3150 Mahyar Nasabi Khashayar Khoshmanesh Francisco J. Tovar-Lopez Kourosh Kalantar-zadeh Arnan Mitchell∗ School of Electrical and Computer Engineering, RMIT University, Melbourne, VIC, Australia
Received May 15, 2013 Revised September 6, 2013 Accepted September 7, 2013
Electrophoresis 2013, 34, 3150–3154
Short Communication
Dielectrophoresis with 3D microelectrodes fabricated by surface tension assisted lithography This paper demonstrates the utilization of 3D semispherical shaped microelectrodes for dielectrophoretic manipulation of yeast cells. The semispherical microelectrodes are capable of producing strong electric field gradients, and in turn dielectrophoretic forces across a large area of channel cross-section. The semispherical shape of microelectrodes avoids the formation of undesired sharp electric fields along the structure and also minimizes the disturbance of the streamlines of nearby passing fluid. The advantage of semispherical microelectrodes over the planar microelectrodes is demonstrated in a series of numerical simulations and proof-of-concept experiments aimed toward immobilization of viable yeast cells. Keywords: Cell trapping / Dielectrophoresis / 3D microelectrodes / Microfluidics / Surface tension assisted lithography DOI 10.1002/elps.201300233
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Dielectrophoresis, the induced motion of suspended polarized particles under nonuniform electric fields, is a versatile mechanism for manipulation, sorting, and immobilization of cells in microfluidic systems [1, 2]. The performance of DEP systems strongly depends on the configuration of microelectrodes. A variety of microelectrode configurations has been proposed by different groups to improve and diversify the performance of DEP systems, as discussed in a comprehensive review [1]. While the majority of DEP systems have utilized 2D microelectrodes [3–6], very few of them are equipped with 3D microelectrodes. The main advantage of 3D microelectrodes is their capacity to generate strong electric field gradients over a larger area of the channel cross-section. This improves the sorting efficiency of the DEP system, enabling the sorting of rare target cells against a heterogeneous population of suspending cells. Moreover, it enhances the trapping efficiency of the DEP system, allowing for immobilization, stimulation, and characterization of large clusters of target cells for a range of bioapplications. These advantages come at a price, since the fabrication of 3D microelectrodes is more complicated. The desire for 3D microelectrodes has led to innovative microfabrication techniques by different groups. For exam-
Correspondence: Mahyar Nasabi, School of Electrical and Computer Engineering, RMIT University, GPO Box 2476, Melbourne, VIC 3001, Australia E-mail: [email protected]
Abbreviation: STAL, surface tension assisted lithography C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
ple, Chen et al. created a semi-3D DEP array by patterning 2D microelectrodes on the top and bottom surfaces of the channel [7]. However, this leads to generation of weak electric fields along the channel centerline, which in turn limits the height of the channel. Alternatively, Iliescu et al. developed extruded silicon microelectrodes by etching a silicon wafer, and sandwiching it between two glass substrates, which formed the bottom and top surfaces of the channel [8]. Nevertheless, the etching and bonding of silicon wafer is relatively time consuming and expensive. As an alternative, Voldman et al. fabricated cylindrical gold microelectrodes by electroplating gold into a geometry defined by a layer of thick photoresist [9]. This leads to formation of sharp electric fields along the corners of the top face of the cylinder and also can disturb the intrinsic velocity profile of the flow. Alternatively, Wang et al. fabricated 3D vertical microelectrodes along the sidewalls of the channel by electroplating gold and later embedding the electrodes in SU-8, which formed the sidewalls [10]. However, this leads to generation of weak electric fields along the middle of the channel, which limits the channel width. A further study by Martinez-Duarte et al. created pillar-type carbon microelectrodes by thermal degradation of SU-8 at 900⬚C in an inert atmosphere [11]. Although this technique provides an inexpensive alternative to metal microelectrodes, it is restricted to planar structures that result in sharp electric fields ∗
Additional corresponding author: Dr. Arnan Mitchell, E-mail: [email protected]
Colour Online: See the article online to view Figs. 1–3 in colour.
www.electrophoresis-journal.com
Electrophoresis 2013, 34, 3150–3154
along the corners of their top faces. Moreover, the fluorescent nature of SU-8 can interfere with fluorescent microscopy [12]. The difficulties associated with fabrication of 3D microelectrodes have even encouraged researchers to conceive electrodeless DEP systems. Lapizco-Encinas et al. pioneered the concept of insulator-based dielectrophoresis by positioning an array of PDMS barriers along the channel and providing an electric field along the channel’s two ends to create sharp electric fields across the small gaps between the neighboring barriers [13, 14]. Despite simple fabrication, the length of the channel was limited to produce strong electric fields. Moreover, the cell samples could be contaminated at the reservoirs due to contact with metallic electrodes. To overcome the operating issues of the latter approach, Shafiee et al. introduced the concept of contactless dielectrophoresis by inserting metallic electrodes into conductive channels, which were orthogonal to the main channel and were separated from it by a thin layer of PDMS membrane [15, 16]. However, this led to formation of weak electric fields along the channel centerline, similar to the SU-8 embedded electrode [10]. We have recently reported a novel technique, called surface tension assisted lithography (STAL) for fabrication of 3D curved structures [17]. STAL provides independent control over the height and diameter of the curved structures. In this paper, we propose the fabrication of semispherical microelectrodes using STAL technique for DEP manipulation of yeast cells. The semispherical shape of microelectrodes avoids the formation of undesired sharp electric fields along the structure and also minimizes the disturbance of the streamlines of nearby passing fluid. Figure 1 illustrates the fabrication process of 3D microelectrodes using STAL technique [17]. STAL in principle is a combination of two well-established techniques: soft lithography and photo lithography. Soft lithography is used to pattern the photoresist while photo lithography is utilized to define a container by exposing the photoresist outside the perimeter enclosing the patterned area. Heat treatment is used to
Microfluidics and Miniaturization
3151
reflow the patterned and unexposed photoresist confined by the container (exposed). The shape of the final 3D structure is governed by surface tension, which enables microfabrication of smooth 3D structures. In Fig. 1A and B, array of surface recesses were imprinted on a film of negative photoresist (SU-8) using elastomeric PDMS mold. The surface recesses were cylindrical with a diameter of 35 m and a center-to-center separation of 100 m. Next, a container (reflow boundary) was defined by selectively exposing the sample to UV light through a photomask (Fig. 1C). The photomask contained circular opaque patterns with a diameter of 50 m and a center-to-center separation of 100 m, which were aligned to the imprint features to protect them from exposure. Then a thermal treatment was performed to simultaneously cross-link the exposed photoresist and reflow the unexposed regions to form concave 3D curved structures, or so called “dimples” (Fig. 1D). A final flood exposure step following the thermal treatment was necessary to solidify the dimples such that they would retain their shape. Next, a layer of PDMS with a thickness of 500 m was spun onto the fabricated sample to obtain the convex 3D structures (Fig. 1E). The PDMS film was then placed on a microscope glass slide. A thin layer of gold was then deposited on the PDMS film using electron beam evaporation (Fig. 1F). Next, the microelectrodes were patterned on the gold film using conventional gold etching techniques (Fig. 1G). The final structure, as presented in SEM image (Fig. 1H), composed of an array of convex microelectrodes with a diameter of 50 m, an average maximum height of 30 m, and a separation gap of 50 m between the opposite poles. The spacing of 50 m was chosen to avoid merging of the electric field between the opposite microelectrodes, while producing a strong nonuniform electric field, and consequently DEP force within the microfluidic channel. Moreover, it provides enough space for immobilization of cell clusters between the microelectrodes under positive DEP force, as will be shown in the results.
Figure 1. Schematics of the fabrication method of 3D microelectrodes using STAL technique. (A and B) Patterning SU-8 film using soft lithography technique, (C) defining a container for 3D structure using photo lithography, (D) reflowing of unexposed region to form curved 3D structures (dimple) within the exposed area container, (E) obtaining the bulged 3D structures on a PDMS film by casting, (F) deposition of a thin gold layer on the PDMS by electron beam evaporation, (G) patterning of gold microelectrodes by etching, and (H) SEM image of fabricated 3D microelectrode.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
3152
M. Nasabi et al.
Electrophoresis 2013, 34, 3150–3154
Figure 2. Comparing the DEP performance of 2D and 3D microelectrodes using numerical simulations: (A and B) electric field contours for 2D and 3D configurations, (C and D) variations of electric field along different heights of the channel for 2D and 3D configurations, (E and F) contours of ∇ E 2 at the bottom surface of the channel.
The slight misalignment observed at the end of the microelectrodes (Fig. 1H) could be due to the nonplanarity of the PDMS at alignment stage. Finally, a PDMS microfluidic channel with dimensions of 600 × 100 m (width × height) was aligned and placed on top of the microelectrodes to create the DEP platform. If it is assumed that yeast cells are spherical structures, the time-averaged DEP force applied on them is calculated as below: 2 , F¯DEP = 2r 3 medium Re[ f CM ]∇ E rms
(1)
where r is the radius of cells, medium is the permittivity of the medium, Erms is the root-mean-square of the applied electric field, and fCM is the Clausius–Mossotti factor of the cells, describing their relative polarization with respect to the surrounding medium, as detailed in the Supporting Information S1. For quantitative comparison of 2D and 3D microelectrodes, we calculated the contours of electric field and electric filed gradient using the ANSYS Fluent 6.3 software package [18]. In doing so, the Laplace equation (∇ 2 rms = 0) was solved within the channel by applying appropriate electric potentials at the microelectrodes while enforcing a boundary condition of zero electric flux at other surfaces of the channel, C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
including the bottom, top, and sidewalls. Next, the electric field was calculated by differentiating the electric potential (E = −∇rms ) Finally, the DEP forces were obtained by calculating the gradient of electric field square (FDEP ) ∝ ∇ E 2 , as given in Eq. (1). Figure 2A and B illustrates the contours of electric field produced by the 2D and 3D microelectrodes, respectively, at the cross-section of the channel that intersects with the middle plane of microelectrodes. To highlight the difference between the two designs, the minimum threshold of the contours is set to 25 kV/m while the maximum threshold is set to 400 kV/m. The 2D microelectrodes produce a strong electric field at the bottom of the channel with a maximum of 400 kV/m along their edges. However, the electric field sharply reduces along the height of the channel and reaches to 25 kV/m at an approximate height of 50 m (Fig. 2A). Similar to the 2D electrode, the 3D microelectrodes produce a maximum electric field of 350 kV/m, however, unlike the 2D microelectrode, this strong electric field is distributed across the sides of the hemisphere that confront the opposite microelectrode. This electric field reduces to 25 kV/m at an approximate height of 75 m (Fig. 2B), which is significantly deeper into the channel than for the 2D microelectrode. For further comparison, Fig. 2C and D shows the variations of electric field across the microchannel at the www.electrophoresis-journal.com
Electrophoresis 2013, 34, 3150–3154
heights of 0, 20, 30, and 40 m, produced by the 2D and 3D microelectrodes, respectively. Figure 2C clearly shows the sharp reduction of electric field along the height, as it reduces from a peak value of 400 kV/m at z = 0 m to 100 kV/m at z = 20 m, 55 kV/m at z = 30 m, and 40 kV/m at z = 40 m. In comparison, Fig. 2D clearly indicates the smooth reduction of electric field along the height. There is no significant difference between the peak electric field values, 350 kV/m, at z = 0, 20, and 30 m that is the height of the 3D microelectrodes, whereas it reduces to 100 kV/m at z = 40 m. Finally, Fig. 2E and F illustrates the contours of ∇E 2 that generates the DEP force, as given in Eq. (1), produced by the 2D and 3D microelectrodes, respectively. To highlight the difference, the minimum threshold of the contours is set to 5 × 1014 V2 /m3 while the maximum threshold is set to 8.1 × 1015 V2 /m3 . Under the appropriate combination of medium flow rate and applied AC signals, the area covered by these contours determines the effective trapping area of microelectrodes. Comparing Fig. 2E and F clearly indicates the advantage of 3D microelectrodes as their peak ∇E 2 is 2.1 times bigger than that of 2D microelectrodes. More importantly, the area of the contour is 1.85 times larger than that of 2D design (subtracting the surface area of microelectrodes this ratio reduces to 1.77). Generation of strong fields by microelectrodes heats the surrounding liquid due to Joule heating effect [19]. Numerical simulations reveal maximum temperatures of 28⬚C and 31.8⬚C in 2D and 3D configurations, respectively, obtained at a voltage of 10 V and a medium conductivity of 0.0125 S/m (see Supporting Information S2). This indicates that 3D microelectrodes cause more heat than 2D ones of equal footprint. This temperature rise leads to formation of electrothermal-induced vortices around the microelectrodes, which can be simulated by addition of Coulomb force and dielectric force source terms to the Navier–Stokes equation, as given in Supporting Information S2 [19]. Numerical simulations reveal that using an AC signal of 10 V and 12 MHz, the induced vortices produce a maximum velocity of 7.4 and 28 m/s along the edges of 2D and 3D microelectrodes, respectively. The stronger electrothermal vortices caused by 3D microelectrodes could be beneficial for rapid immobilization and drug stimulation of cells in open-top configurations. In such configurations, the vortices act as conveyer belts and bring the cells to the vicinity of microelectrodes, where they can be trapped [20]. To experimentally assess the advantage of 3D design for immobilization of cells, we compared the cell trapping capability of DEP platforms equipped with 2D and 3D microelectrodes. The cell suspension was prepared by adding 40 mg of Saccharomyces cerevisiae yeast powder (Sigma) to 100 mL of low electrical conductivity buffer (8.5% w/v sucrose and 0.3% w/v dextrose). The electrical conductivity of the cell buffer was set to 0.125 S/m by adding PBS to the suspension. The flow rate was set to 2 L/min, which corresponds to an average velocity of 0.55 mm/s within the used channel. The lateral flow velocity caused by electrothermal-induced C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Microfluidics and Miniaturization
3153
Figure 3. Comparing the trapping of viable yeast cells obtained by 3D and 2D microelectrodes. The flow rate and electrical conductivity of the medium are set to 2 L/min and 0.125 S/m, respectively, while the frequency of the AC signal is set to 12 MHz. Images are taken 10 min after the application of cells. Trapped cells can be observed as bright clusters between the microelectrodes, especially at the tip region. (A and B) At 5 V, the area of bright region, which represents trapped cells, is 25 260 ± 2400 m2 for 2D microelectrodes while it is 30 670 ± 3100 m2 for 3D microelectrodes. (C and D) At 10 V, the area of bright region is 70 350 ± 7400 m2 for 2D microelectrodes while it is 13 4400 ± 2980 m2 for 3D microelectrodes.
vortices (see Supporting Information S2) is at least 20 times smaller than the above average flow velocity and therefore cannot affect the motion of suspending cells. The amplitude of the AC signal was set to 5 and 10 V while the frequency was set to 12 MHz. This frequency was chosen to maximize the positive DEP response of yeast cells (Re[fCM ] = 0.064, as given in Supporting Information S1) while minimizing the risk of electrolysis and electrothermal vortices at the surface of microelectrodes [20]. In the first set of experiments, the amplitude of AC signal was set to 5 V, as shown in Fig. 3A and B. The weak DEP forces produced at this voltage limited the trapping of cells to the circular or spherical tips of microelectrodes for both 2D and 3D designs. However, the 3D design demonstrated a slightly better trapping performance, as evidenced by bridging of cells between the neighboring opposite microelectrodes, as shown with yellow arrows. Under this condition, both 2D and 3D designs become saturated after 12 min. This procedure was repeated three times and the data presented as mean ± standard error. The area covered by trapped cells reached 25 260 ± 2400 and 30 670 ± 3100 m2 for the 2D and 3D designs, respectively. Assuming a single layer of cells, this corresponds to 560 ± 55 and 680 ± 70 cells, respectively. Although the positive DEP response and consequently the trapping performance of the system can be intensified by reducing the medium’s electrical conductivity (see Supporting Information S1), the value of medium conductivity was set to 0.125 S/m to (i) assess www.electrophoresis-journal.com
3154
M. Nasabi et al.
the trapping performance of 2D and 3D microelectrodes in extreme conditions and (ii) avoid the formation of multilayer cell clusters around the microelectrodes. In the second set of experiments, the amplitude of AC signal was set to 10 V, as shown in Fig. 3C and D. The magnitude of DEP forces was four times stronger than the previous case, which expanded the call trapping area of both 2D and 3D designs. Under this condition, both microelectrodes become saturated after 8–10 min. The area covered by the cells after 10 min reached 70 350 ± 7400 and 134 400 ± 2980 m2 for 2D and 3D designs, respectively. Again by assuming a single layer of cells, this corresponds to 1560 ± 160 and 2980 ± 325 cells, respectively. The 91% increase in cell trapping for the case of 3D design is consistent with the 113% increase of DEP forces, as shown in Fig. 2E and F. Such an improvement in trapping efficiency can play a vital role when dealing with rare circulating tumor cells or low-concentration samples from clinical patients. Moreover, it enables the operation of DEP system at lower voltages while higher medium conductivities are set in order to minimize the damage imposed on cells. In summary, we demonstrated the DEP trapping of viable yeast cells using semispherical microelectrodes realized by STAL technique. Numerical simulations indicate the smooth decrease of electric field along the height of the channel for semispherical microelectrodes. More importantly, the simulations predict a 113% increase in the magnitude of the DEP force compared to that of planar microelectrodes. This is in line with our experiments, showing a 91% increase in the number of trapped cells compared to that of planar microelectrodes. STAL technique can be utilized to engineer 3D microelectrodes to achieve even greater efficiencies. It was shown in [3,4] that it can be of great utility to grade the strength of DEP electrodes in order to control the cell trapping behavior. One of the unique features of the STAL process, which is used to realize the 3D electrodes, is the ability to independently control the height and the perimeter of the 3D structures within the same fabrication step [17]. Thus, it would be possible to grade the electrodes continuously from being flat 2D electrodes to fully 3D electrodes with an arbitrary number of intermediate forms. Optimization of the form of the 3D electrodes and exploitation of the flexibility of grading the 3D geometry is proposed for future work. K. Khoshmanesh acknowledges the Australian Research Council for funding under Discovery Early Career Researcher Award (DECRA) scheme, (project DE120101402). The authors have declared no conflict of interest.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2013, 34, 3150–3154
References [1] Khoshmanesh, K., Nahavandi, S., Baratchi, S., Mitchell, A., Kalantar-zadeh, K., Biosens. Bioelectron. 2011, 26, 1800–1814. [2] Pethig, R., Biomicrofluidics 2010, 4, 022811. [3] Khoshmanesh, K., Baratchi, S., Tovar-Lopez, F. J., Nahavandi, S., Wlodkowic, D., Mitchell, A., Kalantar-zadeh, K., Microfluid. Nanofluid. 2011, 12, 597–606. [4] Tang, S.-Y., Zhang, W., Yi, P., Baratchi, S., Kalantarzadeh, K., Khoshmanesh, K., Electrophoresis 2013, 34, 1407–1414. [5] Khoshmanesh, K., Kiss, N., Nahavandi, S., Evans, C. W., Cooper, J. M., Williams, D. E., Wlodkowic, D., Electrophoresis 2011, 32, 3129–3132. [6] Velugotla, S., Pells, S., Mjoseng, H. K., Duffy, C. R. E., Smith, S., De Sousa, P., Pethig, R., Biomicrofluidics 2012, 6, 44113. [7] Chen, D., Du, H., Microfluid. Nanofluid. 2007, 3, 603–610. [8] Iliescu, C., Yu, L., Tay, F. E. H., Chen, B., Sensor. Actuator. B: Chem. 2008, 129, 491–496. [9] Voldman, J., Gray, M., Toner, M., Schmidt, M., Anal. Chem. 2002, 74, 3984–3990. [10] Wang, L., Flanagan, L. a., Monuki, E., Jeon, N. L., Lee, A. P., Lab Chip 2007, 7, 1114–1120. [11] Martinez-Duarte, R., Gorkin, R. a., Abi-Samra, K., Madou, M. J., Lab Chip 2010, 10, 1030–1043. [12] Pai, J.-H., Wang, Y., Salazar, G. T., Sims, C. E., Bachman, M., Li, G. P., Allbritton, N. L., Anal. Chem. 2007, 79, 8774–8780. [13] Lapizco-Encinas, B. H., Simmons, B. A., Cummings, E. B., Fintschenko, Y., Anal. Chem. 2004, 76, 1571–1579. [14] Lapizco-Encinas, B. H., Simmons, B. A., Cummings, E. B., Fintschenko, Y., Electrophoresis 2004, 25, 1695–1704. [15] Shafiee, H., Sano, M. B., Henslee, E. a., Caldwell, J. L., Davalos, R. V., Lab chip 2010, 10, 438–445. [16] Salmanzadeh, A., Romero, L., Shafiee, H., GalloVillanueva, R. C., Stremler, M. a, Cramer, S. D., Davalos, R. V., Lab chip 2012, 12, 182–189. [17] Nasabi, M., Tovar-Lopez, F. J., Khoshmanesh, K., Kostovski, G., Mitchell, A., J. Mater. Chem. C 2013, 1, 401–405. [18] Khoshmanesh, K., Zhang, C., Nahavandi, S., TovarLopez, F. J., Baratchi, S., Hu, Z., Mitchell, A., Kalantarzadeh, K., Electrophoresis 2010, 31, 1366–1375. [19] Morgan, H., Green, N. G., AC Electrokinetics: Colloids and Nanoparticles, Research Studies Press, Philadelphia, PA 2003. [20] Khoshmanesh, K., Akagi, J., Nahavandi, S., Skommer, J., Baratchi, S., Cooper, J. M., Kalantar-zadeh, K., Williams, D. E., Wlodkowic, D., Anal. Chem. 2011, 83, 2133–2144.
www.electrophoresis-journal.com
Received May 15, 2013 Revised September 6, 2013 Accepted September 7, 2013
Electrophoresis 2013, 34, 3150–3154
Short Communication
Dielectrophoresis with 3D microelectrodes fabricated by surface tension assisted lithography This paper demonstrates the utilization of 3D semispherical shaped microelectrodes for dielectrophoretic manipulation of yeast cells. The semispherical microelectrodes are capable of producing strong electric field gradients, and in turn dielectrophoretic forces across a large area of channel cross-section. The semispherical shape of microelectrodes avoids the formation of undesired sharp electric fields along the structure and also minimizes the disturbance of the streamlines of nearby passing fluid. The advantage of semispherical microelectrodes over the planar microelectrodes is demonstrated in a series of numerical simulations and proof-of-concept experiments aimed toward immobilization of viable yeast cells. Keywords: Cell trapping / Dielectrophoresis / 3D microelectrodes / Microfluidics / Surface tension assisted lithography DOI 10.1002/elps.201300233
Additional supporting information may be found in the online version of this article at the publisher’s web-site
Dielectrophoresis, the induced motion of suspended polarized particles under nonuniform electric fields, is a versatile mechanism for manipulation, sorting, and immobilization of cells in microfluidic systems [1, 2]. The performance of DEP systems strongly depends on the configuration of microelectrodes. A variety of microelectrode configurations has been proposed by different groups to improve and diversify the performance of DEP systems, as discussed in a comprehensive review [1]. While the majority of DEP systems have utilized 2D microelectrodes [3–6], very few of them are equipped with 3D microelectrodes. The main advantage of 3D microelectrodes is their capacity to generate strong electric field gradients over a larger area of the channel cross-section. This improves the sorting efficiency of the DEP system, enabling the sorting of rare target cells against a heterogeneous population of suspending cells. Moreover, it enhances the trapping efficiency of the DEP system, allowing for immobilization, stimulation, and characterization of large clusters of target cells for a range of bioapplications. These advantages come at a price, since the fabrication of 3D microelectrodes is more complicated. The desire for 3D microelectrodes has led to innovative microfabrication techniques by different groups. For exam-
Correspondence: Mahyar Nasabi, School of Electrical and Computer Engineering, RMIT University, GPO Box 2476, Melbourne, VIC 3001, Australia E-mail: [email protected]
Abbreviation: STAL, surface tension assisted lithography C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
ple, Chen et al. created a semi-3D DEP array by patterning 2D microelectrodes on the top and bottom surfaces of the channel [7]. However, this leads to generation of weak electric fields along the channel centerline, which in turn limits the height of the channel. Alternatively, Iliescu et al. developed extruded silicon microelectrodes by etching a silicon wafer, and sandwiching it between two glass substrates, which formed the bottom and top surfaces of the channel [8]. Nevertheless, the etching and bonding of silicon wafer is relatively time consuming and expensive. As an alternative, Voldman et al. fabricated cylindrical gold microelectrodes by electroplating gold into a geometry defined by a layer of thick photoresist [9]. This leads to formation of sharp electric fields along the corners of the top face of the cylinder and also can disturb the intrinsic velocity profile of the flow. Alternatively, Wang et al. fabricated 3D vertical microelectrodes along the sidewalls of the channel by electroplating gold and later embedding the electrodes in SU-8, which formed the sidewalls [10]. However, this leads to generation of weak electric fields along the middle of the channel, which limits the channel width. A further study by Martinez-Duarte et al. created pillar-type carbon microelectrodes by thermal degradation of SU-8 at 900⬚C in an inert atmosphere [11]. Although this technique provides an inexpensive alternative to metal microelectrodes, it is restricted to planar structures that result in sharp electric fields ∗
Additional corresponding author: Dr. Arnan Mitchell, E-mail: [email protected]
Colour Online: See the article online to view Figs. 1–3 in colour.
www.electrophoresis-journal.com
Electrophoresis 2013, 34, 3150–3154
along the corners of their top faces. Moreover, the fluorescent nature of SU-8 can interfere with fluorescent microscopy [12]. The difficulties associated with fabrication of 3D microelectrodes have even encouraged researchers to conceive electrodeless DEP systems. Lapizco-Encinas et al. pioneered the concept of insulator-based dielectrophoresis by positioning an array of PDMS barriers along the channel and providing an electric field along the channel’s two ends to create sharp electric fields across the small gaps between the neighboring barriers [13, 14]. Despite simple fabrication, the length of the channel was limited to produce strong electric fields. Moreover, the cell samples could be contaminated at the reservoirs due to contact with metallic electrodes. To overcome the operating issues of the latter approach, Shafiee et al. introduced the concept of contactless dielectrophoresis by inserting metallic electrodes into conductive channels, which were orthogonal to the main channel and were separated from it by a thin layer of PDMS membrane [15, 16]. However, this led to formation of weak electric fields along the channel centerline, similar to the SU-8 embedded electrode [10]. We have recently reported a novel technique, called surface tension assisted lithography (STAL) for fabrication of 3D curved structures [17]. STAL provides independent control over the height and diameter of the curved structures. In this paper, we propose the fabrication of semispherical microelectrodes using STAL technique for DEP manipulation of yeast cells. The semispherical shape of microelectrodes avoids the formation of undesired sharp electric fields along the structure and also minimizes the disturbance of the streamlines of nearby passing fluid. Figure 1 illustrates the fabrication process of 3D microelectrodes using STAL technique [17]. STAL in principle is a combination of two well-established techniques: soft lithography and photo lithography. Soft lithography is used to pattern the photoresist while photo lithography is utilized to define a container by exposing the photoresist outside the perimeter enclosing the patterned area. Heat treatment is used to
Microfluidics and Miniaturization
3151
reflow the patterned and unexposed photoresist confined by the container (exposed). The shape of the final 3D structure is governed by surface tension, which enables microfabrication of smooth 3D structures. In Fig. 1A and B, array of surface recesses were imprinted on a film of negative photoresist (SU-8) using elastomeric PDMS mold. The surface recesses were cylindrical with a diameter of 35 m and a center-to-center separation of 100 m. Next, a container (reflow boundary) was defined by selectively exposing the sample to UV light through a photomask (Fig. 1C). The photomask contained circular opaque patterns with a diameter of 50 m and a center-to-center separation of 100 m, which were aligned to the imprint features to protect them from exposure. Then a thermal treatment was performed to simultaneously cross-link the exposed photoresist and reflow the unexposed regions to form concave 3D curved structures, or so called “dimples” (Fig. 1D). A final flood exposure step following the thermal treatment was necessary to solidify the dimples such that they would retain their shape. Next, a layer of PDMS with a thickness of 500 m was spun onto the fabricated sample to obtain the convex 3D structures (Fig. 1E). The PDMS film was then placed on a microscope glass slide. A thin layer of gold was then deposited on the PDMS film using electron beam evaporation (Fig. 1F). Next, the microelectrodes were patterned on the gold film using conventional gold etching techniques (Fig. 1G). The final structure, as presented in SEM image (Fig. 1H), composed of an array of convex microelectrodes with a diameter of 50 m, an average maximum height of 30 m, and a separation gap of 50 m between the opposite poles. The spacing of 50 m was chosen to avoid merging of the electric field between the opposite microelectrodes, while producing a strong nonuniform electric field, and consequently DEP force within the microfluidic channel. Moreover, it provides enough space for immobilization of cell clusters between the microelectrodes under positive DEP force, as will be shown in the results.
Figure 1. Schematics of the fabrication method of 3D microelectrodes using STAL technique. (A and B) Patterning SU-8 film using soft lithography technique, (C) defining a container for 3D structure using photo lithography, (D) reflowing of unexposed region to form curved 3D structures (dimple) within the exposed area container, (E) obtaining the bulged 3D structures on a PDMS film by casting, (F) deposition of a thin gold layer on the PDMS by electron beam evaporation, (G) patterning of gold microelectrodes by etching, and (H) SEM image of fabricated 3D microelectrode.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.electrophoresis-journal.com
3152
M. Nasabi et al.
Electrophoresis 2013, 34, 3150–3154
Figure 2. Comparing the DEP performance of 2D and 3D microelectrodes using numerical simulations: (A and B) electric field contours for 2D and 3D configurations, (C and D) variations of electric field along different heights of the channel for 2D and 3D configurations, (E and F) contours of ∇ E 2 at the bottom surface of the channel.
The slight misalignment observed at the end of the microelectrodes (Fig. 1H) could be due to the nonplanarity of the PDMS at alignment stage. Finally, a PDMS microfluidic channel with dimensions of 600 × 100 m (width × height) was aligned and placed on top of the microelectrodes to create the DEP platform. If it is assumed that yeast cells are spherical structures, the time-averaged DEP force applied on them is calculated as below: 2 , F¯DEP = 2r 3 medium Re[ f CM ]∇ E rms
(1)
where r is the radius of cells, medium is the permittivity of the medium, Erms is the root-mean-square of the applied electric field, and fCM is the Clausius–Mossotti factor of the cells, describing their relative polarization with respect to the surrounding medium, as detailed in the Supporting Information S1. For quantitative comparison of 2D and 3D microelectrodes, we calculated the contours of electric field and electric filed gradient using the ANSYS Fluent 6.3 software package [18]. In doing so, the Laplace equation (∇ 2 rms = 0) was solved within the channel by applying appropriate electric potentials at the microelectrodes while enforcing a boundary condition of zero electric flux at other surfaces of the channel, C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
including the bottom, top, and sidewalls. Next, the electric field was calculated by differentiating the electric potential (E = −∇rms ) Finally, the DEP forces were obtained by calculating the gradient of electric field square (FDEP ) ∝ ∇ E 2 , as given in Eq. (1). Figure 2A and B illustrates the contours of electric field produced by the 2D and 3D microelectrodes, respectively, at the cross-section of the channel that intersects with the middle plane of microelectrodes. To highlight the difference between the two designs, the minimum threshold of the contours is set to 25 kV/m while the maximum threshold is set to 400 kV/m. The 2D microelectrodes produce a strong electric field at the bottom of the channel with a maximum of 400 kV/m along their edges. However, the electric field sharply reduces along the height of the channel and reaches to 25 kV/m at an approximate height of 50 m (Fig. 2A). Similar to the 2D electrode, the 3D microelectrodes produce a maximum electric field of 350 kV/m, however, unlike the 2D microelectrode, this strong electric field is distributed across the sides of the hemisphere that confront the opposite microelectrode. This electric field reduces to 25 kV/m at an approximate height of 75 m (Fig. 2B), which is significantly deeper into the channel than for the 2D microelectrode. For further comparison, Fig. 2C and D shows the variations of electric field across the microchannel at the www.electrophoresis-journal.com
Electrophoresis 2013, 34, 3150–3154
heights of 0, 20, 30, and 40 m, produced by the 2D and 3D microelectrodes, respectively. Figure 2C clearly shows the sharp reduction of electric field along the height, as it reduces from a peak value of 400 kV/m at z = 0 m to 100 kV/m at z = 20 m, 55 kV/m at z = 30 m, and 40 kV/m at z = 40 m. In comparison, Fig. 2D clearly indicates the smooth reduction of electric field along the height. There is no significant difference between the peak electric field values, 350 kV/m, at z = 0, 20, and 30 m that is the height of the 3D microelectrodes, whereas it reduces to 100 kV/m at z = 40 m. Finally, Fig. 2E and F illustrates the contours of ∇E 2 that generates the DEP force, as given in Eq. (1), produced by the 2D and 3D microelectrodes, respectively. To highlight the difference, the minimum threshold of the contours is set to 5 × 1014 V2 /m3 while the maximum threshold is set to 8.1 × 1015 V2 /m3 . Under the appropriate combination of medium flow rate and applied AC signals, the area covered by these contours determines the effective trapping area of microelectrodes. Comparing Fig. 2E and F clearly indicates the advantage of 3D microelectrodes as their peak ∇E 2 is 2.1 times bigger than that of 2D microelectrodes. More importantly, the area of the contour is 1.85 times larger than that of 2D design (subtracting the surface area of microelectrodes this ratio reduces to 1.77). Generation of strong fields by microelectrodes heats the surrounding liquid due to Joule heating effect [19]. Numerical simulations reveal maximum temperatures of 28⬚C and 31.8⬚C in 2D and 3D configurations, respectively, obtained at a voltage of 10 V and a medium conductivity of 0.0125 S/m (see Supporting Information S2). This indicates that 3D microelectrodes cause more heat than 2D ones of equal footprint. This temperature rise leads to formation of electrothermal-induced vortices around the microelectrodes, which can be simulated by addition of Coulomb force and dielectric force source terms to the Navier–Stokes equation, as given in Supporting Information S2 [19]. Numerical simulations reveal that using an AC signal of 10 V and 12 MHz, the induced vortices produce a maximum velocity of 7.4 and 28 m/s along the edges of 2D and 3D microelectrodes, respectively. The stronger electrothermal vortices caused by 3D microelectrodes could be beneficial for rapid immobilization and drug stimulation of cells in open-top configurations. In such configurations, the vortices act as conveyer belts and bring the cells to the vicinity of microelectrodes, where they can be trapped [20]. To experimentally assess the advantage of 3D design for immobilization of cells, we compared the cell trapping capability of DEP platforms equipped with 2D and 3D microelectrodes. The cell suspension was prepared by adding 40 mg of Saccharomyces cerevisiae yeast powder (Sigma) to 100 mL of low electrical conductivity buffer (8.5% w/v sucrose and 0.3% w/v dextrose). The electrical conductivity of the cell buffer was set to 0.125 S/m by adding PBS to the suspension. The flow rate was set to 2 L/min, which corresponds to an average velocity of 0.55 mm/s within the used channel. The lateral flow velocity caused by electrothermal-induced C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Microfluidics and Miniaturization
3153
Figure 3. Comparing the trapping of viable yeast cells obtained by 3D and 2D microelectrodes. The flow rate and electrical conductivity of the medium are set to 2 L/min and 0.125 S/m, respectively, while the frequency of the AC signal is set to 12 MHz. Images are taken 10 min after the application of cells. Trapped cells can be observed as bright clusters between the microelectrodes, especially at the tip region. (A and B) At 5 V, the area of bright region, which represents trapped cells, is 25 260 ± 2400 m2 for 2D microelectrodes while it is 30 670 ± 3100 m2 for 3D microelectrodes. (C and D) At 10 V, the area of bright region is 70 350 ± 7400 m2 for 2D microelectrodes while it is 13 4400 ± 2980 m2 for 3D microelectrodes.
vortices (see Supporting Information S2) is at least 20 times smaller than the above average flow velocity and therefore cannot affect the motion of suspending cells. The amplitude of the AC signal was set to 5 and 10 V while the frequency was set to 12 MHz. This frequency was chosen to maximize the positive DEP response of yeast cells (Re[fCM ] = 0.064, as given in Supporting Information S1) while minimizing the risk of electrolysis and electrothermal vortices at the surface of microelectrodes [20]. In the first set of experiments, the amplitude of AC signal was set to 5 V, as shown in Fig. 3A and B. The weak DEP forces produced at this voltage limited the trapping of cells to the circular or spherical tips of microelectrodes for both 2D and 3D designs. However, the 3D design demonstrated a slightly better trapping performance, as evidenced by bridging of cells between the neighboring opposite microelectrodes, as shown with yellow arrows. Under this condition, both 2D and 3D designs become saturated after 12 min. This procedure was repeated three times and the data presented as mean ± standard error. The area covered by trapped cells reached 25 260 ± 2400 and 30 670 ± 3100 m2 for the 2D and 3D designs, respectively. Assuming a single layer of cells, this corresponds to 560 ± 55 and 680 ± 70 cells, respectively. Although the positive DEP response and consequently the trapping performance of the system can be intensified by reducing the medium’s electrical conductivity (see Supporting Information S1), the value of medium conductivity was set to 0.125 S/m to (i) assess www.electrophoresis-journal.com
3154
M. Nasabi et al.
the trapping performance of 2D and 3D microelectrodes in extreme conditions and (ii) avoid the formation of multilayer cell clusters around the microelectrodes. In the second set of experiments, the amplitude of AC signal was set to 10 V, as shown in Fig. 3C and D. The magnitude of DEP forces was four times stronger than the previous case, which expanded the call trapping area of both 2D and 3D designs. Under this condition, both microelectrodes become saturated after 8–10 min. The area covered by the cells after 10 min reached 70 350 ± 7400 and 134 400 ± 2980 m2 for 2D and 3D designs, respectively. Again by assuming a single layer of cells, this corresponds to 1560 ± 160 and 2980 ± 325 cells, respectively. The 91% increase in cell trapping for the case of 3D design is consistent with the 113% increase of DEP forces, as shown in Fig. 2E and F. Such an improvement in trapping efficiency can play a vital role when dealing with rare circulating tumor cells or low-concentration samples from clinical patients. Moreover, it enables the operation of DEP system at lower voltages while higher medium conductivities are set in order to minimize the damage imposed on cells. In summary, we demonstrated the DEP trapping of viable yeast cells using semispherical microelectrodes realized by STAL technique. Numerical simulations indicate the smooth decrease of electric field along the height of the channel for semispherical microelectrodes. More importantly, the simulations predict a 113% increase in the magnitude of the DEP force compared to that of planar microelectrodes. This is in line with our experiments, showing a 91% increase in the number of trapped cells compared to that of planar microelectrodes. STAL technique can be utilized to engineer 3D microelectrodes to achieve even greater efficiencies. It was shown in [3,4] that it can be of great utility to grade the strength of DEP electrodes in order to control the cell trapping behavior. One of the unique features of the STAL process, which is used to realize the 3D electrodes, is the ability to independently control the height and the perimeter of the 3D structures within the same fabrication step [17]. Thus, it would be possible to grade the electrodes continuously from being flat 2D electrodes to fully 3D electrodes with an arbitrary number of intermediate forms. Optimization of the form of the 3D electrodes and exploitation of the flexibility of grading the 3D geometry is proposed for future work. K. Khoshmanesh acknowledges the Australian Research Council for funding under Discovery Early Career Researcher Award (DECRA) scheme, (project DE120101402). The authors have declared no conflict of interest.
C 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Electrophoresis 2013, 34, 3150–3154
References [1] Khoshmanesh, K., Nahavandi, S., Baratchi, S., Mitchell, A., Kalantar-zadeh, K., Biosens. Bioelectron. 2011, 26, 1800–1814. [2] Pethig, R., Biomicrofluidics 2010, 4, 022811. [3] Khoshmanesh, K., Baratchi, S., Tovar-Lopez, F. J., Nahavandi, S., Wlodkowic, D., Mitchell, A., Kalantar-zadeh, K., Microfluid. Nanofluid. 2011, 12, 597–606. [4] Tang, S.-Y., Zhang, W., Yi, P., Baratchi, S., Kalantarzadeh, K., Khoshmanesh, K., Electrophoresis 2013, 34, 1407–1414. [5] Khoshmanesh, K., Kiss, N., Nahavandi, S., Evans, C. W., Cooper, J. M., Williams, D. E., Wlodkowic, D., Electrophoresis 2011, 32, 3129–3132. [6] Velugotla, S., Pells, S., Mjoseng, H. K., Duffy, C. R. E., Smith, S., De Sousa, P., Pethig, R., Biomicrofluidics 2012, 6, 44113. [7] Chen, D., Du, H., Microfluid. Nanofluid. 2007, 3, 603–610. [8] Iliescu, C., Yu, L., Tay, F. E. H., Chen, B., Sensor. Actuator. B: Chem. 2008, 129, 491–496. [9] Voldman, J., Gray, M., Toner, M., Schmidt, M., Anal. Chem. 2002, 74, 3984–3990. [10] Wang, L., Flanagan, L. a., Monuki, E., Jeon, N. L., Lee, A. P., Lab Chip 2007, 7, 1114–1120. [11] Martinez-Duarte, R., Gorkin, R. a., Abi-Samra, K., Madou, M. J., Lab Chip 2010, 10, 1030–1043. [12] Pai, J.-H., Wang, Y., Salazar, G. T., Sims, C. E., Bachman, M., Li, G. P., Allbritton, N. L., Anal. Chem. 2007, 79, 8774–8780. [13] Lapizco-Encinas, B. H., Simmons, B. A., Cummings, E. B., Fintschenko, Y., Anal. Chem. 2004, 76, 1571–1579. [14] Lapizco-Encinas, B. H., Simmons, B. A., Cummings, E. B., Fintschenko, Y., Electrophoresis 2004, 25, 1695–1704. [15] Shafiee, H., Sano, M. B., Henslee, E. a., Caldwell, J. L., Davalos, R. V., Lab chip 2010, 10, 438–445. [16] Salmanzadeh, A., Romero, L., Shafiee, H., GalloVillanueva, R. C., Stremler, M. a, Cramer, S. D., Davalos, R. V., Lab chip 2012, 12, 182–189. [17] Nasabi, M., Tovar-Lopez, F. J., Khoshmanesh, K., Kostovski, G., Mitchell, A., J. Mater. Chem. C 2013, 1, 401–405. [18] Khoshmanesh, K., Zhang, C., Nahavandi, S., TovarLopez, F. J., Baratchi, S., Hu, Z., Mitchell, A., Kalantarzadeh, K., Electrophoresis 2010, 31, 1366–1375. [19] Morgan, H., Green, N. G., AC Electrokinetics: Colloids and Nanoparticles, Research Studies Press, Philadelphia, PA 2003. [20] Khoshmanesh, K., Akagi, J., Nahavandi, S., Skommer, J., Baratchi, S., Cooper, J. M., Kalantar-zadeh, K., Williams, D. E., Wlodkowic, D., Anal. Chem. 2011, 83, 2133–2144.
www.electrophoresis-journal.com
