Ga naar hoofdinhoud

Serendipity

Serendipity

For many years, Richard Norte has been on a quest to develop the most precise nanoscale sensors possible. The Assistant Professor (Precision and Microsystems Engineering, Faculty of 3mE) works with tiny sensors that fit on microchips and vibrate, and uses the vibrations to measure small forces that would not be sensed otherwise. It was when he joined strengths with a machine learning expert, Miguel Bessa, that the research made a huge leap forward – through the unexpected combination of nature and artificial intelligence. When the two scientists started collaborating, bringing together their fields of expertise – nanotechnology and machine learning – they were not certain it would yield any significant results. “In the field of microchip technology, we mainly worked on intuition, based on what we knew about vibration,” says Norte. “Could we use machine learning to go beyond intuition? We first took the best existing designs and tried using machine learning to improve them, but we were only able to make very marginal improvements.” 

Then the spider’s web came along. One day, during the COVID-19 lockdown, Norte noticed a lot of spiders’ webs on his terrace. “It occurred to me that we could use a spider’s web as the basic design. Spiders’ webs are essentially nature’s vibration sensors and have been evolving and perfecting themselves for millions of years. If a fly hits the web, it vibrates and sends a signal to the spider.” “We didn’t know if it’d be successful,” adds Miguel Bessa, who is now Associate Professor of Engineering at Brown University but remains affiliated with TU Delft. “We fed the idea to a machine learning system, which – to our surprise – came up with a simple and elegant design for a microchip sensor that used different ways of vibrating. It took us a year before we could validate it, but eventually we were able to show that it was the most precise microchip sensor of its kind out there.” 

Amazing things

Although it might be too soon to say where exactly the development will lead, it is safe to assume that these miniature devices can help us unravel some mysteries of the universe. “At the end of the day, if you want to learn about the world, you need to be able to sense and measure small things and forces,” says Norte. “These microchips can do amazing things. The spider’s web sensor can vibrate without using energy for a very long time – it rings 100,000 times per second for over five hours. This is equivalent to you pushing a swing once and it continuing to swing for 100 years without stopping.” One example of the sensors’ use could be to study gravity and dark matter, explains Norte. “Quantum mechanics deals with very tiny particles, atoms and molecules, while gravity is a weak force that is usually studied at the scale of a whole planet. The idea of measuring gravity at a small scale is extremely difficult. Combining quantum mechanics and gravity has been a big quest in physics – and our sensor could play a role here.” Knowing how to produce and utilise microchips could also have major geopolitical consequences, believes Norte. “Many countries are racing to learn about microchips. Knowing how to design and manufacture them gives us a leg up.”

Last, but not least, the microchip sensors are not only incredibly precise, but they also work at room temperature. “This is incredibly important because other top sensors need cryogenic fridges to function properly,” says Norte. Being able to work at room temperature helps make the research easier, more affordable, and much more energy efficient. The scientists are continuing to work with the ‘spider’s web sensor’ but are also taking other steps forward. Recently they once again used machine learning to design an even more precise microchip sensor. “Before, our goal was to make the smallest sensors possible. Now we are looking at larger structures and using a completely different design and machine learning approach, but our sensors can measure the tiny gravitational pull between two people 100 km apart.”

Stumbling onto something unexpected is a common occurrence for scientists. This time: how combining spiders’ webs and machine learning led to some of the world’s most precise microchip sensors. 

Text Michaela Nesvarova

Richard Norte (left) and Miguel Bessa (right) demonstrate a new type of sensor in the lab, inspired from spider webs.

© Frank Auperlé

Researcher Richard Norte created this artistic picture that illustrates the concept of his research.

© Richard Norte

How the technology works

The web-shaped microchip sensor is a nanomechanical sensor that can vibrate in extreme isolation from everyday noise and function at room temperature. In essence, it is a very small device that keeps vibrating and does not lose energy even after the vibration is stopped. This is crucial because if a sensor maintains its energy, external forces and noise from room temperature heat cannot interfere with it. This isolation from the environment allows for precise measurements.  

Stumbling onto something unexpected is a common occurrence for scientists. This time: how combining spiders’ webs and machine learning led to some of the world’s most precise microchip sensors. 

Text Michaela Nesvarova

For many years, Richard Norte has been on a quest to develop the most precise nanoscale sensors possible. The Assistant Professor (Precision and Microsystems Engineering, Faculty of 3mE) works with tiny sensors that fit on microchips and vibrate, and uses the vibrations to measure small forces that would not be sensed otherwise. It was when he joined strengths with a machine learning expert, Miguel Bessa, that the research made a huge leap forward – through the unexpected combination of nature and artificial intelligence. When the two scientists started collaborating, bringing together their fields of expertise – nanotechnology and machine learning – they were not certain it would yield any significant results. “In the field of microchip technology, we mainly worked on intuition, based on what we knew about vibration,” says Norte. “Could we use machine learning to go beyond intuition? We first took the best existing designs and tried using machine learning to improve them, but we were only able to make very marginal improvements.” 

Then the spider’s web came along. One day, during the COVID-19 lockdown, Norte noticed a lot of spiders’ webs on his terrace. “It occurred to me that we could use a spider’s web as the basic design. Spiders’ webs are essentially nature’s vibration sensors and have been evolving and perfecting themselves for millions of years. If a fly hits the web, it vibrates and sends a signal to the spider.” “We didn’t know if it’d be successful,” adds Miguel Bessa, who is now Associate Professor of Engineering at Brown University but remains affiliated with TU Delft. “We fed the idea to a machine learning system, which – to our surprise – came up with a simple and elegant design for a microchip sensor that used different ways of vibrating. It took us a year before we could validate it, but eventually we were able to show that it was the most precise microchip sensor of its kind out there.” 

Richard Norte (left) and Miguel Bessa (right) demonstrate a new type of sensor in the lab, inspired from spider webs.

© Frank Auperlé

Amazing things

Although it might be too soon to say where exactly the development will lead, it is safe to assume that these miniature devices can help us unravel some mysteries of the universe. “At the end of the day, if you want to learn about the world, you need to be able to sense and measure small things and forces,” says Norte. “These microchips can do amazing things. The spider’s web sensor can vibrate without using energy for a very long time – it rings 100,000 times per second for over five hours. This is equivalent to you pushing a swing once and it continuing to swing for 100 years without stopping.” One example of the sensors’ use could be to study gravity and dark matter, explains Norte. “Quantum mechanics deals with very tiny particles, atoms and molecules, while gravity is a weak force that is usually studied at the scale of a whole planet. The idea of measuring gravity at a small scale is extremely difficult. Combining quantum mechanics and gravity has been a big quest in physics – and our sensor could play a role here.” Knowing how to produce and utilise microchips could also have major geopolitical consequences, believes Norte. “Many countries are racing to learn about microchips. Knowing how to design and manufacture them gives us a leg up.”

Last, but not least, the microchip sensors are not only incredibly precise, but they also work at room temperature. “This is incredibly important because other top sensors need cryogenic fridges to function properly,” says Norte. Being able to work at room temperature helps make the research easier, more affordable, and much more energy efficient. The scientists are continuing to work with the ‘spider’s web sensor’ but are also taking other steps forward. Recently they once again used machine learning to design an even more precise microchip sensor. “Before, our goal was to make the smallest sensors possible. Now we are looking at larger structures and using a completely different design and machine learning approach, but our sensors can measure the tiny gravitational pull between two people 100 km apart.”

Researcher Richard Norte created this artistic picture that illustrates the concept of his research.

© Richard Norte

How the technology works

The web-shaped microchip sensor is a nanomechanical sensor that can vibrate in extreme isolation from everyday noise and function at room temperature. In essence, it is a very small device that keeps vibrating and does not lose energy even after the vibration is stopped. This is crucial because if a sensor maintains its energy, external forces and noise from room temperature heat cannot interfere with it. This isolation from the environment allows for precise measurements.