Microfluidics and Nanofluidics

, Volume 3, Issue 2, pp 217–225

Lab-on-a-display: a new microparticle manipulation platform using a liquid crystal display (LCD)


  • Wonjae Choi
    • Department of BioSystemsKorea Advanced Institute of Science and Technology (KAIST)
  • Se-Hwan Kim
    • Department of Information DisplayKyung Hee University
  • Jin Jang
    • Department of Information DisplayKyung Hee University
    • Department of BioSystemsKorea Advanced Institute of Science and Technology (KAIST)
Research Paper

DOI: 10.1007/s10404-006-0124-5

Cite this article as:
Choi, W., Kim, S., Jang, J. et al. Microfluid Nanofluid (2007) 3: 217. doi:10.1007/s10404-006-0124-5


This paper reports a new portable microfluidic platform, “lab-on-a-display,” that microparticles are manipulated by optoelectronic tweezers (OET) on a liquid crystal display (LCD). The OET has been constructed by assembling a ground layer, a liquid chamber, and a photoconductive layer. Without lens or optical alignments, the LCD image directly forms virtual electrodes on the photoconductive layer for dielectrophoretic manipulation. The lab-on-a-display was first realized by a conventional monochromatic LCD module and a light source brighter than 5,000 lux. It was successfully applied to the programmable manipulation of 45 μm polystyrene beads; more than 100 particles were transported with an optical image-driven control, following the moving edge of the image at every moment. The effects of bead size and bias voltage on the manipulation speed were also investigated. Due to the portability and compatibility for disposable applications, this new platform has potential for programmable particle manipulation or chip-based bioprocessing including cell separation and bead-based analysis.


MicrofluidicsLab-on-a-displayLiquid crystal display (LCD)Optoelectronic tweezersDielectrophoresisParticle manipulation

1 Introduction

The manipulation of biological cells and microparticles plays an important role in many biological applications (Verpoorte 2003; Dittrich and Manz 2006). Dielectrophoresis (DEP) is a favorable phenomenon for the manipulation of various dielectric particles (Hughes 2002; Krupke et al. 2003; Washizu et al. 1994). Generally, DEP utilizes the interaction force between non-uniform electric field and the induced dipole moment of the particle. Due to its ability to handle cells or particles without any modification of them, it has become one of the most attractive manipulation techniques in a microfluidic device or a lab-on-a-chip. In recent studies, many research groups have revealed the dielectric properties of polystyrene microbeads (Abe et al. 2004; Choi and Park 2005), DNA (Lao and Hsing 2005), bacteria (Li and Bashir 2002; Lagally et al. 2005), yeast (Perch-Nielsen et al. 2003; Doh and Cho 2005), leukocytes (Wang et al. 2000), and erythrocytes (Minerick et al. 2003). The dielectrophoretic force is described as follows:
$$ {\mathbf{F}}_{{{\text{DEP}}}} = 2\pi a^{3} \varepsilon _{{\text{m}}} \text{Re} [f_{{{\text{CM}}}} ]\,\nabla {\mathbf{E}}^{2} , $$
$$ f_{{{\text{CM}}}} = \frac{{\varepsilon ^{*}_{{\text{p}}} - \varepsilon ^{*}_{{\text{m}}} }} {{\varepsilon ^{*}_{{\text{p}}} + 2\varepsilon ^{*}_{{\text{m}}} }}, $$
where a is the diameter of microparticle; εm, permittivity of media; fCM, the Clausius–Mossoti factor; ε*, complex permittivity \( (\varepsilon * = \varepsilon - j\sigma /\omega ) \); σ, the conductivity; ε is the electric field frequency.

A programmable DEP electrode array has become one of the key issues in DEP-based manipulation techniques. Manaresi et al. (2003) reported a complementary metal-oxide-semiconductor (CMOS) circuit-driven technique. It was demonstrated that the programmable DEP manipulator with individually addressable two-dimensional electrode array could be used for the parallel manipulations of biological cells and microscopic particles. Since the activated electrode patterns are movable and reconfigurable, it has many advantages such as single-particle addressing and selection, grap-and-drag motion, parallel manipulation, robustness from clogging and channel-less structure. However, it has a potential drawback for disposable applications due to high manufacturing cost. Integration of on-chip circuits increases the cost of the device, making it less attractive for disposable applications.

To deal with this drawback, optoelectronic tweezers (OET) were proposed by replacing the patterned electrodes with a pattern-less photoconductive layer (Chiou et al. 2005). In their approach, light-induced virtual electrodes on a photoconductive layer were used to make a DEP force by using the transmitted light signal generated from a digital micro-mirror device (DMD). This technique is potentially useful for biomedical applications, because the OET part can be used as a disposable cartridge-type for sample handling. The only electrical connections for the OET part would be two wires to supply an AC bias. However, the optical structure for DMD projection leads to difficulties in portable applications. The reflective structure resulted in complex structure. Additionally, the optical lens between DMD and OET requires optical alignment to focus the projected image on the photoconductive layer.

This paper aims to demonstrate a new lab-on-a-display platform for microscopic particle manipulation using a new optical structure. Unlike the projection display method based on DMD, we utilize direct image transfer on a liquid crystal display (LCD). A lab-on-a-display has no optical component between display (LCD) and OET, thus the OET part is just placed on the display device. Due to the elimination of lens and optical alignment, it would be more suitable for portable applications. In addition, this platform is relatively thin and tolerant to vibrations in real world application.

In this study, the lab-on-a-display was first realized by a conventional monochromatic LCD module. Prior to conducting the manipulation of microparticles, the feasibility of the lab-on-a-display was evaluated by electrical field simulation. We have also investigated various OET properties which depend on the light source, the LCD type (color or monochromatic), bias voltage, and microparticle size. Detailed experimental procedures and results of the lab-on-a-display prototype are reported herein.

2 Materials and methods

2.1 Design and microfabrication of a lab-on-a-display

Figure 1 shows the schematic configuration of a lab-on-a-display, which has an OET part on the top of a LCD. In the OET, the liquid containing microparticles is sandwiched between the photoconductive layer and the ground layer. When an AC bias voltage is applied between two layers, the LCD represents an image and transmits it to the photoconductive layer. Consequently, the image forms virtual electrodes on the surface of photoconductive layer, resulting in an electric field gradient in the liquid. This electric field gradient generates dipole moments of neutral particles, which causes a DEP force for microparticle manipulation.
Fig. 1

Schematic of a lab-on-a-display. Microparticles-containing liquid was sandwiched between the middle photoconductive layer and the top ground layer. A bottom LCD makes an image and transmits it to the photoconductive layer. When an AC bias voltage is applied between the photoconductive and the ground layers, this image forms virtual electrodes on the surface of photoconductive layer, which results in the electric field gradient for DEP manipulation

The transparent and conductive ground layer was an indium tin oxide (ITO) layer. Glass substrates coated with a 180-nm-thick ITO layer were purchased from Samsung-Corning Precision Glass (Asan, Korea). After dicing into 37.5 mm × 25.0 mm size, a wrapping wire was connected for biasing as a ground layer. The photoconductive surface on a glass substrate was comprised of four layers: a 180-nm-thick ITO layer, a 50-nm-thick n+ doped hydrogenated amorphous silicon (n+ a-Si:H) layer, a 1-μm-thick intrinsic hydrogenated amorphous silicon layer (intrinsic a-Si:H, photoconductor), and a 20-nm-thick silicon nitride (SiNx) layer (Table 1). Triple layers of n+ a-Si:H, intrinsic a-Si:H, and SiNx were consecutively deposited by plasma enhanced chemical vapor deposition (PECVD) method on an ITO-coated glass substrate in a single chamber reactor. The n+ a-Si:H was deposited from a gas ratio of 1.5% PH3 in SiH4 and then intrinsic a-Si:H was deposited from a gas mixture of 20% SiH4/He = 300 sccm and H2 = 100 sccm at 280°C. The SiNx layer was deposited by a SiH4, NH3 and N2 mixture. Then, some regions were etched by reactive ion etch (RIE) to expose the ITO layer for bias connections. After dicing the photoconductive layer, a wire was connected for biasing like the ground layer.
Table 1

Thickness comparisons of each component in the lab-on-a-display


Thickness (μm)


Ground layer

Glass substrate





Liquid chamber




Photoconductive layer



Intrinsic a-Si:H


n+ a-Si:H




Glass substrate




2.2 Experimental setup

A film mask made from silver halide emulsion on a polyester film substrate (Han&All Tech Co., Ansan, Korea) was used as a static image pattern. Two types of LCD were investigated in this study. A color LCD (800 × 3 × 480 pixel array with 100 × 300-μm pixel size) was picked out from a 7-in LCD panel (TX18D11VM1CAA; Hitachi, Japan). Its dimensions were 102 mm in length and 163 mm in width. The thickness of the color LCD without backlight and with backlight was 2 and 11 mm, respectively. A 1.3-in monochromatic LCD (800 × 600 pixel array with 33-μm pixel pitch) module was taken out of a conventional projector (EMP-5300; Epson, Japan). Its dimensions were 42 mm in length, 40 mm in width, and 5 mm in thickness. The LCD module was operated by the LCD driver circuit of the projector. The image area of the LCD module was 26.4 mm × 19.6 mm, and the LCD images were drawn by using standard presentation software (Microsoft PowerPoint) on a computer.

Plain polystyrene beads (PolySciences, PA, USA) were used for particle manipulation. The sample was prepared by dilution with deionized water to the final concentration of about 2.5 × 105 particles/mL. A sample droplet of 5–7 μL was sandwiched between the ground layer and the photoconductive layer using double-stick tape as a spacer, ensuring the liquid chamber of 120 μm thick. The electric bias voltage produced from a function generator (MXG-9802A; Seowon Family Co., Korea or AGF3022; Tektronix, USA) was applied across the ground layer and the photoconductive layer of the lab-on-a-display.

The movements of beads were observed and recorded using an upright microscope (Zeiss Axioskop 40; Carl Zeiss, Germany) with a camera (Coolpix5400; Nikon, Japan). We used two illumination of the microscope: one for actuation and the other for observation (Fig. 2). The downside illumination with high intensity was used for actuation, i.e., to create the image for virtual electrodes. The upside illumination with low intensity was used for observation, because it was difficult to see particles in a dark region without this upside illumination. To investigate the effects of bias voltage and bead size on the bead velocity, we recorded the bead movements and analyzed the video images frame by frame. The bead speed was calculated from at least ten beads.
Fig. 2

Experimental setup. The image was drawn using presentation software (Microsoft PowerPoint) on a computer. The image is electronically transferred from the computer to the LCD module. The movements of beads were observed using an upright microscope with two illuminations: one for actuation and the other for observation. A voltage produced from a function generator was applied across the ground layer and the photoconductive layer

3 Results and discussion

3.1 Simulation of electric field distribution

As a proof-of-concept, we simulated an electric field distribution in the liquid layer of the lab-on-a-display. The electric field was calculated by using a commercial CFD solver (CFD − ACE+; ESI US R&D Inc., Huntsville, AL, USA) in the condition of 22 V AC bias at 100 kHz. Because the photoconductivity of hydrogenated amorphous silicon photoconductor was larger than the dark conductivity for more than three orders of magnitude (Ryu et al. 2001), we assumed the illuminated region and non-illuminated region of photoconductive layer as a conductor and an insulator, respectively.

Figure 3a is a cross-sectional view of lab-on-a-display when the LCD displays an image, which the left and right side of the LCD is bright and dark, respectively. It was noticed that the electric field gradient was relatively larger at the localized region around the image edge. Since DEP force is proportional to the gradient of the square of the electric field, the DEP force at the edges of the image is stronger than elsewhere. Therefore, particles can move faster in these regions than other regions. In negative DEP condition, microparticles would move in the direction from the illuminated side to the non-illuminated side (rightward direction in the figure). There is also a vertical difference of DEP velocity. At the same lateral position, the particles at different vertical positions would move with different velocities; the particles near the top ground layer (e.g., 90 μm above the photoconductive layer) would move slower than other particles near the photoconductive layer (e.g., 30 μm above).
Fig. 3

Distribution of the square of the electric field in the liquid layer of the lab-on-a-display. a A cross-sectional view of lab-on-a-display when the photoconductive layer is illuminated by LCD image, where the left side is bright and the right side is dark. b Schematic views representing the liquid layers at three sequential moments while the bright image extends from the left side to the right side. The electric field was calculated by using CFD − ACE+, and the DEP force is proportional to the gradient of this distribution

In order to transport the particle across a long distance, the LCD image should be changed dynamically according to a proper program sequence. If the image is static, a particle would move only for a limited distance. For a long movement the image pattern needs to be continuously changed because a trapped particle would move according to the edge of moving image. For example, a particle can move from the left end to the right end of the liquid chamber if the image edge moves from the left to the right of the LCD. Figure 3b shows the liquid layers at three sequential moments while the bright image extends from the left side to the right side.

3.2 Light-induced virtual electrode generation

The photoconductivity of hydrogenated amorphous silicon was characterized using a test device, in which two coplanar aluminum electrodes were fabricated on the hydrogenated amorphous silicon layer (Fig. 4a). Here, the distance between electrodes was 360 μm, the electrode width was 4.9 mm, and the intrinsic a-Si:H thickness was 400 nm. A white light source with uniform intensity was used for the measurement. Figure 4b shows the measured photoconductivity versus light intensity. The dark conductivity was 1.0 × 10−9 S/cm, which was increased by almost 10,000 times under illumination of light with an intensity of 20,000 lux. This conductivity difference makes the virtual electrode light-inducible because the conductivity of photoconductive layer in the illuminated regions is much larger than that in the non-illuminated regions. The virtual electrodes are able to create non-uniform electric fields to manipulate particles by DEP forces. From the result shown in Fig. 4b, we selected the light intensity of more than 5,000 lux in order to induce virtual electrodes effectively. To confirm the generation of virtual electrodes on the fabricated photoconductive layer, the light-induced DEP experiments were conducted using a static image pattern created by a film mask. The beads in liquid layer were successfully moved in this experiment (data not shown).
Fig. 4

Photoconductivity measurement of hydrogenated amorphous silicon layer. a A test device used for the measurement of photoconductivity and b photoconductivity versus light intensity

To investigate the generation of virtual electrodes using color and monochromatic LCDs, light-induced DEP experiments were also conducted in a similar way. In Fig. 5, the left figures are schematic and the right figures show the transmitted image through the glass substrate (the upper side) or the photoconductive layer (the lower side). At first, we utilized the color LCD with its backlight (Fig. 5a). There was no visible pixel through the photoconductive layer, while all pixels through the glass substrate were visible. This result indicates that the backlight is weak, which is not intense enough to make virtual electrodes. Therefore, we utilized the color LCD with an external light source, making the illumination intense enough to pass through the photoconductive layer (Fig. 5b). When white light was illuminated, each pixel of the LCD represented one of three colors (red, green, and blue) due to color filters inside the LCD. Through the photoconductive layer, however, only red pixels were visible. The reason why only the red light among three colors was visible was thought to be that the light absorption through the amorphous silicon layer depends on its wavelength. It seems that the red light with a longer wavelength was relatively less absorbed. When an AC voltage was applied, the beads around the visible red pixel were moved with the direction away from the red pixel (data not shown). Although the LCD light-induced DEP was successful, it was hard to implement the bead movement across several pixels since two thirds of LCD pixels (green and blue) did not make virtual electrodes.
Fig. 5

Comparison of virtual electrodes using various LCDs. Left figures are schematic and right figures show the transmitted image through the glass substrate (the upper side) or the photoconductive layer (the lower side). a Using color LCD with its backlight, the light intensity of LCD backlight was not enough to pass through the photoconductive layer. b Using color LCD and external light source, only red light among three colors (red, green and blue) passed through the photoconductive layer. c While white light was illuminated by monochromatic LCD and external light source, the color of illuminated photoconductive region was red

Instead of the color LCD, we thus tested a monochromatic LCD module which has no color filter (Fig. 5c). The LCD image was visible through both the glass side and the photoconductive side, and the beads were transported by the LCD light-induced DEP. When white light was illuminated, the color of bright regions through the photoconductive side was also red. Here, the monochromatic LCD gave two additional changes such as the reduction of the LCD pixel size and unwanted image pattern. Since the pixel size was shrinked from 100 μm × 300 μm to 33 μm × 33 μm, smaller virtual electrodes would enable more precise control of position. As shown in Fig. 5c, the image pattern (“I”) on the LCD module was not the same with the intended image pattern. Two pixels in every six pixels were always bright and dark, respectively, while the rest four pixels in every six pixels were controllable according to the programmed image by a computer. In addition, the dark pixels out of the image pattern were visible as a black vertical line through the photoconductive layer; however, the bright pixels in the image pattern were invisible through the photoconductive layer. The difference of this microscopic image pattern was inferred to have its origin in the LCD driver circuits, because the image without microscope looked like the original on computer screen: the black vertical lines by the dark pixels were invisible.

3.3 Microparticle manipulation using dynamic LCD image pattern

As a demonstration of a lab-on-a-display manipulation, a number of 45 μm diameter size beads were transported to finally form a letter. The red (bright) image was made on the LCD and DEP force was activated by 22 V AC bias. During 60 s, the image gradually appeared, finally making the letter “I” of 1.8 mm × 2.4 mm in size (Fig. 6). It was observed that the beads followed the moving edge of the image at every moment. Under negative DEP regime, particles in the lab-on-a-display traveled from the bright region to the dark region. These movements were occurred at the localized region near the edge of light pattern as expected in the simulation. After continuous transportation for 60 s, more than 100 beads were gathered to make an arrangement of the letter “I” as the image on LCD. After the image disappeared (at 65 s), the beads stayed in the same place and represented the alphabet letter. Some beads were hard to manipulate although the most of beads were manipulated according to the moving image edge with a speed of 7 μm/s. Possible reasons for this phenomenon are explained as follows: the vertical differences of electric field distribution and sticking of bead to the substrate.
Fig. 6

An example of LCD-driven optical manipulation of microbeads with a diameter of 45 μm. Through white backlight is illuminated, the color of bright region was red because of the photoconductive layer. During 60 s, an image gradually appears, finally making the letter “I” of 1.8 mm × 2.4 mm in size. The beads still represent the letter “I” after the image disappears (at 65 s). The video clip is available at http://nanobio.kaist.ac.kr/Papers/ESI/LCD45polystyrene.mpg (7,412 kB)

To investigate the relation of manipulation speed and the bias voltage, velocities of beads were measured against different bias voltage (Fig. 7a). The velocities only around the image edge were measured, because the beads moved fastest there. The bead velocity increases as the bias voltage increases, finally reaching the speed of 35.7 ± 7.9 μm/s when the 75 μm of bead was used in the condition of 20 V bias at 100 kHz. This result is in a good agreement with the equation (3), implying that velocity, v, is proportional to the square of the bias voltage, V. This is a similar result obtained from yeast cells (Lu et al. 2005).
$$ \Delta {\mathbf{v}} \propto {\mathbf{F}}_{{{\text{DEP}}}} \propto \nabla {\mathbf{E}}^{2} \propto V^{2} . $$
Velocity measurements of different size beads were also conducted in the same condition of 20 V bias at 100 kHz. As shown in Fig. 7b, the bead velocity increases as the bead diameter increases. This is due to the fact that DEP force increases as the bead diameter increases (Eq. 1). However, the velocity of bead over than 75-μm diameter was decreased rapidly. This size dependent phenomenon was resulted from the chamber height as well as the dielectrophoretic force. For larger beads (e.g., beads of 90 μm diameter), the chamber height of 120 μm would disturb the bead movement. When the chamber height increased into 210 μm, the velocity of 90-μm bead increased more than nine times (data not shown).
Fig. 7

a Bead velocity with respect to bias voltage. The bias frequency was 100 kHz, and the beads of 75 μm diameter were used. b Bead velocity with respect to bead diameter. The bias was 20 V at 100 kHz. The velocities only around the image edge were measured, because the beads move fastest there

Overall thickness of the lab-on-a-display was about 6.5 mm (Table 1). This dimension may be applicable to the portable applications. The thickness of the prototype is determined mostly by the thicknesses of LCD and glass substrate; the LCD was 5 mm thick and the two glass substrates are 0.7 mm each. In addition, the weight of lab-on-a-display was mainly contributed to the LCD. The LCD was 14.9 g in weight, while the other layers were about 3.4 g in weight. Therefore, a thinner LCD would lead to reduce the size and weight of lab-on-display device.

In this study, the manipulation area of the lab-on-a-display was 26.4 mm × 19.6 mm, which larger than other programmable DEP manipulators. The reported manipulation areas of CMOS technique and DMD technique were 8 mm × 8 mm and 1.3 mm × 1.0 mm, respectively. The manipulation area can be more extended than current prototype without technical difficulties, since a lot of larger LCD panels which can be implemented after removing color filters are commercially available already. Larger manipulation areas or a larger number of virtual electrodes are affordable to many potential applications, e.g., high throughput application in cell separation or drug discovery.

However, current prototype was thought to be hard for single particle manipulation because of image blurring. The image from the 33-μm-size pixel displays up to 200 μm wide, when a microscopic focus is set at the beads in the liquid chamber. It seems that the light from the LCD is scattered to all direction while it pass through glass substrate of the LCD or the OET. To conduct single particle (or cell) manipulation, this image blurring should be reduced to less than or similar with the particle (or cell) size. One of the possible methods to decrease image blurring is using a thinner LCD module. Another approach is the use of a thinner substrate for the photoconductive layer in the OET (e.g., 150-μm-thick plastic film substrate instead of 700-μm-thick glass substrate).

There are several issues to improve the performance of the lab-on-a-display for successful applications. Some issues including the image blurring, the property of the photoconductive layer, the integration of light source, and the conductivity of buffer solution should be explored to improve the device performance. The specifications of current lab-on-a-display are summarized in Table 2.
Table 2

Specifications of the demonstrated lab-on-a-display


LCD technology

High temperature poly silicon


42 mm × 40 mm × 5 mm


14.9 g

Image area

26.4 mm × 19.6 mm (1.3 in diagonal)

Pixel pitch

33 μm × 33 μm

Pixel array size

800 × 600


Die size

37.5 mm × 25.0 mm

Key layer of the photoconductive layer

1-μm-thick hydrogenated amorphous silicon

Actuation voltage

22 V @ 100 kHz

Microchamber height

120 μm


Manipulation area

26.4 mm × 19.6 mm

Maximum number of virtual electrodes


Overall thickness

6.5 mm

Overall weight

18.3 g

4 Conclusion

The lab-on-a-display platform based on a new optical structure was designed and demonstrated. Numerical simulation and experimental results indicated that the direct transfer of a LCD image could enable the OET manipulation. From the photoconductivity measurement experiment, the fabricated photoconductive layer showed good property for virtual electrode formation. A number of microparticles were transported by the monochromatic LCD and light-patterned virtual electrodes. As the bead velocity increased proportionally to the voltage squared, there was a microparticle size dependency of its speed. This new platform is anticipated to be useful for lab-on-a-chip and diagnostic instruments since it enables programmable 2-D manipulation as well as provides portability and compatibility for disposable applications.


This research was supported by the Nano/Bio Science & Technology Program (M10536090002-05N3609-00210) of the Ministry of Science and Technology (MOST), Korea. The authors also thank CHUNG Moon Soul Center for BioInformation and BioElectronics, KAIST. The microfabrication works were performed at the Digital Nanolocomotion Center (KAIST, Daejeon, Korea) and the TFT-LCD Research Center (Kyung Hee University, Seoul, Korea).

Copyright information

© Springer-Verlag 2006