The evolution and impact of photonics components

Transcription

The evolution and impact of photonics components
The evolution and impact of
photonics components
David N Payne
Director: ORC Southampton University
Chairman: Southampton Photonics Inc
Optical Fibre Group 1972
Telecom growth remains healthy – UK Oftel
US internet traffic still growing at x4 per year – Roberts et al. 2001, Caspian Networks
The art of prediction
“Prediction is very difficult, especially about the future”
Niels Bohr, Nobel Laureate
"The future isn't what it used to be !"
IBM PC executive 1992
Great predictions of the 20th century
ƒ
"Stocks have reached what looks like a permanently high plateau.“
-- Irving Fisher, Professor of Economics, Yale University, 1929
ƒ
"Everything that can be invented has been invented."
-- Charles H. Duell, Commissioner, U.S. Office of Patents, 1899
ƒ
"The herd instinct among forecasters makes sheep look like
independent thinkers. " --Edgar R. Fiedler
ƒ
“There's no-one who likes the PC who doesn't like Microsoft"
--Bill Gates, Free Market and the LA Times
“Bandwidth is like health-care – you can never have too much”
-- David Payne ECOC 2000
19th Century technical predictions
Laszlo Solymar:
Solymar: Getting the message
Dubious investments of the twentieth century?
“This 'telephone' has too many shortcomings to be seriously considered as a
means of communication. The device is inherently of no value to us.”
-- Western Union internal memo, 1876.
Laszlo Solymar:
Solymar: Getting the message
1000
ECOC/OFC
Reported fibre
capacity Gbit/s
(growth 1.7/year)
100
Te
ch
no
Bu
lo
gy
si
ne
ss
10,000
Sound investments of the twenty-first century?
Key issues in optical fibre telecommunications
•
•
•
•
•
•
•
WDM and TDM - The optimum mix
Broadband amplifiers for 1.2 – 1.75 µm
Dispersion and polarization management
Optical regeneration
Optical networks
IP on optics
Integration and cost reduction
Optical spectral efficiency
ƒ Fibre transmission window from 1.1µm – 1.7µm
ƒ ~ 90 THz available
ƒ Amplifiers are the limiting technology (Fibre, Raman, Diode?)
ƒ Towards 3 bit/Hz requires better wavelength control
Spectral Region covered by Fiber Amplifiers
PDFA : Praseodymium-doped FA
TDFA : Thulium-doped FA
EDTFA : Erbium-doped Tellurite FA
35
1.55µm
(C-band)
EDFA
1.65µm
Band
EDTFA
1.2
Gain (dB)
30
1.0
25
0.8
20
1.3µm Band
PDFA
ORC TDFA
15
0.6
1.58µm
(L-band)
EDFA
0.4
Loss < 0.35dB/km
380 nm (52 THz)
10
0.2
5
0
0
1.2
1.3
1.4
1.5
Wavelength (µm)
1.6
1.7
Fiber Loss (dB/km)
1.45µm Band
TDFA
Thulium-doped fibre amplifier
16
thulium
erbium
1.0
Internal gain (dB)
Normalised intensity
1.2
0.8
0.6
150nm
0.4
12
8
Length 70cm
NA 0.4
4
0.2
0.0
0
1300
1400
1500
1600
1700
1460
1480
wavelength (nm)
• Telluride glass fibre
• 800nm/1064 double pump scheme
1500
1520
Wavelength (nm)
1540
The Inventory Problem
The problem
ƒ
Components (transmitters, add/drops, routers) are
DWDM channel-specific
ƒ
Large inventory required for 160-channel systems
The solution
ƒ
ƒ
ƒ
Make everything tunable (or pre-settable)
The plug-and-play network?
New fail-safe component configurations
Remote provisioning?
Tunable Lasers
ƒ
ƒ
ƒ
ƒ
ƒ
Wide tunability allows wavelength self-routing
Reduces the need for space switching
Must be discretely tuned to the ITU grid
Dark while tracking
No mode-hopping, drift or increase in noise
ƒ These are tough requirements!
SG-DBR Tunable Laser Source
Module 9831L Tuning Comb; Superimposed
0
-5
-10
-15
Output (dB rel.)
-20
-25
-30
-35
-40
-45
-50
-55
-60
1525
1530
1535
1540
1545
1550
Wavelength (nm
1555
1560
1565
1570
1575
Tunable grating filters and DFB lasers
•
•
5
0.6
L-band
C-band
0
0.4
-5
0.2
-10
0.0
-15
-0.2
-20
-25
Res.BW
= 0.05 nm
-30
1530
1550
1570
1590
1610
-0.4
-0.6
1630
Compression-tunable fibre DFB-laser
Continuous hop-free tuning
40
- 22.5 nm
30
+ 4.5 nm
20
10
Power [dBm]
Compression-tunable Bragg grating filter
Record 70nm continuous tuning-range
(1544nm – 1614nm)
3-dB bandwidth change (nm)
normalized reflectivity (dB)
•
0
-10
-20
-30
-40
-50
-60
-70
1520
1525
wavelength (nm)
M. Ibsen et al., ORC , 2002.
M. Ibsen et al., “Advanced fibre Bragg grating devices”, AIP-ACOFT’02.
1530
1535
1540
1545
Wavelength [nm]
1550
1555
Optical Switching
•
Large NxN switch matrix required (128x128) for networks
– speed 1ms
•
Fast (ps), sensitive switches required for processing
– reshaping, switching
Slow optical switching
■
■
■
■
The most effective optical switch is mechanical!
The smaller the mass the faster the switch
Speed in ms range at present
VOA and NxN switches commercial
Ultra-fast switching:
Holey Fibres at the ORC…
Background
Fibre core
Silica holey fibre
•Core formed by solid region
•Air holes act as fibre cladding
•Wavelength-scale features lead to
novel optical properties
•Made by stacking capillaries
2
Small core (3um )
→Tight mode confinement
+
large air-fraction cladding →Low switching threshold
Switching/regeneration
Results
All optical switch using 3m HF, 16W
switching power
Output Power (W)
(a)
noisy data
0.04
(b)
10%
0.03
10%
0.02
Transfer function of
the data regeneration
device
0.01
(c)
regenerated data
0.00
0
10
20
30
40
Input Power (W)
• Further large reductions in switching power by using soft glass
Soft-glass preform extrusion
heat
ram
bulk
glass
Advantages:
•Controllable/Reproducible
•Good surface quality
•Efficient use of glass
die
• Fewer interfaces c.f. stacking
• Wider range of profiles possible
• Suitable for many materials
SF57 Extruded preform
@1550nm
• Long
(>10cm) complex
extruded preform
• Core isolated by long, thin
supports
0.37mm
• Good uniformity and
straightness
• Excellent surface quality
OD = 16mm
Can be polarization preserving
Dispersion ≈100ps/nm/km (anomalous)
SF57 glass has nonlinearity 20x silica
PC PM Isolator PC
High Power
Er/Yb Amp
PC
1600~1640nm
Tunable Laser
1530/1630
WDM Coupler
Lenses
75m Holey Fibre
Photo-receiver
Oscilloscope
50/50
Coupler
Lenses
AOTF
Internal Gain (dB)
Pump Laser
λ: 1536nm
45
45
40
40
35
35
30
30
25
25
20
20
15
15
10
10
5
5
0
0
1605 1610 1615 1620 1625 1630 1635 1640 1645
Wavelength (nm)
Optical
Spectrum
Analyser
• High nonlinearity fiber gives reduced device length/pump power
• Pumped using high power amplifier
• Gains up to 42dB, noise figure 6dB demonstrated
Internal Noise Figure (dB)
Holey Fibre Raman Amplifier
Wavelength Conversion through
a Nonlinear Optical Loop Mirror
DSF 1 km
λo: 1550nm
PC
50:50
coupler
10GHz Control Pulses
Wavelength
Converted Pulses
EDFA
Continuous Wave
Filter
-15
1.0
Original Input Pulse
Wavelength Converted Pulse
SHG Intensity (arb. unit)
-20
Arb. Amplitude (dB)
-25
-30
-35
-40
-45
3.5ps
0.5
2.5ps
-50
0.0
-55
1540
1541
1542
1543
1544
1545
Wavelength (nm)
1546
1547
1548
-25
-20
-15
-10
-5
0
5
Time Delay (ps)
10
15
20
25
Pervasive power
■
■
■
The cost of pump diodes is dropping rapidly
Costs drop per Watt for large-area high-power devices
Costs will be driven by the emerging industrial solid-state
laser market (cutting, welding, defense)
Opportunities for: parallel-pumping of telecom devices
short-pulse instrumentation
X-ray generation
The Optical Power Rail
Advantages
ƒ
ƒ
ƒ
ƒ
Power management at convenient locations
Cost savings by sharing reduced no. of SDLs
Built-in SDL pump redundancy
Inventory problem solved by non-channel-specific SDLs
Cladding Pumped Fibre Device
•
•
•
•
•
4 July 2001
Large inner cladding guides MM pump beam
RE-doped core converts pump to high brightness signal
Core diameter – 5- 50 mm
Pump cladding diameter 50 – 500 mm
Device length 1 – 100 m (depends on core/cladding ratio)
Optical Amplifiers and Their Applications
Stresa, Italy
976 nm Fibre Laser:
Yb-Doped Fibre
Core
Clad
4 July 2001
•
•
•
•
•
Pump cladding – 35 µm
Pump NA= 0.4 nm
Fibre efficiency > 70%
Device length – < 40 cm
2W output at 976nm
Optical Amplifiers and Their Applications
Stresa, Italy
8-channel DFB Array
M odule Layout
977nm CW Yb-fibre-laser
Wall plug efficiency to
be added here
980nm 1 x 8 splitter 1550nm fibre DFB-laser 50GHz array
(3dB couplers @ 980nm)
Channel 1
Channel 2
Channel 3
Channel 5
Outputs
Channel 4
Channel 6
Channel 7
•
•
•
Single fiber pump source
10 – 16 dBm channel output
Shared pump prevents channel failure
DFB laser power [m W ] Output power [dBm ]
Channel 8
Chan 1
a)
10
0
-10
-20
-30
-40
-50
-60
-70
1550.4
Chan 8
1551.2
1552.0
1552.8
1553.6
1554.4
W avelength [nm ]
b)
10
8
6
4
2
0
0
100
200
300
400
500
600
700
800
977nm pum p incident in 1x8 splitter [m W ]
Strictly Private. Not to be distributed without prior consent of Southampton Photonics, Inc.
900
1000
Saturated Output Power, W
High Power Er/Yb Fibre Amplifier
6
4
•
•
2
•
•
5.5 W output power at 1550nm
Efficiency close to theoretical
limit
Small signal gain >50dB
Noise < 5dB 1533 -1560nm
0
0
2
4
6
8
10
12
Absorbed Power, W
Scalable to 1kW and above?
4 July 2001
Optical Amplifiers and Their Applications
Stresa, Italy
Optical Component Cost Is the Problem
ƒ
Long-Haul is relatively insensitive to cost
ƒ
Metro and FTTH are very cost sensitive
ƒ
Optical components and subsystems assembly is still done manually!
ƒ
New low-cost manufacturing techniques - replication, stamping, polymers
ƒ
Automated process machinery becoming available
There is no optical equivalent of silicon!
How far have we got with partial integration?
Just how integrated is electronics?
PC motherboard
Photonics in its infancy
1948 Manchester Mk1 valve computer
Photonic integration
• 3-dimensional optical circuits give increased freedom
New planar waveguide technologies
• Parallelism of WDM has driven planar optical circuits
• New low-cost technologies needed (polymers, replication etc)
• Thermal stability a major issue
• 3-dimensional waveguides?
• Why not compound glass?
•Will we sacrifice performance for cost?
Silica-on-silicon glass Deposition and Core Structures
Glass deposition in FHD
Core ridge structures in
directional coupler after RIE
Embedded core
Close-up View of AWG
Phase-Error Compensation using UV Irradiation
UV laser light
1st and 2nd
metal masks
Half-wave plate
Metal mask
Lk
#k
Array waveguides
(a) Collective phase trimming
(b) UV irradiation through metal masks
(c) Transmission characteristics of 160ch-10GHz AWG
NTT
Photonics Labs.
PLC Micro-mirror for Surface Coupling
SMF
z PLC
d
x
y
Cross-sectional configuration
‚š
‚™
‚˜
Bird’-eye view
Cross-sectional photograph
Silica Platform + Flip-chip MO E M S
Silica waveguides
8 array M E M S chip
Philippe LEOP O LD – PLM Hybrids – August 2002
8 array M E M S chip flip-chip
mounted on SiO2/Si
hybridisation platform for 8
channel VOA
Passive/MOE M S Hybrid Integration
A W G +VOA+TD A
TDA
Philippe LEOP O LD – PLM Hybrids – August 2002
VOA
AWG
Glass or semiconductor technology?
Cheap optical fibres
Expensive components
• Can we bridge the gap?
Planar Amplifiers
Particularly
suited to high
concentration
rare-earths
Signal
Grating gainmonitor
flattening or ASE
filter
Output
Pump
monitor
1-10 cm
Rare-earth doped
compound-glass substrate
Pump
Signal
Hot Dip Spin Coating of
Planar Glass Waveguides
Boron nitride
vacuum chuck
o
430C
Nd-doped
spin coated film
o
1000C
Undoped molten
overcladding glass
Undoped glass
substrate
Neodymium doped
glass film (5µm)
Undoped overcladding
Glass (20µm)
•
UV writing lowers refractive index of channel.
•
Use 2-Step process to produce 5µm thick planar waveguide.
•
Guide between channels pumped at 800nm (100mW).
•
Lasing at 1317nm observed.
or photorefractivity
•
Cerium doping leads to
strong UV absorption
•
CW exposure with
frequency doubled argon
ion laser and 6u spot
•
Writing speed 1.2630mm/min
•
Laser power 0.5W
CD and DVD rewritable technology is thermal!
Channel Waveguides in GLS glass
a)
0.03m/min
0.3m/min
0.5m/min
(a)
(c)
1.0m/min
0.6
0.4
0.2
h
(µ
m
1.5m/min
(b)
Beam Depth ~ 7.21 µm
Beam Width ~ 7.79 µm
0.8
(norma
Beam Intensity
lised)
1.0
)
0.7m/min
Losses 0.38 ± 0.05 dB/cm
Width
(µm)
Be
Beam
am
De
pt
0.0
Simultaneously UV-Written Planar Waveguides
and Bragg Gratings
Taper - width +
refractive index
Tilted grating
add/drop
Encoded grating
Fibre connected first,
then waveguides
written to the fibre
core
Phase shifts
•
•
•
Gratings can be chirped, appodised and contain phase shifts
Waveguides can be aligned to the fibre rather than the fibre to the waveguide
Mode converting tapers
UV-written waveguides with Bragg gratings
Planar silica waveguide with UV written Bragg grating
65
1565
55
Peak Reflected Wavelength / nm
Transmission/reflection
60
50
45
40
35
30
25
20
15
10
5
0
1557.6
1557.8
1558.0
1558.2
1558.4
1558.6
1558.8
1559.0
Wavelength (nm)
• Silica-on-silicon waveguides
1560
1555
1550
1545
1540
1535
1530
526
528
530
532
534
Strobe Period / nm
Wavelength shift by
write-strobing as
phase mask moves
536
538
Combining fibre and planar technology
for low-cost production
Up to 64
doped waveguides
Slab
waveguide
preform
10cm x 1cm
Etched pit for
field access
Bragg
gratings
Overlay for
post-processed
coupler
Furnace
Fibre
ribbon
connector
5mm x 0.5mm
Ribbon draw
Femtosecond Direct-writing:
The Principle
• Tight focusing of laser into glass
• High intensity leading to multiphoton absorption
• Structural changes in matter
confined to focal volume due to
short pulse duration – 3-D
• Photosensitivity not required
• -ve or +ve index changes
Intensity ~ 1014 W/cm2
Temperature ~ 106 K
Pressure ~ 106 bar
Hirao Active Glass Project
Microscope Images of ‘Fiber Gratings’
Viewing microscope focused:
Gratings :
Period = 2 µm
Depth = 5 µm
Form:
•1-d
•2-d
•Hexagonal
5 µm below surface
λc = 600 nm
d = 125 µm
on surface
λc = 600 nm
d = 125 µm
5 µm below surface
2 µm
Velocity = 200 µm/s
2 µm
Shutter open 3 ms
5 µm below surface
λc = 1250 nm
d = 125 µm
2 µm
Shutter open 3 ms
Prospects for full
integration
• Full integration requires light generation, modulation,
switching, routing, filtering and detection
• An isolator is really difficult
• No current photonics platform satisfies these
requirements
• What are the possibilities?
(Silicon) Industry Roadmap v time
Optical fibre/
Human hair
smaller
10
Photonic Crystals
5
Quantum Dots
(IC Physics)
1
0.5
Molecular scales
0.1
Nanotechnology
0.05
1970
1980
Acknowledgements: W J Stewart
1990
2000
2010
Photonic band gap integration
technology
• High confinement implies dense circuits
• Could match electronic densities
Problems:
• Very high precision required to avoid spurious resonances
• High levels of backscatter make amplifiers/lasers difficult
Acknowledgement: W J Stewart
Photonic Integrated Circuits
Joannopoulos, MIT
Conclusions
• Despite progress large-scale component integration
remains a dream
• There is no winning platform technology for
integration
• Glass and III/IV semiconductors are the current
leaders
• Thermal stabilization remains a problem
• Hybrid technologies offer better performance and
higher power
• Cost is still an issue and can be reduced by by new
approaches