A novel MEMS platform for a cell adhesion tester Ethan Abernathey Jeff Bütz Ningli Yang Instructor: Professor Horacio D. Espinosa ME-381 Final Project, Dec 1, 2006
TABLE OF CONTENTS Page Number Table of Contents 1 Project Summary 2 Project Description 3 References Cited 14 Biographical Sketches 15 1
PROJECT SUMMARY This project proposes the use of a novel MEMS platform for the purpose of testing cell adhesion forces. At the level of individual cells, quite little is known about a cell s mechanical properties. These mechanical properties can affect and be affected by chemical changes in the cell. These changes can even be different for uniaxial vs. biaxial stimulation [1]. Since the development of MEMS, research in this area has been increasing. This is because of the scale of MEMS. The typical cell is tens of microns in diameter. As the name implies, MEMS devices act on the micro scale providing the perfect platform for carrying out experiments on individual cells. There are currently many devices for applying and measuring forces and calculating rough cell adhesion. These include cantilever beams, post arrays, clamps, atomic force microscopy, magnetic twisting cytometry, micropipettes, and optical tweezers [2]. The biaxial cell stretcher design explored in this project [3] uses the unique application of biaxial stress to reveal more about a cell in such a stress state and offers to solve some of the limitations of previous designs that have limited the exact measurement of the adhesion force. Electrostatic actuators such as comb drives exhibit fast response times and are easily integrated into microsystems. Although they have been used extensively in dry MEMS, they have received less attention in microfluidic systems probably because of challenges such as electrolysis, surface tension and electrical conductivity. This report will outline the design, manufacturing and operation of the biaxial cell stretcher. First we will discuss the selection and advantages of biaxial stress application, followed by the basic functioning of the device. We will then discuss coatings for the device that can be selected to optimize force measurement and the detailed manufacturing of the device. The final portion of the report discusses the measurement of the force and the detailed operation of the device, in both air and water. 2
PROJECT DESCRIPTION Design Selection With a variety of tools available for the investigation of cell adhesion forces, the selection of a specific design requires careful consideration. After choosing a design that would apply force to stretch the cell it was necessary to determine how exactly forces would be applied. The biaxial cell stretcher design is a promising solution, benefiting from the advantages of biaxial force application on the cell. MEMS cell stretching devices thus far have been limited to uniaxial force application. The limitation of this approach is in the flexibility of cells. Cells can undergo elongation of over 50% [4], and this can make applying a force using traditional MEMS devices difficult due to the large displacements required. Furthermore, the force response to deformation in living cells can appear linear [4,5], meaning that applying a force to a cell can result in the cell reducing its stiffness. This can be seen in the two graphs in Figure 1. Figure 2 demonstrates a possible explanation of the linear behavior, showing internal buckling in the cytoskeleton that reduces the stiffness of the cell [5]. Figure 1 Cell force response to deformation showing linear behavior Figure 2 Proposed explanation of linear behavior 3
These large deformations can be reduced by applying biaxial stress to a cell. Like most elongated materials, cells experience lateral strain, reducing in one dimension while increasing in another. This can clearly be seen in Figure 3 in the cells that were not treated with Cyto D [4]. The lateral strain can be prevented by applying a tensile force in the same direction. This reduces the overall elongation of the cell so equivalent force can be applied with smaller displacements. Applying biaxial stress thus allows the use of more common MEMS actuators, such as comb drives, which wouldn t work with the uniaxial elongation of a cell. Structure and Basic Operation Figure 3 Demonstration of lateral strain in the y direction when cell is stretched in the x direction The selected design is a biaxial cell stretcher as pictured in Figure 4. The center circle, where the cell is placed, is divided into four quadrants. The top quadrant is stationary, the side quadrants move to the side and downward, and the bottom quadrant moves downward. This movement is driven by a single comb drive actuator at the bottom of the device with several folded springs used as a restoring force. The side quadrants movement is accomplished by the X structure, with the top portion of the quadrant linkages anchored to the substrate and the bottom linkages moving with the shuttle attached to the comb drive. 4
Figure 4 Biaxial cell stretcher structure The device is operated by applying a voltage (either DC or AC) to the comb drive, pulling the bottom quadrant down and the side quadrants downward and outward. The displacement of the quadrants was found to be within 5% of perfectly biaxial, demonstrating the effectiveness of the design. This displacement is measured using an optical microscope. The comb drive was able to achieve a maximum force of 60 µn at 100 V DC, causing a displacement of 3.4 µm. The force is more than enough to measure the adhesion force of a cell, typically between 10 to 100 nn, with most of the force absorbed by the stiffness of the device. Stretcher Surface Coating The displacement of the device is small for testing cell adhesion, even with the biaxial nature of the system. This can be solved by using particular coatings for the stretcher surface to change the nature of adhesion with the cell. In a study on cell adhesion with gold coated surfaces, three different coatings, 1-dodecanethiol (C12H26S, DDT), 1-hexadecanethiol (C16H34S, HDT) and 1-octadecanethiol (C18H38S, ODT), were tested to lower the adhesion force [6]. Detachment tests were performed using atomic force microscopy and the ODT was found to reduce adhesion force by about 63%, with the other two coatings resulting in intermediate reduction in adhesion force (38% for DDT, 47% for HDT). Use of these coatings would allow adjustment of the required force to measure the cell adhesion with a known scale factor, allowing the small displacement of the system to apply enough force for the application. Microfabrication of the Cell Adhesion Tester 5
The described cell adhesion tester is a design that is complicated by its use of multiple layers. This complexity is best addressed through the use of a PolyMUMPS process of microfabrication [2]. The PolyMUMPS process of microfabrication is a multilevel process utilizing layers of polycrystalline silicon and sacrificial material to build up MEMS structures [7]. To make description of the process easier and better focus on the complexity of the actual testing device, the fabrication description has been broken down in to two distinct parts. The two parts are fabrication of the test platform (testing device) and the fabrication of the comb drive. The following sections refer to different layers as PolyS 0, 1, or 2 or PSG 1 or 2; these layers are the same for both the test device and the comb drive. 1. Fabrication of the Test Platform The process begins with an n-type (100) silicon wafer. To prevent charge feed through from the device to the silicon surface, the wafer is n-doped further with POCl 3. An insulating layer of Si3N4 is deposited with low pressure chemical vapor deposition (LPCVD). Following this, the initial layer of polycrystalline silicon (PolyS 0) is deposited with LPCVD. Using photolithography along with PolyS_Mask_#1 (Figure 5) the five stationary posts of the device are created. The process for this photolithography requires that a photoresist layer be deposited on top of the PolyS 0, the layers be exposed through PolyS_Mask_#1, and then the unexposed material be etched away with a Reactive Ion Etch (RIE). The photoresist can then be washed away. Figure 5 - PolyS_Mask_#1 The next step in building up this structure is to coat the surface with a sacrificial layer of PhosphoSilicate Glass (PSG). This first layer, PSG 1, is then patterned with photolithography and PSG_Mask_#1 (Figure 6) in order to leave the five posts of PolyS 0 surrounded but not covered by PSG. This will leave a surface on a new layer of PolyS can be deposited and formed. 6
Figure 6 PSG_Mask_#1 The second layer of PolyS, PolyS 1, is then deposited with LPCVD and patterned with another photolithographic process. The mask PolyS_Mask_#2 (Figure 7) is used to pattern this level of PolyS into the adhesion testing platforms on which the cells will sit. The photolithographic process again follows the procedures previously outline. Again, as before with the PSG 1 layer, a second sacrificial layer, PSG 2, is deposited and patterned to surround the PolyS 1 layer but not cover it. The patterning of PSG 2 is done through PSG_Mask_#2 (Figure 8). Figure 7 PolyS_Mask_#2 Figure 8 PSG_Mask_#2 The final step to creating the cell stretcher is to form the linkage arms that pull the side portions of the platform outwards as the moving transverse is pulled downwards by the connected comb drive. This is simply done through the same means as the previous steps with a LPCVD of a PolyS 2 layer, which is patterned by PolyS_Mask_#3 (Figure 9). For this step, no further PSG layer is needed because this is the top level of PolyS and no further building is needed on top of this layer. 7
Figure 9 PolyS_Mask_#3 Figure 10 is the side view of final layered cell stretcher device. Figure 10 Side view of final layered cell stretcher device 2. Fabrication of the Comb Drive The process of fabricating a comb drive is analogous to the fabrication of the test device. The process begins with the posts of the folded springs and stator bases being patterned onto the wafer surface in the PolyS 0 layer. A PSG 1 sacrificial layer is used to provide support for the deposition and patterning of all the comb heads of the stator and rotor portions of the drive and the actual folded springs out of the PolyS 1 layer. Above this layer, there is no more structure on the comb drive. Thus, all further steps in producing the test device, such as use of PSG 2 and PolyS 2 layers, are not important and will be fully etched away on the comb drive, leaving a fully mobile comb drive actuator connected to the transverse of the test device. The appropriate masks and final layered cross section are shown from Figure 11 to Figure 13. 8
To release the device, a 49% HCl mixture in water has traditionally been found to achieve the best results [3]. This leaves the platform suspended above the wafer and allows for free motion of the linkage arms and comb drive. Figure 11 Side view of final layered comb drive actuator Figure 12 PolyS 0 mask for comb drive Figure 13 PSG 1 mask for comb drive Force Calculation A single cell is adhered to the substrate of the proposed cell stretcher device is shown in Figure 1. Cell adhesion force between the cell and the substrate is estimated using F = F comb - k x [8], where F comb is the force of the comb drive, k is the stiffness of the MEMS and x is the cell displacement. At the heart of the device is a circular four quadrant sectioned plate that works as the seating platform for the cell under test. This platform ideally needs proper surface functionalization to promote cell adhesion. The platform is connected to a comb drive actuator via a four link mechanism. The equivalent kinematic scheme of the mechanism is shown in Figure 2. In order to 9
gain an insight into the kinematic behavior of the device, each beam which constitutes the cell stretcher is modeled by means of a rigid beam. Moreover rigid beams are linked to each other by means of revolute joints, which accounts for the on-plane compliance of the real beam. Figure 2 represents the right half of the upper part of the complete cell stretcher (Figure 1). Points A, B and C in Figure 2 correspond to the same points of Figure 1. The slide allows the vertical displacement of point C; point B represents the position of the right quadrant. Notice that beams are tilted at 45 and 135 respectively with respect to the horizontal in the undeformed configuration (solid line in Figure 14). When a force is applied to the cell stretcher (dashed line in Figure 14), causing a downward motion C y of point C, the point B moves towards the right and downwards, respectively of a distance B x and B y. Figure 14 Kinematic principle of the biaxial cell stretcher platform (letters A, B and C referred to the same locations of Figure 1) If small displacements are assumed, it can be inferred that B x C y /2 By = Cy/2 (1) Such relationships are satisfied if and only if the two beams have the same length and the angle they form is 90. Considering the cell stretcher symmetry, the distance x and y between the tips of the two pairs of opposite quadrants is: x = x 0 + 2 B x x 0 + C y and (2) y = y 0 + 2 B y = y 0 + C y respectively, where x 0 and y 0 are tip distances in the undeformed configuration. Equation (1) do not hold for large displacements since nonlinearities occur; in particular, for high values of C y, B x tends to decrease with respect to B y. As it will be shown, for small displacements, both Finite Elements Analysis FEA and experimental results validate these initial assumptions. This platform ideally needs proper surface functionalization to promote cell adhesion. The platform is connected to a comb drive actuator via a four link mechanism. Six folded springs are connected to the central bar of the vertical moving structure of the device to provide restoring force. Additionally, these springs prevent the comb drive from exhibiting lateral instability. According to Saint-Venant linear beam theory [9], the spring stiffness can be described by k b = 24EJ/(l 3 1 + l 3 2 ) (3) where E is the Young s modulus, J is the moment of inertia, and l 1 and l 2 are the lengths of the 10
beams constituting the folded spring. The six springs should have a combined spring constant six times that of a single beam. In the other hand, the stiffness k x of the stretcher structure, which is approximately eight times of the on of each single folded spring, can not be neglected, assuming plane stress and sress-strain linearity conditions. Totally, the stiffness k is the sum of the all seven contributions. A comb drive actuator is employed to operate the cell stretcher. We use comb drive based on its several advantages; low power consumption, high speed, moderate driving voltage, low cost, high accuracy. The actuation force of a comb drive actuator is [10] F act =NεtV 2 /g (4) where N is the number of comb electrodes, ε is the permittivity constant, t is the comb electrode thickness, V is the driving voltage and g is the comb electrode gap. Operation in Air To operate the MEMS in air, a DC power supply is wired to the support plate connectors. Since from an electrical point of view, comb drives are equivalent to ideal capacitors, they are able to operate at DC conditions without drawing energy from the power supply: in fact the circulating current is effectively zero if the surrounding medium is an insulator, as is this case. In this case it is possible to use a low-power, high output impedance power supply, composed of a high voltage generator in series with a voltage divider with a total resistance of 10 MΩ. The latter is realized using a highly sensitive potentiometer, whose center tap was directly connected to the comb drive stator. Using a high voltage generator it is possible to collect displacement information, while reading the actual voltage by means of a high input impedance multimeter. To observe and record its behavior, the MEMS device is placed on the stage of an optical microscope equipped with a digital camera. Underwater Operation Since the relative permittivity ε r of pure water is about 80, which is higher than that of air (ε r =1), a great enhancement in the displacement per volt will be observed. Therefore it is better to do the measurements underwater. Although electrostatic actuation has found extensive use in dry MEMS, it has received little attention in microfluidic systems. The three main challenges of operating an electrostatic actuator underwater are electrolysis, surface tension and electrical conductivity [11]. These challenges and the proper solutions are discussed respectively. 1. Electrolysis Electrolysis can occur in pure, de-ionized water if a DC voltage greater than 1.23 V is applied across two submerged electrodes [12]. This process, by which water is broken down into hydrogen and oxygen at the anode and cathode, respectively, can produce large amounts of gas underwater, which unavoidable would lead to device failure due to bubbling. The de-ionization process ensures that the electrical conductivity of the water is low and varies as little as possible between trials. The lower this conductivity is, the lower the rate of electrolysis becomes. Some researchers have avoided these problems by isolating the actuator from the microchannel. This requires a mechanical seal [13] or a flexible actuator [14], which are difficult to accomplish in silicon MEMS and limit the potential applications. AC driving system can be used to prevent electrode screening and electrolysis and thus 11
enable electrostatic actuation in many liquids [15], at potentials low enough to avoid electrochemistry. Equivalent circuit model for a silicon electrostatic actuator in a resistive conducting fluid is shown in Figure 15. Figure 15 Equivalent circuit model for a silicon electrostatic actuator in a resistive conducting fluid It consists of a signal generator a high-frequency ac square wave that was set to drive the comb with a 1 MHz square wave signal with an average voltage of 0 V. This technique has been termed root mean square (RMS) operation. RMS is defined to be the time average of an AC signal that has the same power as a dc input. Although any type of ac signal could work, the square wave was chosen because the RMS value of a square wave is approximately equal to its amplitude. The rapid alternation of the position of anode and cathode inhibits a net development of gas at either electrode, thus allowing the device to work properly for a prolonged time. Since the frequency response of the mechanical device drops rapidly at frequencies well below that of the input signal, and since the ideal transfer function of the comb drive depends on the square of the applied voltage, a net motion can be obtained, virtually with no traces of the high frequency carrier signal. Electrostatic actuation may be suitable in many electrolytes typically used in microfluidic applications. 2. Surface tension Water is prevented from flowing under the PolyS 1 layer (which is the bottom part of the moving layer), since the silicon-water interface tension is high, which in turn causes the silicon surface to behave hydrophobically [11]. Therefore the resulting stiction forces across water-air meniscus reduce device performance. In addition, certain actuator designs can keep air trapped when water is placed over the entire chip. The hydrophobicity can be deduced by the observed spherical shape of water drops over the device. If the surface tension of the water is eliminated as a factor by operating completely submerged, no significantly greater stiction problems are encountered compared to air. Drying the actuators caused complete sticking to the substrate but rewetting them would generally free them for other trials. Consider that surfactants can reduce the surface tension of water by adsorbing at the liquid-gas interface, we can add a surfactant (sodium laureth sulphate) to reduce the silicon-water interface tension till the silicon surface became hydrophilic. 12
3. Electrical conductivity If the medium is electrically conductive, current can bypass the actuators and the power available to the actuators is reduced, negatively affecting actuator efficiency. Electrostatic actuators are the most vulnerable to the electrical conductivity of the water because the large electrodes in close proximity mean that the resistance between the capacitor plates is relatively small. The electrical conductivity will be reduced as much as possible by the use of de-ionized water. Using deionized water allowed the comparison of water properties such as thermal conductivity and dielectric constant without unusually large current bypassing the actuators. Underwater measurements on the same device can be performed applying a small drop of deionized water over the entire surface of the chip and covering it with a microscope slide glass window. This spread the water relatively evenly on the chip and flattened the surface so that devices could be viewed without distortion. The displacements are measured using the same optical equipment already described, and an oscilloscope was used for the acquisition of the effective amplitude signal. Conclusion This project outlined the use of a novel biaxial cell stretcher to measure the adhesion force of a single cell. The selection of biaxial stress application helps to solve the large elongation found in uniaxially stressed cells, allowing the use of an extensively tested MEMS actuator (comb drive). The selection of surface coatings allows the adjustment of adhesion force to optimize measurement. The PolyMUMPS manufacturing process outlined provides an easy, repeatable way to manufacture the device. Finally, the calculation for the adhesion force was described and the operation in air and water analyzed, completing a description for the manufacture and use of the biaxial stretcher to measure cell adhesion. 13
REFERENCES CITED [1] Wang, JHC, and BP Thampatty. An introductory review of cell mechanobiology. Biomechanics and Modeling in Mechanobiology, 5 (1): 1-16 Mar 2006. [2] Van Vliet, K.J., G. Bao and S. Suresh. The biomechanics toolbox: experimental approaches for living cells and biomolecules. Acta Materialia, 51 (19): 5881-5905 Nov 25 2003. [3] Scuor, N., et al. Design of a novel MEMS platform for the biaxial stimulation of living cells. Biomedical Microdevices, 8 (3): 239-246 Sep 2006. [4] Yang, Shengyuan, and Taher Saif. Reversible and repeatable linear local cell force response under large stretches. Experimental Cell Research, Apr 2005, Vol. 305 Issue 1, p42. [5] Saif, MTA, CR Sager and S Coyer. Functionalized biomicroelectromechanical systems sensors for force response study at local adhesion sites of single living cells on substrates. Annals of Biomedical Engineering, 31 (8): 950-961 Sep 2003. [6] Chang, CH, et al. Cell adhesion and related phenomena on the surface-modified Audeposited nerve microelectrode examined by total impedance measurement and cell detachment tests. Nanotechnology, 17 (10): 2449-2457 May 28 2006. [7] Koester, D., A. Cowen, R. Mahadevan, M. Stonefield and B. Hardy. Polymumps design handbook (Memscap 2003). [8] Sager, C., P. LeDuc, and T. Saif. 1st Annual International IEEE-EMBS Special Topic Conference on Microtechnologies in Medicine & Biology October 12-14, 2000. [9] Gere, J. M., and S. P. Timoshenko. Mechanics of Materials (Chapman & Hall, 1991). [10] Jaecklin, V. P., C. Linder, N. F. de Rooij and J. M. Moret. Micromechanical comb actuators with low driving voltage. Journal of Micromechanics and Microengineering, 2 (4): 250-255 Dec 1992. [11] Sameoto, D., T. Hubbard and M. Kujath. Operation of electrothermal and electrostatic MUMPs microactuators underwater. Journal of Micromechanics and Microengineering, 14 (10): 1359 1366 Oct 2004. [12] Zumdahl, S. S. Chemical Principles 2nd edn (DC: Heath and Company, 1995) 477 485. [13] Okandan, M., P. Galambos, S. Mani, and J. Jakubczak. Surface micromachined cell manipulation device for transfection and sample preparation. MicroTAS, Monterey, CA, 2001. [14] Francais, O. and I. Dufour. Dynamic simulation of an electrostatic micropump with pull-in and hysteresis phenomena. Sensors and Actuators, 70 (1): 56 60 Oct 1998. [15] Sounart, T. L., T. A. Michalske, and K. R. Zavadil. Frequency-dependent electrostatic actuation in microfluidic MEMS. Journal of Microelectromechanical Systems 14 (1): 125 133 Feb 2005. 14
BIOGRAPHICAL SKETCH Ethan Abernathey Will graduate in June of 2008 with a B.S. degree in Materials Science and Engineering from Northwestern University. He is directing his degree in an area of specialization in the area of materials nano-fabrication. Jeff Bütz B.S. student in the department of Mechanical Engineering at Northwestern University, graduating in June of 2007. He is a member of the Tau Beta Pi and Pi Tau Sigma engineering honor societies. Ningli Yang Ph.D student in the department of Mechanical Engineering at Northwestern University. Received the B.S. degree in department of Electrical engineering in Nanjing University, China in 2003, and the M.S. degree in department of Electrical engineering in Nanjing University, China in 2006, respectively. 15