Modulators for Optical Arbitrary Waveform Generator

The goal of this project is to create an optical arbitrary waveform generator. To achieve this goal, light from a laser that generates ultra short pulses with a high repetition rate is first broken down into its various frequency components. By modifying the phase and amplitude of each frequency component, the resulting pulse of the recombined components results in an optical pulse with an arbitrary amplitude and phase. The aim of this research is to design, develop and fabricate a modulator. By creating an array of these modulators, the phase and amplitude of the various frequency components from an ultra short pulse, high repetition rate laser can be modified in order to create an optical arbitrary waveform generator.

To create an arbitrary optical waveform at wavelengths that are centered at 800nm, ultra broad band modulator arrays are required. Since these modulators are to operate around 800nm, the material choices are limited to relatively high Al content AlGaAs and to In0.5 (GaxAl1-x)0.5P layers that are lattice-matched to GaAs. In addition, since GaAs absorbs light with a wavelength less than 870nm, the lower cladding layer of the modulator must be relatively thick in order to isolate the modulator from the GaAs substrate. To create the largest mode possible and to minimize the coupling loss, the index contrast between the waveguiding layers and the cladding layers should be minimized. To minimize the index contrast, a dilute waveguide structure in which thin layers of high index material are embedded in a low index material is employed. The resulting layered structure has an effective index slightly higher than the low index material and is determined by the layer thicknesses as well as the refractive index of the two materials that comprise the dilute waveguide.

The modulator structure that was grown by molecular beam epitaxy is an Al0.8Ga0.2As-based structure in which the dilute waveguide consists of alternating layers of Al0.8Ga0.2As and InGaP. The structure is challenging in terms of the epitaxial growth. Although the use of Al0.8Ga0.2As for the cladding layer minimized the lattice-mismatch problem, achieving high quality, high Al content AlGaAs cladding layers is difficult due to the low Al adatom mobility on the surface during growth. To minimize free carrier loss, P-I-N structures are employed in which the Si and Be dopants are graded from the contact layers to the dilute waveguide region. Photoluminescence (PL) measurements from the arsenide-based structure show a weak PL peak at ~650nm from the InGaP layers in the dilute waveguide. The Al0.8Ga0.2As and Al0.5Ga0.5As layers as well as the InAlP layers have indirect band gaps and hence do not exhibit photoluminescence. Due to the high etch selectivity between the arsenide and phosphide layers, the uppermost high index layer of the dilute waveguide also acts as an etch stop.

In addition to this original structure, a second Metal-Oxide-Semiconductor-type structure has also been grown which differs from the previous design by the addition of two oxidized AlAs layers, enabling a strongly confined optical mode in the middle of the structure. The AlxOy layers will allow the device to be capable of withstanding higher operating voltages. Furthermore, the device can be unipolar. The structure also contains an InAlP etch stop to facilitate fabrication.

The optical properties of the dilute waveguide in both structures have been simulated using OptiBPM (Optiwave Corporation). The Al0.8Ga0.2As-based structure is designed to support a single optical mode within a 2 micron wide ridge waveguide; the fundamental mode for the arsenide-based structure is roughly 2 micron x 1 micron (W x H). The MOS-type structure is also designed to support a single optical mode, which is roughly 1.5 micron x 1 micron (W x H) as simulated by OptiBPM. If the dilute waveguide of the Al0.8Ga0.2As-based structure is not completely etched, due to the low index contrast of the dilute waveguide, the bending radius is quite large, on the order of a millimeter. Ultimately, the modulator will the incorporated into an array waveguide grating, therefore the loss needs to be considered.

A new self-aligned fabrication process, which defines both the passive devices and the powered modulators in the same step, has been developed that is compatible with both the MOS-type structure and the Al0.8Ga0.2As-based design. The only difference in the fabrication process is the addition of the AlAs oxidation step that is inserted after the reactive ion etching that is used to define the waveguides. The mask set associated with this process has been designed and fabricated. The mask set contains Mach Zehnder interferometer modulators of various lengths with multimode interference couplers or Y-splitters. The Mach Zehnder interferometer modulators as well as conventional modulators are oriented both parallel and perpendicular to the major flat of the 2" GaAs (100) wafers. The mask set also contains a variety of passive components such as Y-splitters, multimode interference couplers as well as straight and curved waveguides.

Arbitrary waveform generation is obtained by the phase and amplitude modulation of the individual frequency components within a frequency comb. Hence, optical wavelength demultiplexers and multiplexers are necessary for the spatial separation and recombination of wavelength components prior to and following modulation. Therefore, the structure and performance of arrayed waveguide gratings (AWG) have been modeled and a mask containing the AWG is currently being designed. The AWG has eight input and output waveguides that are each 2 microns wide. As the input aperture of the free propagation region (FPR) is approached, the waveguide width gradually increases to 3 microns over a length of 50 microns. The output waveguides taper in width at the output aperture, scaling back from 3 microns to 2 microns over a similar length. Adjusting the waveguide width, allows the optical mode to smoothly transition from the confined waveguides to the dispersive free propagation region. The thirty waveguides in the phased array section similarly taper from a width of 4 microns to 3 microns. At the first FPR output, where the waveguides are 4 microns wide, there is no space between the waveguides, encouraging full transmission of the diffracted power from the first FPR to the phased array waveguides and on to the second FPR. The AWG is designed and simulated to have 10 GHz channel spacing with -30dB to -40dB of optical cross-talk between output waveguides.