Lecture: Coupling light to silicon photonic circuits coupling.pdf · Lecture: Coupling light to...

Lecture: Coupling light tosilicon photonic circuits

Wim BogaertsVersion: November 2009

Credit where it’s due...

This lecture material was only possible thanks toSeveral people in the photonics group:

Dirk Taillaert, Gunther Roelkens, Pieter Dumon, Frederik Van Laere, Stijn Scheerlinck, Shankar Kumar Selvaraja, Diedrik Vermeulen, Marie Verbist, Karel Van Acoleyen, Kei Watanabe

Lars Zimmermann, Karsten Voigt (TU Berlin)Tolga Tekin (IZM)Cary Gunn (Luxtera)Badhise Ben Bakir, Jean-Marc Fédéli (CEA-LETI)

Silicon Photonics –PhD course prepared within FP7-224312 Helios project

http://www.luxtera.com/index.php?PHPSESSID=88afeb7e4ccac719aeec80337a72ca33�

Integration of circuits

Photonics is going through the same evolution as electronics

smaller building blocksmore functions on a chip

intel 40041971

pentium 42002

transistor radio1954

discrete elements

photonic IC

nanophotonic ICSilicon Photonics –PhD course prepared within FP7-224312 Helios project

Increasing Index Contrast

Low Contrast - Fiber Matched(silica or polymer based)

Bend Radius ~ 5 mmSize ~ several cm2

Medium Contrast (InP-InGaAsP)

Bend Radius ~ 500µm

5 mm

Ultra-high Contrast (SOI based)

Bend Radius < 5µm

50 µ

m5

cm


Coupling to nanophotonics

The problemefficient coupling between a submicrometre waveguide and a fiberspot-size converter needed :

• in plane (horizontal)

• out-of-plane (vertical) : more difficult

the polarization problem

Single-mode fiber

1µm

SOI wire


Overview

IntroductionHorizontal incoupling: inverted tapersVertical incoupling: grating couplers

operating principlesthe base lineimproving efficiency, bandwidth, polarizationfiber arrays and packaging

Using grating couplersbasic measurements with fiberswafer-scale testingfor free-space communication

ConclusionSilicon Photonics –PhD course prepared within FP7-224312 Helios project

Fiber-chip couplingRegular (planar) taper

Multi-modeFacet coating requiredNo vertical matching

3-D taperDifficult to fabricateWafer scale?


The inverted taper

Narrow waveguide tipmode is ‘squeezed out’ of corecaptured by overlay waveguide

High NAfiber

inverted taper

to circuit


The inverted taper

Broad wavelength rangeSingle modeEasy to fabricate (if you can make the tips)Low facet reflections

inverted taper

to circuit

Shoji et al. EL 38, p.1669 (2002)McNab et al. OpEx 11(22), P. 2927 (2003)


Coupling for both polarizations

(lensed fiber mfd 3.5µm)


XY alignment tolerances SiNx waveguide (3µm x 3µm) 3µm fiber-spot

1dB: 1µm3µm

3µm

XY Alignment tolerances


Both polarizations?

Inverted taper coupler couples both polarizationsTE and TM in the photonic wire

BUT: photonic wires are very polarization sensitiveYou want just one polarization in your wire

Solutionpolarization splitterpolarization-diversity approach

inverted taper

to circuit


Polarization diversity

Process both polarizations separatelysplit polarizationsconvert to the same polarization on the chipcombine polarization back into the fiber

two identical circuits

TE/TM

TMTE

TETM

TE/TM TE/TM


Polarization splitter/rotator

Short, efficient deviceNot trivial to make (multi-layer, sharp tips)

Watts et al, OL 30(9), p.937 (2005)TE/TM

TM

TETE


http://web.mit.edu/site/aboutsite.html�

Horizontal coupling

High coupling efficiencyBroadband operationWorks for both polarizations

But:Large footprint on a ‘nanophotonic’ chipRequires post-processing

• dicing and polishing• no wafer-scale testing possible

Alignment tolerances• larger spot is larger gives better tolerances• larger spot is harder to fabricate• larger spot needs longer taper

inverted taper

to circuit


Overview





10

TE

Grating Fiber CouplerCompatible with SMF-28

No need for a polished facet

Wafer-scale testing Wafer-level packaging

Flexible and cheap!

adiabatic taper

10µm wide waveguidegrating

single-mode fibre,

Taillaert et al, JQE 38(7), p.949 (2002)Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Operating principle

superstrateradius = k0nsup

K=2π/Λ

substrateradius = k0nsub

kz

Λ

βz

β

1-D grating coupler

Experimental results (λ=630nm,depth=70nm, TE pol.)31 % efficiency (5.1 dB coupling loss)40nm 1dB bandwidth

Also acts as a broadband filter

deep trench

shallow grating

Taillaert et al. JJAP 45(8A), p.6071 (2006)Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Bogaerts et al. JLT 23(1), p.401 (2005)

Fabrication

Fabrication with CMOS toolsdeep UV litho (193nm) + dry etch + strip

• shallow etch (gratings)• deep etch (waveguides)

Overlay alignment: ~15nm

deep trench

shallow grating


10

TE

Grating Fiber Coupler

What aboutEfficiency?Alignment tolerance?Optical bandwidth?Polarization dependence?Footprint? Packaging?

adiabatic taper


single-mode fibre,


Coupling efficiencyEfficiency is determined by two factors

Directionality: Fraction sent to the right diffraction orderMode matching: diffracted field profile vs. fiber mode

Taillaert et al., OL 29(23), p.2749 (2004) Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Directionality

Second-order grating: K = βElegant: Vertical outcouplingBackreflection!

superstrate

substrate

K=2π/Λ

β

BACKREFLECTION!

TE


Cancel out backreflection

Additional backreflector in front of gratingcancels out grating reflectionwavelength dependentvery accurate positioning required!

Roelkens et al, OL 32(11), p.1495 (2007)

incident

coupled

backreflected

additionalreflector


10

TE

Directionality

Detuned second-order grating: K < βOutcoupling in the same direction as waveguide

square grating: most coupling in first ordersome residual coupling in second order

superstrate

substrate

K=2π/Λ

β


Directionality

Detuned second-order grating: K > βOutcoupling in the opposite direction as waveguide

No second order to couple to

superstrate

substrate

K=2π/Λ

β

10

TE


Directionality

First diffraction order: to top and bottom

Select only the top oneSuppress the bottom one

superstrate

substrate

K=2π/Λ

β

2µm SiO2

silicon substrate

220nm Si


Directionality

Extended grating teethextra degree of freedom, break symmetry

Roelkens et al, OpEx 14(24), p.11622 (2006)

2µm SiO2

silicon substrate

220nm Si

380nm teeth


Directionality condition

Constructive Interference Δφ = 2π

Δφ = πDestructive Interference

D1 D2W2W1

Δφ = 2π

Δφ = π


Extended grating teeth: epitaxy Uniform grating defined in Silicon overlay

simulations: 80% coupling efficiency possible First Demonstration with good results:

Epitaxially grown Silicon (LETI) 55% coupling efficiency -1dB/3dB bandwidth of 50nm/100nm

Roelkens et al., IPNRA 2008Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Extended teeth: deposition + etching

2µm SiO2

silicon substrate

220nm Siβ

amorphous/poly-silicon

D. Vermeulen, GFP 2009, PD1

MeasurementsCoupling efficiency = -1.6 dB (68%)

-5.0

-4.5

-4.0

-3.5

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

1510 1520 1530 1540 1550 1560 1570

Inse

rtio

n lo

ss fo

r 1 c

oupl

er [d

B]

Wavelength [nm]


Poly-Si vs. amorphous Si Wavelength shift due to refractive index difference a-Si poly-Si Coupling efficiency is of the same order

-6

-5.5

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

1470 1490 1510 1530 1550 1570

Cou

plin

g Ef

ficie

ncy

(dB

)

Wavelength (nm)

a-Si overlay poly-Si overlay


Directionality

“Recycling” mirror:recycle bottom diffraction to the topmirror distance is critical: constructive interference

Van Laere et al., JLT 25(1), p.151 (2007)

2µm SiO2

silicon substrate

220nm Si

mirror


Adding a “bottom” mirror

Make a top mirrorBCB spin coating + gold evaporationTransfer circuit to other carrier + substrate removal


BCB spin-coat

2µm SiO2

silicon substrate

220nm Si

Gold mirror


BCB

silicon carrier




BCB spin-coat

2µm SiO2

silicon substrate

220nm Si

Gold mirror


BCB

silicon carrier




BCB spin-coat

220nm Si

Gold mirror


Bottom gold mirror

Van Laere et al., JLT 25(1), p.151 (2007) Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Bottom gold mirrorGold bottom mirror:

mirror on top of gratingupside down on carriersubstrate removal

70% efficiency (-1.5dB)

Van Laere et al., JLT 25(1), p.151 (2007) Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Silicon bottom mirror

Bottom Distributed Bragg mirrorNot possible with crystalline SOIPossible with high-quality amorphous silicon

silicon substrate

220nm aSi

Selvaraja et al. CLEO/IQEC 2009

aSi DBR


Silicon bottom mirrorSilicon bottom mirror: DBR

Replacement of gold by a SiO2/a-Si DBR Deposited amorphous Silicon waveguide (no bonding)Coupling efficiency: 70% (-1.5dB)

DBR mirror

Selvaraja et al. CLEO/IQEC 2009Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Coupling efficiencyEfficiency is determined by two factors

Directionality: Fraction sent to the right diffraction orderMode matching: diffracted field profile vs. fiber mode


Modal overlap

Mismatch between diffracted field and fiber mode

fiber mode

diffractedintensity

Penalty > 1dB


Optimized onuniform grating

Optimized : > 90% efficiency

Challenging to fabricatesmallest groove width is 30nm, but can be increased to 60nm with only 1-2 percent decreased efficiencybottom reflector (when using regular SOI the efficiency is 63%)

narrow grooves(weak coupling) broad grooves (50%)

DBR for directionality


Optimized nonuniform grating

Efficiency versus wavelength

1500 1520 1540 1560 1580 1600

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1SOISOI+DBR

wavelength (nm)

effic

ienc

y

1dB coupling loss

> 90%

> 60%


10

TE


What aboutEfficiency?Alignment tolerance?Optical bandwidth?Polarization dependence?Footprint? Packaging?

adiabatic taper


single-mode fibre,


Diffraction grating: alignmentGood alignment tolerances are achieved since we are aligning modes with a mode field diameter of about 10µm!

Alignment sensitivity measurement

diameter of the -0.5dB contour is 2µm

10

TE

Grating Coupler: Footprint

adiabatic taper


single-mode fibre,

10µm → 0.5µmMin length ~150µm


Adiabatic taper

“Adiabatic” Taper of 10µm to 1µm

0 10 20 30 40 50 60

0.2

0.4

0.6

0.8

1

Taper length [µm]

Tran

smis

sion

parabolic

linear


Focusing grating couplersCurved gratings: focus light in submicron waveguides

No adiabatic transition neededGrating in linear taperGrating in slab, focus on low-contrast aperture

Van Laere et al, PTL 19(23), pp. 1921 (2005)Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Focusing grating couplers

Grating in taperfocusing on taper point

Grating in slabFocusing on waveguide aperture (shallow etched)


Focusing grating couplers


Diffraction gratings: bandwidthTypical bandwidth of a grating coupler

Gaussian coupling spectrum1dB bandwidth of a single grating coupler: 40-50nm3dB bandwidth of a single grating coupler: 50-100nm

1dB bandwidth

3dB bandwidth


Diffraction gratings: bandwidthBandwidth is sufficient for a single wavelength bandSome applications require two wavelength bands

• e.g. in FTTH, use of 1310nm and 1490nm/1550nm for upstream and downstream communication

Roelkens et al.OpEx 15(16), p.10091 (2007)

see posterD. Vermeulen


FTTH Duplexer

Using a simple grating coupler(only 70nm etch, no overlay)Overall coupling efficiency: -7dB/coupler

-16

-14

-12

-10

-8

-6

-4

-2

0

1200.00 1300.00 1400.00 1500.00 1600.00wavelength [nm]

coup

ling

effic

ienc

y [d

B]

1490nm

1550nm1310nm

Vermeulen et al. ECOC2008 Tu.3.C.6

see posterD. Vermeulen


10

TE

Grating Coupler: Polarization

adiabatic taper


single-mode fibre,

Only one fiberpolarization is

coupled


10

TM

Grating Coupler for TM

Different parameters thanfor TE gratings

TM mode has lower βTM grating has larger pitch:

1040nm instead of 630nm


TM coupler - efficiency

-30

-25

-20

-15

-10

1500 1520 1540 1560 1580 1600

Wavelength [nm]

Effi

cien

cy [d

B]

Fiber to fiber efficiency

25% per coupler


2-D grating + polarization splitter

TE

TE

10

Taillaert et al, PTL 15(9), p.1249 (2003)

Fiber-to-waveguide interface for polarization independent photonic integrated circuit

2-D grating, 2 waveguidescouples each fiber polarizationin its own waveguidein the waveguides thepolarization is the same (TE)


2-D fiber couplers

Efficiency: -6.7dB (21%)Extinction ratio: > 18dB3dB bandwidth: 60nm

-11

-10

-9

-8

-7

-6

1510 1520 1530 1540 1550 1560 1570 1580wavelength [nm]

tran

smis

sion

of s

ingl

e fib

er c

oupl

er [d

B]

60nm

-6.7dB

10µm

Bogaerts et al.OpEx 15(4), p.1567 (2007) Silicon Photonics –PhD course prepared within FP7-224312 Helios project

on-chip components are polarisation dependent

fiber-to-fiber transmission is polarisation independent

Polarisation Diversity Circuit

y-polarization

split polarisations

light in

identicalcircuits

x-polarization

x

yz

xy

light out

single-modefiber

2-D grating

combine polarisations

2-D grating

10

TE


What aboutAlignment toleranceEfficiencyOptical bandwidthPolarization dependenceFootprint

We can fix each of these, but fix all together?adiabatic taper


single-mode fibre,

YES! Different tricks can be combined


2-D focusing grating couplers

2-D grating = 2 1-D grating1-D curved grating linesPlace hole at intersection25% efficiency!

Van Laere et al, JLT, to be publishedSilicon Photonics –PhD course prepared within FP7-224312 Helios project

2-D focusing grating couplers

Experimental results25% coupling efficiency(better than non-curved)

Van Laere et al, JLT, to be publishedSilicon Photonics –PhD course prepared within FP7-224312 Helios project

10um fiber core

waveguides

Grating coupler

Luxtera gratings


Luxtera grating couplers

Demonstrated performance:– 0.8 dB loss to/from fiber

Grating Coupler Evolution

Wavelength (μm)

Inse

rtion

Los

s (d

B) Today’s

D.O.R.

Firstdesign

Seconddesign

Thirddesign

Bestdesign


10

TE

Grating Coupler: Packaging

Vertical fiber: problems

Mechanical stability?Incompatible with planar packages?

How to attach the fibers to the chip?adiabatic taper


single-mode fibre


Just glue?

Very accurateActive alignmentUV curing(even through the fiber)

ButNot very practicalMechanically unstable


10

TE

Grating Coupler: Packaging

Vertical fiber: problemsMechanical stability?Incompatible with planar packages?

For many purposes, the fibers should be horizontal

adiabatic taper


single-mode fibre


Horizontal fiber

Use angle-polished fiber (array)

TE

10 total internal

reflection

50 polishangle


10

TE

Many fibers?

Multiple fibers?

adiabatic taper


single-mode fibre


Fiber arrays

1-D rows of fibersTypical pitch = 125µm or 250µmPositioning must be accurateexample: Silicon V-grooves


Array-connected SOI circuit4 x 4 wavelength router

Standard connector with 8 fibers (4 input, 4 output)Vertical fiber couplers4 x 4 AWG

Dumon et al.,OpEx 14(2), p.664 (2006) Silicon Photonics –PhD course prepared within FP7-224312 Helios project

• coupling to fully passive SOI chip

T. Tekin et al. ECOC 2008,P.2.21

Fiber array package - concept


uv-curingepoxy

fiber-array

SOI chip

Si v-groovebase

glass

uv-curingepoxy

fiber-array

SOI chip

Si v-groovebase

glass

uv-curingepoxy

fiber-array

SOI chip

uv-curingepoxy

uv-curingepoxy

fiber-array

SOI chip

uv-curingepoxy

T. Tekin et al. ECOC 2008,P.2.21

8 fiber array package

Example – g-Pack


Alignment of 1D grating coupler array vs fiber array

T. Tekin et al. ECOC 2008,P.2.21Dumon et al.,OpEx 14(2), p.664 (2006)

Fiber array alignment


V-groove based 8 fiber array - specifications

1 2 3 4 5 6 7 8

Fiber array tolerances


Fiber array – position scanXY-Scanned Fiber Positions1-3 and 6-8

Fiber array tolerances


Grating Coupler: Density

Fiber density?Fiber pitch: 125 or 250µmCoupler pitch can be as low as 25µmWaste of expensive chip real-estateImpact on yield

Useful chip area

Waste of space


Silicon vs. silica density

Silicon photonics40 channel AWGFootprint 1 1mm2

fiber array width: 5-6mm2

Area loss: 90%

Silica technologysimilar core/spacing as fibersNo fan-out required

Fiber pitch = 127 µm ~ 6mm


Photonic interposer

Silica-based interposer chipFan-out of photonic waveguides to fiber arrayFan-out of electronic connection to wire bond pads

Silicon chip can be kept small (cost, yield)One interposer design can serve many chip layouts


Coupling to the interposer

Use TIR mirror to couple to the silica waveguideDifficult to fabricate!

silicon carrier

SOI photonic chip


Photonic interposer

Si photonics device

Photonic interposer chips

Fiber block

Electric pads


Overview





Using vertical coupling

Easy measurements

Wafer-scale testingWith array probeWith grating probe

Free-space couplingCamera measurementsRetroreflective chipsBeam steering and forming

10

TE

adiabatic taper


single-mode fibre


Start measuring in 5 minutes


Wafer-scale testingDiffraction gratings allow for wafer-scale testing of photonic integrated circuits

• A modified electrical wafer prober can be used for this purpose, e.g. Luxtera

• This does not allow probing of individual components in a complex photonic integrated circuit!


Optical probing

10

TE

Fixed location: can only testcircuits which have fiber

coupler input/outputs


Optical probeNeed for a true equivalent of an electrical probe

Allows for testing of individual components in a PIC without the need for dedicated coupling structures on the PICOptical fiber with a diffraction grating (gold stripes) defined on the core of the optical fiberFabricated using a nano-imprint and transfer technique

Scheerlinck et al.APL 92(3), p.031104 (2008)Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Optical probes

Scheerlinck et al.APL 92(3), p.031104 (2008)c

Fiber core

Waveguide

a b

Metal gratingattached to the fiber

Optical probes15% coupling efficiency from a single mode

fiber probe to a 3µm wide SOI waveguide demonstrated

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

1500 1510 1520 1530 1540 1550 1560 1570 1580

Wavelength (nm)

Out

put (µW

)

Scheerlinck et al.APL 92(3), p.031104 (2008)Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Fibers are just cables

Fibers havemore bandwidthless weightno interference

than electrical cables

BUT FIBERS ARE STILL CABLES


Wireless photonics?

Free-space interfaces for photonic ICs?

Useful forremote readoutmassive parallel sensingtracking, identificationmedical diagnostics...

Main limitationsline of sightpower budget

Using an IR cameraIR Camera with microscope

observe many grating couplerswith tunable laser: sweep wavelength

tuneable lasertilt stage

microscope

XenICsIR camera

polarisation controller

lamp


Camera setup


Camera measurements

Readout system

massively parallelNo density problem2-D arrays possible


Remote readout

Contact between reader and chip should beangle-independentwork with limited power budget

Retroreflective operation


Free-space retroreflectivity

Retroreflective: use same grating coupler for input and output

Verbist et al. IEEE/LEOS Benelux Symposium 2008 p.171Silicon Photonics –PhD course prepared within FP7-224312 Helios project

Butterfly chip

Solution: duplicate filter circuits

curvedgrating couplersfor different angles

identical filtercircuits

varyorientation

vary tilt angle


Results

Clear resonancespectrum from fiber coupler


Results

Resonance is evenvisible in multiplecouplers and from the entire chip

Coupler ACoupler B (other tilt)Entire chip

AB


Reconfigurable beams

Dynamically reconfigure the beamshapedirection

wide-angle beamfor seeking andidentification

narrow beamfor communication


Phased arrays

1 grating coupler = patch antennaMany grating couplers = synthetic antennaPhase between different couplers controls

direction (beam steering)shape (beam shaping)

grating

heaters forphase control

Silicon


Beam shaping

16 grating couplersenvelope = angular spectrum of one gratingpeak spacing determined by grating spacing


Beam Steering

Current through heaters shifts beam


Horizontal vs. VerticalHorizontal Vertical

Alignment tolerance + ++Efficiency +++ ++(+)Polarisation +++ ++(+)Bandwidth ++ +Fabrication + ++Packaging +++ ++Wafer-scale testing --- +++Footprint --- +++Density --- +++

Free-space -- +++

Summary

Coupling to submicron silicon waveguides = difficultHorizontal: mode expandersVertical: grating couplers

Functional issues: polarisation, efficiency, bandwidthhorizontal still outperforms verticalvertical has different tricks up its sleeve

Practical issues: tolerances, footprint, testingvertical outperforms horizontal

Applicationsfiber connectionsfree space


Collaboration works…

Much of this work was donein collaborative projects…

The European Union • IST-PICCO • IST-PICMOS• IST-ePIXnet• ICT WaDiMOS• ICT HELIOS

The European Space AgencyBelgian Science Policy

• IAP-V/18 PHOTON network• IAP-6/10 Photonics@be


http://www.helios-project.eu/�

Coupling light is not trivial…

So you’re convinced that there should be

some light coming through?


Lecture: Coupling light to silicon photonic circuits coupling.pdf · Lecture: Coupling light to...

Documents

Transcript of Lecture: Coupling light to silicon photonic circuits coupling.pdf · Lecture: Coupling light to...