A Spanier está desenvolvendo 'cérebros' de grafeno para impulsionar a Inteligência Artificial?

Spanish engineer Mario Lanza is leading the application of two-dimensional materials to create the most advanced “hard drives” from Suzhou, China.

Artificial Neural Networks

Heaven has paradise; earth, Suzhou and Hangzhou.” With this saying, China has two of its most beautiful enclaves. In the first, Spaniard Mario Lanza has found his very own Eden. In the city that even Marco Polo mistook for Venice, the researcher devotes himself to his two great passions: his family and nanoscience. With more than 100 scientific articles, a book, four patents and three companies, Lanza is now developing the new “hard drives” that will make the explosion of artificial intelligence possible.

“I had always thought that I should gain experience to find a better job in Spain,” he tells INNOVADORES.

That idea led the young Lanza from Salesian High School in Sarrià to Germany. He was lucky that the German university where he studied was very close to the headquarters of chipmaker Infineon Technologies, with which his supervisor had a cooperation agreement. That’s how the electronics engineer discovered research and designed his first paper – while still an Erasmus student. “It’s quite unusual,” he admits.

He returned to Spain to pursue a doctorate, but the technology he needed for his work wasn’t that advanced. “In nanoscience, if you don’t have access to the best machines, repeating experiments is much more complex,” he says. “I couldn’t stay.” He packed his bags again, went back to Germany and then jumped to England. In the last year of his doctorate, a curious thing happened to him: he had achieved so many scientific results that he was able to push ahead with writing his dissertation, so he decided to use the last months of his studies to learn Chinese.

He agreed with his professors that he could move to Peking University to learn the language in the morning and finish his research in the afternoon. That was a decade ago. A real feat. Not many people have lasted that long in the Asian country. “People tend to give up, the competition there is brutal”.

In 2012, Lanza took a year and a half off to do research at Stanford University. Upon his return to China, he received a juicy offer from Suzhou University and started a new life in the “most beautiful city in China.” Not only did he receive a very generous grant for his research project, but he also received a 200,000-euro grant to settle in the country. “This is common practice.” From there, he now leads research on a device readers may never have heard of, but one that will play a key role in the near future.

It’s called a memristor, and it’s basically a super-strong hard drive. “It’s a revolutionary device capable of storing and processing huge amounts of information in a very small space,” he explains. Traditionally, these operations have been performed using transistors, which can emulate the ones and zeros of binary code by storing electrical charge. Memristors, however, store bits in a radically different way: by changing the electrical resistance of insulating materials embedded in nanometer-sized capacitors.

Lanza’s research group has pioneered the application of new compounds to create more efficient memristors. In particular, it has introduced two-dimensional materials such as graphene and boron nitride that act as electrodes and insulators, respectively. “They have very exotic properties, for example, high mechanical strength, flexibility, transparency and very high thermal conductivity,” the engineer explains.

His team has discovered other singularities related to the electrical properties of the memristor. Two-dimensional materials have the ability to control the electrical resistance of the memristor very precisely, making it much easier to tune its speed. “They can create more progressive transitions.”

And to what end? This control over conductivity enhances the performance of the devices. So much so that they can even simulate the function of neural connections. “Neurons are connected by synapses and the memristor works in the same way.” Artificial neural networks are already being made using this technology (IBM is a leader in this field). So Lanza is leading the design of the future physical brains (chips) that will run artificial intelligence.