Low-Power Sensors Could Last 10 Years, Providing Surveillance, Security
Converting Chemical to Electronic Signals
The Sandia team collaborated with a professor and graduate student at the University of Virginia to design a readout integrated circuit that can detect changes in the sol-gel and convert them into useful signals while consuming minimal power, Mieko said.
This was important because a readout integrated circuit typically consumes the most power in a sensor system, said Jesse Moody, a Sandia sensor engineer who led the circuit design.
“Essentially, we needed to develop a device that can detect very minute changes in that capacitive sensing film and convert that into useful digital information in an extremely low-power manner,” Jesse said. “That was the main electronics challenge of the project.”
The circuit was designed using structures on the scale of 65 nanometers, 1,500 times smaller than a human hair and three times smaller than the smallest transistor available from Sandia’s Microsystems Engineering, Science and Applications Complex’s fabrication facilities. This size was chosen to allow the sensor system to operate faster while using less power, Mieko said.
The circuit can quickly check the status of 10 chemical sensing channels in a few thousandths of a second, Jesse said. The development of the ultra-low-power readout integrated circuit built on prior LDRD and DOD projects in low-power sensing, he added.
The circuit design was fabricated at Taiwan Semiconductor Manufacturing Co., one of the world’s most advanced microchip fabrication facilities. Typically, it costs about $1 million to have a custom microchip fabricated at such a foundry; however, for this project, the Sandia team shared the initial cost with other companies as part of a multi-project wafer, Mieko said.
Usually, these multi-project wafers are diced, and each company receives only the square millimeter segments — about the size of a sharpened pencil tip — containing their design, Jesse said. However, these tiny pieces are challenging to handle and build on, even when inserted into a slightly larger package with connections called a silicon interposer, Mieko said.
For this project, the researchers had the other companies’ designs removed with a high-power laser and developed an innovative method to fill in the gaps with a polymer, allowing them to work with and build on a full silicon wafer, Mieko said.
Testing the Full Sensor
After resolving issues with the polymer planarization, the team assembled the sensor system. They compared the sensor system constructed via the heterogeneous integration pathway — from the tiny chips in a silicon interposer — with a monolithic sensor where everything was built on the same wafer.
The final monolithic sensor system measured 1 by 1 inch, while the heterogeneous test system was 10 times larger, Mieko said. She added that had the team optimized the heterogeneous sensor system, they could have reduced its size to about twice that of the monolithic sensor.
The monolithic sensor system used 30.9 nanowatts of power per sensing channel and had an area of 43 microns per channel. It ultimately consumed slightly more power per channel than the heterogeneous system due to a power leakage issue that the team didn’t have time to fully resolve, Mieko said. Once they optimize the monolithic system further, it should use less power than the heterogeneous system because direct connections require less power.
Since the proof-of-concept sensor system was successful, the team is exploring additional funding sources to continue developing ultra-low-power, long-lasting chemical sensors, Mieko said. The sensor system could be adapted with other chemically selective materials to detect multiple chemicals of concern in the same device, Philip said.
“The novelty of integrating the low-power microcontroller and the sol-gel sensor was a really cool project to work on,” Philip said. “Doing system-level sensor work is hard and doing it in this innovative way is especially challenging. I think we have a nice path forward to realize a functional device with additional support.”
Multi-project Wafers Made Easy(er)
Multi-project wafers significantly reduce the initial cost of fabricating microchips at external foundries, but they can also introduce challenges. Typically, these wafers are diced, and each company receives only the square millimeter segments containing its design. Handling these tiny pieces during further processing can be difficult, said Mieko Hirabayashi, a Sandia electronics engineer and project lead. For this project, the team used wafers from which other companies’ designs had been removed with a high-power laser.
“The biggest success we had was with the planarization of the multi-project wafer,” Mieko said. “Initially, the project proposal called for a layer of oxide to planarize. But they didn’t fully understand the height that needed to be planarized, and that amount of oxide would have cracked under stress. I spent a lot of time researching different polymers that could give us that layer that we needed for the planarization.”
Mieko discovered that one polymer, SU8, effectively filled the gaps created by laser etching out the other companies’ intellectual property in the commercially produced wafers.
“Thankfully, we got a new tool called the nanoform, which is a purely mechanical processing tool. A mechanical etch works much better for full planarization. It created a very planarized layer, similar to what we have with a regular wafer. That tool made a huge difference.”
The nanoform mechanical processing tool is located in the lab of materials scientist Christian Arrington, and the process was refined by laboratory support technologist Chris St. John, Mieko said.
“There were a couple of tricks we had to implement to make sure we didn’t introduce too much heat at once because you’re working with pretty thick material at that point that could crack or become stressed. That was one of the key developments that allowed us to get it to function.”
Mollie Rappe is Principal Corporate Communications Specialist at Sandia National Laboratories. The article was originally posted to the website of Sandia National Laboratories.
