Researchers Put Electronic “Brains” Onboard Walking Microbots
While micrometer-scale robots are usually controlled externally, Cornell's prototypes include a circuit onboard the autonomous microrobots.
A group of researchers at Cornell University have successfully installed "brains" onboard an intelligent microscale robot powered by solar energy.
The silicon wafer containing the finished CMOS product, the "brains" of the microbot. Image used courtesy of Cornell University
These microbots measure from 100 to 250 micrometers—nearly the width of a single strand of human hair. While robots this size are usually controlled externally, Cornell's prototypes include an onboard circuit to control movements.
Creating a Wireless Walking Robot
In recent years, researchers at Cornell have developed microscopic machines that can crawl, walk, swim, and form bends in their shape. These microbots' movements were dictated by wires or focused laser beams, however.
Itai Cohen, a professor of physics in the college of arts and sciences at Cornell, brought together a group of researchers to find ways to remove external controls so the microbots could move freely on their own. In the past, researchers struggled to scale microrobots to micron levels without wire bonding and multichip stacking. Typical CMOS circuitry includes various wire bonds for different functionalities, but this increases the depth, height, and weight of the IC, making it difficult to scale down. Another major challenge is heterogeneously integrating CMOS ICs and microactuators to become a single, fully-functional robot.
Cornell researchers combined surface electrochemical actuators (SEAs) and silicon photovoltaics (PVs) to overcome these challenges. Once a voltage signal is received from the CMOS circuit, oxygen molecules cling and expand to the layer, causing the actuator to bend the robot’s legs.
The "Brain" and Fabrication of the Microbot
Cornell researchers claim the "brain" of the microbot is a simple CMOS clock circuit that implements an array of transistors, diodes, resistors, and capacitors. This IC generates a signal to produce a series of phase-shifted square wave frequencies that control a robot’s gait. During the testing phase, several microbots could move between 4 to 20 micrometers per second.
Shown above are the different stages of photolithography the team encountered. The finished product allowed the robot’s legs to move independently via exposed platinum ground electrodes. Image used courtesy of Cornell University
The last hurdle the research team overcame was the fabrication process. This process required 13 layers of photolithography, 12 etches, and 11 depositions from 10 different materials. Using Cornell’s Nanoscale Science and Technology Facility (CNF), the researchers finally completed an 8-inch silicon-on-insulator wafer. With this wafer, the microbot's PV power supply and CMOS clock circuit did not interfere with the robot’s leg movement.
Solar Powering the Microbot
With two pairs of photovoltaic-based power supplies, the microbot had dedicated power for leg movement and the clock circuit. The CMOS clock circuit consisted of a relaxation oscillator, which served as a clock output and a frequency driver. The frequency driver comprised a series of D-type flip-flops that converted pulses to a 50% duty cycle, allowing the robot to easily send signals throughout each output pin.
Even with onboard circuitry, Cornell's microbot is only the width of a human hair. Screenshot courtesy of Cornell University
During the testing phase, Cornell researchers discovered with about 1 and 3 kW/m2 of continuous light intensity, the microbot reached 12 μm/s in speed, traveling about 3-body lengths a minute. While the researchers could have increased the speed by pushing the input frequency to 4 Hz, the team kept operating frequencies between 1 and 2 Hz, so the microbot wouldn't slip on flat surfaces.
The Microbot's Potential for Cleanup, Monitoring, and Even Surgery
Cornell researchers believe these microbots could be used for environmental cleanup, medical substance monitoring, and microscopic surgery. For example, these microbots might be used as intelligent robots for wound healing and tissue morphogenesis.
Cornell researchers are looking to incorporate machine learning algorithms, so the microbots can learn and adapt independently. The team hopes to expand on the microbot family with more complex tasks and AI-based algorithms in future studies.