SurveyonPhotonicsandNovel Op6cal’Materials’ · SurveyonPhotonicsandNovel Op6cal’Materials’ Minghao(Qi(Purdue(University(DLA2011 SLAC(Naonal(Accelerator(Laboratory(September15,2011

Survey  on  Photonics  and  Novel  Op6cal  Materials  

Minghao  Qi  Purdue  University  

DLA  2011  

SLAC  Na=onal  Accelerator  Laboratory  

September  15,  2011  

Photonics  and  Novel  Op=cal  Materials  

•  Outline  –  On-­‐chip  hollow  TM  structures  –  On-­‐chip  accelerator  based  on  Omniguide  waveguides  

–  Fiber  to  waveguide  couplers,  power  spliJers  – Materials  and  Damage    

–  New  fabrica=on  approach  to  Woodpile  3D  photonic  crystals  

On-­‐Chip  TM-­‐Mode  Waveguides  •  Must  be  hollow  •  Need  to  have  a  phase  speed  matched  with  the  speed  of  the  bunch  – At  high  par=cle  energy,  phase  velocity  must  match  c  

•  Prefer  to  be  strongly  confined  – Higher  gradient  

•  Omniguide  fiber,  PhC  fibers,  3D  photonic  crystals  – Tunable  to  tolerate  fabrica=on  varia=ons  – Low  nonlinearity?  

Omni-­‐Guide  Fibers  

•  Direct  analogy  with  hollow  metallic  waveguides  –  Cylindrical  symmetry  may  facilitate  TM  modes  

•  Also  called  Bragg  Fibers  

•  We  need  to  bring  them  onto  chips  

S.  G.  Johnson,  et  al,  Op=cs  Express,  9,  748-­‐779,  (2001)  

Engineering  Op=ons  

•  Indices:  Red:  2.8,  blue:  1.5  •  TM  band  gap  intersects  light  line.  

T.  D.  Engeness,  et  al,  Op=cs  Express,  10,    1175-­‐1196,  (2003)  

Hybrid  Modes  can  Intersect  Light  Line  

T.  D.  Engeness,  et  al,  Op=cs  Express,  10,    1175-­‐1196,  (2003)  

Higher  Order  TM  Modes  Could  Intersect  with  Light  Line  

G.  Ouyang,  et  al,  Op=cs  Express,  10,    899-­‐908,  (2002)  

Proposal  of  an  On-­‐Chip  Accelerator  

Electron  bunches  

Beam  posi=on  sensor  

Accelera=on  module  

Focusing  module  

Laser  input  fiber,  l1  

Laser  input  fiber,  l2  

Si

W SiO2

•  Inner  diameters  can  be  different  and  controlled  

How  to  Fabricate  It?  

•  Standard  CMOS  process  except  wafer  bonding  •  Does  not  require  deep  submicron  technology  

•  Can  control  the  inner  diameter  

Si W

Si W

SiN SiN

Oxida=on  to  Achieve  Circular  Shape  

•  Require  aligned  wafer  bonding  (but  just  once)  •  Tungsten  as  quadruple-­‐poles  to  withstand  high  temperature.  

SiN SiN SiN SiN

Si

W

Si

W

Chop  the  hollow  waveguide  to  right  length  

•  Right  module  length  for  accelera=on  and  focusing  •  Short  enough  for  atomic  layer  deposi=on  to  work  

ALD  to  coat  the  inner  Bragg  layers  

•  Atomic  Layer  Deposi=on  is  extremely  uniform  

Short distance to waveguide terminals

Previous  Demonstra=on  on  Chip  

T.  C.  Shen,  et  al,  Journal  of  Lightwave  Technology,  28,  1714,  (2011)  

G.  R.  Hadley,  et  al,  Op=cs  LeJers,  29,    809,  (2004)  

•  Not  using  ALD  •  With  deposi=on  of  Si  followed  by  oxida=on  of  Si  

Other  Approaches  

Polymer protecting sidewall in Bosch process used as mask for isotropic etch with xenon difluoride Image courtesy of Carnegie Mellon University MEMs Laboratory

D. Gaugel, K. Gabriel, "CMOS-Compatible Micro-Fluidic Chip Cooling Using Buried Channel Fabrication," Proceedings of IMECE '02, New Orleans, 2002

Exposing  the  Quadruple  Poles  for  beam  sensing    

•  Or  not  necessary,  if  we  only  need  to  measure  the  posi=ons?  

•  Unlikely  to  be  able  to  focus  or  deflect  beam  bunches?  

Fiber  Pigtailing  

•  The  hollow  coupler  could  be  short  and  tapered  •  Fiber  =ps  could  be  tapered  •  Add  heater  to  achieve  tunability  

TE mode

Short coupler

Fiber  to  Waveguide  Coupling  •  Fiber  splicing  <  0.1dB  (>  97.7%  power  coupling)  

•  Fiber  to  waveguides  on  chip  – Pigtailed  fiber  in  V-­‐grooves  –  Inverse  taper  

•  Gra=ng  couplers  – ~70%  coupling  in  a  CMOS  line  using  Si  based  structures  

– SiN  gra=ng  couplers  with  ~60%  efficiency  – 82%  in  theory  

Inverse  Taper  for  Fiber  to  Waveguide  Coupling  

•  ~  1dB  loss  per  facet  is  predicted  •  ~1.6dB  per  facet  loss  realized  

overcladding Silicon  or  SiN  core

Silicon  waveguides

Polymer  waveguides Doped  Silicon  dioxide  core  

Undercladding

Gra=ng  Coupler:  Ver=cal  to  Horizontal  coupling  

Gra=ng  coupler  achieving  -­‐1.6dB/coupler  

•  D.  Vermeulen,  et  al,  Op=cs  Express,  18,  18278,  2010  

220nm

150nm 220nm

370nm

240nm

Design  op=miza=on  

Ziran’s Design

Fabricated  Si  Gra=ng  Coupler  

Tes=ng  setup:  fiber  bundle  

Material  Guidelines  •  High  power  handling  capability  

– High-­‐damage  threshold  

– Low  nonlinearity  •  Conduc=vity:  avoid  electron  trapping  

– Dielectrics  – Semiconductor  

– Metal  – Graphene?  

•  CMOS  compa=bility  – SiO2,  SiN,  Si  

•  Other  semiconductors  or  exo=c  materials?  

Material  Damage  •  Con=nuous  wave  laser  characteriza=on  of  gra=ng  couplers  – Can  extract  power  enhancement  factor  from  resonant  structures.  

•  Si  has  high  two-­‐photon  absorp=on  probability  – Generates  free  carriers  – Absorbs  light  – Heat  up  structures  

•  Silicon  nitride  has  larger  band  gap  and  does  not  suffer  from  two-­‐photon  absorp=on  – Expected  to  have  much  higher  damage  threshold  

Silicon  Nitride  Waveguides  •  570nm Si3N4 by LPCVD •  3um buried silicon dioxide (BOX) •  4.5um Top Oxide Cladding by PECVD

SEM Picture is taken after Si3N4 etch. Sidewall has a slope of about 78°. HSQ is the etch mask

Oxide  Cladding  

BOX  

Si3N4  

Si  

Si3N4

HSQ

BOX

Etching Profile

Nitride  Ring  R=40um  with  Taper  

•  WG  width  =  1um  

•  Gap  =  700nm  

•  3dB bandwidth at 1558nm is about 7pm. •  Q ≈ 223K low propagation loss ~ 2dB/cm •  Grating couplers with SiN is being fabricated

On-­‐Chip  power  splipng:  SOI  Y-­‐junc=on  

•  Flat  power  splipng  across  a  large  bandwidth  

•  Arbitrary  splipng  ra=o?  

•  Post-­‐fabrica=on  trimming?  

Port#1

Port#2

Port#3

10dBm input power

Woodpile 3D Photonic Crystals •  Best flexibility in

designing hollow waveguides

•  17 layers are needed

Layer  7

300  nm Si  substrate

HSQ

6

1 2 3 4

5

100  nm

100  nm 300  nm

Membrane Transfer Technique 1. Fabricate all 17 layers in one step

+

+

+

2. Assemble layers to form 3D photonic crystal

waveguide

Woodpile structures before release

•  Grating lines may stick together due to capillary forces –  May need more spacers

Structures Released Successfully and Can be Stacked up

•  The residuals are water debris and can be eliminated. •  2nd layer does not have the same debris.

Two layer Woodpile Structures with Reasonable Alignment

•  Small particles within the membrane region. Can be avoided, we think.

Advantages •  Every layer has the exact thickness

–  They are from the same film

•  No patterning for each layers –  One can produce 100s or 1000s layers

in one wafer

•  No stress problem –  Assembly done in room temperature

•  For 9mm (CO2 laser) operating wavelength, can use optical lithography for patterning and alignment –  Period is 4.2 mm, rod width 1.2 mm,

and layer thickness 1.6 mm

500  nm  Silicon  

Layer  1  Layer  2  

3 Layer Membrane Bonding

•  High precision alignment is required

1  2  Layer  3  Layer  1  

Layer  2  Layer  3  

Questions?