Silicon Microrobots

The solar-powered 10 mg silicon microrobot was used to demonstrate that an autonomous robot could be built from silicon fabrication technologies. The project was led by Seth Hollar (graduated 2003) from Pister's lab with Anita Flynn from Micropropulsion Corp. The advantage of such a fabrication scheme allows the potential for building many microrobots at a relatively low cost. Most of the information on this web site was taken from this paper. Other contributing researchers were Colby Bellew and Sarah Bergbreiter.

Building the Microrobot

The microrobot was built in three separate pieces. While a unified process that combined electronics and MEMS together seemed appealing at first glance, the time and energy required for such a development would have been beyond the limit of a few researchers. We, therefore, developed a robot design that would be fabricated from three separate chips with a final assembly step. The first piece included the legs, motors, and frame of the robot; the second included the solar cells and high voltage buffers while the third was the digital logic chip for controlling the walking motion.

Two one-DOF (degree-of-freedom) legs are used in a coordinated manner to create forward motion. A simulation of the walking can be seen here.

The first piece of the robot (legs, motors, and skeleton) took advantage of two-key developments. First, in designing microrobots we needed high force actuators that operated at low power. Richard Yeh (previously at Berkeley) developed an electrostatic inchworm motor that could be fabricated in a single mask SOI process. By using electrostatic gap-closing actuators, inchworm motors combine large forces with large displacements by continually repeating small steps much like how a person climbs a ladder. The motors were fabricated using a high-aspect-ratio Bosch process silicon etch in a 50 mm device layer SOI wafer.

The second development involved combining the inchworm motors with compliant structures to create out-of-plane motion. We developed the SOI / Poly Process in which electrostatic inchworm motors in thick SOI can be fabricated with thinner polysilicon structures on top. A key challenge of the process was adding structural polysilicon after the primary Bosch process SOI etch. This was accomplished using Azzam Yasseen's method of thick glass planarization (JMEMS paper from when he was at Case Western Reserve University). By using glass frit mixed with an organic transport, a slurry is created that can be spun on a wafer much like how photoresist is spun on. By firing the glass frit at 890 degrees Celsius, the glass coalesces and forms a single network filling in the deep trenches. Afterwards, the surface is CMP'd to the silicon interface, yielding a perfectly flat surface ready for surface micromachining. Two structural polysilicon layers are then added creating polysilicon pin-microhinges for the robot. With such a process, we designed and fabricated the frame of the robot along with its legs, hinges, and inchworm motors.

For the second piece of the robot, we needed a power supply to run the robot autonomously. To overcome this obstacle, we used a solar-cell process developed by Colby Bellew (formerly at Berkeley). The process included isolation trenches which allowed the electrical isolation of blocks of silicon on an SOI wafer. Using this process, we could connect solar cells in series, effectively adding the voltages of individual cells. In all, 90 solar cells were connected in series, generating approximately 50 V in an area no large than 2 mm2. Finally, CMOS was integrated into this 8-mask process to create high voltage switching-buffers for the high voltage supply.

The last piece of the robot was fabricated by National Semiconductor Corporation in their 0.25 micron, 5-metal-layer process. This chip was a simple digital sequencer that included the logic to create the walking motion of the robot.

Once fabricated, the three chips were assembled together and wirebonded. Because the robot had to be able to get up and walk, we had to use assembly methods that did not require permanently fixing the chips to die packages, as is done in traditional wirebonding processes. Using wafer mounting wax we were able to bond the chips to substrates temporarily to wirebond. The chips were subsequently released by dissolving the wax in acetone.

Autonomous Mobility

The assembled robot weighed no more than 10 mg and during operation consumed 2.6 microwatts of power. The robot lifted itself successfully and went through walking motions for approximately 250 cycles (videos click here). Instead of going forward, however, the robot would slowly veer to the right as its feet slipped on the surface, shuffling to the side a few millimeters. A fault in the motor design limited the overall output force, which yielded sporadic results in the robots mobility. However, future motor designs can substantially increase the reliability of the legs. With a few improvements on the design, the robot could attain forward speeds up to 1.3 mm/s.

Most importantly, however, we have established a firm foundation for launching new microrobot designs. Recently demonstrated two-DOF legs (see figure) can be combined with microprocessor-based intelligence, sensors, and communication to create insect-sized microrobots.

Furthermore, the demonstration of one autonomous mobile microrobot begs the question of "what next?" With silicon fabrication technologies, the possibility of making one robot easily extends to the fabrication of many robots. With a community of robots in mind, much like distributed sensor networks, we can imagine a plethora of future applications.

Microrobot Applications

Let us imagine, for the moment, future microrobots: ones equipped with a choice of sensors, behavioral processing capabilities, and reasonable communication systems. With not just one such robot but a whole colony of robots, the plethora of design variations and applications instantly becomes apparent. While microrobots can operate in a haphazard manner monitoring the environment or performing search and rescue missions, they can also work as a cooperative community to perform high level tasks. Much like a community of ants building a nest, microrobots could work together to form structures or perform tasks that the individual robots could not do themselves. An application would be the deployment of microrobots to the surface of the planet Mars where they gather environmental data and construct a satellite antenna (see figure below).

Honestly told, however, while Mars exploration is a fascinating application for microrobots, there are many other applications that have yet to be imagined. Much like with the invention of the cell-phone and the photo copier, which were paradigm shifts in their own right, microrobots, too, could espouse a new way of interacting with the world. While my colleagues and I have demonstrated a microrobot from a 3-chip hybrid; cheap, disposable robots will come when the robot is fully fabricated all in a monolithic die, costing but a few pennies.

More Information

For all the videos click here.

For SEM pictures click here.

Related Publications

S. Hollar, A Solar-Powered, Milligram Prototype Robot from a Three-Chip Process, Dissertation, UC Berkeley, June 2003: PDF color version (18MB) or black and white (4MB)

S. Hollar, S. Bergbreiter, K.S.J. Pister, "Bidirectional Inchworm Motors and Two-DOF Robot Leg Operation," Transducers 2003, Boston, June 2003. (paper.pdf) (presentation.pdf)

C. Bellew, S. Hollar, K.S.J. Pister, "An SOI process for fabrication of integrated solar cells, transistors and electrostatic actuators," Transducers 2003, Boston, June 2003. (paper.pdf) (presentation.pdf)

S. Hollar, A. Flynn, C. Bellew, K.S.J. Pister, "Solar Powered 10mg Silicon Robot," MEMS 2003, Kyoto, Japan, January 19-23, 2003. (paper pdf) (presentation pdf)

R. Yeh, S. Hollar, K.S.J. Pister, "Single mask, large force, and large displacement electrostatic linear inchworm motors," JMEMS, pp.330-36, August, 2002.

S. Hollar, A. Flynn, S. Bergbreiter, K.S.J. Pister, "Robot leg motion in a planarized-SOI, 2 poly process," Hilton Head 2002 Workshop, Hilton Head Island, S.C., June 2-6, 2002. (paper pdf) (presentation pdf)

R. Yeh, R.; Hollar, S.; and Pister, K.S.J. "A single mask, large force, and large displacement electrostatic inchworm motor," MEMS 2001, Interlaken, Switzerland, pp.260-64, Jan. 21-25, 2001. (pdf)

Comments? Questions? Send email to: shollar@alum.mit.edu

Last Updated: 5/2011

Seth Hollar