8 May 2020

AMF team members and manufacturing equipment

© Advanced Micro Foundry

The second most abundant element in the earth’s crust is also at the heart of every electronic chip. Open up your phone or laptop and you will see it: silicon, in chips cut from large, crystalline wafers. As a semiconductor, silicon has proven to be an ideal material for processing electronic signals. As our ability to manipulate electrons at ever smaller scales grew over the decades, computing power steadily increased, following a dictum known as the Moore’s Law.

However, all good things must come to an end. Electronic foundries are fast approaching a hard physical limit to how much further they can extend the Moore’s Law, as shrinking chips further becomes increasingly impractical due to the limits imposed by fundamental physics and or the prohibitive high cost in building it. As a means of circumventing these limits, researchers have been exploring the use of photons or light particles rather than electrons to transmit information since the 1980s.

Based on laser and optical fiber technologies, optical or light-based communication systems boast of better speeds and higher energy efficiency. But before and after photons travel through an optical fiber network, they still need to be converted to electronic signals so that they can be processed by our computers; such conversion represents a major bottleneck in terms of speed and power efficiency, cost and size of the converting equipment.

The solution comes back to silicon. Since it was first started in 2006 at A*STAR’s Institute of Miroelectronics (IME), now later as a Singapore-incorporated spin-off in 2017, Advanced Micro Foundry (AMF) has been at the forefront of developing silicon photonics where silicon-based microchips use photons as the transmitting and processing mediums. However, the team led by AMF Co-founder & President Dr Patrick Lo, faced a daunting challenge: what to do next once a technology has been successfully developed? How do you take it to the next level?


Let there be light

As an expert in complementary metal oxide semiconductor (CMOS) devices, Lo found himself at a career crossroads 15 years ago when faced with the decision of whether he should continue to research Moore’s scaling in CMOS or if he should consider ‘more-than-Moore’s’ technologies as a viable option.

At the time, Lo and his team were IME scientists. They were tasked by then-IME Executive Director Professor Dim-Lee Kwong to examine feasible alternatives to electronics for research directions—silicon photonics being one of them—only to encounter even more questions. Was it possible to manipulate light at such a small scale and in an integrated fashion? And perhaps, more importantly, could even silicon do the job?

“These were the questions that triggered us to think of an area of research that would best utilize the expertise and facilities that IME had,” Lo shared. “So silicon photonics at IME was started as a thematic strategic research program funded by A*STAR with a strong commitment from IME management to this technology over the next 10 years.”

As mentioned previously, silicon was not the immediate choice for a photonic system as it does not emit light efficiently nor manipulate the light. The key technical challenge, Lo said, was to find out how to integrate such functions onto a silicon chip. “For light emitting, at the very beginning, we tried to use silicon germanium and silicon nanocrystals as the material to emit light,” he said. “Although the outcome was interesting from a scientific viewpoint, the material properties were not suitable for commercialization.”

Subsequently, there were many breakthroughs such as high speed modulating and detecting devices, a unique coupling solution that allowed the team to capture light coming from a traditional laser and large dimension fiber into a small waveguide with minimal loss penalty.

“This innovation was the fruit of multi-disciplinary brainstorming sessions and rapid development that leveraged IME’s experience in microelectromechanical systems (MEMS) and system-level design,” Lo added. “Our group came up with several out-of-the-box ideas such as hollowing the structure to receive light input power with minimal leak.”

While the IME team worked hard to lay the groundwork for next-generation technology, they were not the first to delve into silicon photonics research. Large multinational companies such as Intel and IBM in the US and research teams in Europe, China and Japan were also interested, though the exact applications of the technology were still not well understood at the time.

“We kept pushing the device boundaries, constantly testing out the industry to refine the roadmap dynamically until the industry was ready to start considering it. Through this exercise we built a rock-solid technical foundation for our team in recognizing what the material could and could not do, with the mindset of pursuing the possibilities in front of us instead of banging our heads on what could not be done,” said Lo.

For instance, over the next decade, it slowly became clear that silicon photonics was critical for long haul data transfer, and would also gradually find its way into many other key applications. “The same technology that we developed for data transfer and interconnects started finding new applications that we could not have anticipated at the beginning, in sensing, free-space detecting ranging and even computing, for example. Needless to say, it is also becoming an indispensable option for quantum technology,” Lo added.

In 2010, this research was recognized for its groundbreaking contribution to the semiconductor industry development and awarded the President’s Technology Award in Singapore.


Spinning off to scale up

While working with silicon itself was one part of the challenge, another, almost unexpected problem arose: scaling up.

“One of the key features with our technology is that it worked and it worked ahead of its time. Most new partners would approach us expecting pre-competitive analysis but our team had a ready-to-use platform that worked. ” Xianshu Luo, AMF’s Manager for Research & Development explained.

“Technically, we could solve all the device- and circuit-level issues and raise the maturity level of the technology in a year’s time for each customer. So our customers needed the immediate next step: a commercial volume fab to manufacture their photonic chips, as a product, and no longer as a research concept,” added Kavitha Buddharaju, Senior Manager for Business Development at AMF.

The next issue was that IME lacked a commercial entity that would be able to take the silicon photonics technology they had developed and bring it into a viable production path. “By this, I mean that there was no commercial semiconductor fabrication plant that was willing to take up our technology and supply our customers who wanted to launch products immediately,” Lo said. “As we were at the beginning stages of a disruptive new technology, the volume required by each customer was very low and unattractive to large fabs.”

Since actualizing their promising silicon photonic technology required an approach that did not fit the existing business model of traditional manufacturers—relying on the economies of scale for large volume production—IME sensed an opportunity to launch a high mix, low volume fabrication plant, and anchoring in Singapore.

“Here was a technology that had everything going for it: industry pull, a great team and an existing customer base. We assessed that a high mix, low volume foundry coupled with the broad knowledge base of our team and the business awareness accumulated over many years would be able to cater to the needs of the new technologies being developed not only in silicon photonics but also in other adjacent areas,” said Luo.

To capture this opportunity, IME and A*STAR put together AMF’s founding team and linked them to commercialization experts. The company successfully raised private funding to expand its technology offerings and is now 160-members strong.

“It was an exceptionally challenging task as we needed to privatize a government-funded group and grow it into a successful company while serving existing customers at the same time,” Lo shared. “One of our lead customers likened it to repairing a race car while driving it on a highway! It was the most challenging phase of our careers but also the most enriching; to see our passion project launched as a private entity.”


Forging the future

A major focus area for Lo and his team is also to look ahead and build for the future. “While we serve the immediate customer base in volume manufacturing on 5G infrastructures and data center interconnect technologies, we need to ensure that the next wave of technologies such as LIDAR, 3D sensing, bio/chemical sensing and computing are also equally supported through new developments and ready-to-use technologies,” said Chao Li, senior manager of Technology Development at AMF.

Lo hopes that these emerging technologies will quickly transition into easily scalable ones too—something he believes is essential for the long-term success of AMF and for Singapore semiconductor and manufacturing industry.

“To be a one-stop-shop that provides value-added services will require a great deal of effort from AMF, but also strong partnerships with other players in the ecosystem, including support and endorsement from the public sector,” Lo said. This, he believes, will be the critical piece to complete the puzzle to take advantage of the excellent work being done by the research community in Singapore and bring it to commercialization. “We have helped start-ups globally, why not locally for Singapore?” Lo added.

“AMF is and will be an international company, and our vision is to make AMF Singapore the de-facto solution provider and silicon photonics volume house, world-wide. Technology innovation, scaling up with quality and customer satisfaction will remain the key focus areas for AMF through this exciting journey,” Lo said.

“Silicon photonics has gone through several phases—from excitement to hype to disappointment, and eventually to reality. Not all paths were rosy. It has quietly gone through functional verification from prototyping stage to customer sampling and scale verification of volume, to finally becoming an adopted technology. To be a critical part of this development was exhilarating for me, as well as for our team,” he concluded. “We owe a lot of thanks to our customers who believed in us, and helped us grow; as well as to our shareholders who trusted our vision and our competitors who challenged us. We consider ‘competitors’ as our partners to grow this young industry collectively.”

https://research.a-star.edu.sg/articles/features/innovation-at-the-speed-of-light/