Douglas Yu, an executive at TSMC, the world’s largest chip manufacturer, clearly articulates the disruptive potential of photonic integrated circuits: “If we succeed in implementing a good system for silicon photonics integration, we will unleash a new paradigm . We are likely at the dawn of a new era.”
Silicon photonics aims to harness the technology of this chemical element to optimize the transformation of electrical signals into light pulses. The most apparent application for this innovation is the development of high-performance links that can be used to manage communication between multiple chips as well as enhance the flow of information between various machines.
Advanced packaging technologies utilized by major semiconductor manufacturers like TSMC, Intel, and Samsung stand to benefit significantly from a high-performance inter-chip communication mechanism. Furthermore, large data centers requiring connections among numerous machines could substantially gain from this technology. However, one domain that could particularly capitalize on the advantages presented by silicon photonics is artificial intelligence (AI) .
China Plans to Utilize This Technology in AI, 6G Communications, and Quantum Computers
Intel and TSMC have been working on developing their silicon photonics-related technologies for several years. As anticipated, this innovation has not gone unnoticed by Chinese companies and research centers. In mid-May 2024, the Shanghai Institute of Information Technology and Microsystems, in collaboration with the Swiss Federal Institute of Technology in Lausanne, achieved a crucial milestone. Until that point, one essential ingredient for photonic integrated circuits was lithium niobate.
Lithium tantalate allows for large-scale photonic chip manufacturing at significantly lower costs
This synthetic salt plays a pivotal role in the production of photonic integrated circuits because its physicochemical properties optimize the conversion of electricity into light. However, it has a drawback: the high cost of each wafer and their respective sizes. What these scientists have accomplished is the substitution of lithium niobate with another semiconductor material that possesses even more attractive properties: lithium tantalate (LiTaO3).
Ou Xin, one of the leading scientists on this project, asserts that, in addition to outperforming lithium niobate, lithium tantalate enables the manufacture of photonic integrated circuits at scale and at much lower costs . This cost efficiency is attributed to the similarity of its manufacturing processes to those already employed for producing conventional silicon semiconductors.
In this context, the Shanghai Jiao Tong University’s Integrated Photonic Chip Center (CHIPX) has recently announced its commencement of six-inch wafer production for photonic chips . Interestingly, this production line still employs lithium niobate, indicating that there is still room for advancement in leveraging the properties of lithium tantalate. Regardless, Professor Jin Xianmin, director of CHIPX, claims that photonic integrated circuits possess immense potential not only in training and inferring AI models , classical supercomputing, and quantum computing but also in advancing 6G communications .
Image | TSMC
More Information | SCMP
In Xataka | Today, China and the U.S. have set aside their differences for a good reason: they will collaboratively analyze the risks of AI.