Polymer-Based Wireless Resonant Magnetic ... - Semantic Scholar
26
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Polymer-Based Wireless Resonant Magnetic Microrobots Hsi-Wen Tung, Massimo Maffioli, Dominic R. Frutiger, Kartik M. Sivaraman, Salvador Pan´e, and Bradley J. Nelson, Fellow, IEEE
Abstract—This paper presents a new generation of wireless resonant magnetic microactuator (WRMMA) type microrobot, named PolyMite. A polymer is used as a spring material in the design of the PolyMite. Compared with the previous generation of such devices, the use of a polymer, i.e., SU-8, as a spring material enables the design of a stable spring system without a significant increase in stiffness. A speed of over 20 mm/s (40 body lengths per second) was measured when the PolyMite was driven in air. Micromanipulation abilities of the PolyMites in wet environments have also been demonstrated. The agents were capable of transporting microobjects such as light polystyrene beads and heavier glass beads in water. The large output force, remote actuation, and biocompatible outer surface (SU-8 and gold) make the PolyMite an exciting candidate for biomedical applications. Index Terms—CoNi alloy, micromanipulation, SU-8, wireless mobile microrobots, wireless resonant magnetic microactuators (WRMMA).
I. INTRODUCTION OWER transmission is one of the primary challenges for wireless controlled microrobots; due to the unfavorable scaling of energy storage at the microscale, the optimal strategy is to design these devices to harvest energy from their environment. Strategies to harvest energy are mainly focused on electrostatic force [2], [3], thermal-induced deformations of the structures [4], chemical fuel from the environments [5]–[7], and magnetic force or torque [8]–[13]. Magnetic fields have emerged as the most favorable option, especially for biomedical applications, because they are capable of penetrating nonmetallic materials with little or no interaction and are nearly harmless to most living organisms. In our previous research, we have developed a concept called wireless resonant magnetic microactuators (WRMMAs) [14], [15]. The model of the WRMMA can be simplified as a twomass-spring oscillator system (see Fig. 1). One mass called the “body” stands on the substrate, while the other mass called the
P
“hammer” is supported by the spring and suspended above the substrate. Both the body and the hammer are made of a softmagnetic material. Because of magnetic shape anisotropy, the masses are magnetized in the direction of the easy axis (the long axis) of the combined magnetic masses when exposed in an external magnetic field. This magnetization induces an attractive force between the two masses to pull them toward each other, and a torque to rotate a WRMMA agent to align its easy magnetic axis with the field. When the magnetic field is turned OFF, the attraction force vanishes and the restoring force of the spring pushes the two masses away. By placing the device in an oscillating magnetic field whose frequency is close to the resonant frequency of the device, the response of the body and the hammer to these forces can be magnified to generate a net displacement of the device A microrobot based on the WRMMA concept can be driven on a flat, nonengineered substrate, in dry or wet environments. A precursor to this microrobot, called MagMite, can move controllably at high speeds when driven in air (12.5 mm/s or 42 body lengths per second), and the force it generates is strong enough to push micron-sized objects such as glass beads and metal disks. MagMite can be potentially used for sorting and extracting delicate biological and biochemical samples. However, the fabrication process which consists of four-layer stacks of electroplated metals totaling up to 70 μm, including one layer of 25 μm thick high-aspect-ratio gold spring, is very expensive, complex, and laborious. We introduce here our new microrobots that are produced by an easier, faster, and cheaper fabrication process in the Section II. The details of the fabrication process are presented in the Section III. The characterization of the new agents, including a more reliable motion thanks to the new design of the spring system, and the micromanipulation abilities of the new agents are discussed in Section IV. II. DESIGN AND MODELING
Manuscript received April 22, 2013; revised September 16, 2013; accepted October 9, 2013. Date of publication November 13, 2013; date of current version February 3, 2014. This paper was recommended for publication by Guest Editor A. Ferreira upon evaluation of the reviewers’ comments. An early version of this work was described the 2012 Proceedings of the IEEE International Conference on Robotics and Automation. This work was supported by the Swiss National Science Foundation and ETH Zurich. The authors are with the Institute of Robotics and Intelligent Systems, ETH Zurich CH-8092, Switzerland (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]). This paper has supplementary downloadable material available at http:// ieeexplore.ieee.org. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TRO.2013.2288514
A. Design There are two important characteristics to be considered when designing a new WRMMA microrobot: eigenfrequencies and eigenmodes of the agent, and the displacement of the hammer induced by a magnetic field. Limited by the working range of the magnetic field generator in the lab, the eigenfrequency of the eigenmode that causes the propulsion of the agent should be less than 10 kHz, and the displacement of the hammer under a uniform magnetic field of 8 mT should be sufficient to cause a net motion when driving the WRMMA agent at resonant frequency. The stiffness of the spring system of a WRMMA
1552-3098 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
27
Fig. 1. (a) Schematic diagram of the physical model of WRMMAs. (b) When exposed to a magnetic field, the two soft-magnetic masses are magnetized. Attractive interbody force is induced and pulls the two masses close to each other. (c) When the magnetic field is removed, the interbody force vanishes and the restoring force of the spring pushes the two masses away.
SU-8, which is an epoxy-based polymer, was used due to its low Young’s modulus (a few gigapascal), the ease of patterning it with photolithography, and its suitability for high-aspect-ratio structures [17]–[20]. Electrodeposited cobalt–nickel (CoNi) alloy was used to form the soft ferromagnetic masses due to its high magnetic permeability, high magnetic saturation, and low remanence (a saturation magnetization of 0.85 T and a coercivity of 20 Oe) [21]. The high magnetic permeability allows the PolyMite to generate higher interbody forces in a stronger external magnetic field. The low remanence reduces the residual magnetic force acting to the spring after the magnetic field is OFF, and thus, it maintains a consistent behavior of the PolyMite following exposure to strong magnetic fields. B. Modeling
Fig. 2. PolyMite with the thickness of 50 μm, and a cross-section from A-A shows the dimple feet.
microrobot dominates these characteristics. Due to the high Young’s Modulus of gold, only one gold spring was used on the MagMite to keep the stiffness in a reasonable range. However, a single spring system was too weak to constrain the motion of the hammer that resulted in the drift motion of the MagMite. The PolyMite, a new type of WRMMA microrobot, has been developed based on a design that uses polymer for the springs. The illustration of the design of the PolyMite is shown in Fig. 2. Polymers are suitable for flexible components in microactuators because of their low Young’s modulus [16]. Young’s modulus of a polymer is more than one order of magnitude smaller than that of gold. By replacing gold with polymer as the material of the spring system, the PolyMite is able to have two symmetric springs to connect the body and the hammer. Thus, the movement of the hammer is more stable in the lateral direction. The magnetic masses are present on the same plane as the springs, thus ensuring that the magnetic attraction forces and the restoration forces of the spring are in the same plane. This is in contrast with the older MagMite devices, where the presence of the magnetic masses and the spring in different planes created a bending moment on the device. The body stands on a small dimple feet, while the hammer is suspended above the substrate. The dimple feet reduces the contact area between the PolyMite and the substrate and, thus, reduces friction. Additionally, the curved shape of the PolyMite is helpful to release objects from the agent during micromanipulation tasks.
The interbody force induced due to applied magnetic fields, the stiffness of the springs, and the resonant frequencies of the PolyMites were investigated by finite-element analysis using COMSOL Multiphysics. A standard design of the PolyMites had a diameter of 500 μm and a thickness of 50 μm. The dimensions of the two magnetic masses, the body and the hammer, were 225 μm × 170 μm × 50 μm and 170 μm × 170 μm × 50 μm, respectively. The distance between the two masses was 45 μm. The attraction force induced between the body and the hammer for different magnetic flux densities was simulated, and the stiffness of the spring system was designed in a range that allows tens of nanometers of relative displacement of the two masses at the initial state. The springs were designed with the dimensions 1272 μm (unfolded length) ×10 μm × 50 μm for each side. The resonant frequencies were investigated for a tethered PolyMite. The major resonant mode driving the PolyMite was the third resonant mode (the in-plane compression mode) at 5.6 kHz, which is far away from the first resonant mode (out-ofplane waving) and the second mode (in-plane shifting) that were at around 2.1 kHz (see Fig. 3). For an untethered PolyMite sliding on a substrate, the resonant frequency of the two mass-spring √ oscillator increased by a factor of 1 + α, where α = mH /mB is the mass ratio of the body and the hammer [22]. From this calculation, the resonant frequency of an untethered PolyMite was predicted to be around 7.4 kHz. III. FABRICATION PROCESS To combine a polymer and a ferromagnetic metal on the same platform, a new fabrication process was developed that connects the two materials without any adhesion problems between them.
28
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Fig. 3. FEM-derived resonant modes of a tethered PolyMite. The first and the second modes exist around 2.1 kHz, while the third mode is found at 5.7 kHz. Different colors indicate relative displacement of each element in the structure, where dark blue stands for zero displacement and dark red represents maximum displacement.
A single-side polished silicon wafer is used as a substrate of the process. A silicon dioxide (SiO2 ) layer is used as a hard mask for dimple holes etching process. The etching windows are patterned on the SiO2 layer by photolithography and reaction-ionetching (RIE) process [see Fig. 4(a)]. After removing the photoresist, the wafer is immersed in potassium hydroxide (KOH) solution to etch pyramidal dimple holes into the silicon wafer [see Fig. 4(b)]. The SiO2 hard mask is removed after the etching process [see Fig. 4(c)], and another thin layer of SiO2 is deposited on the wafer as a sacrificial layer [see Fig. 4(d)]. The electrodes are deposited by evaporating titanium (Ti)/gold (Au) bilayer on the substrate and patterned by the lift-off method, thus rendering only a part of the substrate electrically conductive [see Fig. 4(e)]. SU-8 2025 is applied onto the substrate and patterned into the shapes of springs, frames for electrodeposited metals, and tethers [see Fig. 4(f)]. The patterned wafer is then immersed into different electrolytes to deposit CoNi alloy and Au layers by electrodeposition. The metals selectively deposit on the patterned electrodes and the shapes are defined by the SU-8 frames [see Fig. 4(g)]. Unlike most current fabricating methods, where magnetic metals and nonmagnetic structures are fabricated separately before assembling them, metals in our process are selectively electrodeposited within the polymer frames of the device structure. Therefore, the issue of assembly is solved and the magnetic bodies are well aligned with the nonmagnetic structures. The CoNi microstructures exhibit similar thickness as the SU-8 ones, while the Au layer is less than 1 μm thick. The Au layers protect the CoNi alloy from being damaged in the releasing solution in the following step. To release the devices from the substrate, the wafer is immersed in buffered hydrofluoric acid to etch SiO2 and the evaporated titanium away [see Fig. 4(h)]. Finally, the devices are cut from their tethers using a microlaser milling tool [see Fig. 4(i)]. Fig. 5 shows a released PolyMite together with micro polystyrene beads. Due to the outer surfaces being made of biocompatible materials, namely SU-8 and Au, the devices are suitable for biomedical applications.
Fig. 4. Fabrication flow of the PolyMite. (a) Thin layer of SiO2 is grown on the Si wafer as a hard mask for etching and the etching windows of dimple holes are patterned on the SiO2 layer. (b) Dimple holes are etched anisotropically into the silicon wafer by KOH solution. (c) SiO2 layer is etched away. (d) Another layer of SiO2 is deposited as the sacrificial layer. (e) Ti 10 nm/Au 200 nm layers are deposit by evaporation and patterned to form the electrodes. (f) SU-8 is patterned to form springs, frames of electroplated metals, tethers, etc. (g) CoNi alloy and Au are deposited by electroplating. (h) Wafer is immersed in the buffered hydrofluoric acid to etch the sacrificial layer away and release the whole structures from the Si wafer. (i) Microlaser milling tool is applied to clean and release the PolyMites from their tethers.
Fig. 5. Fully released PolyMite with polystyrene beads of approximately 120 μm in diameter.
IV. EXPERIMENTAL RESULTS The control system of the PolyMites consists of two pairs of Helmholtz coils nested orthogonally to generate a uniform
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
29
TABLE I DESIGN PARAMETERS OF DIFFERENT TYPES OF POLYMITES
Fig. 6.
Helmholtz coils utilized to drive the PolyMites. Fig. 7. Frequency response of three different designs of PolyMites. Speeds of more than 20 mm/s were measured when driving agents at 4 mT.
magnetic field at their center in a direction parallel to the surface of manipulation. The custom-made current control electronics send a rectangular waveform signal to the coils to generate oscillating magnetic field. The orientation of the field is set by linear superposition of the two orthogonal magnetic fields (see Fig. 6). A computer vision tracker was built to trace the motion of the PolyMites from the images captured from an overhead camera. The refresh rate of the camera was set to 60 fps. A. Characterization of PolyMites The motion of the PolyMites is primarily controlled by the frequency and the magnitude of the magnetic field, and the orientation of the PolyMites is changed by changing the direction of the magnetic field. Three different designs of the PolyMites were tested. The agents differ from each other in the spring length, distance between the two magnetic masses, and the weight of the two masses. The details of the difference in the three PolyMites are listed in Table I. Fig. 7 shows the frequency sweep of the three PolyMites on a Ti-coated Si wafer in air. The M1A01 agent was the standard design used to model the PolyMite. The peak of response was at 7.3 kHz, very close to our prediction from the simulation result. The maximum speed of the M1A01 agent was measured from the data collected by the tracker as over 20 mm/s at 4 mT, which is equal to 40 times the body length per second. The M1B04 agent had the same geometry of the springs as the M1A01 agent, but the weight of the two magnetic masses is about 40% heavier. This caused the resonant frequency of the M1B04 to shift 1.2 kHz lower. Additionally, a minor motion caused by
the first and the second resonant mode of the PolyMites was observed at around 2 kHz on the frequency response plots of the M1A01 and M1B04 agents, respectively. The M1B05 agent had stiffer spring and less weight of the two magnetic masses than the M1B04 agent, and therefore its resonant frequency was higher. Moreover, the bandwidth of the frequency response of the M1B05 agent was nearly a half of the bandwidths of the M1A01 and M1B04. The shift of the resonant frequency of individual agents and narrow bandwidth of the frequency response are favorable in multiagent control by driving agents with signals composed of different frequencies. The response of the PolyMites to the magnitude of the magnetic field was measured at a frequency slightly shifted from the resonant frequency of each agent because the speed of the agents was too fast to be tracked at high magnitude of the magnetic field at resonant frequency. Fig. 8 depicts the response of the M1B04 and M1B05 agents to the magnitude of the magnetic field. The M1B04 agent started to move at 5 mT and the M1B05 agent started at 7 mT. Some movement was observed at lower fields but is excluded in the figures because the agents were more sensitive to asperities and cleanliness of the substrates at lower fields and got stuck on the substrates easily. Excluding these measurements, the speed of the agents increases linearly with the magnitude of the magnetic field. The M1B05 agent required higher energy to overcome the friction and to generate a smooth net motion because of its stiffer spring and the larger distance between the two masses. This explains why the bandwidth of the M1B05 agent is narrower.
30
Fig. 8. Response of (a) the M1B04 agent and (b) the M1B05 agent to the magnitude of the magnetic field.
As mentioned in Section II-A, the PolyMite uses a spring system consisting of two symmetric springs that provides equal support on both lateral sides of the hammer and increases the stability of the in-plane motion of the hammer. The improvement in stability was evaluated by comparing the drift motion of the PolyMite and the MagMite. The drift angle is defined as the angle between the forward orientation of the agent and the direction of the movement. Fig. 9 shows the trajectories of a MagMite and a PolyMite tested under different frequencies. The direction of the driving field was along the x-axis. The drift angles of the MagMite varied with the exciting frequencies, and the agent moved sideways at certain frequencies [23]. Although drift was observed in the motion of the PolyMite, it was significantly lower than that of the MagMite. The drift motion of the PolyMite may come from some defects introduced during the fabrication, small particles stuck to the bottom of the agents that changed the contact to the substrate, or other system setup issues. The drift motion is not a serious issue since the direction of the motion can be corrected easily by manual or PID control. The wear of the dimple feet of the PolyMite can change the friction between the PolyMite and the substrate. One PolyMite was driven for an hour in air, and its dimple feet were inspected
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Fig. 9. Trajectories of (a) a MagMite driven by frequency sweep 5.1–5.7 kHz [23] and (b) a PolyMite driven by frequency sweep 1–9 kHz. The orientation of the magnetic field is pointed in the x-direction. The drift motion of the PolyMite is improved compared with the motion of the MagMite thanks to the symmetric springs of the PolyMite that stabilize the in-plane motion of the hammer.
in the scanning electron microscope. There was no significant difference in the shape and height of the dimple feet between the tested agent and another new agent from the same batch of fabrication process. Therefore, the wear of the dimple feet was not considered in our experiments. B. Micromanipulation Micromanipulation in wet environments is of general interest to many scientists since most biomolecules function in a liquid medium. To demonstrate the abilities of micromanipulation in wet environments, the PolyMite was driven by manual control to push microbeads in water. Fig. 10 shows a PolyMite collecting polystyrene beads of diameter 105–125 μm one by one to the left corner of the stage. (In addition, see the supplementary material S1.) Thanks to the curved contour of the PolyMites, polystyrene beads were released from the PolyMite easily by rapidly turning the agent. In the rare cases, where the bead remained stuck to the PolyMite, the drag force generated by the quick turning motion
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
31
Fig. 10. Image sequence showing how a PolyMite manipulates polystyrene beads with a diameter of 105–125 μm in water. The agent collects randomly located beads on the right and moves them one by one to the left corner of the stage. The numbers labeled identify individual beads, and dash lines indicate the trajectory of the PolyMite.
Fig. 11. Image sequence showing how a PolyMite pushes a glass bead with a diameter of 60 μm in water on a structured substrate. The PolyMite pushes the glass bead toward the hole at the edge of the stage. After the bead falls down into the hole, the agent turns and departs. The circles mark the position of the pushed bead, and dash lines indicate the trajectory of the PolyMite.
was strong enough to break the adhesion between the agent and the bead. Heavier samples such as glass beads could also be manipulated by the PolyMite. Fig. 11 depicts a PolyMite driven on a structured substrate with a hole at the edge of the stage. A glass bead with a diameter of around 60 μm was transported by the agent to the top of the hole. After the force of gravity caused the bead to fall into the hole, the PolyMite was driven away (see supplementary material S2). V. CONCLUSION A new generation of WRMMA type microrobot—the PolyMite—has been developed. The design of the PolyMite reduces the time, the cost, and the complexities of the fabrication process of its predecessor, i.e., the MagMite, with no loss in performance. The outer surfaces of the PolyMite consist of gold and SU-8 which are inert to most chemicals and safe for use in biomedical applications. The fabrication process of the PolyMite combines polymer and ferromagnetic material with no problem of adhesion and alignment between them. The PolyMite is capable of being operated on a plane surface in both dry and wet environments. A speed of over 20 mm/s (40 body lengths per seconds) was reached when the agent was driven in air on a titanium-coated silicon wafer with an applied magnetic flux density of 4 mT. The new design of the symmetric spring system reduces drift motion. The PolyMite is capable of transporting micropolystyrene beads and microglass beads in water to a predefined location. In future, we envisage the PolyMite
selecting or extracting micron-sized samples that are not easy to be handled by human hands. It can be a group of microorganisms or protein crystals on a stage, and the PolyMite can push the selected sample toward a hole or another macro-sized tool for extraction.
REFERENCES [1] H. W. Tung, D. R. Frutiger, S. Pane, and B. J. Nelson, “Polymer-based wireless resonant magnetic microrobots,” in Proc. IEEE Int. Conf. Robot. Autom., 2012, pp. 715–720. [2] B. R. Donald, C. G. Levey, C. D. McGray, I. Paprotny, and D. Rus, “An untethered, electrostatic, globally controllable MEMS micro-robot,” J. Microelectromech. Syst., vol. 15, no. 1, pp. 1–15, 2006. [3] B. R. Donald, C. G. Levey, and I. Paprotny, “Planar microassembly by parallel actuation of MEMS microrobots,” J. Microelectromech. Syst., vol. 17, no. 4, pp. 789–808, 2008. [4] O. J. Sul, M. R. Falvo, R. M. Taylor, S. Washburn, and R. Superfine, “Thermally actuated untethered impact-driven locomotive microdevices,” Appl. Phys. Lett., vol. 89, no. 203512, 2006. [5] S. Fournier-Bidoz, A. C. Arsenault, I. Manners, and G. A. Ozin, “Synthetic self-propelled nanorotors,” Chem. Commun., pp. 441–443, 2005. [6] Y. Wang, S. T. Fei, Y. M. Byun, P. E. Lammert, V. H. Crespi, A. Sen, and T. E. Mallouk, “Dynamic interactions between fast microscale rotors,” J. Amer. Chem. Soc., vol. 131, no. 29, pp. 9926–9927, 2009. [7] A. A. Solovev, Y. F. Mei, and O. G. Schmidt, “Catalytic microstrider at the air-liquid interface,” Adv. Mater., vol. 22, no. 39, pp. 4340–4344, 2010. [8] J. J. Abbott, O. Ergeneman, M. P. Kummer, A. M. Hirt, and B. J. Nelson, “Modeling magnetic torque and force for controlled manipulation of softmagnetic bodies,” IEEE Trans. Robot., vol. 23, no. 6, pp. 1247–1252, Dec. 2007. [9] L. Zhang, J. J. Abbott, L. X. Dong, B. E. Kratochvil, D. Bell, and B. J. Nelson, “Artificial bacterial flagella: Fabrication and magnetic control,” Appl. Phys. Lett., vol. 94, no. 064107, 2009.
32
[10] A. Ghosh and P. Fischer, “Controlled propulsion of artificial magnetic nanostructured propellers,” Nano Lett., vol. 9, no. 6, pp. 2243–2245, 2009. [11] C. Pawashe, S. Floyd, and M. Sitti, “Modeling and experimental characterization of an untethered magnetic micro-robot,” Int. J. Robot. Res., vol. 28, no. 8, pp. 1077–1094, 2009. [12] S. Martel, M. Mohammadi, O. Felfoul, Z. Lu, and P. Pouponneau, “Flagellated magnetotactic bacteria as controlled MRI-trackable propulsion and steering systems for medical nanorobots operating in the human microvasculature,” Int. J. Robot. Res., vol. 28, no. 4, pp. 571–582, 2009. [13] L. Zhang, T. Petit, K. E. Peyer, and B. J. Nelson, “Targeted cargo delivery using a rotating nickel nanowire,” Nanomed.-Nanotechnol. Biol. Med., vol. 8, no. 7, pp. 1074–1080, 2012. [14] K. Vollmers, D. R. Frutiger, B. E. Kratochvil, and B. J. Nelson, “Wireless resonant magnetic microactuator for untethered mobile microrobots,” Appl. Phys. Lett., vol. 92, no. 144103, 2008. [15] D. R. Frutiger, K. Vollmers, B. E. Kratochvil, and B. J. Nelson, “Small, fast, and under control: Wireless resonant magnetic micro-agents,” Int. J. Robot. Res., vol. 29, no. 5, pp. 613–636, 2010. [16] S. Keller, D. Haefliger, and A. Boisen, “Fabrication of thin SU-8 cantilevers: Initial bending, release and time stability,” J. Micromech. Microeng., vol. 20, no. 045024, 2010 [17] J. Hammacher, A. Fuelle, J. Flaemig, J. Saupe, B. Loechel, and J. Grimm, “Stress engineering and mechanical properties of SU-8-layers for mechanical applications,” Microsyst. Technol.-Micro-Nanosyst.-Inf. Storage Process. Syst., vol. 14, no. 9–11, pp. 1515–1523, 2008.
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
[18] N. Chronis and L. P. Lee, “Electrothermally activated SU-8 microgripper for single cell manipulation in solution,” J. Microelectromech. Syst., vol. 14, no. 4, pp. 857–863, 2005. [19] T. Fujita, K. Maenaka, and Y. Takayama, “Dual-axis mems mirror for large deflection-angle using SU-8 soft torsion beam,” Sens. Actuat., A. Phys., vol. 121, no. 1, pp. 16–21, 2005. [20] D. Bachmann, B. Schoberle, S. Kuhne, Y. Leiner, and C. Hierold, “Fabrication and characterization of folded SU-8 suspensions for mems applications,” Sens. Actuat., A. Phys., vol. 130, pp. 379–386, 2006. [21] O. Ergeneman, P. Eberle, M. Suter, G. Chatzipirpiridis, K. M. Sivaraman, S. Pane, C. Hierold, and B. J. Nelson, “An in-plane cobalt-nickel microresonator sensor with magnetic actuation and readout,” Sens. Actuat. A, Phys., vol. 188, pp. 120–126, 2012. [22] Z. Nagy, R. I. Leine, D. R. Frutiger, C. Glocker, and B. J. Nelson, “Modeling the motion of microrobots on surfaces using non-smooth multi-body dynamics,” IEEE Trans. Robot. Autom., vol. 28, no. 5, pp. 1058–1068, Oct. 2012. [23] D. R. Frutiger, “Magmites design, fabrication, and control of wireless resonant magnetic micromachines for dry and wet environments,” Ph.D. dissertation, ETH Zurich, Switzerland, 2010
Authors’ photographs and biographies not available at the time of publication.
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Polymer-Based Wireless Resonant Magnetic Microrobots Hsi-Wen Tung, Massimo Maffioli, Dominic R. Frutiger, Kartik M. Sivaraman, Salvador Pan´e, and Bradley J. Nelson, Fellow, IEEE
Abstract—This paper presents a new generation of wireless resonant magnetic microactuator (WRMMA) type microrobot, named PolyMite. A polymer is used as a spring material in the design of the PolyMite. Compared with the previous generation of such devices, the use of a polymer, i.e., SU-8, as a spring material enables the design of a stable spring system without a significant increase in stiffness. A speed of over 20 mm/s (40 body lengths per second) was measured when the PolyMite was driven in air. Micromanipulation abilities of the PolyMites in wet environments have also been demonstrated. The agents were capable of transporting microobjects such as light polystyrene beads and heavier glass beads in water. The large output force, remote actuation, and biocompatible outer surface (SU-8 and gold) make the PolyMite an exciting candidate for biomedical applications. Index Terms—CoNi alloy, micromanipulation, SU-8, wireless mobile microrobots, wireless resonant magnetic microactuators (WRMMA).
I. INTRODUCTION OWER transmission is one of the primary challenges for wireless controlled microrobots; due to the unfavorable scaling of energy storage at the microscale, the optimal strategy is to design these devices to harvest energy from their environment. Strategies to harvest energy are mainly focused on electrostatic force [2], [3], thermal-induced deformations of the structures [4], chemical fuel from the environments [5]–[7], and magnetic force or torque [8]–[13]. Magnetic fields have emerged as the most favorable option, especially for biomedical applications, because they are capable of penetrating nonmetallic materials with little or no interaction and are nearly harmless to most living organisms. In our previous research, we have developed a concept called wireless resonant magnetic microactuators (WRMMAs) [14], [15]. The model of the WRMMA can be simplified as a twomass-spring oscillator system (see Fig. 1). One mass called the “body” stands on the substrate, while the other mass called the
P
“hammer” is supported by the spring and suspended above the substrate. Both the body and the hammer are made of a softmagnetic material. Because of magnetic shape anisotropy, the masses are magnetized in the direction of the easy axis (the long axis) of the combined magnetic masses when exposed in an external magnetic field. This magnetization induces an attractive force between the two masses to pull them toward each other, and a torque to rotate a WRMMA agent to align its easy magnetic axis with the field. When the magnetic field is turned OFF, the attraction force vanishes and the restoring force of the spring pushes the two masses away. By placing the device in an oscillating magnetic field whose frequency is close to the resonant frequency of the device, the response of the body and the hammer to these forces can be magnified to generate a net displacement of the device A microrobot based on the WRMMA concept can be driven on a flat, nonengineered substrate, in dry or wet environments. A precursor to this microrobot, called MagMite, can move controllably at high speeds when driven in air (12.5 mm/s or 42 body lengths per second), and the force it generates is strong enough to push micron-sized objects such as glass beads and metal disks. MagMite can be potentially used for sorting and extracting delicate biological and biochemical samples. However, the fabrication process which consists of four-layer stacks of electroplated metals totaling up to 70 μm, including one layer of 25 μm thick high-aspect-ratio gold spring, is very expensive, complex, and laborious. We introduce here our new microrobots that are produced by an easier, faster, and cheaper fabrication process in the Section II. The details of the fabrication process are presented in the Section III. The characterization of the new agents, including a more reliable motion thanks to the new design of the spring system, and the micromanipulation abilities of the new agents are discussed in Section IV. II. DESIGN AND MODELING
Manuscript received April 22, 2013; revised September 16, 2013; accepted October 9, 2013. Date of publication November 13, 2013; date of current version February 3, 2014. This paper was recommended for publication by Guest Editor A. Ferreira upon evaluation of the reviewers’ comments. An early version of this work was described the 2012 Proceedings of the IEEE International Conference on Robotics and Automation. This work was supported by the Swiss National Science Foundation and ETH Zurich. The authors are with the Institute of Robotics and Intelligent Systems, ETH Zurich CH-8092, Switzerland (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]). This paper has supplementary downloadable material available at http:// ieeexplore.ieee.org. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TRO.2013.2288514
A. Design There are two important characteristics to be considered when designing a new WRMMA microrobot: eigenfrequencies and eigenmodes of the agent, and the displacement of the hammer induced by a magnetic field. Limited by the working range of the magnetic field generator in the lab, the eigenfrequency of the eigenmode that causes the propulsion of the agent should be less than 10 kHz, and the displacement of the hammer under a uniform magnetic field of 8 mT should be sufficient to cause a net motion when driving the WRMMA agent at resonant frequency. The stiffness of the spring system of a WRMMA
1552-3098 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
27
Fig. 1. (a) Schematic diagram of the physical model of WRMMAs. (b) When exposed to a magnetic field, the two soft-magnetic masses are magnetized. Attractive interbody force is induced and pulls the two masses close to each other. (c) When the magnetic field is removed, the interbody force vanishes and the restoring force of the spring pushes the two masses away.
SU-8, which is an epoxy-based polymer, was used due to its low Young’s modulus (a few gigapascal), the ease of patterning it with photolithography, and its suitability for high-aspect-ratio structures [17]–[20]. Electrodeposited cobalt–nickel (CoNi) alloy was used to form the soft ferromagnetic masses due to its high magnetic permeability, high magnetic saturation, and low remanence (a saturation magnetization of 0.85 T and a coercivity of 20 Oe) [21]. The high magnetic permeability allows the PolyMite to generate higher interbody forces in a stronger external magnetic field. The low remanence reduces the residual magnetic force acting to the spring after the magnetic field is OFF, and thus, it maintains a consistent behavior of the PolyMite following exposure to strong magnetic fields. B. Modeling
Fig. 2. PolyMite with the thickness of 50 μm, and a cross-section from A-A shows the dimple feet.
microrobot dominates these characteristics. Due to the high Young’s Modulus of gold, only one gold spring was used on the MagMite to keep the stiffness in a reasonable range. However, a single spring system was too weak to constrain the motion of the hammer that resulted in the drift motion of the MagMite. The PolyMite, a new type of WRMMA microrobot, has been developed based on a design that uses polymer for the springs. The illustration of the design of the PolyMite is shown in Fig. 2. Polymers are suitable for flexible components in microactuators because of their low Young’s modulus [16]. Young’s modulus of a polymer is more than one order of magnitude smaller than that of gold. By replacing gold with polymer as the material of the spring system, the PolyMite is able to have two symmetric springs to connect the body and the hammer. Thus, the movement of the hammer is more stable in the lateral direction. The magnetic masses are present on the same plane as the springs, thus ensuring that the magnetic attraction forces and the restoration forces of the spring are in the same plane. This is in contrast with the older MagMite devices, where the presence of the magnetic masses and the spring in different planes created a bending moment on the device. The body stands on a small dimple feet, while the hammer is suspended above the substrate. The dimple feet reduces the contact area between the PolyMite and the substrate and, thus, reduces friction. Additionally, the curved shape of the PolyMite is helpful to release objects from the agent during micromanipulation tasks.
The interbody force induced due to applied magnetic fields, the stiffness of the springs, and the resonant frequencies of the PolyMites were investigated by finite-element analysis using COMSOL Multiphysics. A standard design of the PolyMites had a diameter of 500 μm and a thickness of 50 μm. The dimensions of the two magnetic masses, the body and the hammer, were 225 μm × 170 μm × 50 μm and 170 μm × 170 μm × 50 μm, respectively. The distance between the two masses was 45 μm. The attraction force induced between the body and the hammer for different magnetic flux densities was simulated, and the stiffness of the spring system was designed in a range that allows tens of nanometers of relative displacement of the two masses at the initial state. The springs were designed with the dimensions 1272 μm (unfolded length) ×10 μm × 50 μm for each side. The resonant frequencies were investigated for a tethered PolyMite. The major resonant mode driving the PolyMite was the third resonant mode (the in-plane compression mode) at 5.6 kHz, which is far away from the first resonant mode (out-ofplane waving) and the second mode (in-plane shifting) that were at around 2.1 kHz (see Fig. 3). For an untethered PolyMite sliding on a substrate, the resonant frequency of the two mass-spring √ oscillator increased by a factor of 1 + α, where α = mH /mB is the mass ratio of the body and the hammer [22]. From this calculation, the resonant frequency of an untethered PolyMite was predicted to be around 7.4 kHz. III. FABRICATION PROCESS To combine a polymer and a ferromagnetic metal on the same platform, a new fabrication process was developed that connects the two materials without any adhesion problems between them.
28
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Fig. 3. FEM-derived resonant modes of a tethered PolyMite. The first and the second modes exist around 2.1 kHz, while the third mode is found at 5.7 kHz. Different colors indicate relative displacement of each element in the structure, where dark blue stands for zero displacement and dark red represents maximum displacement.
A single-side polished silicon wafer is used as a substrate of the process. A silicon dioxide (SiO2 ) layer is used as a hard mask for dimple holes etching process. The etching windows are patterned on the SiO2 layer by photolithography and reaction-ionetching (RIE) process [see Fig. 4(a)]. After removing the photoresist, the wafer is immersed in potassium hydroxide (KOH) solution to etch pyramidal dimple holes into the silicon wafer [see Fig. 4(b)]. The SiO2 hard mask is removed after the etching process [see Fig. 4(c)], and another thin layer of SiO2 is deposited on the wafer as a sacrificial layer [see Fig. 4(d)]. The electrodes are deposited by evaporating titanium (Ti)/gold (Au) bilayer on the substrate and patterned by the lift-off method, thus rendering only a part of the substrate electrically conductive [see Fig. 4(e)]. SU-8 2025 is applied onto the substrate and patterned into the shapes of springs, frames for electrodeposited metals, and tethers [see Fig. 4(f)]. The patterned wafer is then immersed into different electrolytes to deposit CoNi alloy and Au layers by electrodeposition. The metals selectively deposit on the patterned electrodes and the shapes are defined by the SU-8 frames [see Fig. 4(g)]. Unlike most current fabricating methods, where magnetic metals and nonmagnetic structures are fabricated separately before assembling them, metals in our process are selectively electrodeposited within the polymer frames of the device structure. Therefore, the issue of assembly is solved and the magnetic bodies are well aligned with the nonmagnetic structures. The CoNi microstructures exhibit similar thickness as the SU-8 ones, while the Au layer is less than 1 μm thick. The Au layers protect the CoNi alloy from being damaged in the releasing solution in the following step. To release the devices from the substrate, the wafer is immersed in buffered hydrofluoric acid to etch SiO2 and the evaporated titanium away [see Fig. 4(h)]. Finally, the devices are cut from their tethers using a microlaser milling tool [see Fig. 4(i)]. Fig. 5 shows a released PolyMite together with micro polystyrene beads. Due to the outer surfaces being made of biocompatible materials, namely SU-8 and Au, the devices are suitable for biomedical applications.
Fig. 4. Fabrication flow of the PolyMite. (a) Thin layer of SiO2 is grown on the Si wafer as a hard mask for etching and the etching windows of dimple holes are patterned on the SiO2 layer. (b) Dimple holes are etched anisotropically into the silicon wafer by KOH solution. (c) SiO2 layer is etched away. (d) Another layer of SiO2 is deposited as the sacrificial layer. (e) Ti 10 nm/Au 200 nm layers are deposit by evaporation and patterned to form the electrodes. (f) SU-8 is patterned to form springs, frames of electroplated metals, tethers, etc. (g) CoNi alloy and Au are deposited by electroplating. (h) Wafer is immersed in the buffered hydrofluoric acid to etch the sacrificial layer away and release the whole structures from the Si wafer. (i) Microlaser milling tool is applied to clean and release the PolyMites from their tethers.
Fig. 5. Fully released PolyMite with polystyrene beads of approximately 120 μm in diameter.
IV. EXPERIMENTAL RESULTS The control system of the PolyMites consists of two pairs of Helmholtz coils nested orthogonally to generate a uniform
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
29
TABLE I DESIGN PARAMETERS OF DIFFERENT TYPES OF POLYMITES
Fig. 6.
Helmholtz coils utilized to drive the PolyMites. Fig. 7. Frequency response of three different designs of PolyMites. Speeds of more than 20 mm/s were measured when driving agents at 4 mT.
magnetic field at their center in a direction parallel to the surface of manipulation. The custom-made current control electronics send a rectangular waveform signal to the coils to generate oscillating magnetic field. The orientation of the field is set by linear superposition of the two orthogonal magnetic fields (see Fig. 6). A computer vision tracker was built to trace the motion of the PolyMites from the images captured from an overhead camera. The refresh rate of the camera was set to 60 fps. A. Characterization of PolyMites The motion of the PolyMites is primarily controlled by the frequency and the magnitude of the magnetic field, and the orientation of the PolyMites is changed by changing the direction of the magnetic field. Three different designs of the PolyMites were tested. The agents differ from each other in the spring length, distance between the two magnetic masses, and the weight of the two masses. The details of the difference in the three PolyMites are listed in Table I. Fig. 7 shows the frequency sweep of the three PolyMites on a Ti-coated Si wafer in air. The M1A01 agent was the standard design used to model the PolyMite. The peak of response was at 7.3 kHz, very close to our prediction from the simulation result. The maximum speed of the M1A01 agent was measured from the data collected by the tracker as over 20 mm/s at 4 mT, which is equal to 40 times the body length per second. The M1B04 agent had the same geometry of the springs as the M1A01 agent, but the weight of the two magnetic masses is about 40% heavier. This caused the resonant frequency of the M1B04 to shift 1.2 kHz lower. Additionally, a minor motion caused by
the first and the second resonant mode of the PolyMites was observed at around 2 kHz on the frequency response plots of the M1A01 and M1B04 agents, respectively. The M1B05 agent had stiffer spring and less weight of the two magnetic masses than the M1B04 agent, and therefore its resonant frequency was higher. Moreover, the bandwidth of the frequency response of the M1B05 agent was nearly a half of the bandwidths of the M1A01 and M1B04. The shift of the resonant frequency of individual agents and narrow bandwidth of the frequency response are favorable in multiagent control by driving agents with signals composed of different frequencies. The response of the PolyMites to the magnitude of the magnetic field was measured at a frequency slightly shifted from the resonant frequency of each agent because the speed of the agents was too fast to be tracked at high magnitude of the magnetic field at resonant frequency. Fig. 8 depicts the response of the M1B04 and M1B05 agents to the magnitude of the magnetic field. The M1B04 agent started to move at 5 mT and the M1B05 agent started at 7 mT. Some movement was observed at lower fields but is excluded in the figures because the agents were more sensitive to asperities and cleanliness of the substrates at lower fields and got stuck on the substrates easily. Excluding these measurements, the speed of the agents increases linearly with the magnitude of the magnetic field. The M1B05 agent required higher energy to overcome the friction and to generate a smooth net motion because of its stiffer spring and the larger distance between the two masses. This explains why the bandwidth of the M1B05 agent is narrower.
30
Fig. 8. Response of (a) the M1B04 agent and (b) the M1B05 agent to the magnitude of the magnetic field.
As mentioned in Section II-A, the PolyMite uses a spring system consisting of two symmetric springs that provides equal support on both lateral sides of the hammer and increases the stability of the in-plane motion of the hammer. The improvement in stability was evaluated by comparing the drift motion of the PolyMite and the MagMite. The drift angle is defined as the angle between the forward orientation of the agent and the direction of the movement. Fig. 9 shows the trajectories of a MagMite and a PolyMite tested under different frequencies. The direction of the driving field was along the x-axis. The drift angles of the MagMite varied with the exciting frequencies, and the agent moved sideways at certain frequencies [23]. Although drift was observed in the motion of the PolyMite, it was significantly lower than that of the MagMite. The drift motion of the PolyMite may come from some defects introduced during the fabrication, small particles stuck to the bottom of the agents that changed the contact to the substrate, or other system setup issues. The drift motion is not a serious issue since the direction of the motion can be corrected easily by manual or PID control. The wear of the dimple feet of the PolyMite can change the friction between the PolyMite and the substrate. One PolyMite was driven for an hour in air, and its dimple feet were inspected
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
Fig. 9. Trajectories of (a) a MagMite driven by frequency sweep 5.1–5.7 kHz [23] and (b) a PolyMite driven by frequency sweep 1–9 kHz. The orientation of the magnetic field is pointed in the x-direction. The drift motion of the PolyMite is improved compared with the motion of the MagMite thanks to the symmetric springs of the PolyMite that stabilize the in-plane motion of the hammer.
in the scanning electron microscope. There was no significant difference in the shape and height of the dimple feet between the tested agent and another new agent from the same batch of fabrication process. Therefore, the wear of the dimple feet was not considered in our experiments. B. Micromanipulation Micromanipulation in wet environments is of general interest to many scientists since most biomolecules function in a liquid medium. To demonstrate the abilities of micromanipulation in wet environments, the PolyMite was driven by manual control to push microbeads in water. Fig. 10 shows a PolyMite collecting polystyrene beads of diameter 105–125 μm one by one to the left corner of the stage. (In addition, see the supplementary material S1.) Thanks to the curved contour of the PolyMites, polystyrene beads were released from the PolyMite easily by rapidly turning the agent. In the rare cases, where the bead remained stuck to the PolyMite, the drag force generated by the quick turning motion
TUNG et al.: POLYMER-BASED WIRELESS RESONANT MAGNETIC MICROROBOTS
31
Fig. 10. Image sequence showing how a PolyMite manipulates polystyrene beads with a diameter of 105–125 μm in water. The agent collects randomly located beads on the right and moves them one by one to the left corner of the stage. The numbers labeled identify individual beads, and dash lines indicate the trajectory of the PolyMite.
Fig. 11. Image sequence showing how a PolyMite pushes a glass bead with a diameter of 60 μm in water on a structured substrate. The PolyMite pushes the glass bead toward the hole at the edge of the stage. After the bead falls down into the hole, the agent turns and departs. The circles mark the position of the pushed bead, and dash lines indicate the trajectory of the PolyMite.
was strong enough to break the adhesion between the agent and the bead. Heavier samples such as glass beads could also be manipulated by the PolyMite. Fig. 11 depicts a PolyMite driven on a structured substrate with a hole at the edge of the stage. A glass bead with a diameter of around 60 μm was transported by the agent to the top of the hole. After the force of gravity caused the bead to fall into the hole, the PolyMite was driven away (see supplementary material S2). V. CONCLUSION A new generation of WRMMA type microrobot—the PolyMite—has been developed. The design of the PolyMite reduces the time, the cost, and the complexities of the fabrication process of its predecessor, i.e., the MagMite, with no loss in performance. The outer surfaces of the PolyMite consist of gold and SU-8 which are inert to most chemicals and safe for use in biomedical applications. The fabrication process of the PolyMite combines polymer and ferromagnetic material with no problem of adhesion and alignment between them. The PolyMite is capable of being operated on a plane surface in both dry and wet environments. A speed of over 20 mm/s (40 body lengths per seconds) was reached when the agent was driven in air on a titanium-coated silicon wafer with an applied magnetic flux density of 4 mT. The new design of the symmetric spring system reduces drift motion. The PolyMite is capable of transporting micropolystyrene beads and microglass beads in water to a predefined location. In future, we envisage the PolyMite
selecting or extracting micron-sized samples that are not easy to be handled by human hands. It can be a group of microorganisms or protein crystals on a stage, and the PolyMite can push the selected sample toward a hole or another macro-sized tool for extraction.
REFERENCES [1] H. W. Tung, D. R. Frutiger, S. Pane, and B. J. Nelson, “Polymer-based wireless resonant magnetic microrobots,” in Proc. IEEE Int. Conf. Robot. Autom., 2012, pp. 715–720. [2] B. R. Donald, C. G. Levey, C. D. McGray, I. Paprotny, and D. Rus, “An untethered, electrostatic, globally controllable MEMS micro-robot,” J. Microelectromech. Syst., vol. 15, no. 1, pp. 1–15, 2006. [3] B. R. Donald, C. G. Levey, and I. Paprotny, “Planar microassembly by parallel actuation of MEMS microrobots,” J. Microelectromech. Syst., vol. 17, no. 4, pp. 789–808, 2008. [4] O. J. Sul, M. R. Falvo, R. M. Taylor, S. Washburn, and R. Superfine, “Thermally actuated untethered impact-driven locomotive microdevices,” Appl. Phys. Lett., vol. 89, no. 203512, 2006. [5] S. Fournier-Bidoz, A. C. Arsenault, I. Manners, and G. A. Ozin, “Synthetic self-propelled nanorotors,” Chem. Commun., pp. 441–443, 2005. [6] Y. Wang, S. T. Fei, Y. M. Byun, P. E. Lammert, V. H. Crespi, A. Sen, and T. E. Mallouk, “Dynamic interactions between fast microscale rotors,” J. Amer. Chem. Soc., vol. 131, no. 29, pp. 9926–9927, 2009. [7] A. A. Solovev, Y. F. Mei, and O. G. Schmidt, “Catalytic microstrider at the air-liquid interface,” Adv. Mater., vol. 22, no. 39, pp. 4340–4344, 2010. [8] J. J. Abbott, O. Ergeneman, M. P. Kummer, A. M. Hirt, and B. J. Nelson, “Modeling magnetic torque and force for controlled manipulation of softmagnetic bodies,” IEEE Trans. Robot., vol. 23, no. 6, pp. 1247–1252, Dec. 2007. [9] L. Zhang, J. J. Abbott, L. X. Dong, B. E. Kratochvil, D. Bell, and B. J. Nelson, “Artificial bacterial flagella: Fabrication and magnetic control,” Appl. Phys. Lett., vol. 94, no. 064107, 2009.
32
[10] A. Ghosh and P. Fischer, “Controlled propulsion of artificial magnetic nanostructured propellers,” Nano Lett., vol. 9, no. 6, pp. 2243–2245, 2009. [11] C. Pawashe, S. Floyd, and M. Sitti, “Modeling and experimental characterization of an untethered magnetic micro-robot,” Int. J. Robot. Res., vol. 28, no. 8, pp. 1077–1094, 2009. [12] S. Martel, M. Mohammadi, O. Felfoul, Z. Lu, and P. Pouponneau, “Flagellated magnetotactic bacteria as controlled MRI-trackable propulsion and steering systems for medical nanorobots operating in the human microvasculature,” Int. J. Robot. Res., vol. 28, no. 4, pp. 571–582, 2009. [13] L. Zhang, T. Petit, K. E. Peyer, and B. J. Nelson, “Targeted cargo delivery using a rotating nickel nanowire,” Nanomed.-Nanotechnol. Biol. Med., vol. 8, no. 7, pp. 1074–1080, 2012. [14] K. Vollmers, D. R. Frutiger, B. E. Kratochvil, and B. J. Nelson, “Wireless resonant magnetic microactuator for untethered mobile microrobots,” Appl. Phys. Lett., vol. 92, no. 144103, 2008. [15] D. R. Frutiger, K. Vollmers, B. E. Kratochvil, and B. J. Nelson, “Small, fast, and under control: Wireless resonant magnetic micro-agents,” Int. J. Robot. Res., vol. 29, no. 5, pp. 613–636, 2010. [16] S. Keller, D. Haefliger, and A. Boisen, “Fabrication of thin SU-8 cantilevers: Initial bending, release and time stability,” J. Micromech. Microeng., vol. 20, no. 045024, 2010 [17] J. Hammacher, A. Fuelle, J. Flaemig, J. Saupe, B. Loechel, and J. Grimm, “Stress engineering and mechanical properties of SU-8-layers for mechanical applications,” Microsyst. Technol.-Micro-Nanosyst.-Inf. Storage Process. Syst., vol. 14, no. 9–11, pp. 1515–1523, 2008.
IEEE TRANSACTIONS ON ROBOTICS, VOL. 30, NO. 1, FEBRUARY 2014
[18] N. Chronis and L. P. Lee, “Electrothermally activated SU-8 microgripper for single cell manipulation in solution,” J. Microelectromech. Syst., vol. 14, no. 4, pp. 857–863, 2005. [19] T. Fujita, K. Maenaka, and Y. Takayama, “Dual-axis mems mirror for large deflection-angle using SU-8 soft torsion beam,” Sens. Actuat., A. Phys., vol. 121, no. 1, pp. 16–21, 2005. [20] D. Bachmann, B. Schoberle, S. Kuhne, Y. Leiner, and C. Hierold, “Fabrication and characterization of folded SU-8 suspensions for mems applications,” Sens. Actuat., A. Phys., vol. 130, pp. 379–386, 2006. [21] O. Ergeneman, P. Eberle, M. Suter, G. Chatzipirpiridis, K. M. Sivaraman, S. Pane, C. Hierold, and B. J. Nelson, “An in-plane cobalt-nickel microresonator sensor with magnetic actuation and readout,” Sens. Actuat. A, Phys., vol. 188, pp. 120–126, 2012. [22] Z. Nagy, R. I. Leine, D. R. Frutiger, C. Glocker, and B. J. Nelson, “Modeling the motion of microrobots on surfaces using non-smooth multi-body dynamics,” IEEE Trans. Robot. Autom., vol. 28, no. 5, pp. 1058–1068, Oct. 2012. [23] D. R. Frutiger, “Magmites design, fabrication, and control of wireless resonant magnetic micromachines for dry and wet environments,” Ph.D. dissertation, ETH Zurich, Switzerland, 2010
Authors’ photographs and biographies not available at the time of publication.
