Advanced circuit platforms are moving beyond Si for electronics and are moving to partially or fully optical platforms.
These new platforms need to use certain structures for optical signal processing without conversion between optical and electrical domains.
Photonic systems will rely on new circuit architecture that is compatible with CMOS processes to ensure scalability, commercialization, and standardization.
Newer circuit platforms will integrate all of these optician and electronic components into a single board/chip package.
The future of computing may very well be all-optical and will rely on a bevy of optical signal processing components. RF engineers and board designers need to know the basics of analog routing, but analog systems generally no longer use analog techniques for complex signal processing tasks. Aside from heterodyning, harmonic generation, superheterodyne reception, or other effects involving interference, signal processing is normally performed in the digital domain by converting the analog signal into a digital signal.
In the realm of optics, even in microwave photonics, the focus is on all-optical signal processing, where one or more optical signals are manipulated directly as one would do with an electrical analog signal. Conceptually, analog signal processing tasks used in electronic systems are the same as those used in optical signal processing. However, the structure of different circuit elements and the materials involved are very different. Here’s how optical signal processing works at the IC level, and how PCBs will need to change to accommodate these advanced circuits.
Electrical vs. Optical Signal Processing
It is important to remember that electrical signals and optical signals are one and the same; any time-varying electrical signals generating an oscillating electromagnetic field, which is light by definition. For this reason, it’s easy to think about all-optical signal processing in the same way as electrical signal processing in the RF domain. Many of the tasks you would need to perform for RF signals (heterodyning, modulation/demodulation, etc.) can also be performed for continuous-wave optical signals.
So what makes all-optical signal processing different from the analog and digital signal processing implemented as firmware and software? The difference is in the structures that enable these techniques, as well as the frequencies involved. In addition, the physical mechanisms involved in each domain are different because, again, we are working at different frequencies. The table below summarizes some similarities and differences between analog and optical signal processing systems.
The goal in optical signal processing is to eliminate the need for optical → electrical → optical conversion. Instead, structures are designed to perform standard signal processing tasks directly on optical signals. In this way, optical signal processing circuits are totally passive systems; they require no input power other than that used in any emitters and detectors. Just like signal processing in electronics, optical signal processing can fall into the digital or analog regimes, depending on the task involved.
Analog vs. Optical Signal Processing
Performing analog signal processing with optical signals follows the same logic as in electronics, but the circuits, on-chip structures, and frequencies involved are very different. When planning out a system to perform standard analog signal processing tasks, your functional block diagrams for optical and analog electronics systems would be basically identical.
The materials used in optical circuits need to be nonlinear, meaning their optical properties are a nonlinear function of the electric or magnetic field at the particular operating wavelength. In electronics, this is provided by transistors, diodes, or other components built from doped semiconductors. In optics, the designer needs to take advantage of optical nonlinearities, where the dielectric constant is a function of the strength of the electric field. A Taylor series is normally used to represent the dielectric constant.
Taylor series expansion for the dielectric function of a material. This equation describes a material’s nonlinear response and the radiation it produces as part of optical signal processing.
This equation is the basis for describing all nonlinear optical phenomena upon which optical signal processing relies. Designers normally focus on the 2nd order nonlinearity, which produces the Kerr effect in optical systems. The 3rd order nonlinearity is also used to produce higher-order modulation. These nonlinearities allow harmonics of an optical signal to be generated and manipulated, just like in electronics. When combined with group velocity dispersion, stable pulses that resemble digital electronic pulses can be generated.
Digital vs. Optical Signal Processing
The block diagram for a digital signal processing circuit for executing numerical algorithms will look identical to that used in optical signal processing. Everything is based on logic circuits built from nonlinear materials, just as is the case with transistors built from doped Group IV, III-V, or II-VI semiconductors. A digital logic circuit uses a digital signal as binary data, whereas optical pulses are used in optical signal processing circuits.
These systems are much more complicated as maintaining stable pulses requires routing through a transparent nonlinear medium with no dispersion. The gate element used in one of these optical signal processing circuits needs to have a nonlinear response in order to provide the necessary logic functions. This then requires bonding two different semiconductors together to place a detector and logic circuit alongside routing structures in a PIC/EPIC. This has led to many fabrication difficulties that are yet to be solved in PICs/EPICs.
PICs and EPICs for optical signal processing can be fabricated from the same materials as standard semiconductor devices.
What’s Next in PICs/EPICs?
Devices that will perform optical signal processing tasks are still being commercialized, and the semiconductor industry is focused on developing standards for these new systems. Si is the most likely material to dominate the PIC/EPIC landscape, as it is compatible with existing foundry processes. In addition, current IC design tools can be easily used to start building PICs/EPICs for optical signal processing. The open-source community has responded with a range of free design tools for optical ICs.
PICs and EPICs are definitively the next generation of technology for computing and data processing, and electronics designers need the right EDA tools to help them stay at the cutting edge. If you’re designing boards to support PICs/EPICs, integrated simulation can give you a platform for optical IC design and optical circuit board design.