We describe theoretically planar waveguide structures that accomplish the Fabry-Perot filtering functinality by using wavelength selective coupling between antiresonant reflecting optical waveguides (ARROW). In one of the structures studied, a thin cladding ARROW is coupled with a thick cladding ARROW. The thick cladding layer can be viewed as a Fabry- Perot interferometer operating at grazing-incidence. This Fabry-Perot cladding layer defines a comb of uniformly spaced optical frequencies for which input light is coupled over (filtered out) into the `drop' output port. Using parameters pertinent to a polymer-based configuration, we obtain a wavelength resolution of 3 nm for a free-spectral range of 29.7 nm. A second configuration, which we have studied, features dual output. Side-lobe suppression ratios fall in the range of 9-14 dB. Higher suppression ratios and finesse could be achieved by cascading many identical devices. The corresponding `add' functions can be achieved with these devices by simply reversing the propagation direction of light. These devices can be used as multi/demultiplexers or as optical filters depending on the applications. Computations in this work are for the 1500 nm communications band, but results can be translated to other communications bands.

An optically tunable filter device is considered based on a Bragg reflection grating in a nonlinear waveguide. An external light beam of high intensity controls the refractive index via nondegenerate optical Kerr effect and tunes the grating to a selected wavelength. The signal wavelengths are different from the excitation, and typical values of the signal power do not affect the refractive index of the material. Assuming an instantaneous response of the Kerr material and neglecting absorption, we apply the general predictions of the local normal mode expansion theory (LNME) for an arbitrary angle of incidence. A design example is given in a waveguide based on InGaAs on a AlGaAs substrate. The spectral response and tuning range of the waveguide filter device is examined.

A simple theory is presented for the analysis of optical channel-dropping filters for dense WDM applications. It combines the orthogonal coupled-mode theory based on normal modes and the nonorthogonal coupled-mode theory based on isolated waveguide modes. Analytical solution is obtained for the power dropped in the resonant channel. The interplay between mutual coupling and direct Bragg reflection is discussed.

Appropriate apodization or tapering of the interaction strength in codirectional couplers leads to the suppression of sidelobe levels. It is shown that taper shape optimization may lead to sidelobe suppression on the order of - 125 db and lower. However, lower sidelobe levels are accompanied by a broadening of the passband width. It is shown that there are distinct coupler shapes which minimize the passband width for a desired sidelobe level. The theoretical results are shown to be in excellent agreement with beam propagation method simulations.

We calculate the expected number of internodal hops for a network established with a wavelength-division-multiplexed (WDM) 2-D multiple-plane optical interconnection. This WDM optical interconnection incorporates WDM pixels consisting of multiple-wavelength vertical-cavity surface-emitting laser (VCSEL) arrays and wavelength-selective detectors. The WDM interconnection can support simultaneous and reconfigurable communication among a network of nodes. Using the expected number of hops as a measure of internodal switching delay, we show that the integration of WDM into the interconnection results in a significantly reduced delay as compared to single-wavelength systems. Substantial delay reduction results even when the number of wavelengths is small relative to the number of 2-D planes. We analyze the bus, dual-bus, and ring architectures since they define the means of communication between pixels. For each architecture, we analyze three configurations which provide each node access to: (1) an entire plane of pixels, (2) a row (or column) of pixels, or (3) an individual pixel. When each network node has access to an entire plane of pixels, the proposed WDM interconnection incurs substantially shorter delay than single-wavelength optical solutions. By allowing a node to access an entire row or column of pixels, the interconnection benefits from the incorporation of spatial-division-multiplexing (SDM) and the number of interconnections can grow substantially with negligible added delay. Finally, when a node can access only a single pixel, a large number of independent processors can be interconnected exhibiting far less switching delay than other electronic or optical interconnections of comparable size.

The inherent information bandwidth of optical fibers between the wavelengths 1.3 and 1.6 micrometers is in the terahertz range. One obvious way to exploit this bandwidth is to use wavelength-division multiplexing (WDM). The Canadian Solid State Optoelectronics Consortium (SSOC), an association of industry, university, and federal government research laboratories, has been developing the component technologies required to demonstrate the operation of an eight channel WDM system. This paper discusses the integration of the transmitter (Tx) and the receiver (Rx) modules using a thin film process on alumina substrates. The Tx module contains a fully integrated eight channel DBR laser array with two quad-laser driver circuits. The signal from the lasers is combined into a single waveguide and is then carried off-chip via a polarization maintaining optical fiber. The Rx module is made up of an integrated receiver circuit, and a series of amplifiers providing the gain required for signal and clock recovery. The receiver circuit consists of an echelle grating which disperses the eight distinct wavelengths into a bank of InGaAs metal-semiconductor-metal (MSM) detectors. Some of the performance parameters of the Tx and Rx modules are presented.

A review of multi-wavelength gain-coupled DFB laser arrays for wavelength division multiplexing is given, with emphasis on the application of the gain-coupling mechanism and the effective index changes by varying the ridge waveguide width. Fabrication and characterization results of long-wavelength InGaAsP/InP multiquantum-well gain-coupled DFB lasers and laser arrays are presented. Features of the devices are interpreted and simulated using theoretical models. It is demonstrated that due to the gain-coupling effect, seven singlemode lasing wavelengths around 1.55 micrometers with 1 - 2 nm spacing are obtained simultaneously. Despite the significant change in the ridge width, low threshold currents and small beam divergence are achieved by careful designs. Issues regarding practical applications of the array devices are discussed.

We report on the design, growth, fabrication and characterization of monolithic wavelength division multiplexed (WDM) laser array transmitter and receiver chips produced by the Canadian Solid State Optoelectronics Consortium. The transmitter chip includes multiple, discrete wavelength, distributed Bragg reflector (DBR) laser diodes monolithically integrated with waveguide combiners fabricated using an InGaAs/GaAs heterostructure. The corresponding wavelength demultiplexer unit is based on a Rowland circle grating spectrometer monolithically integrated with a metal-semiconductor-metal (MSM) detector array fabricated on an InGaAs/AlGaAs/GaAs heterostructure. The epitaxial layer wafers for both transmitter and receiver modules were grown in single molecular beam epitaxy (MBE) runs.

We have carefully observed on the spectral characteristics of two-segment 1.55 micrometers InGaAsP/InP ridge waveguide multi-quantum-well DFB lasers and found two types of wavelength switching. The first type occurs between the longitudinal-modes on the opposite sides of the DFB stopband, while the second type on the same side of the stopband. The physical mechanism is correlated to a slight difference of the effective grating period between the two inhomogeneously injected laser segments.

A rapidly tunable receiver intended for wavelength-division multiple-access systems is constructed from an integrated optic grating demultiplexer, photodetector array and an amplifier/selector chip.

A technique for fabricating transparent waveguides on the same wafer as a quantum well (QW) DBR laser array has been developed. High [MeV] energy ion implantation is used to create a large number of vacancies and interstitials throughout the active region of the device. Upon annealing, these entities enhance the intermixing of the QW and barrier materials resulting in a blue shift of the QW bandgap. Energy shifts (measured using low temperature photoluminescence spectroscopy) of greater than 60 meV can be achieved. Room temperature waveguide absorption measurements verify the shift in the bandgap energy and confirm that the waveguide is now effectively transparent in the wavelength range of the QW lasers. This technique is being used in an eight wavelength WDM transmitter array in which the waveguiding region is selectively implanted and blue shifted.

We analyze the system performance of gain-saturated SOA-based all-optical wavelength shifting with respect to its dynamic characteristics. When considering the risetime, contrast ratio and intersymbol interference, we find that there exists an optimal probe power and wavelength for high-speed wavelength shifting which reduces the power penalty by 3 dB. The minimum power penalties for the data rates of 10 Gb/s and 20 Gb/s are 2.5 dB and 5 dB, respectively.

Design and development of basic optical components are vital for the future development of all optical and optoelectronic circuits. Optical components fabricated on Si are restricted to operation at wavelengths below 1.1 micrometers . However, the incorporation of SiGe in Si devices enables operation to be extended to the communications wavelengths of 1.3 micrometers and 1.5 micrometers . Optical modulation in a Si/Si_0.85Ge_0.15/Si, MBE grown, waveguide p-i-n diode was experimentally investigated at a wavelength of 1.3 micrometers . When the waveguide p-i-n diode is forward biased, the injected free carriers absorb the light propagating in the intrinsic layer thereby giving a strong modulation of the device output intensity. The modulator, grown on a 2 X 10¹⁹ cm^-3 n⁺⁺-Si substrate, consists of a 2.5 micrometers thick, 7 X 10¹⁵ cm^-3 n-Si layer, a 0.1 micrometers thick i-Si_0.85Ge_0.15 guiding layer, a 2 mumm thick, 2.5 X 10¹⁶ cm^-3 p-Si layer and a 0.1 micrometers thick, 2 X 10²⁰ cm^-3 p++)-Si contact layer. Under forward bias the p-i-n diode achieves a carrier density of 5 X 10¹⁸ cm^-3 in the guiding layer of the waveguide. A maximum modulation depth of 66% was obtained for a 2 mm long, 37.5 micrometers wide ridge waveguide at a peak pulse current density of 2700 A/cm².