Perspectives on Photonic Integration T. L. Koch Prof. of ECE and - - PowerPoint PPT Presentation

▶
perspectives on photonic integration
SMART_READER_LITE
LIVE PREVIEW

Perspectives on Photonic Integration T. L. Koch Prof. of ECE and - - PowerPoint PPT Presentation

Dec 22, 2022 192 likes •561 views

Center for Optical Technologies Perspectives on Photonic Integration T. L. Koch Prof. of ECE and Physics Director Center for Optical Technologies Lehigh University Bethlehem, Pennsylvania, USA 2005 WOCC Newark April 22, 2005 Center for

slide-1
SLIDE 1

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

  • T. L. Koch
  • Prof. of ECE and Physics

Director Center for Optical Technologies

Lehigh University Bethlehem, Pennsylvania, USA

Perspectives on Photonic Integration

slide-2
SLIDE 2

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

OUTLINE:

Photonic Integration:

  • what it isn’t
  • what it is

Status, market-induced trends, challenges

slide-3
SLIDE 3

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

The greatest story of Integration …

  • Sketch from Jack Kilby’s lab

notebook of IC concept at TI

  • Use semiconductor for circuit

elements

  • p-n junctions for capacitors
  • bulk semi for resistors, etc.
  • Demonstrated simple circuits

like flip-flop, oscillators, etc., in Ge

slide-4
SLIDE 4

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Silicon Planar Processing …

  • Noyce at Fairchild

simultaneously files IC concept in Si

  • Hoerni introduces planar

process in Si

  • The race is on …
slide-5
SLIDE 5

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

The Digital Revolution Ramps Up!

  • First commercial IC in 1961
  • 2 logic gates ( 4 bipolar

transistors, 4 resistors)

  • Rapid industry advances

ensue:

  • Linear ckts, op amps, etc.
  • Digital logic gates

1966 TTL logic chip

slide-6
SLIDE 6

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Relentless Advance of IC Integration

  • Moore’s Law (1965)
  • Doubling in # of transistors

every 18 months (60% CAGR)

  • Example (close to true!):
  • 1965 most complex digital

chip had 64 transistors

  • 2000 intro of Pentium IV

processor w/est. 42 million transistors

  • DRAM tracking also
slide-7
SLIDE 7

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Optics Progress: Even Faster than Moore’s Law!

10 YEAR IMPROVEMENT IC Density : × 100 Fiber Capacity: × 200 !

10,000 300 100 30 10 84 86 88 90

Year Capacity (Gb/s)

92 94 96 98 Single Channel (ETDM) Multi-Channel (WDM) Single Channel (OTDM) WDM + OTDM WDM + Polarization Mux Single Channel (ETDM) Multi-Channel (WDM) 80 82 3 1 0.3 0.1 0.03 Experimental Commercial 1000

*

*

3000 00 02

*

EXPERIMENTAL COMMERCIAL

slide-8
SLIDE 8

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Get a Grip! Capacity growth is not traceable to integration -

  • Photonic Integration offers promise of cost, power, size reduction

however ….

  • Massive, repetitive digital blocks in electronic IC’s are truly,

fundamentally, application enabling Power of digital processing & especially software in volume markets are drivers for Moore’s Law

slide-9
SLIDE 9

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

So Turn Back The Clock, reset our expectations …

  • 1964 early linear IC
  • Matched actives &

passives on chip

  • 1965 first commercial

fully integrated OP AMP These analog circuits would not have driven Moore’s Law & $B fabs … but they do target clear volume applications and are powerful and enabling in their reduced cost, size, and power!

slide-10
SLIDE 10

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Photonic Integrated Circuit Vision

INPUT FIBER POLARIZATION CONTROLER WAVELENGTH AGILE LASER AMPLIFIER FILTER COUPLER DETECTORS OUTPUT FIBER DETECTOR SWITCHES MODULATOR WAVELENGTH AGILE LASER

  • Fig. from S. K. Korotky

Exemplary PIC with large variety of guided-wave elements: Monolithically interconnected optical and optoelectronic components fabricated on a common substrate

slide-11
SLIDE 11

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Monolithically Integrated Balanced Heterodyne Receiver PIC

Koch, et al PTL 2, 1990 p. 577

Why hasn’t this taken off?

slide-12
SLIDE 12

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005 CW DFB Laser Module 10 Gb/s LiNbO3 Module 10 Gb/s Integrated Modulator Laser Module Using InP-based PIC

The Value Proposition:

The merits of integration When two or more elements are optically interconnected to form a functional subsystem:

  • we pay for two or more packages

(packaging can be 50-80% of cost, even more compelling than IC’s!)

  • we get larger subsystems; devices

constrained to use long optical paths

  • we lose power from fiber coupling

efficiencies (lower SNR)

  • we incur instabilities or fluctuations

from coupling – efficiency, phase, reflections …

Systems vendors require reductions in cost, size, power; improvements in performance

Can these be so dramatic as to be market enabling?

  • ultimately yes, but in the near term we have …

The challenging work of implementing a replacement technology!

slide-13
SLIDE 13

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

  • Replacing discrete optical solution & offering

improvements in cost, size and/or power.

  • Replacing FUNCTIONALITY previously done with

electronics using optical networking architectures and

  • ptical technologies

Two kinds of replacement technology… What needs to be integrated …

slide-14
SLIDE 14

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

  • Replacing discrete optical solution & offering

improvements in cost, size and/or power.

  • Replacing FUNCTIONALITY previously done with

electronics using optical networking architectures and

  • ptical technologies

Two kinds of replacement technology… But there’s also some subtlety …

slide-15
SLIDE 15

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Optically Amplified WDM Transmission System

  • Data In

Data Out

λA ~ 80-120 km

OA

Erbium-Doped Fiber Amplifier

OA OA

RCVR RCVR λ1 λ2 λN

35 nm Saturated Small-Signal 1550 nm

λ

Gain Spectra

Gain λ1, λ2 ... λN Er-Doped Fiber ΣPl ~ N µW Pump Laser (0.98 mm)

Amplified (Non-regenerated) Transmission Line

OA

λ1, λ2 ... λN ΣPl ~ N mW

RCVR λ1 λ2 λN

D M U X O M U X O

XMTR XMTR XMTR

slide-16
SLIDE 16

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Integrated DFB Laser/EA Modulator by SAG

DFB Laser Section EA Modulator Section n-InP Substrate InGaAsP Grating Fe:InP Blocking p-InGaAs/InP Cap Selective-Area MOCVD Grown MQW-SCH HR AR

  • Data In

Data Out

λA ~ 80-120 km

OA OA OA

RCVR RCVR λ1 λ2 λN

OA

RCVR λ1 λ2 λN

D M U X O M U X O

XMTR XMTR XMTR

slide-17
SLIDE 17

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

What about the receive demux side?

  • Data In

Data Out

λA ~ 80-120 km

OA OA OA

RCVR RCVR λ1 λ2 λN

OA

RCVR λ1 λ2 λN

D M U X O M U X O

XMTR XMTR XMTR

Discrete solutions: Cascaded Thin-Film Filters

slide-18
SLIDE 18

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Typical TFF Demux Configuration Ex: 100 GHz 16-Channel What about 160-channel?

Input Port

29 27 30 28 32 25 31 26 22 24 21 23 35 33 36 34

Band-Pass Separator Band-Pass Separator

slide-19
SLIDE 19

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

slide-20
SLIDE 20

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Si/SiO2 Waveguide Grating Router:

Readily scalable to large

channel count

Wafer scale processing Stable, inexpensive optical

interconnections

Input Star Coupler Output Star Coupler

  • Data In

Data Out

λA ~ 80-120 km

OA OA OA

RCVR RCVR λ1 λ2 λN

OA

RCVR λ1 λ2 λN

D M U X O M U X O

XMTR XMTR XMTR

slide-21
SLIDE 21
  • n-chip loss: 4 dB

responsivity: 0.4 A/W crosstalk:

  • 35 dB

Today's technology for WDM integration: Today's technology for WDM integration:

World's smallest integrated World's smallest integrated AWGs AWGs: 40 channels integrated : 40 channels integrated

9 arrayed waveguide gratings+ 40 Photodetectors

3 1

4.8 mm

4

4.6 mm

Chip (mag 5x) Component Module

slide-22
SLIDE 22

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

  • Data In

Data Out

λA ~ 80-120 km

OA OA OA

RCVR RCVR λ1 λ2 λN

OA

RCVR λ1 λ2 λN

D M U X O M U X O

XMTR XMTR XMTR Packaged Module

Receive Receive

slide-23
SLIDE 23

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

  • Replacing discrete optical solution & offering

improvements in cost, size and/or power.

  • Replacing FUNCTIONALITY previously done with

electronics using optical networking architectures and

  • ptical technologies

Two kinds of replacement technology… But there’s also some subtlety …

slide-24
SLIDE 24

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

RECONFIGURABLE ACTIVE WDM ADD/DROP

Traditional Single-Channel Add/Drop

Low-Speed Electronic Add/Drop High-Speed Traffic High-Speed Electronic Add/Drop

High-Capacity Traffic

Local Traffic

Space Switch Electronic Add/Drop MUX/DMUX

λ1, λ2, .….. λn λ1 λ2 λn

1xN

λ-MUX

Nx1

λ-DMUX λ1, λ2, .….. λn

  • Reduced cost dynamic

provisioning at λ granularity without full OEO

x x

OEO Interfaces

slide-25
SLIDE 25

Center for Optical Technologies

8-λ 1 × 9 Wavelength Selective Cross-connect

8.6 cm

Doerr et al, Bell Labs

Features:

64 switches, 80 shutter/VOA’s Any λ to any port All paths have double rejection in both space and wavelength Smaller and fewer activated switches than classic split-and-combine

4λ, 1x5 example

slide-26
SLIDE 26

Advantages of Monolithic Integration

Widely Tunable SG-DBR Laser with integrated SOA and EAM

Light Out Front Mirror Gain Phase Rear Mirror

SG-DBR Laser

Amplifier Modulator

EA Modulator SOA

Advantages: smaller space (fewer packages) lower cost (fewer package components) lower power consumption (lower coupling losses) high reliability (fewer parts)

slide-27
SLIDE 27

Full C Full C-

  • band, high power wavelength selectable laser

band, high power wavelength selectable laser

1 x 12 DFB MMI SOA S-Bent

PMF output power > 30 mW at TLD=50 °C

  • SMSR > 50 dB
  • ∆λ=3.5 nm, σ < 0.1nm
  • Tunability ~ 40 nm

IDFB=150 mA

  • 50
  • 40
  • 30
  • 20
  • 10

10 20 1520 1530 1540 1550 1560 1570 Wavelength [nm] Intensity [dBm]

(reported in ECOC (reported in ECOC’ ’2003) 2003)

slide-28
SLIDE 28

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Optical ASICs

High Volume Batch Fabrication Dense Functionality on a Chip

(Tunable High order microring resonators, switches, VOAs, polarization beam splitters, polarization rotators, mode transformers, delay lines, all pass filters)

1 cm (Hundreds of chips per wafer)

slide-29
SLIDE 29

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Channels = 100 Avg IL = 4.1 dB Avg PDL < 0.5 dB Avg 1 dB BW = 28 Ghz Avg 3 dB BW = 33.0 Ghz

IL (dB) f (Thz) 194.6 194.65 194.7 194.75 194.8

  • 70
  • 60
  • 50
  • 40
  • 30
  • 20
  • 10

Full C-band (or C+L) Tunable Filters

High Performance & Functionality

f (Thz) IL (dB)

192 193 194 195 196

  • 60
  • 50
  • 40
  • 30
  • 20
  • 10
slide-30
SLIDE 30

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Optical regeneration

Optoelectronic All-optical

20 cm

Front-end Driver CW-laser

  • r EML

MZI

λin λout

Non-linear device: SOA 2 mm

CW-laser SOA-MZI

λin λout

Similar techniques have worked at speeds to 160 Gb/s!

slide-31
SLIDE 31

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Resonant-Enhanced Functionality:

  • Q’s of 108 !!
  • Raman & parametric oscillators,

amplification in µ-resonators

  • SiO2-based resonators
  • Extend to Si-based resonators?

(i.e., SOI) Need ultra-low loss!

Single longitudinal mode Raman lasing in a 40-µm diameter microsphere, exhibiting 16% pump

  • conversion. Inset: ηd = 36%.

Vahala et al, Caltech

Think analog IC’s & microwave mixing techniques

slide-32
SLIDE 32

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Yield & Performance

Why does this happen?

  • Huge leverage in core of

network for incremental improvements in performance

  • High willingness to pay

due to tremendous cost sharing

  • That was then, this is

now …

Performance

  • No. of Units

New design breakthrough (MQW’s, etc.)

slide-33
SLIDE 33

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Today’s Market Realities …

  • Severe pressure on service providers; fighting

for piece of limited disposable income in the face of escalating bandwidth & connectivity!

  • New deployment targets:
  • Larger relative investments in metro & access
  • Less ability to share cost at edge of network
  • Systems houses less likely to pay for marginal

increases in performance (standardization)

  • All asking for dramatic reductions in cost, size, power

(really cost, cost, cost)

  • Photonic Integration can deliver what today’s market is

asking for

slide-34
SLIDE 34

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Important future directions:

  • Drive refinement, stabilization of process technologies
  • in III-V’s, refine use of QWI, more planar process technologies
  • reliable process/device/integration modeling tools on stable platforms
  • piggyback on CMOS, explore functionality of Si and SOI
  • for passives, push limits of high-index-contrast integration
  • true integration with CMOS electronics
  • Packaging of photonic IC’s
  • RF & thermal challenges of high-density; I/O bottlenecks
  • better array solutions; passive or MEMs alignment
  • Really understand cascadability of active functionality
  • Limits of 2-R, 3-R regeneration, ultrafast nonlinear SOA dynamics
  • Power consumption, density limits, noise
  • Drive ultra-low loss fab techniques; understand potential of

ultra-high Q resonant structures

  • nonlinear functionality (Raman, parametric)
  • highly cascaded passives
  • higher-density integration
slide-35
SLIDE 35

Center for Optical Technologies

2005 WOCC – Newark April 22, 2005

Already exceeding complexity of 1st generation analog IC’s!

For the naysayers … ..