Monolithically Integrated Mach-Zehnder Interferometer Wavelength Converter and Widely-Tunable Laser in InP Milan L. Mašanović, Vikrant Lal, Jonathon S. Barton, Erik J. Skogen, Daniel J. Blumenthal, Larry A. Coldren
Summary of Work Objective/scope Approach Major accomplishments Issues Demonstrate InP monolithic integration of a widely tunable laser and all-optical wavelength converter for digital and analog wavelength conversion. Approach Utilize a common offset quantum well integration platform and a combination of passive, active and filter waveguides to implement on a single chip a Mach-Zehnder SOA Interferometer and widely tunable SGDBR laser. Major accomplishments Demonstration and testing of world’s first monolithically integrated tunable laser and AO wavelength converter Wide conversion range: 50+nm in, 22nm (L-Band) 2.5GB/s error free operation Issues SGDBR on-chip output power and mirrors Passive waveguide losses, chip insertion losses SOA speed (gain recovery) C-Band operation
Monolithically Integrated InP AOTWC Sampled Grating DBR Laser Mach-Zehnder Interferometer Input Amplifier
Integration Challenges Enable a Common Fabrication Platform Offset Quantum Wells Tradeoffs Between Laser and MZI Performance Optical Isolation Lasers Highly Sensitive to Coherent Reflections Carefully Design Active/Passive Interfaces Amplified Reflections Facet Reflections Need to be Suppressed Effectively (10-5 or better) Processing Issues Material Quality - Uniformity Processing Uniformity - Very Long Devices
SGDBR Laser Background 4 Section Device Wide Tunability, High Power Suitable for Integration Passive – Active Waveguide Combination Does not Require Facet Reflection to Lase Realized in Several Integration Platforms(*) Offset Quantum Wells Burried Ridge Stripe (OQW, QWI) Front Mirror Gain Phase Rear Mirror Offset QW Device Try not to spend too much time on this slide *Beck Mason, Erik Skogen, Larry Coldren
Interferometer - Cross Phase Modulation π phase shift CW in No CW light out Converted Signal Out Data in Cross-Phase Modulation Principle Semiconductor optical amplifiers used to achieve phase shift Incoming data disturbs phase balance data conversion Transfer Function SOA Current Optical Power Inverting Operation Non Inverting Operation
Interferometer Designs Two different interferometer realizations MMI/S Bend Design 2 Stage MMI Design 1 - 1x2 MMI Splitter/Combiner 2 - S Bend 3 - MQW SOA 1 2 3 1 - 1x2 MMI Splitter/Combiner 2 - 2x2 MMI Coupler 3 - MQW SOA 3 1 2
MMI Components Function Properties N*N power splitting/coupling Most common components: 1x2, 2x1 splitter/combiner 2x2 (3dB) coupler Properties Simple Structure and Fabrication Low Inherent Loss Large Bandwidth Low Polarization Dependence 1515 nm 1545 nm 1575 nm
Facet Reflections Depending on the SOA Gain, Maximum Tolerable Reflection R=0.251.25 10-4 Optimized Output Waveguide Required Multilayer (3) AR Coating Width Taper 5 µm 7° Angle
Thermal Consideration – Interferometer Design Is Zhaoyang speaking to this point on thermal? If yes, can you push this slide to him? It also needs some bullets.
Epitaxial Heterostructure
Offset Quantum Well Process Active – Passive Removal Most Mature SGDBR Fabrication Technology Requires Single MOCVD Regrowth Grating Formation InP/InGaAs Regrowth Metalization/Anneal Passivation/Implant Ridge Etch
Critical Steps - Verification Active Regions Removal PL Line Scan Gratings Etch AFM Scan Active Regions Actual Device Layout (Active and Ridge Layer Shown) This slide is somewhat busy and confusing. Is there any way to simplify and clean up?
First Generation Tunable Wavelength Converter 5 um What are you trying to show here? There is a lot going on and cluttered.
Gain Bandwidth and Tuning Range Gain Peak – 1560nm Tuning Range – 22nm
Tuning Efficiency This needs a clearer statement of what you are showing here.
Mirror Design Why new design is better? Current Design New Design
Electrical/Optical Control Maybe don’t need this slide Optical Control Electrical Control
DC Extinction Map This is good Higher Extinction For low bias currents
Power-Extinction Analysis Difference in SOA Output Power Strongly Affects Extinction Keep Powers Equal in Both Branches and Change Phase Independently
Phase-Extinction Relation SOA Phase Change of Pi Required Extend the SOA Length
Test Setup EOM Data Gen. λin EDFA BPF Device Pin PD BER Error Rate λout Back-to-back Attenuator PC
2.5GB/s Conversion – Fixed SGDBR Wavelength 1535nm 1565nm Input Eye Diagram PRBS 231-1 Maybe need to make this more readable. 1545nm 1575nm Converted Data 1555nm 1585nm
2.5GB/s Conversion – Fixed Input Wavelength Input Eye Diagram PRBS 231-1 1557nm 1566nm Converted Data 1577nm 1570nm
2.5GB/s Bit Error Rate Testing Fixed SGDBR Output Wavelength PRBS 231-1 Input L-band (due to filter/receiver) Indicate power penalties on graph ~2dB Power Penalty
First Generation WC Performance Limits 5GB/s Eye λin=1590nm +4mW (SGDBR Power Difference
Summary of Issues to be Resolved Mirror Design Low MZI Input Power (<0.8mW) Improve conversion efficiency (input to output) Measure and understand analog characteristics Speed up Carrier Dynamics Non Optimum Electrical and Optical Extinction
Second Generation WC Design Improvements Redesigned Mirrors Wider tuning range Better ‘behavior’ Output Coupler Light Evacuation Phase Control Improved Extinction Input SOAs Boost Input Power Reduce Lifetime 1.4Q Waveguide Higher Output Power Fixed !
Second Generation WC with TIR Mirror This is cool!
Carrier Recovery Time Slide (QWI)
Summary Fabricate Second Generation Devices Finish Analog Characterization for Gen I Devices Compare Performance (Digital/Analog) Can you give Zhaoyang your submount pictures to include in the packaging talk?