Transcript Park, Namkyoo - Quantum Electronics Group

Next Generation Optical Amplifiers
requirements, bottlenecks, possible resolutions
(Approach concerned on Cost, Footprint , Functionality
rather than Efficiency, utilizing nano-photonics)
Namkyoo Park
Nanoscale Energy Conversion and Information Processing Devices
September 24 th, 2006
Photonic Systems Laboratory
School of EE, Seoul National University
http://stargate.snu.ac.kr
[email protected]
Photonic Systems Lab
School of EECS, S.N.U.
Research Topics in PSL : Past / Present

Photon Generation







Transient control & amplified transmission line design
Polarization Mode Dispersion tolerant transmission format
Multi-level Optical Transmission
Photon Control – Coding, Detection, Logic






 This presentation
Photon Transport


Raman Amplifier
Erbium Amplifier
Thulium Amplifier
nano-Photonics : Er / Raman based Si Amplifier / Laser)
Optical Coding (CDMA, Noise reduction)
Super-resolution Techniques (2D / 3D Imaging)
Surveillance system for FTTH network
Distributed / Multi-port Temperature sensor
Semiconductor Amplifier & SOA based Logic Gates
Integration with / Applications to IT, BT & NT


Tunable Optical devices (including Photonic Crystals, MEMS)
Application to Medical-Photonics (3-D Tomography)
Photonic Systems Lab
School of EECS, S.N.U.
Network Evolution – Market Calls for METRO and Below
Backbone
Metro
Continent to Continent
Coast to Coast all over
Fiber at 10 Gbps & up
City to City-Town to Town
all over Fiber
at 1Gbps
10 Gbps
Access To the optically Fibered
World “First Mile / Last Mile”
56kbps
1 Gbps
Desktop to Desktop –
Floor to Floor
10 Mbps
1 Gbps
LAN
The First Mile
Jonathan Thatcher, OFC2002, Tutorial Sessions(2002)
 Long Haul

More of the same (higher speed, more wavelength, longer reach…)
 Metro/Access will shape the next wave of innovative components

Tunable, intelligent, distributed amplification
 The “siliconization” of photonics


Drive scalable manufacturing and cost efficiency
Push optics further into the network and ensure sustainable growth of the industry
Photonic Systems Lab
School of EECS, S.N.U.
Network Evolution – Challenges in Technology
OITDA2000
Photonic Systems Lab
School of EECS, S.N.U.
Network Evolution – Challenges : Met for METRO network
Now in the Market ! (2002)
Photonic Systems Lab
School of EECS, S.N.U.
Network Evolution – Beyond Metro : How far ?
You will need more photons for
Your desktop PC / Processors
Electronics-photonics must converge !
Photonic Systems Lab
School of EECS, S.N.U.
Two pillars of information revolution: Si IC & photonics
“A chip that can transfer data using laser light”, NYT, 20060918

WDM
By Intel and UC-Santa Barbara
“$6 million project to develop silicon-based laser”, 20060804


By MIT, the Microphotonics Center from US DoD
Using nanocrystalline silicon as an sensitizer for Er
“Electronics-photonics must converge”, MIT, 20050520

By MIT from 3-year study
vs
Photonic Systems Lab
School of EECS, S.N.U.
Status of Photonics – Compared to Electronics
Electronics : Vacuum tube
Transistor
$’s per
IC
0.1M
VLSI
100’s M
Common User
Gb Memory
TO OPEN THE PHOTONIC AGE, compatible to that of semiconductor industry,
Need Cost reduction
Need Smaller Footprint
Need Integrated functionality
Need optical power lines (amplification function)
 25% of the material cost is in the package
Optical
Sub-Assembly
 Another 25% is in the assembly
 Beyond automation and off-shore assembly
 Packaging really hasn’t advanced much until recently



Costly 70’s-era technology
Poor signal integrity
Poor Thermal properties
Electronics
Sub-Assembly
Zolo Technologies
 Need : New materials, Athermal designs, and Packaging standards
Photonic Systems Lab
School of EECS, S.N.U.
Challenges and Promises

Challenges in achieving Photonic Age -- if it comes ^^;




Cost
Footprint
Functionality
10’s of $
Size of PCMCIA
More than Serial integration
Promises made to meet above challenges, with some technologies










OEIC
PLC
MEMS
EDWA
Hybrids
Si-Photonics
Plasmonics
Ph-Xtals
…..
….
mostly for active devices
mostly for passive devices
mostly for switching devices
mostly for amplification devices
Compound semiconductor
Silica, Polymer
Silicon, Glass, etc..
Silica

How far do we need to go ?

Is it really possible to meet these Promises ?

Let’s sit back and look at the status of Silica based technologies (EDWA / PLC )
Photonic Systems Lab
School of EECS, S.N.U.
Si Photonics – Optical Communication in the Chip
Leveraging
Photonics
Passives,
the astronomical Si processing technology for photonics
based on Si-based materials and Si-compatible processes
Modulators, Detectors but still missing Photon Generators
0.5 m bend
0.5 m splitter
Photonic Systems Lab
School of EECS, S.N.U.
What are we missing ? - Functionality / Size
Cost
Gain
E
D
F
A
E
D
W
A
Discrete
??????
Not allowed by physics
E
D
A
Integration
Discrete
components
Integration /
Functionality
Loss
 For Photonics, there is lack of optical power lines which is compatible to that of electrical PCB
 For Increased data rates, we need more photon (per bit) : Photon generator (amplifier, laser)
 For SOA, with its strong electron - electron interaction and high noise figure
 For EDWA, true integration is impossible after certain level (e.g. Splitting OA) : Only serial
Photonic Systems Lab
School of EECS, S.N.U.
Challenges in Amplifier Technology
OITDA2000
Gbps/user  need more photons/bit, but not with $1K/unit nor at current size !!
Photonic Systems Lab
School of EECS, S.N.U.
Amplifier evolution before millennium
Proposal for optical amplifier first published
J.E. Geusic and H.E.D. Scovil at ATT
Stimulated Raman scattering first observed
E. Woodbury and Won Ng
1964
optical fiber amplifier demonstrated
E. Snitzer using Nd doping for 1060-nm signals
1966
Proposal for glass light waveguides
By K.C. koa and G. A. Hockman
1970
First continuous operation of diode laser at room
temperature simultaneously demonstrated
Hayashi and M. panish at ATT.
and Z.I. Alferov at loffe institute(USSR)
Mass production of quality optical fiber
Corning
1976
First major trial of commercial lightwave system
Atlanta, Georgia(USA), without optical amplifiers
1983
First demonstration of doped single-mode fiber
By ATT
First demonstration of 1550-nm operation
Without optical amplifiers
5 wavelength CWDM in 1310-nm range
By Toshiba, using 5-nm spacing
First EDFAs simultaneously developed
R.J. mears, D. Payne. et al at university of
Southampton, and E. Desurvire, et al, at ATT
Diode-pumped EDFA demonstrated
By m. Nakazawa
First commercial EDFA introduced
By Oki Electric
First commercial SOA introduced
By BT&D Technologies (now Agilent)
1993
First major installation of optical amplifiers
By MCI
1996
First installation of EDFAs into undersea links
TPC-5 and TAT-12,13
1999
First EDWAs introduced
MOEC and Teem Photonics
1962
1984
1987
1989
Photonic Systems Lab
School of EECS, S.N.U.
Relative Cost (A.U.)
Amplifier – Bandwidth and Cost
4.0
3.0
2.0
1.0
Nortel Networks,1999
0
0
10
20
30
40
50
60
70
80
90
100
110
120
Bandwidth (nm)
Pump laser diode
Photodiode
Gain media (whatever)
Photodiode
DGFF
Tap
Isolator
Photonic Systems Lab
Pump
MUX
Tap
Isolator
School of EECS, S.N.U.
Amplifiers – Any challenges left ?

Optical Amplifier now 40 + year old, mature technology

Researchers have touched most issues on amplifiers






Gain flattening
Transient
Temperature
Power Conversion efficiency
Noise, Scattering, Fiber structure, Host materials, Co-Dopants
Various types of OAs have been commercialized, by numerous vendors






EDFA
TDFA
Raman
Hybrid
EDWA
SOA (bulk, QW, QD)

Not much issues left for OAs, especially for LH, trunk line applications (personal opinion)

Let’s sit back and look at the Technology / Bottlenecks of OA for Metro and beyond
Photonic Systems Lab
School of EECS, S.N.U.
Status of Amplifier – for Metro and beyond, to FTTH
Photonics : Different material / structure for each function, with high losses
1st Generation of Integration : Parallel (Laser, Detector, VOA arrays)
2nd Generation of Integration : Serial (ILM, Router, Receiver..)
Amplifier : 10-20 components with intensive package SERVING JUST ONE FUNCTION
NOT have been integrated with any other functional devices
Size reduction achieved to reasonable level to Metro, but not yet enough
Cost reduction achieved to reasonable level to Metro, but not yet enough
Is it possible to achieve above requirements with EDWA ?
Isolators
Er-doped Waveguide Gain
Block Array
Pump LD
For N=8 : 80  1 part
Hybrid integrated actives
Photodiode
Pump laser die
xN
PD
Photonic Systems Lab
Yields ? Delivery ?
School of EECS, S.N.U.
The current status of EDWA
 Exactly the same schematic as that of an EDFA
 Efforts on the amplifying section only : > 10 M$ to make the cheapest part cheaper
 Larger than the smallest available conventional EDFA
 Need every component in one plane
: severely restricts further reduction in size
 Integration no more than an addition : Amplifying splitter = AMP + splitter
Photonic Systems Lab
School of EECS, S.N.U.
Story behind the Limitation - Cost
Er energy level diagram
1.53m
0.98m
0.80m
0.66m
Energy level determined by QM

4F
9/2
4I
9/2
4I
11/2
4I
13/2
4I
Pump laser diode
$
15/2
Cost-centered view of an EDA
For EDFA & EDWA, optical excitation occurs through direct photon absorption


Very small absorption cross section (210-21 cm2)
 Requires a long interaction length btw pump and signal  Efficiency, Size
Narrow absorption band – requires finely tuned lasers
 Requires an expensive pump LD with wavelength (temperature) control  Cost
Photonic Systems Lab
School of EECS, S.N.U.
Story behind the Limitation - Structure
Reflecting
mirror
Half
MUX/
DMUX
Amplifying
Section
Splitting
Section
Serial Integration : Lower Yield
2-D Structure : Point Access Impossible

PLC and EDWA shares the same platform but the real integration is Difficult

Limited to serial integration

Point amplification impossible  Series of resistors and filters adds in system noise

Photonic Lightwave circuit without optical power line, EDWA as a mimic of EDFA
Photonic Systems Lab
 Marginal reduction in the footprint with lower chip yield
School of EECS, S.N.U.
Story behind the Limitation - Material

PLC is a stabilized, patterned fiber arrays using the same material

Mode size limitation dictates the minimum device size (much bigger than memory chip)

Wafer uniformity affects the yield of the chip  higher index for smaller device size

To keep the Er numbers same within smaller volume, concentration have to be much higher

Increased Er concentration  much lower ( ~ x 2 ) PCE from the quenching process
Photonic Systems Lab
School of EECS, S.N.U.
Faults of Integrated Amplifiers proposed so far

EDFA on a substrate



Still requires an expensive pump LD




Dictates the smallest possible size of EDWA
Not different at all when compared to EDFA again
Current OA technology not enough to support for metro – access network




Transfers the control over the final price of the device to LD suppliers
The better you are, the worse this problem gets!
The smaller you get, you lose more pump power from Er quenching


Similar properties under similar conditions
 competes against established products with only an incremental advantage
Can never be integrated with anything else
 can never truly “siliconize” photonics
Cost
Footprint
Functionality
Too high Photon Price, dictated by Electrical – Optical – Optical pumping
Limited by Erbium on Silica wafer
Limited by 2-D structure
Any solutions… ?
Photonic Systems Lab
School of EECS, S.N.U.
Contemplations on Photon Generators

Compound Semiconductor








Silica base Rare Earth







Bandgap - electrical : fast, strong interaction  modulation, switching
Strong interaction  Smaller device size
Energy source (electrical pump) independent from signal plane
Feedback structure : LED  FP, DFB but at much increased cost
Bandgap engineering  Wider, adjustable bandgap
Difficulties in pigtailing  Cost
Differences in refractive index with fiber  AR coating for SOA
Atomic level - optical : slow, weak interaction  amplification without crosstalk
Weak interaction  Larger device size, Low efficiency
Energy source (optical pump) requires waveguide
Feedback structure : Fiber laser but no modulation capability
Bandgap engineering  None
Compatibility in Pigtailing
Next generation Optical Amplifier – photon wavelength converter



Eliminate an expensive LD source : just need to provide inversion
Require dimensional separation of Pump and Signal plane
Need stronger interaction mechanism for the excitation
Photonic Systems Lab
 COST
 FUNCTIONALITY
 FOOTPRINTS
School of EECS, S.N.U.
Si-Photonic Optical Amplifier ?
Photonic Systems Lab
School of EECS, S.N.U.
Nanocrystal-Si sensitized EDWA
Amplifier is Wavelength
Converter in its nature
Pump photons
Interacting medium
Signal photons
Conversion
mechanism
Si nanoclusters
Why do we use expensive
coherent photons ?
Signal photons
Er ions
20 nm
SiO 2
(host matrix)
Photonic Systems Lab
School of EECS, S.N.U.
Material properties
5
4
10
PL Intensity (a.u.)
Intensity (a.u.)
10
with 477 nm pump
with 980 nm pump
0
10
-1
10
-2
10
0
10 20 30 40
Time (msec)
3
10
2
10
1.30
1.35
1.40
1.45
1.50
1.55
1.60
1.65
Wavelength (m)
 Continuous excitation spectra from IR to UV: anything bluer than green works!
 No need of Frequency control (or cooling)
 >100 times Er3+ luminescence intensity even with half the photon flux
 nc-Si completely dominates excitation (100 x larger with 477nm than 980nm)
Photonic Systems Lab
School of EECS, S.N.U.
SRSO / Er layer deposition

Deposition process




ECR – PECVD (electron cyclotron resonant plasma enhanced chemical deposition)
Silicon rich silicon oxide to construct silicon nanocluster
Silicon contents control : Ar , SiH4, O2 (automated MFC)
Evaporation / sputtering  Er
Microwave
Ar
plasma
Er Target
Ar gas
(with negative bias)
(99.9999%)
Load lock
Chamber
SiH 4, O2 gas
(99.999%)
Sample
holder
TMP
Photonic Systems Lab
School of EECS, S.N.U.
Silicon nanocrystal control

Silicon nanocrystal

Controlled by RTA annealing condition / silicon contents

Best energy coupling condition to Er ions

Size control : Quantum energy state of nanocrystal

State control : Crystal or amorphous
Photonic Systems Lab
School of EECS, S.N.U.
Bulk performance measurement

PL measurement

PL intensity and lifetime for various pump wavelength (980nm for Er, 477nm for NC-Er)


Activity of silicon nanocluster and Er, coupling efficiency
RBS measurement

Atomic composition estimation for layer depth
Photonic Systems Lab
School of EECS, S.N.U.
Waveguide characterization (amplification)
H 2 mm
LED array
W 1,000 m x L 5 mm
Ridge Waveguide
W 8 m x L 11 mm
 Butt-coupled tapered fibers for signal input and output
 15mm linear array of commercial, 470 nm LEDs
 Need to clear the fibers and cover glass: 2mm separation between LED and waveguide,
pump only center 5mm portion of the waveguide
Photonic Systems Lab
School of EECS, S.N.U.
2
2
Laser on (24W/cm )
2
Laser on (16W/cm )
LED on
Pump off
Signal Intensity
3
Signal Intensity (a.u.)
Signal Change (dB/cm)
Wavelength-dependence of signal change
1
0
-1
-2
1292
1294
Wavelength (nm)
LED On
LED Off
a)
-3
1515
1520
1525
1530
1535
1540
1545
1532
Wavelength (nm)
1533
Wavelength (nm)
 Signal change: Itrans(P)/Itrans(0): typical inversion curves for Er3+
 With LED pump: low pump density due to unoptimized alignment
 lower inversion, optical gain at 1545 nm
 Simulation of high LED pump power with 477 nm laser:
 full inversion with 3 dB/cm optical gain at 1533 nm
Photonic Systems Lab
School of EECS, S.N.U.
Numerical Assessment / Design

Parameters from experimental results




Much larger effective excitation cross-section and signal absorption cross-section
Emission cross-section from PL measurement
Absorption cross-section from McCumber relation
Simulation scheme (top pumped NC-Si EDWA)


2-D propagation equations (with 10x10x400 segments)
1500 ~ 1610 nm with 1nm resolution
Photonic Systems Lab
School of EECS, S.N.U.
Population inversion characteristics of NC-Si Er

Much larger pump absorption cross-section than signal emission cross-section
Over 50% inversion with small # of pump photons (Left-shift of red region in below figures)
 Top pumping scheme
 Large doping area than conventional EDF



Doping area confined to the center of the fiber core for high inversion (conventional EDF)
Large doping area Enhancement of overlap factor with signal  high gain per length
Inversion of NC-Si EDWA
Signal intensity (dBm/cm2)
Signal intensity (dBm/cm2)
Inversion of conventional EDFA
Population
inversion
Pump intensity (dBm/cm2)
Photonic Systems Lab
Pump intensity (dBm/cm2)
School of EECS, S.N.U.
Device Structure & Feasibility

Performance comparison
4 cm EDWA without coupling loss
 NC-Si EDWA with type A core (7x7 μm2)
 Type A core EDWA with bottom mirror (100% reflection)



Large gain by reusing of wasted pump power
Adiabatic designed large core (type B, 100x7 μm2)


Saturation gain enhancement by increasing pump collection area
Small signal gain enhancement by overlap factor enhancement
Type A
Photonic Systems Lab
w/ mirror
School of EECS, S.N.U.
Optimization : Pump LEDs

High intensity visible (blue) pump LED

Chip LED for illumination application (Cree)

Max 25W/cm2 (hard contact)

Easy to align (waveguide width 50um < LED 250um)
7mW(at 20mA) x 64
Chip size : 300um x 300 um
Array size : 0.03(cm) x 3(cm)
Total Power : 5W/cm2
LED Die-Bonding pattern
Photonic Systems Lab
Emission of LED Array(0.3x3 cm)
School of EECS, S.N.U.
Optimization : Material Composition
Evaporation
Sputtering
Estimated result
Gain: 2.4dB/cm
PL : 93000
LT : 6.3 ms
x 17
x 12
Experimental result
Gain: 0.2dB/cm
PL : 8000
LT : 9.3 ms
Photonic Systems Lab
School of EECS, S.N.U.
Examples & Implications in the applications
Amplifying
Section
Amplifying
Section
Pump WDMs
Splitting
Section
Pump & Signal
Totally NEW concept, Reduced Complexity & Higher Chip Yield !
LED Pump array
SRSO wafer
Amplifying
Splitter
Pump light
Ultra-compact, low-cost
Schematic of a SRSO based VCPAC (pump WDM removed)
Photonic Systems Lab
True integration for Active PLC
VCPAC Splitter (Splitting section = Amplifying section)
School of EECS, S.N.U.
Summary
Much things you can do with NANO Si !!
That’s a good news for Photonics Engineers
# of Amplifier Worldwide
# of nano-particle Worldwide
# of Amplifier Engineers
Photonic Systems Lab
School of EECS, S.N.U.
Photonic Systems Lab
School of EECS, S.N.U.