Resonant Cavity Nonlinear Detectors for Optical Signal Processing

In recent years, there has been enormous progress in the ability to transmit information over optical fiber. As optical networks increase in speed and scope, there is also an urgent need for techniques that sense, modify, or process signals at intermediate points. In present-day networks, this processing is performed with high-speed electrical circuits, where well-established integrated circuit technology can be employed. Although electronic circuits will continue to have a role in optical networks, optical signal processing through nonlinear interactions could greatly simplify several important functions, such as pattern/address recognition, quality monitoring, and clock recovery.

The goal of this work is to design and build a new type of resonant nonlinear photodetector that could be used for optical signal processing. The physical mechanism that enables nonlinear photodetection is two-photon absorption (TPA): a process in which two photons are simultaneously absorbed in a photodiode to generate a single electron-hole pair. Unlike conventional photodetection, two-photon absorption produces a photocurrent proportional to the square of the optical power. Because of this quadratic nonlinearity, two-photon absorption can be used to measure the correlation between high-speed optical signals, in much the same way that sum-frequency generation is used in optical autocorrelators to diagnose pulses that are too short to be measured electrically.

Fig. 1. (a) In a conventional (linear) photodetector,the photocurrent that is proportional to the optical power P. (b) When the photon energy is less than the bandgap (but larger than E_g/2), two-photon absorption can occur, yielding a photocurrent proportional to P².

The challenge faced by nonlinear photodetectors is that they often require impractically high optical intensity to achieve a photocurrent with sufficient signal-to-noise ratio. Many have sidestepped this problem of low sensitivity by using low duty-cycle optical pulses for which the peak power is orders of magnitude larger than the average power. However, optical communication channels employ densely-populated signals for which the peak power and average power are similar in magnitude. To be useful for high-speed optical signal processing, the sensitivity of nonlinear photodetectors must be improved.

Fig. 2. Illustration of a proposed device (for clarity, not all layers are depicted.) An optical signal (wavelength = $ 1.55 µm) is coupled into a reverse-biased GaAs/AlGaAs waveguide p-i-n photodiode. An nonlinear optical photocurrent is produced because of two-photon absorption in the waveguide core. The nanofabricated Bragg gratings form the mirrors of a resonant cavity, which greatly enhances the nonlinear interaction and reduces the overall device size.

In this project, we plan to enhance the sensitivity of two-photon absorption by confining the light in waveguide-based resonant cavities. Waveguides will significantly increase the rate of two-photon absorption by confining the optical signal to a small area, thereby overcoming the diffraction-limited performance of bulk photodetectors. Resonant cavities provide additional confinement in the third dimension, which will further increase efficiency at the expense of wavelength dependence and temporal resolution . In applications that do not require femtosecond resolution, the resonant cavity provides a way to trade response time for improved sensitivity.

Preliminary research indicates that by using microresonators it will be possible to construct nonlinear photodetectors with sufficient temporal resolution to process 400 Gb/s optical signals, a feat that is impossible with today's linear photodetectors. The sensitivity of two-photon absorption in microresonators is predicted to be several orders of magnitude higher than in conventional nonlinear photodetectors, and should enable practical devices that need only micro-Watts of optical input power.

Ultra-sensitive nonlinear detectors could become critical building-blocks for a variety of applications. Some of the specific applications that we are currently exploring include optical clock recovery systems, high-speed optical sampling for quality assessment, and optical address recognition. In each of these applications, the improved efficiency provided by microresonators will enable a new generation of high-speed optical processing components that operate with micro-Watt input powers and do not require bulky and expensive optical amplifiers.

Fig. 3. (a) Structure of proposed GaAs/AlGaAs p-i-n optical waveguide and (b) calculated fundamental TE mode profile (contours labelled in dB relative to peak value.)

Note: This is not a comprehensive list of references such as you might find in a well-written journal article; it is provided only as a suggested starting point for visitors interested in learning more about this subject.