John Bowers: We are still at the dawn of photonics

Professor John Bowers has been a key contributing figure in the development of silicon photonics. In an interview, he reflects on his career, the technical advancements shaping silicon photonics, and its expanding role.

After 38 years at the University of California, Santa Barbara (UCSB), Professor John Bowers (pictured) is stepping away from teaching and administrative roles to focus on research.

He welcomes the time it will free for biking and golf. He will also be able to linger, not rush, when travelling. On a recent trip to Saudi Arabia, what would have centered around a day-event became a week-long visit.

Bowers' career includes significant contributions to laser integration and silicon photonics, mentoring some 85 PhD students, and helping found six start-ups, two of which he was the CEO.

Early Influences

Bowers’ interest in science took root while at high school. He built oscilloscopes and power supplies using Heathkits, then popular educational assemblies for electronics enthusiasts. He was also inspired by his physics and chemistry teachers, subjects he majored in at the University of Minnesota.

A challenging experience led him to focus solely on physics: “I took organic chemistry and hated it,” says Bowers. “I went, ‘Okay, let’s stick to inorganic materials.’”

Bowers became drawn to high-energy physics and worked in a group conducting experiments at Fermilab and Argonne National Laboratories. Late-night shifts - 10 PM to 6 AM - offered hands-on learning, but a turning point came when his mentor was denied tenure. “My white knight fell off his horse,” he says.

He switched to applied physics at Stanford, where he explored gallium arsenide and silicon acoustic devices, working under the supervision of the late Gordon Kino, a leading figure in applied physics and electrical engineering.

Bowers then switched to fibre optics, working in a group that was an early leader in single-mode optical fibre. “It was a period when fibre optics was just taking off,” says Bowers. “In 1978, they did the first 50-megabit transmission system, and OFC [the premier optical fibre conference] was just starting.”

Bell Labs and fibre optics

After gaining his doctorate, Bowers joined Bell Labs, where his work focused on the devices—high-speed lasers and photodetectors—used for fibre transmission. He was part of a team that scaled fibre-optic systems from 2 to 16 gigabits per second. However, the 1984 AT&T breakup signalled funding challenges, with Bell Labs losing two-thirds of its financial support.

Seeking a more stable environment, Bowers joined UCSB in 1987. He was attracted by its expertise in semiconductors and lasers, including the presence of the late Herbert Kroemer, who went on to win the 2000 Nobel Prize in Physics. Kroemer developed the double heterostructure laser and played a big part in enticing Bowers to join. Bowers was tasked with continuing the laser work, something he has done for the last 40 years.

“Coming to Santa Barbara was brilliant, in retrospect,” says Bowers, citing its strong collaborative culture and a then newly formed materials department.

Professor Bowers lecturing

Integrated lasers

At UCSB, Bowers worked on integrated circuits using indium phosphide, including tunable lasers and 3D stacking of photonic devices.

At the same time, the field of silicon photonics was starting after Richard Soref wrote a seminal paper proposing silicon as an optical material for photonic integrated circuits (PIC).

“We all knew that silicon was a terrible light emitter because it is an indirect band-gap material,” says Bowers. “So when people started talking about silicon photonics, I kept thinking: 'Well, that is fine, but you need a light source, and if you don't have a light source, it'll never become important.’”

Bowers tackled integrating lasers onto silicon to address the critical need for an on-chip light source. He partnered with Intel’s Mario Paniccia and his team, which had made tremendous progress developing a silicon Raman lasers with higher powers and narrower linewidths.

“It was very exciting, but you still needed a pump laser; a Raman laser is just a wavelength converter from one wavelength to another,” says Bowers. “So I focused on the pump laser end, and the collaboration benefitted us both.”

Intel commercialised the resulting integrated laser design and sold millions of silicon-photonics-based pluggable transceivers.

“Our original vision was verified: the idea that if you have CMOS processing, the yields will be better, the performance will be better, the cost will be lower, and it scales a lot better,” says Bowers. “All that has proven to be true.

Is Bowers surprised that integrated laser designs are not more widespread?

All the big silicon photonics companies, including foundry TSMC, will incorporate lasers into their products, he says, just as Intel has done and Infinera before that.

Infinera, an indium phosphide photonic integrated circuit (PIC) company now acquired by Nokia, claimed that integration would improve the reliability and lower the cost, says Bowers: “Infinera did prove that with indium phosphide and Intel did the same thing for silicon.”

The indium phosphide transceiver has a typical failure rate of 10 FIT (failures per ten billion hours), and if there are 10 laser devices, the FIT rises to 100, he says. By contrast, Intel’s design has a FIT of 0.1, and so with 10, the FIT becomes on the order of 1.

Silicon lasers are more reliable because there’s no III-V material exposed anywhere. Silicon or silicon dioxide facets eliminate the standard degradation mechanisms in III-V materials. This enables non-hermetic packaging, reducing costs and enabling rapid scaling.

According to Bowers, Intel scaled to a million transceivers in one year. Such rapid scaling to high volumes is important for many applications, and that is where silicon photonics has an advantage.

“Different things motivate different people. For me, it's not about money, it's more about your impact, particularly on students and research fields. To the extent that I've contributed to silicon photonics becoming important and dynamic, that is something I'm proud of.”

Professor John Bowers

Optical device trends

Bowers notes how the rise of AI has surprised everyone, not just in terms of the number of accelerator chips needed but their input-output (I/O) requirements.

Copper has been the main transmission medium since the beginning of semiconductor chips, but that is now being displaced by optics - silicon photonics in particular - for the communications needs of very high bandwidth chips. He also cites companies like Broadcom and Nvidia shipping co-packaged optics (CPO) for their switching chips and platforms.

“Optics is the only economic way to proceed, you have to work on 3D stacking of chips coupled with modern packaging techniques,” he says, adding that the need for high yield and high reliability has been driving the work on III-V lasers on silicon.

One current research focus for Bowers is quantum dot lasers, which reduce the line width and minimise reflection sensitivity by 40dB. This eliminates the need for costly isolators in datacom transceivers.

Quantum dot devices also show exceptional durability, with lifetimes for epitaxial lasers on silicon a million times longer than quantum well devices on silicon and 10 times less sensitivity to radiation damage, as shown in a recent Sandia National Labs study for space applications.

Another area of interest is modulators for silicon photonics. Bowers says his group is working on sending data at 400 gigabits-per-wavelength using ‘slow light’ modulators. These optical devices modulate the intensity or phase, of light. Slowing down the light improves its interaction in the material, improving efficiency and reducing device size and capacitance. He sees such modulators is an important innovation.

“Those innovations will keep happening; we're not limited in terms of speed by the modulator,” says Bowers, who also notes the progress in thin-film lithium niobate modulators, which he sees as benefiting silicon photonics, “We have written papers suggesting most of the devices may be III-V,” says Bowers, and the same applies to materials such as thin-film lithium niobate.

“I believe that as photonic systems become more complex, with more lasers and amplifiers, then everyone will be forced to integrate,” says Bowers. 

Other applications

Beyond datacom, Bowers sees silicon photonics enabling LIDAR, medical sensors, and optical clocks. His work on low-noise lasers, coupled to silicon nitride waveguides, reduces phase noise by 60dB, enhancing sensor sensitivity. “If you can reduce the frequency noise by 60dB, then that makes it either 60dB more efficient, or you need 60dB less power,” he says.

Applications include frequency-based sensors for gas detection, rotation sensing, and navigation, where resonance frequency shifts detect environmental changes.

Other emerging applications include optical clocks for precise timing in navigation, replacing quartz oscillators. “You can now make very quiet clocks, and at some point we can integrate all the elements,” Bowers says, envisioning chip-scale solutions.

Mentorship and entrepreneurial contributions

Bowers’ impact extends to mentorship, guiding so many PhD students who have gone on to achieve great success.

“It’s very gratifying to see that progression from an incoming student who doesn’t know what an oscilloscope is to someone who’s running a group of 500 people,” he says.

Alan Liu, former student and now CEO of the quantum dot photonics start-up Quintessent, talks about how Bowers calls on his students to ‘change the world’.

Liu says it is not just about pushing the frontiers of science but also about having a tangible impact on society through technology and entrepreneurship.”

Professor John Bowers at his recent retirement celebration. “I had about 85 Ph.D. students, many of whom are tremendously successful and have done great things. It's very gratifying to see.”

Bowers co-founded UCSB’s Technology Management Department and taught entrepreneurship for 30 years. Drawing on mentors like Milton Chang, he focused on common start-up pitfalls: “Most companies fail for the same set of reasons.”

His own CEO start-up experience informed his teaching, highlighting interdisciplinary skills and team dynamics.

Mario Paniccia, CEO of Anello Photonics, who collaborated with Bowers as part of the Intel integrated laser work, highlights Bowers’ entrepreneurial skills.

“John is one of the few professors who are not only brilliant and technically a world expert - in John’s case, in III-V materials - but also business savvy and entrepreneurial,” says Paniccia. “He is not afraid to take risks and can pick and hire the best.”

Photonics’ future roadmap

Bowers compares photonics’ trajectory to electronics in the 1970s, when competing CMOS technologies standardised, shifting designers’ focus from device development to complex circuits. “Just like in the 1970s, there were 10 competing transistor technologies; the same consolidation will happen in photonics,” he says.

Standardised photonic components will be integrated into process design kits (PDKs), redirecting research toward systems like sensors and optical clocks.

“We’re not at the end, we’re at the beginning of photonics,” emphasises Bowers.

Reflections

Looking back, would he have done anything differently?

A prolonged pause follows: “I’ve been very happy with the choices I have made,” says Bowers, grateful for his time at UCSB and his role in advancing silicon photonics.

Meanwhile, Bowers’ appetite for photonics remains unwavering: “The need for photonic communication, getting down to the chip level, is just going to keep exploding,” he says.