18-06-2021 | | By Sam Brown
Recently, MIT announced that it had successfully embedded sensor circuits consisting of multiple ICs and connectors into fibres suitable for clothing. What did the researchers demonstrate, how does the demonstration solve challenges faced by wearable electronics, and could it be the key to wearable electronics?
Recently, researchers from MIT demonstrated how micro-electronics can be embedded into material fibres. They start by placing chip-scale devices (i.e. chips only), and other electronics onto a preform which then has four tungsten wires placed in a line going across each chip. From there, the entire setup is heated such that the preform melts, and this setup is then drawn into a fibre. At the same time, the tungsten wires are pressed and bonded to the chips creating a 4-wire I2C bus with two carrying power, one carrying clock, and the last carrying data.
The resulting wire (which has the electronics embedded inside it), can be bent around a radius of 12mm with no damage to the internal electronics. In contrast, damage is only observed when bending at a radius less than 3mm. Furthermore, 90% of the chips in the fibre successfully bonded to the tungsten wires, and multiple circuits have been demonstrated to function.
Examples of ICs that have been successfully embedded into the fibre include a Microchip 24CW1280X EEPROM, and a Maximum MAX31875 temperature sensor, while the end of the wire was terminated to an STM32F401. One example developed by the researchers was a length of fibre with multiple EEPROMs connected in parallel which had a memory density of 767kbit per meter. The researchers demonstrated the fibres ability to store information by saving a 480KB file for two months without power.
Another example demonstrated by the researchers was integrating multiple temperature sensors into fibres which were then sewn into a participants shirt. Various temperature readings were made from all over the body during exercise, and this data was later fed into a neural network. Running the trained neural network back onto the shirt enabled for reliable detection of various exercises showing how the shirt could be used to track activities accurately.
One of the greatest challenges faced by wearable electronics is the need for flexibility, and this flexibility is greatly tied to comfortability. Because electronics are mostly ridged by design, this means that they cannot bend or change shape to fit a particular surface. Most wearable electronics consist of a ridged circuit connected to flexible straps (such as a smartwatch).
The threads developed by MIT can be sewn into the material and worn without any notice. This enables everyday clothing to be integrated with electronics unknowingly to the users and enables comfortable wearable electronics. Furthermore, the threads designed by MIT are more than a few trivial sensors being I2C driven off-the-shelf chips with memory and other advanced capabilities. This enables clothing made with the threads to have real computational capabilities and thus demonstrate the possibility of commercial devices in the future.
While many researchers are showing promising results in the development of wearable technologies, the research conducted by MIT researchers is arguably the most promising. Flexible displays made by other projects are often low-resolution and can only be in laboratory conditions while other flexible circuits using printed transistors have the processing capability of a 7400 logic chip.
However, the threads developed by MIT are using off-the-shelf chips that use I2C for communication, meaning that these threads have far more processing capabilities right off the bat. Furthermore, the I2C protocol could be replaced with a differential bus that could allow for higher data transmission, and if made fast enough, could enable the creation of woven displays. This would enable a piece of clothing to have an entire smart device integrated into its threads, and the use of off-the-shelf technology makes the development of such systems practical.
The researchers now need to focus on is embedding a processor into the threads. Having sensors and I2C devices is all fine and dandy, but integrating a processor into the threads would remove the need for an external processing unit that is bulky and uncomfortable. Of course, one solution around this could be to integrate the processor into the clothing tag, which would also house a small coin-cell battery, but until then, these threads will remain attached to the laboratory.