Programmable PICs to propel photonics prototyping

Share this on social media:

A new type of PIC has been developed, the properties of which can be programmed using a combination of light and heat

Photonic integrated circuits (PICs) are the light-based equivalent of the electronic circuits used in the majority of today’s technologies. However, they offer lower energy consumption, faster operation, and enhanced performance compared to their electronic counterparts, making them more suited to emerging demanding applications, such as quantum computing and environmental sensing.

It’s not just emerging applications such as these that PICs are used in, however. A great number of today’s applications already greatly rely on the technology. In fact, most of us – at least indirectly – reap the benefits of PICs every single day. This is because the technology forms part of the backbone of the entire internet. Any communication that takes place over a few kilometres is done via optical fibres. To send information down these fibres, you need lasers, and to make lasers, you need PICs. It is therefore because of PICs that we are able to stay in touch with our families, friends and colleagues using video conferencing during this trying period of lockdown.

A challenge associated with PICs is that their fabrication methods currently leave a lot to be desired. They result in large variability between the fabricated devices, which translates to limited yield, long delays between a conceptual idea and a working device, and a lack of configurability.

Part of the issue is that PICs are based on very sophisticated structures on the scale of hundreds of nanometres. As the behaviour of the circuit is fully dependent on the geometry and refractive index of these structures, any small deviation from their design – even by a matter of nanometres – will yield a non-optimal device.

Thankfully, due to the volume of demand for PICs currently being at least three orders of magnitude lower than microelectronics, it can still be met even with the limited yield of existing production methods. However, as demand for photonics grows, the pressure on price will grow with it, leaving manufacturers no choice but to explore how they can increase yield.

A new age of programmable PICs

Enter Oded Raz, an associate professor at the department of electrical engineering at the Eindhoven University of Technology, together with colleagues, has devised a new process for the fabrication of PICs that he says could be of paramount importance to the future of PICs. The new process addresses the critical issues of current methods, while enabling novel, reconfigurable (programmable) circuits to be produced. 

The concept of programmable PICs is not entirely new, however. Previous attempts to devise them have suffered from issues such as volatility and/or high optical signal losses – both of which negatively affect a circuit’s ability to maintain its programmed state.

To overcome this, Raz’s group has used hydrogenated amorphous silicon to build the circuits, a material originally used in thin-film silicon solar cells. The optical properties of amorphous silicon can be changed via light exposure and heating, and so with it the researchers have designed a new fabrication process that enables stable programmable PICs to be realised. Not only that, but the new process also addresses the invariability and yield shortfalls of current PIC fabrication methods, as manufacturers will be able to easily tune functionality and correct for small errors during fabrication.

This is the world’s first working demonstration of a reconfigurable PIC, where the material chosen for making the integrated optical circuit is being programmed,’ said Raz.

To program the circuits, the optical properties (refractive index) of the amorphous silicon are tuned by exposing it to a combination of heat and light, via a heater and a high intensity LED lamp. Once programmed, a circuit can then be reset (returned to a programmable state) by again exposing it to heat. This approach therefore promotes the reusable and sustainable use of materials, the researchers say.

What will the impact be? 

To help explain the significance of this development, Raz highlighted current issues that are often encountered when firms are looking to adopt photonics technology.

‘Many people in industry have heard about the benefits that photonics can offer. But in order to convince them of this they need to see it in action,’ he said. ‘In order to see it in action they have to go through the whole process of designing, simulating, fabricating, testing, re-designing, re-simulating, and re-fabricating before they finally get a working PIC specific to their application. This can take anywhere between 9 to 18 months, which can be very disencouraging to those looking to adopt photonics.’

By creating a new generation of programmable ‘generic’ PICs, Raz believes that these issues can be addressed and that the market volume of photonics can be increased significantly.

‘While previously customers would have had to design a PIC for their specific application, and then wait 9 to 18 months before it finally becomes a product, in the future they could instead buy a generic off-the-shelf PIC, program it, test their idea, and then reset and reprogram the PIC if it fails,’ he explained. ‘That’s a quantum leap in terms of thinking.’

A microscope image showing a typical device fabricated by the researchers using amorphous silicon.

Such off-the-shelf PICs would, for example, explain on their packaging that they can accommodate up to 10 filters, 10 lasers, 10 detectors and a variety of interconnect options. The customer would then have the freedom to decide how many lasers they want to activate and at which wavelength, in addition to how these lasers will be combined and what kind of filters will be used alongside them. ‘This can all be done via programming,’ said Raz. ‘What will then happen is that PIC manufacturers will be able to create large amounts of these circuits to service any customer.’

He likened the impact of generic PICs in photonics to that of FPGAs in microelectronics: ‘FPGAs are general purpose, programmable devices. They are not designed to be optimal in any way in terms of energy consumption, surface area, or speed – they are designed to offer flexibility. If a company makes a design with an FPGA, they will then tapeout an ASIC, which is more energy efficient, lower area, up to one tenth of the cost, and will do exactly the function that the company needs. It will not be general-purpose anymore.’

The same would therefore be the case when using programmable PICs, Raz continued: ‘Once a firm has designed a suitable chip for its application using a programmable PIC, and has a suitable business case, they can then have it produced as a specific ASIC photonic chip for far less cost.’

The introduction of a new generation of programmable PICs would therefore not herald the end of photonic ASIC chips. On the contrary, according to Raz, as he is convinced that introducing programmable PICs will actually increase the volumes of photonic ASICs produced, as the new PICs would help encourage more firms to adopt photonics technology.

‘The time, the investment needed and the uncertainty connected to deploying a photonic solution is often so large, that firms often delay it,’ Raz said. ‘If they instead had an option for quick prototyping when determining whether photonics is the right solution, they will more quickly move on to taking the long path of developing a specific ASIC photonic chip for their application.’

This is just the beginning

While this development is indeed very promising, in a paper written for Advanced Optical Materials, Raz and his team have been very transparent in that the technology and processes developed are not currently without limitations.

‘There are things that need to be improved,’ Raz confirmed. ‘It currently takes about 100 hours to erase a programmed circuit, which is not a realistic prospect and won’t yet be enough to make people excited. We are convinced as a research group, however, that this is just the tip of the iceberg. If people spend the energy and time to further investigate it, then these times can be shortened, the energy required to initiate programming will be reduced, and maybe even a different material will be shown to achieve a similar result.’

This revolutionary approach could therefore herald a wave of further investigations on reconfigurable PICs and their advantages, according to the researchers. ‘I believe that the long-term perspective of PICs will have a very strong component which is programmable,’ Raz concluded.



Synopsys: Featured product

Synopsys is accelerating the adoption of photonic and photonic integrated circuit (PIC) technologies with its comprehensive Photonic Solutions portfolio, which includes the RSoft™ Photonic Device Tools, the Photonic System Tools, and the PIC Design Suite.

Offering the industry’s only seamless design flow, the Synopsys Photonic Solutions portfolio enables innovations in consumer and industrial communication, sensing, and imaging applications – from concept to manufacturing.

FormFactor's fast photonic wafer probers enable rapid tests through faster photonic alignment than previously possible. (Credit: FormFactor Inc)

07 September 2021