Design, simulation and manufacturing advances continue to push photonic integrated circuits (PICs) to smaller sizes, which provide greater capabilities per chip, writes Liam Critchley.
Advanced optics and photonics where light is manipulated and transmitted has already revolutionised communications through optical fibres and amplifiers. PICs are the next frontier of advanced photonic systems and are much smaller than many other optical systems. PICs are essentially the photonic equivalent of very small electronic semiconducting chips and are getting smaller and more advanced as their development continues.
What are PICs?
A PIC, also known as a planar lightwave circuit or an integrated optical circuit, is a microchip that uses photons of light instead of electrons to generate, transport and process information. The electronic equivalent is an integrated circuit (IC). An IC will have embedded electronic components, such as resistors, inductors, transistors and capacitors, that are powered through electrons.
PICs, on the other hand, have their own equivalent components, including waveguides, which are equivalent to wiring, lasers, polarisers, optical amplifiers, photodetectors and phase shifters. This combination of components is responsible for generating, transporting, modifying and measuring visible or near-infrared light across the PIC.
"PICs and conventional electronic ICs share the goal of integrating complex functionality onto compact, scalable chips. However, they differ fundamentally in information carriers, leading to design philosophy, performance limits and applications differences," says Huiyun Liu, professor of semiconductor photonics at UCL. "Electronic circuits use electrons, while PICs use photons guided by optical waveguides."
The materials used between ICs and PICs differ as well. Electronics systems are primarily built using silicon materials using complementary metal-oxide-semiconductor manufacturing. However, PICs are built using a range of electro-optic crystals such as silicon nitride, lithium niobate, gallium arsenide and indium phosphide.
Photons can carry information at extremely high frequencies (hundreds of THz) and are immune to electrical crosstalk and electromagnetic interference. This makes them well-suited for high bandwidth and long-reach data transmission due to electronic interconnects' struggles with resistive and capacitive losses. "Electronics excel at logic, memory and control, but their interconnect bandwidth and power dissipation scale poorly at high data rates," says Liu. "PICs offer ultra-high bandwidth density using wavelength division multiplexing and lower energy per transmitted bit for medium and long-distance links. PICs complement electronics, not replace them."
Where electronic ICs benefit from decades of CMOS scaling and mature libraries of standardised devices, Liu says PICs are younger and face challenges "as silicon is no good for photonic emitters, such as lasers, III–V/Si integration, including quantum dot lasers on silicon, play a crucial role by enabling scalable, on-chip light sources compatible with CMOS manufacturing".
PIC components
The circuity in PICs are optical channels that allow light to be propagated along and manipulated by small-scale components on the chip. The circuity makeup can vary depending on the application of the PIC. As photonic components get smaller and more efficient, being able to integrate more photonic components on the chip increases the value of PICs for optoelectronic devices.
Each component in a PIC is either a source, modulator, amplifier, signal carrier or detector, and are grouped into passive and active components. Passive PIC components have zero electrical input and are purely optical, whereas active components have some kind of electrical input (modulating) or output (sensing) and are optoelectronic in nature.
Why engineers are interested in PiCs
"Interest in photonic integrated circuits has accelerated because data movement, not computation, is now the dominant bottleneck in modern digital systems, particularly in datacentres and AI infrastructure," says Liu. The speed of light compared to electrons allows data to be transmitted much faster and this could help datacentres and AI infrastructure to be a lot more efficient.
Despite not being as mature as electronic ICs, PICs have a number of distinct advantages. Over other optoelectronic systems, a lot more functional components can fit in the same space. The interest for photonic chips here is similar to the interest that electronic chips got for the same reason. The only difference is the medium of data transfer. PICs are smaller than other optic and photonic systems, so can be used to make smaller data communications equipment, and they have fewer failure modes than other photonic systems so are more reliable.
Compared to electronic circuits, not only do photons of light move faster than electrons, but PICs are more efficient than electronic circuits because the electrical resistance in traditional circuits generates more heat that leads to a higher signal strength loss than in PICs. As noted previously by Liu, PICs are complementary to ICs, not a competitor, and that's because PICs can be combined with ICs in monolithic microchips and optoelectronic systems.
Design challenges
Despite the interest and benefits of PICs, they are complex devices that require rigorous engineering design. The behaviour of ultra-fast photons and their interaction with different materials, alongside the modification of light ―that is, changes in the signal's frequency, magnitude, and phase ― inside the PIC makes for complex physics inside the PIC.
"The most significant challenges is that silicon cannot efficiently generate light, yet lasers are essential for optical interconnects," says Liu. "However, III–V/Si integration allows high-performance laser materials to be combined with silicon photonics platforms, wafer bonding and epitaxial growth are now sufficiently mature for volume manufacturing, and quantum dot lasers on silicon are particularly promising, offering lower sensitivity to defects, improved temperature stability and longer device lifetimes."
The cost and complexity of PIC packing is one of the costliest parts of a product containing PICs. Wafer-level and panel-level assembly, alongside new housing/casing materials, are helping to reduce some costs and optimise manufacturing. It's thought in the coming year that PICs could have the same scale and cost structures as other high-tech products, where around 80% of costs come from material and 20% comes from assembly and testing.
Simulation methods are helping to shorten the design cycle and improve the design before a single physical prototype is made. "Dedicated photonic simulation and layout capabilities are needed to meet the unique challenges of PIC design. Having the right tools for modelling and layout implementation not only shortens the design cycle but also reduces the number of cycles necessary for a successful design," says Liu. With advanced simulation software and high powered computing the PIC design challenges are getting easier to navigate.
Industry use
PICs are continually expanding into new application areas as simulation and manufacturing methods become more advanced. These advances help to design and manufacture more compact, more advanced, higher performing and more versatile PICs that is opening the door to new applications across communications, sensing, computing and data processing industries. Here are some of the industries where PICs are starting to gain momentum:
Optical communications: One of the most obvious applications given the reliance on small-scale optical components used in communications. PICs can be used as amplifiers and multiplexers to improve data transmissions through high-speed fibre-optic networks, while PIC-transceivers can connect computers and cell towers. In the future PICs could connect multiple smart vehicles/software defined vehicles in smart city networks using light fidelity.
LiDAR sensing: LiDAR, commonly used in autonomous vehicles, uses pulses of laser light to map the local environments, and PICs are needed to produce the specific light pulses that are sent out by the LiDAR transmitter.
Lab-on-a-chip: PICs can be utilised in lab-on-a-chip diagnostics for detecting different medical ailments. Lab-on-a-chip platforms contain optical and electronic components as light is used to measure fluid samples at the point-of-care and the electronics provide the readable output. As lab-on-a-chip are small devices with smaller channels, the small size of PICs makes them highly beneficial for these diagnostic platforms.
Quantum computing: Quantum computers are being built where the qubits are either based on electrons or photons. For photonic quantum computing, small-scale PICs are going to be critical for controlling and measuring quantum states, as well as providing a bridge between classical computers and quantum computers, and quantum networking between multiple quantum computers.
AI and machine learning (ML): There's a lot of potential for PICs in AI/ML and is potentially one of the biggest growth areas. Currently, PICs are used for efficient optical communication between computers. However, in the future, neural networks could benefit a lot from PICs, and on the simulation side, AI-assisted simulations are helping to develop more effective PICs that are better suited to AI/ML operations/applications.
When asked about what the PIC industry landscape will look like in the coming year, Liu told Electronics Weekly that "In my opinion, the future of PICs lies in progressively moving optics closer to where data is generated and processed. Over the next five years PICs will see continued high-volume deployment in datacentre and AI interconnects by wafer bonding, driven by the need for higher bandwidth and lower energy per bit. Beyond that, optical I/O and tighter electronic – photonic monolithic integration of III-V and Si are likely to reshape system architectures, particularly for large-scale AI with high density and low cost."