In recent years, major advances in nanotechnology have inspired some of the most innovative breakthroughs in various fields, including the development of nanomechanical resonators, such as high-precision sensing and quantum network applications.
Now a team of researchers at TU Delft has designed one of the most advanced and precise nanomechanical resonators, drawing inspiration from both the natural and virtual worlds. Inspired by spider webs and guided by machine learning, the team set about creating a web-shaped device that could exhibit vibrational states isolated from surrounding thermal environments.
One of the most sought after properties of a mechanical resonator is noise insulation from thermal environments, namely at room temperature conditions where thermomechanical noise can dominate.
Dr. Richard Norte, TU Delft
Inspired by nature
Spiderwebs are constructed of unique and complex geometries, making them one of the most well-known and exciting structures in nature. Despite being found almost everywhere and everywhere, scientists from various fields, such as materials science, physics, and biology, are still discovering new and emerging features regarding spider web mechanics.
One of the things you know about spider webs is that they are exceptional, robust insulated vibration sensors. Spiders are able to sense their tangled prey through the design of the web. They detect vibrations coming from the grid, not environmental disturbances around it.
To help their work and get started developing the right kind of nanomechanical resonator, the team used machine learning to guide the optimization of the process using the Bayesian optimization algorithm.
In designing spider tissue nanomechanical resonators, Bayesian optimization is expected not only to explore the design space to find new vibration modes that induce soft clamping with a compact design, but also to use them to achieve high-quality low-frequency factors for a given resonator size.
Dr. Richard Norte, TU Delft
While spider webs are a product of millions of years of extremely sophisticated evolution, the Bayesian optimization model accelerated the process and allowed scientists to quickly make their calculations so they could streamline the design process for their ultra-sensitive device.
The resulting design model was a simplified tissue-like structure consisting of radial and lateral beams with intersections between them. The computer simulations also revealed that the nanomechanical resonator could operate at room temperature under ambient conditions with high vibration energy.
The results also showed that the simple design worked well with extremely low levels of environmental disturbance leaking into the resonator. This led to the team producing an ultra-thin device that demonstrated outstanding performance.
The team was also amazed that there was minimal energy loss outside the device as the vibrations remained mostly confined to the nanomechanical resonator of the spider tissue.
The functionality of the TU Delft team’s nanomechanical resonator can have significant consequences for the quantum calculation area, especially since quantum units generally have to be stored at temperatures below zero (as close to absolute zero as possible), as they are super sensitive to the environment. conditions at room temperature.
However, the costs and techniques required to store and operate quantum units mean that their costs may prove to be obstructive. Therefore, the development of a strategy that could herald the next generation of nanomechanical resonators could help shape the future of quantum computing.
The researchers also say that the design strategy can be applied to a wide range of geometries and design problems involving low-flow simulations or experiments. They anticipate that future developments in machine learning and optimization, along with new manufacturing techniques, may usher in unprecedented advances in nanotechnology within the next decade.
References and further reading
Shin, D. and Cupertino, A., et al., (2021) Spiderweb Nanomechanical Resonators via Bayesian Optimization: Inspired by Nature and Guided by Machine Learning. Advanced materials, [online] p. 2106248. Available at: https://doi.org/10.1002/adma.202106248