In 2015, polystyrene beads (20 μm) and yeast cells (5-6 μm) were captured using a castellated carbon electrodes device [23]. Two electrode designs were used, off-set and non-off-set with an electrode gap between 100-150 μm. Two AC signals were used, 20 Vpp and 10 Vpp both at 100 kHz. 20 Vpp were applied to study polystyrene beads behavior, exhibiting nDEP while 10 Vpp were used for yeast cells, which showed pDEP. The latter allowed for the separation of a mixture of polystyrene beads and yeast cells with separation efficiencies up to 96% with a flow rate of 0.1 ml/h. The
20
same year, Song et al. [24] conducted a study focused on the separation of stem cells from differentiation products (osteoblasts). Stem cells are of interest for regenerative medicine due to their ability to differentiate into other cells with specialized functions.
The device consisted of an array of gold interdigitated electrodes with a 45° inclination angle. It was able to continuously separate human stem cells from osteoblasts. At a flow rate of 1.8 μl/min, an AC signal of 7.2 Vpp oscillating at 3 MHz was used, achieving a collection efficiency of 92% for stem cells and 61% for osteoblasts.
However, when increasing the flow rate to 5.4 μl/min, the voltage had to be raised to 15.4 Vpp (same frequency) to achieve collection efficiencies up to 88% and 69% for stem cells and osteoblasts, respectively. We stress that the voltage was applied in an on-off switching mode to allow particles to flow, instead of getting trapped within electrodes.
The next year, a tunnel dielectrophoresis device was developed [25]. The purpose of this device was to focus particles, in high-speed flows, into a single focal stream. The device achieved that by using an 80 μm wide, 83 μm high, and 6 cm long microchannel. This length is necessary to give particles enough time to migrate to the focal stream. When viewed from a cross-section, electrodes are placed on the four corners of the microchannel and they continue along the channel. Such configuration allows, when particles experience nDEP, to only have one focal stream. Polystyrene beads (9, 15, and 20 μm) were successfully focused when applying 10 Vpp at 1 MHz at an average flow speed of 5 cm/s. Similar experiments were conducted on HeLa cells. Here, AC signals were 15.4 Vpp and 13.8 Vpp, both oscillating at 10 MHz and flow speed was set to 8.7 cm/s and 11 cm/s, respectively. The viability of HeLa cells was estimated at 85.3% after experiments. That year, a device that intended to trap bioparticles when bounded to polystyrene beads (5 μm) was studied [26]. An interdigitated electrode design was used where fingers of each comb had different widths (20 μm and 80 μm, respectively) with a 20 μm inter-electrode gap. First, a 10 V and 5 kHz signal was used, showing no particle trapping effect. Once the voltage was increased to 15 V (same frequency) polystyrene beads were successfully trapped on the edge of slim electrodes. After that, the frequency was increased to 10 MHz, showing a reduction in the trapping effect and highlighting the importance of careful electrode design and voltage selection.
21
In 2017, Barik el at. [27] were able to trap DNA molecules and particles at the nanoscale using voltages with peak amplitude as low as 0.45 V. Polystyrene beads (190 nm) were trapped by pDEP with a peak amplitude of 0.75 V oscillating at 1 MHz.
When increased to 10 MHz, however, particles were released due to nDEP.
Nanodiamonds (40 nm) were also used, first trapping was observed at 0.45 V and 1MHz. However, 2 V at 100 kHz were the conditions where most particles were trapped. Finally, 10 kbp and 500 bp DNA molecules were successfully captured at maximum voltages of 3 and 2.5 V, respectively, both at 1 MHz. All this was achievable due to the device’s design. It consisted of an interdigitated electrode array, which incorporates an atom-thin graphene sheet. This graphene sheet is connected to the electrodes, producing extremely large electric field non-uniformities at low voltages.
The following year, a device capable of manipulating RBCs (5 μm) along both channel height and width (3D switching) using voltages between 1 and 10 Vpp was reported [28]. To achieve 3D switching, two independent sets (top and bottom) of interdigitated electrodes were used. Each featuring 50 μm wide electrodes with 50 μm inter- electrode gaps. When applying 1 Vpp to the top set and 10 Vpp to the bottom set, RBCs were deflected toward the top set of electrodes. When swapping voltages, RBCs were deflected toward the bottom set. Finally, when voltages were 10 Vpp in both sets, RBCs flowed in the center of the channel. In all cases, the frequency to induce nDEP on RBCs was 10 kHz and a flow rate of 5 μl/h was used.
In 2019, Kim et al. [29] reported a DEP device capable of detecting Alzheimer’s disease biomarkers, Aβ42, and tau-441, by capturing them using voltages between 0.5 and 0.9 Vpp. Two designs were used. The first design was an interdigitated electrode array with 10 μm gaps. It was able to capture Aβ42 with 0.8 Vpp and tau- 441 with 0.9 Vpp. The second design incorporated patterned square (3 μm) microstructures within the electrode gaps in the first design. When including the microstructures, the electric field gradient increased, which translates into a lower capturing voltage. With the second design, Aβ42 and tau-441 were captured with 0.5 Vpp and 0.6 Vpp, respectively. A frequency of 50 MHz was used in all cases to induce nDEP on the target particles.
3.1.2 3D electrodes
3D electrodes aim to make the DEP force effective in the entire channel volume, lowering the required input voltage for affecting particles that would otherwise be
22
located far away from the 2D electrode set (Figure 3.1A). The most common 3D electrode fabrication method is photolithography + pyrolysis [22]. This technique uses a polymer, usually SU-8, as a base material to produce carbon electrodes [30].
In 2015, Puttaswamy et al. [31] developed a 3D interdigitated electrode DEP device.
The electrodes (200 μm wide and 40 μm tall) were fabricated from a mixture of silver conductive adhesive and carbon nanopowder. Polystyrene beads (10 μm) were used to test the device by directing them to three outlets. When no voltage was applied, beads headed to the first outlet. After applying a 40 Vpp at 5 MHz signal, particles experienced nDEP and were deflected to the second outlet. Increasing the voltage further to 60 Vpp (same frequency), particles deflected toward the third outlet. A flow rate of 0.2 μl/min was used to introduce the particles into the channel and a flow rate of 0.6 μl/min for the sheath flow.
A couple of years later, a 3D carbon electrode device was used to study the separation of live and dead U937 (human myeloid leukemia) monocytes (23 μm [live] and 22 μm [dead]) under lower than 40 Vpp stimulation voltages [32]. 3D carbon electrodes were fabricated via pyrolysis and their geometry consisted of 100 μm tall cylinders with 50 μm in diameter, placed over an array of interdigitated electrodes. Separation of live and dead cells was achieved with an AC signal of 20 Vpp at 300 kHz and a 1 μl/min flow. At that frequency, live cells experienced pDEP while dead cells showed no response. After separation, 90% of the dead cell were successfully removed from the mixture.
3.1.3 Optically induced (virtual) electrodes
Virtual electrodes (VE) can be defined as electrodes that manifest when a light beam is projected onto a photoconductive surface (Figure 3.1F). The method reported in [33–36] consists of using ITO slides as electrodes in the ceiling and floor of the microchannel. One of the ITO slides was covered by a photoconductive layer.
Therefore, when light meets this layer its impedance is reduced, creating the VE and generating a non-uniform electric field. With this technique, a great degree of flexibility is achieved in the geometry of the electrodes (even making them dynamic) and therefore, it provides enhanced particle manipulation capabilities.
In 2016, Chiu et al. [36] used an ODEP device to isolate circulating tumor cells (CTCs, 23±2.1 μm). The device consisted of a T-shaped microchannel composed of two ITO glass slides as floor and ceiling, one of which was coated with a photoconductive layer.
23
The isolation area was focused on the T intersection, where CTCs were separated from leucocytes. 8 V at 100 kHz were applied to the ITO slides, the CTCs were visually identified, and the image of a ring was projected on the photoconductive layer, creating a non-uniform electric field, trapping CTCs inside the ring. Afterward, a light bar was projected on the T intersection to remove non-target particles. Then, CTCs were moved to a collection area. Next year, a similar device was used to trap and isolate CTCs (18.3±3 μm) from a mixture with leucocytes (8.4±1.2 μm) [35]. However, it was possible to achieve particle trapping with an AC signal of 5 V at 100 kHz. A light bar width of 40 μm was chosen for producing the maximum velocities when moving CTCs and leucocytes, around 148 and 63.8 μm/s, respectively. Up to 100% cell purity was achieved in this experiment through the presence of multiple trapping zones along the microchannel, which allowed reducing the trapping voltage. One year later, Chiu et al.
[34] developed another ODEP device to manipulate CTC clusters (individual CTCs 14- 25 μm). In this case, the VE pattern was a matrix conformed of square elements. The optimal conditions for CTCs cluster manipulation were experimentally determined to be a matrix element of 100 μm by 100 μm, a flow rate of 0.5 μl/min, measured velocity of 110 μm/s, and an AC signal of 5 V at 100 kHz. We note the voltage requirement did not change from the previous work. However, an improvement in the degree of particle manipulation took place, especially due to the dynamic properties of VE—something that is not possible with conventional electrode technologies, either 2D or 3D.
3.1.4 Liquid metal electrodes
Gallium based liquid metal alloys have high electrical conductivity and have been used as electrodes [37]. This material presents low viscosity and high surface tension, allowing for fast fabrication and flexibility in the type of shapes that can be generated.
One example of this work is the creation of liquid metal droplet-shaped Galinstan electrodes (68.5% gallium, 21.5% indium, and 10% tin) using dielectrophoresis. In 2015, Tang et al. demonstrated trapping of 80 nm tungsten trioxide nanoparticles at 15 V and 1 MHz [38]. Another work used liquid metal electrode channels consisting of 67% Ga, 20.5% In, and 12.5% Sn to stretch red blood cells from 6 µm to 8 µm using voltages from 1 Vpp-10 Vpp at 1.5 MHz, providing evidence that this system can be used for cell characterization [39]. Electrohydrodynamic mixing was explored at 100 VDC using eutectic gallium indium liquid channel electrodes contained by PDMS posts. Liquid electrodes are in contact with the main channel through the openings
24
between posts [40]. A microdroplet generation system with electrodes composed of 75.5% Ga and 24.5% In operating between 2600 V and 3000 V achieved the formation of 600 µm long droplets [41]. Studies such as these suggest that the use of liquid electrodes is promising for its versatility and ease of fabrication. Table 3.1 shows a list of the characteristics of eDEP devices.
25
Figure 3.1 Examples of electrode-based microdevices. A 3D electrode array (i) SEM image of the electrodes (ii)-(iv) shows the device performing a separation of HCT116 (green) and lymphocytes (red) [42]. B eDEP device used to concentrate HeLa cells [43]. C interdigitated electrode array in combination with a layer of atom-thin graphene to capture different particles such as polystyrene beads, nanodiamonds, and DNA [27]. D Device with interdigitated- irregular electrodes used for the concentration and detection of CA 19-9, a pancreatic cancer biomarker [44]. E Serpentine-shaped carbon electrodeused to separate live from dead yeast cells [45]. F Array of optically induced (or virtual) electrodes, projected on a photoconductive layer over an ITO surface used to manipulated the trajectory of a cancer cell cluster [34].
26 Table 3.1 Stimulation voltage requirements in current electrode-based devices
DEP type and application
Particle type+size Electrode material Electrode geometry Electrode dimensions
Voltage
eDEP, cell
stretching NB4 leukemia cells Ø
14.04 μm ITO Two rectangular
electrodes 20 µm gap AC: 2-9 Vpp, 1 MHz
[46]
eDEP, patterning Yeast cells Ø2.15 μm PCB standard 16 concentric circles 700 µm surrounding circle Ø; 400 µm center Ø; 100 µm gap
AC: 2-10 V*, 6 MHz [47]
eDEP, trapping dsDNA 10 kbp Quartz with gold coating Nanopipette O.D. 1.0 mm, I.D. 0.5
mm AC: 10-20 Vpp, 0.5-4
MHz [48]
eDEP, patterning Yeast cells PCB (copper) 16 concentric circles 700 µm surrounding circle Ø; 400 µm center Ø; 100 µm gap
AC: 60 Vpp, 6 MHz [49]
eDEP, trapping Polystyrene
beads Ø 5 μm Cr-Au Interdigitated ratios between +/-
20:20 µm, 20:40 µm, 20:60 µm, 20:80 µm
AC: 10-15 V*, 5 kHz- 10 MHz
[26]
eDEP, separation Polystyrene beads Ø
5-10 μm Au Interdigitated 40 μm electrode
length AC: 5.71 Vpp, 60
kHz [6]
eDEP, separation Polystyrene (10 μm) and white blood cells (12.4 μm)
Au-Cr Interdigitated ** AC: 7.5 Vpp, 800
MHz [7]
eDEP, separation Polystyrene (5 μm and 15 μm), yeast cells
Ti-Au Triangle H=1213 μm, Cx=1153
μm & Cy=375 μm AC: 8 Vpp, 16 Vpp, 24 Vpp, 100 kHz (PS) and 1050 kHz (yeast)
[50]
eDEP, separation Polystyrene (7 μm and 20 μm), yeast cells (5-6 μm)
Carbon paste Interdigitated, castellated 100 μm –150 μm AC: 10-20 Vpp, 30 kHz-3 MHz [23]
eDEP, sorting Gold nanoparticles
(10 nm) Gold Arrow-like (with rounded
point) 10 μm gap AC: 18-26 Vpp, 1
kHz-20 MHz [51]
27 eDEP,
separation
‘liquid electrode’
HT-29 (13.2 μm), PLT (1.8 μm), T- Lymph (7 μm), RBC (5 μm), MD 231 (12.4 μm)
Ti-Pt Rectangular, comb-like
structure length varied from 20
μm to 40 μm AC: 6-10 Vpp, 100 kHz [52]
eDEP, trapping Polystyrene (190 nm), nanodiamond (40 nm), DNA (10 kbp, 500 bp)
Ti-Pd gate electrodes, covered by insulator and graphene electrode layer with Cr/ Au ohmic contacts and Cr/Al electrical leads.
Graphene sheet edge Gate and graphene electrode:
interdigitated with 5 μm thick fingers
AC: 0.9-6 Vpp, 1 MHz, 10 MHz [27]
eDEP, isolation HCT116 cells Doped silicon crystals 3D comb array with
electrode digits consisting of castellated blocks
15 pairs of electrode digits, 10 identical pores per digit, each pore 60 μm in diameter
AC: 10-25 Vpp, 60- 400 kHz
[42]
eDEP, patterning MC3T3-E1 bone cells
(7.5 μm) Stainless steel Honey-comb multilayer L:140 μm W:50 μm, mean pore size:140 μm. Thickness: 50 μm
AC: 10 V*, 500 kHz [53]
eDEP, trapping Polystyrene (1 μm) Carbon 3D posts 100 μm height, 50 μm
diameter AC: 28 Vpp, 5-50 kHz [54]
eDEP,
concentration Polystyrene (1 μm), E. coli, MS2 virus vs Troponin I
ITO Coplanar electrodes 100 nm thick ITO layer
and 25 μm gap between electrodes.
AC: 5 Vpp, 10 Vpp, 10 kHz, 100 kHz, 2 MHz [55]
eDEP,
concentration Yeast cells (5 μm) Glass-like carbon Cylindrical 50 μm diameter, 100 μm height, center to center 115 μm
AC: 20 Vpp, 100 kHz [56]
eDEP,
characterization Candida strains - albicans, parapsilosis, tropicalis (5-7 μm)
Carbon Cylindrical 50 μm diameter, 100
μm height, distance between posts 58 μm in all directions
AC: 20 Vpp, 10 kHz - 1 MHz [57]
eDEP, isolation AS2-GFP Au (ground electrode) and
ITO (voltage electrode) Serpentine with “V” shaped
cross-section 700 μm width and 200
μm depth AC: 18 Vpp, 10 kHz [58]
eDEP, sorting Stem cells (hMSCs) Au-Cr Interdigitated 50 μm width and 50
μm gap AC: 7.2 Vpp, 3MHz
[24]
28 eDEP, detection
and quantification Biotin functionalized polystyrene beads (0.74 μm)
Au Interdigitated pearl-shaped ** AC: 10 Vpp, 500 kHz
– 2 MHz [59]
eDEP, sorting RBCs Au-Cr Interdigitated (2 sets: top
and bottom) 50 μm width and 50
μm gap AC: 1 Vpp and 10
Vpp (switching sets), 10kHz (both) [28]
ODEP,
purification PC-3 ITO + photoconductive layer Rectangle and “O” shaped Rectangle width: 150, 200 and 250 μm; “O”
shaped: 40 μm
AC: 8 V*, 100kHz [36]
ODEP,
concentration 5-μm particles ITO + photoconductive layer Square lattice ** AC: 20 Vpp, 250 kHz [33]
ODEP, isolation
and purification PC-3 (18.3 ± 3 μm) ITO + photoconductive layer Interdigitated Width: 40 μm; gap: 80
μm AC: 5 V*, 100 kHz
[35]
ODEP, isolation H209 ITO + photoconductive layer Square matrix 100 x 100μm (by
matrix element) AC: 5 V*, 100 kHz [34]
eDEP, patterning HFF Stainless steel Hexagonal ** AC: 14-56 Vpp,
300kHz [60]
eDEP, focusing PS beads (9, 15, 20
μm), HeLa cells Au Parallel along channel ** AC: 10,13.8, 15.4
Vpp, 1-10MHz [25]
eDEP, sorting PS beads (10 μm) Ag-conductive adhesive and
carbon nanopowder Interdigitated 5000x100x40 μm AC: 40, 60 Vpp, 5MHz [31]
eDEP, separation RBCs Ti-Au Castellated and serpentine ** AC: 8 Vpp, 1MHz
[61]
eDEP, counting
and detection Shewanella
oneidensis Au Interdigitated ** AC: 1, 3, 5 Vpp, 0.1-
1 MHz [62]
eDEP,
concentration PS beads (20 and 40 nm), BSA
(14x4x4nm)
Au on ITO Cone in matrix array Cone: H=100nm, base
diameter= 100nm AC: 10 V*, 2.5 MHz [63]
eDEP,
concentration AuNP (150 nm) Au Planar 10 x 25 μm AC: 17 Vpp, 100kHz
[64]
eDEP, trapping Yeast Au and C Serpentine ** AC: 10 Vpp, 100kHz-
1MHZ [45]
eDEP, separation U937 monocyte cells C Interdigitated with 3D
structures on top 3D structures:100 μm height and 50 μm diameter
AC: 20 Vpp, 50kHz- 1MHz [32]
eDEP,
concentration Aβ42 and tau-441 Ta-Pt Interdigitated with
microstructures in between ** AC: 0.5-0.9 Vpp, 50MHz [29]
29 eDEP, detection Biotinylated DNA +
PS beads (750 nm) Au Irregular interdigitated ** AC: 10 Vpp, 0.5-
2MHz [65]
eDEP, separation K562 cells Cu Interdigitated Height: 25-30 μm,
Thickness: 30 μm, width: 40 μm
AC: 9 Vpp, 48.64 MHz [66]
eDEP, detection Biotinylated CA 19-9 antibody + PS beads (750 nm)
Au Irregular interdigitated ** AC: 10 Vpp, 0.5-2
MHz [44]
eDEP, separation PS beads (2, 5, 8
µm) Cr-Au Insulator microchannel
Electrodes: two linear electrodes with varying distance along channel
Electrodes: width 50 µm,
gap 50 µm (increasing over length of
channel), height 100 nm
AC: 10 Vpp, 200 kHz [67]
eDEP, trapping Tungsten trioxide (WO3)nanoparticle (80 nm diameter)
Galinstan liquid metal electrodes (68.5% gallium, 21.5% indium, and 10% tin)
Circular droplet-shape 3D electrodes(half- sphere)
Electrodes: Cr/Au pad diameter 80 µm, height 30 µm
AC: 15 V*, 1 MHz [38]
eDEP, cell
stretching RBCs Liquid metal consisting of 67% Ga, 20.5% In, and 12.5
% Sn and ITO electrode.
Liquid electrode channel and triangular ITO electrode
Liquid metal electrode:
1000 µm width, 3 cm length
AC: 1 Vpp-10 Vpp at 1.5 MHz
[39]
eDEP, mixing N/A Eutectic gallium indium liquid electrode-EGaIn (Ga 75% In 25% by weight)
Liquid electrode channel walls contained by PDMS posts
1000 µm width, 50 µm
height DC: 100 V [40]
eDEP, microdroplet generation
Water droplets formed in silicon oil (660 µm droplet length)
Ga 75.5In24.5 (75.5 wt.%
Ga, 24.5 wt.% In) Liquid electrode squared
channels 70 µm electrode width DC: 2600 V-3000 V [41]
LFFF-DEP,
isolation CTCs Au Interdigitated electrodes with
channel bonded on top Width of each
electrode ~60 μm and the width of the main channel is ~200 μm
Top set of electrodes at 10 Vpp, 10 kHz while the bottom set
30
of electrodes is held at 15 Vpp, 40 kHz [68]
* the author(s) did not specify whether the voltage was measured as Vrms, Vp, or Vpp.
** the author(s) did not provide detailed information about the electrode dimensions
31
3.2 Voltage reduction strategies in insulator-based devices
In comparison with electrode-based devices, some of the advantages of insulator- based devices are easy and low-cost fabrication [22], and no direct contact between electrodes and sample at the trapping region, which avoids contamination of the sample due to electrolysis or corrosion, or of the electrode due to electrode fouling [21, 69]. A disadvantage of introducing insulating materials in these devices is that the voltage required to achieve particle or cell manipulation is much higher than that required in devices where the electrodes are directly in contact with the sample [9].
It is important to note that, although insulator-based dielectrophoresis (iDEP) has been reported in the literature for two decades, a study published recently by our group [16]
provided evidence that DEP is generally negligible in devices stimulated with DC voltages. Findings suggest that the main contributors to particle movement in such devices are linear and non-linear electrophoresis and electroosmosis. Therefore, insulator-based electrokinetics (iEK) is a more adequate term for this technology.
Nonetheless, two decades of work in the ‘DC-iDEP’ field to decrease input voltage requirements deserve mention.
3.2.1 iEK
In 2000, Cummings and Singh [70] introduced the use of an array of insulating posts rather than only using a single insulator constriction, as had been done in previous works. In the following years, they were also the first to present streaming DEP [71, 72] using insulating structures as an alternative to embedded metal electrodes to produce non-uniform electric fields. Their initial work presented streaming DEP at a field of 800 V/cm and trapping DEP at a field of 1000 V/cm using carboxylated latex spheres. The setup for these devices typically includes one reservoir on each end of the main channel, where wire electrodes are placed so that the voltage is applied for particle manipulation in the channel. In 2004, the first application of iDEP for bacteria and polystyrene beads separation and concentration at fields up to 2000 V/cm was presented [73].
A more recent study presented a device that assessed the DEP response of submicron (100 nm - 1 µm) aminated and carboxylated particles. The study considered the response as a function of size and surface charge at low frequency (<1000 Hz) with AC signals of 2800 Vpp and 3200 Vpp, respectively [9]. The voltages in this work were in the range initially used in iDEP. Although they were manipulating nm-sized
32
particles, their insulating structures were not optimized for this particle size as both the post diameter and the gap were in the µm size range. Furthermore, the array of insulating posts consisted of 18 columns of posts in the channel center, adding more resistance to the channel and thus requiring a higher voltage. In 2015, an iDEP device for trapping yeast cells (~6 µm) and 2 µm polystyrene particles in low concentration from a diluted sample with smaller particles was developed, working at a DC voltage of 600 V with 99% trapping efficiency [74]. Besides using larger particles, this device had 16 columns of posts instead of 18. In the same year, β-galactosidase was concentrated and its electrokinetic response was compared to that of IgG [75]. Since proteins are in the nanometer-size range, an optimized design was necessary to manipulate these molecules at a voltage below thousands of volts. This work used an insulating post design, which included vertically tapered semi-triangular posts.
Because of this design, the channel constriction features were in the order of nanometers, with a 100 nm gap at the base of the post and a 500 nm gap at the top of the post. The device was reported to trap β-galactosidase due to nDEP under an applied electric potential difference of 100 V DC and IgG due to nDEP under an applied electric potential difference of 50 V DC. Another work trapped 80 nm gold nanoparticles in 200 µm circular posts with a 50 µm gap between posts in an 8-column array at voltages ranging from 240-360 V [76]. Two years later, another study [77]
used a 100 Vpp AC signal at frequencies ranging from 0.1 to 20 MHz to manipulate human embryonic kidney cells. The cells were modeled as four-layer dielectric spheres to account for the internal structure of the cell. Results showed that the internal mitochondrial structure affects the behavior of the cell, especially when using a frequency of 0.5 MHz and subjecting the cell to pDEP. More connected structures exhibit a higher pDEP effect than those with fragmented structures.
In 2018, Perez-Gonzalez et al. [15] demonstrated that the voltage required for particle trapping was determined for different microfluidic channel designs by systematically varying the number of columns of insulating posts. In previous works, it was common practice to fabricate channels with 10-20 columns of insulating posts [9, 74]. In this work, the columns of posts were gradually reduced from 21 to 11, 3 and finally leaving only 1 column while, for each of these configurations, the minimum voltage required for trapping of 1 µm polystyrene beads was determined. This strategy led to a significant reduction in the voltage required for particle trapping, which decreased from
33
650 V DC in the channel with 21 columns to 250 V DC in the channel with one column (Figure 3.2B). In the following year, an iDEP creek-gap device to determine the DEP properties of human breast epithelial cells (MCF10A) was developed [78]. The device consisted of a creek-gap pattern created on a 70 µm thick SU-8 layer deposited on top of aluminum electrodes adhered to a glass substrate. At 34 Vpp, cells experienced pDEP at 200 kHz and nDEP at 50 MHz. They were manipulating ~12 µm cells, the creek width was 50 µm and the distance between the electrodes was 250 µm, as opposed to the 1 cm conventional distance in iDEP devices. Another important contribution was published in 2018 [79] in which trapping of 100 nm liposomes in a mixture with 500 nm polystyrene beads was demonstrated. For this, a borosilicate nanopipette design was used to trap the polystyrene beads in the mixture at an applied electric field of 10 V/cm. Liposomes were released by reversing the polarity of the applied field. Furthermore, when studying the effect of nanopipette geometry on particle entrapment, the greatest number of particles trapped was obtained with a 2 µm pore size (compared to lower pore sizes) when polystyrene beads were suspended in 10 mM KCl medium. The next year, a device was able to manipulate the response of nanostructures such as exosomes and extracellular vesicles from pancreatic tumor cells using a 70 Vrms AC signal at frequencies ranging from 0.01 to 1.5 MHz [80]. The nanostructures were treated as dielectric spheres with a conductive shell due to the electric double layer. It was determined that at high frequencies (10 MHz or more) the polarization of the particles should result in nDEP and pDEP from 0.1 to 1 MHz.
Experimentally, it was observed that particles shift from pDEP to nDEP at 0.5 MHz.
Efforts have been made to optimize the post array dimensions in iDEP devices. A study by Mohammadi et al. [81] found that a post radius 40 µm larger than what they defined as K—the transversal distance between posts—enhanced particle trapping, allowing them to trap 6 µm particles at 200 V with 70 µm post diameter and 30 µm gap. Additionally, works on post shape optimization have explored different geometries for their cross-section such as diamonds, circles, and squares. In a publication using diamond and circular posts, the minimum trapping voltage required was 500 V for 1 µm polystyrene beads [82]; after a follow-up work of dimensional optimization, the lowest trapping voltage, using 2 µm polystyrene spheres, was 160 V with square prism insulator posts. This trapping voltage was very close to 170 V using circles of comparable size. A higher voltage (450 V) was needed for trapping using
34
diamond-shaped posts (Figure 3.2A) [83]. Other works have explored combined geometries such as circular and sawtooth on the same constriction, trapping 2 µm particles with voltages ranging from 400 V-1000 V DC [21]. Regarding shape, pointy constrictions as in sawtooth or square, or small gaps for circles tend to be better for trapping.
As mentioned in the introduction for this section, recently, Cardenas-Benitez et al. [16]
proposed a new model to predict particle trapping in DC-stimulated iEK devices. 1.0, 1.9, and 5.1 µm particles were trapped when applying ~942, ~851, and ~629 V DC, respectively, in a device with no post and constant rectangular cross-section. The same experiment was then conducted in a device with the same dimensions that included two circular posts with a diameter of 200 µm and a gap of 25 µm. The posts amplify the magnitude of the electric field at the gap, therefore, voltages of 250, 200, and 150 V DC were applied to produce electric fields with the same magnitudes as those used in the empty channel, also achieving particle trapping. These observations
—which cannot be explained by the traditional iDEP theory because DEP is absent in an empty channel— led to a new theory that presents linear and nonlinear electrophoresis and electroosmosis as the mechanisms responsible for particle manipulation. In 2020, Coll De Peña et al. [18] made use of the new DC-iEK theory to characterize and identify different microorganisms. DC voltages ranging from 100 to 800 V were used to determine the trapping voltage for each microorganism in two different channels. This theory will allow for more rational channel geometry designs, aiming for further input voltage reductions in DC-iEK devices.
3.2.2 cDEP
One of the variations that have been developed in insulator-based devices is contactless dielectrophoresis (cDEP). This approach was developed by the Davalos research group in 2009 [84]. In these devices, the channel, commonly made of PDMS with embedded insulating structures (e.g., posts) is isolated from lateral channels—
filled with a highly conductive solution—that function as two liquid electrodes by a thin membrane. This configuration avoids problems such as electrolysis, electrode fouling, and sample contamination. One additional advantage is that since the system shows a capacitive behavior, an electric field can be set up in the main channel by using AC voltages in the lateral channels. This group also combined AC and DC fields to generate electroosmotic flow and dielectrophoretic manipulation of the sample inside
35
the channel [85]. This combination allowed to use lower applied voltages to control the cells than in traditional ‘DC-iDEP’, demonstrating trapping of 1 µm polystyrene spheres with the simultaneous application of 500 VDC and 250 Vrms at 500 kHz. In a more recent publication [86], similar cancer cells were separated by aggressiveness. This initial study found that it was possible to trap single cells at each post (20 µm in diameter) and separate the more aggressive phenotypes to be cultured under different conditions. This was achieved by balancing the pDEP force with the drag force of the fluid at a voltage of 350 Vrms at 30 kHz. This voltage was selected after studying cell viability [87] over a range of voltages and it was determined to maintain high cell viability. Thus, the flow rate was optimized to balance the drag force and DEP force for trapping without increasing the applied voltage (Figure 3.2C). While the previous studies separated similar cells, a more recent study, using only one population of MOSE cells, determined the trapping voltage to be 150 Vrms at 450 kHz [88]. One of the challenges found in the development of cDEP devices is their fabrication.
Another aspect to take into consideration when dealing with DEP devices are the potential losses that may affect the performance of the device. In the case of cDEP devices, in 2012, Sano et at. [89] studied a multilayer cDEP device to detect several factors affecting device performance (e.g., resistance, resonance, and inductance).
Later in 2020, another study [90] aimed at determining the amount of parasitic voltage drops in cDEP. They were able to do this by modeling the device with an equivalent RC circuit, determining the fraction of the total voltage that a cDEP device uses to achieve electrokinetic manipulation. They defined this fraction as the ratio between the voltage that reached the sample channel and the voltage applied to the device. This study concluded that for an optimal trapping, a voltage fraction equal to or greater than 0.5 is required.
3.2.3 Other variations (c-iDEP curvature induced, r-DEP, g-iDEP)
Currently, there are other modifications to insulator-based devices that have been developed for particle and cell manipulation purposes. One of these variants is gradient insulator-based dielectrophoresis (g-iDEP), which basically adds a geometrical gradient along the channel with a sawtooth shape (the constriction size decreases along the channel). The first work introducing this idea with a sawtooth pattern was published in 2007, achieving separation of live and dead Bacillus subtilis cells—live cells were trapped at wider gaps with lower field strength, and dead cells
36
were trapped later along the channel at narrower gaps with higher field strength [91].
The input voltage for this work was 1000 V. More recent publications demonstrated the capability to trap red blood cells at 400 V (Figure 3.2F) [92] and Sindbis virus at 200 V [93].
Other variations include wall-induced dielectrophoresis, in which the device typically consists of a plain channel with a bifurcation at one of the ends for sorting droplets with different dielectric properties. The principle proposed behind these devices is that when sorting dielectric materials (e.g., oil droplets), the otherwise uniform electric field set along the channel is distorted by the dielectric droplet itself. Hence, this non- uniformity varies by the position along the width of the channel at which the droplet is introduced. A work published in 2017 reported on the separation of Janus and oil droplets [94]. Figure 3.2D shows their experimental results of separating oil droplets of 25 µm and 50 µm diameter by size. They also conducted experiments in which 50 µm and 75 µm Janus droplets were separated by size. Finally, a third phase was carried out in which they separated 50 µm Janus and oil droplets using a voltage range of 200 V – 375 V. Larger lateral migration of droplets was observed at higher electric fields, larger particles, or lower electrokinetic mobility of the droplet. This explained the larger lateral migration of the oil droplet than that of the Janus droplet of the same size.
Reservoir-based dielectrophoresis (rDEP) is another perspective on DEP trapping.
These devices take advantage of the electric field non-uniformity created at the intersection of the microchannel and the reservoirs, which are at each end of the channel to introduce the electrodes. Figure 3.2E shows a schematic of a channel of this type and an example of trapping in one of these devices at 25 VDC+350 Vrms [95].
Rather than using the sharp edge in reservoir-channel junction, curvature-induced dielectrophoresis (c-iDEP) is a technique that uses curved channels to generate a field non-uniformity. The first work involving curvature-induced dielectrophoresis was published in 2009, which reported better focusing at higher field strength as observed from a thinner stream of cells at 100 V/cm than at a 50 V/cm field [96]. A couple of years later, the same group developed a device with one inlet, a spiral in the center of the microchannel, and three outlet branches in which they were able to sort polystyrene particles of the same size (10 µm) [97]. The separation was done in terms of surface charge from a mixed suspension of uncoated particles and carboxyl-coated particles. The initial setup in this publication had a voltage at the inlet channel of 400
37
V while the outlet branches were grounded. Table 3.2 shows a list of the main characteristics of insulator-based examples.
38
Figure 3.2 Insulator-based devices examples. A DEP Trapping with optimized (i) diamond- shaped posts, (ii) circular posts, and (iii) square posts [83]. B DEP trapping in a device with (i) 21 columns of posts and (ii) 1 column of posts [15]. C Trapped cells at 300 Vrms at 30 kHz in (i) a device with 100 µm diameter insulating posts showing pearl formation and (ii) in a device with 20 µm posts where only one or two cells can trap on each post [87]. D Separation of oil droplets by size: (i) separation of 25 µm and 50 µm diameter oil droplets, (ii) separation of 50 µm and 75 µm droplets [94]. E (3D)rDEP device schematic and inset showing 5 µm particle trapping (enclosed in red) at the reservoir-channel junction at an applied voltage of 25 VDC +375 V at 1kHz [95]. F g-iDEP device (i) schematic of the sawtooth geometry of the device with the region observed enclosed, RBCs trapped in a g-iDEP device at 400 V with (ii) MyO protein solution and (iii) H-FABP protein solution [92].
