Heteroepitaxy and selective area heteroepitaxy Lourdudoss/Menu...InP seed ELOG InP Photoluminescenceof multi quantum well laser structure On InP(ref) On Si Big step Towards Integration

25/01/2014

1

Sebastian Lourdudoss

Laboratory of Semiconductor Materials

Department of Materials and Nano Physics

School of Information and Communication Technology

KTH, Electrum 229, 164 40 Kista, Sweden

Heteroepitaxy and selective area heteroepitaxy of III-V compounds on silicon for

silicon photonics

1ADOPT Winter School, Romme, 2014

• Photonic Integrated devices for information highway– Background

• Photonic integration on silicon - Silicon photonics

• Hybrid integration approaches for integrating III-Vs on Si

• Heteroepitaxy and Selective Area Growth : Challenges

• Monolithic Approaches for integrating III-Vs on Si

• QW growth on InP/Si

• Templates for site controlled QD growth

• Summary and conclusions

Outline

2

25/01/2014

2

Background

The information highway used to look like this…

3From Carl Junesand

Background

…But now it looks like this

4From Carl Junesand

25/01/2014

3

Background

But then,

people started to…

…have video conferences…

…play online games…

…stream high‐definition movies 5From Carl Junesand

Background

Computer trends

• multi‐core processing– dual, quad etc.

• Continuing shrinkage– gate length 14 nm

• higher transfer speed– HDMl, USB 3.0 etc.

6

Copper drawbacks:

– Slow

– Crosstalk

– High power consumption

From Carl Junesand

25/01/2014

4

7

Novel area: Silicon PhotonicsMarriage between electronics and photonics

Electronics Data processing

Photonics Data communication

Instrumental for the emerginginformation society. Numerousapplicationspresentlypenetrating all parts of our every-day life.

Silicon-PhotonicsParadigm shift, with major implications

IEEE Spectrum Aug. 2002

Silicon photonics for overcoming the interconnect bottleneck

Intels 50 G Silicon photonics links: Silicon transmitter – receiver system

Mario J. Paniccia, “A perfect marriage: Optics and silicon,”_Optik & Photonik , May 2011 No. 2 , p34.

IBM vision on siliconphotonic chip

8

25/01/2014

5

9

Photonic Integration –Optical Arbitrary Waveform Generation

(OAWG)

Monolithic InP 100-Channel x 10-GHz Device for Optical Arbitrary Waveform Generation, IEEE Photonics Journal, Volume 3, Number 6, 2011

Collaboration:

Background• Why Si?

+ Foundation of the CMOS industry

+ 50 years of process optimization

+ Huge economics of scale

• Si wafers → >100 chips on one wafer

10

25/01/2014

6

11From Carl Junesand

Strengths and weaknesses of silicon

• Excellent processing on very large areas (12” wafer today)

• High density integrated electronics

• SiO2 renders Si an excellent material for electronics

• SiO2 also renders Si an excellent waveguide

• Photodectectors (e.g., Si‐Ge), modulators (e.g., EO polymers

But no light emission!

Solution = III‐Vs on Si!

But……..

25/01/2014

7

Jack Kilby’s integrated circuit at Texas Instruments in 1958 by ”bonding”

Nobel prize in 2000

Integration path in electronics

Intel’s Poulson Itanium Processor in 2012: • 32‐nanometer process• 3.1 billion transistors • 544 mm2

=> ~ 0,5 billion transistors/cm2

http://www.theregister.co.uk/2011/08/22/intel_poul13

Integration path in silicon photonics

Same approach as that of Kilby’sin 1958!

But a goodstarting point!

Monolithic approach through heteroepitaxy more flexible

14

25/01/2014

8

Above‐IC silicon photonics and III‐V/SOI integration scheme

L. Grenouillet et al., Hybrid Integration for Silicon Photonic Applications, Opt. Quant. Electron. (2012).

15

Heterogeneous photonic integration and opticalvertical interconnect access

Qian Wang et al., Heterogeneous Si/III‐V integration and the optical vertical interconnect access, Optics Express, 2012. 16

25/01/2014

9

17

Heteroepitaxy and Selective Area Heteroepitaxyfor Silicon Photonics

Heteroepitaxy: Growth of dissimilar materials often with large lattice mismatch

Selective area heteroepitaxy: Heteroepitaxy conducted selectively on open areas only

Problems:

Dislocations, line defects, planar defects, anti phase domain etc

Concepts: mismatch

• What is mismatched materials?

• Different lattice constant a1 ≠ a2• Different thermal expansion coefficient α1 ≠ α2

a1

a2

25/01/2014

10

Concepts: dislocations

• Line defect

• comes in two types:– Edge: Burgers vector & line vector perpendicular

– Screw: Burgers vector & line vector parallell

• May also be a mix between both19

Concepts: misfit dislocations• a1 ≠ a2

– If no relaxation: strain f =(a2 – a1)/a1– Compressive or tensile strain

– Complete relaxation: dislocations!

a2

a1

25/01/2014

11

Concepts: stacking faults• In FCC lattice, stacking follows ABCABC… along {111} planes

• Insertion of an extra C plane causes an intrinsic fault

21

A

C

B

C

A

C

B

A

Faulty plane

Concepts: Antiphase Boundary

• Atomic steps may introduce Antiphase Boundary

• By off‐cutting with diatomic steps, antiphase boundary removed

[001]

Si substrate

Compound layer

25/01/2014

12

21

APB

Antiphase Boundary or Inversion Domain Boundary or Anti-phase Domain in heteroepitaxy

InP on Si?

Courtesy: Wondwosen Metaferia

Monolithic approaches for III-Vs on Si

Ge seeded growth

App. Phys. Lett. 97, 121913 (2010)

III-VsMetal catalysed nanowire growth

FIB activated growth

App.Phys.Lett. 97, 12 17 (2001)

III-V Quantum dots on Si

J. Nanophotonics, 3, 031602 (2009)

Conformal growth

Appl. Phys. Lett. 68 ,19(1996)

S.Lourdudoss, Current Opinion in Solid State and Materials Science, 16, 2(2012)91-99 24

25/01/2014

13

25


26

25/01/2014

14


27


28

J.Yang, P. Bhattacharya, Integration of epitaxially-grown InGaAs/GaAsquantum dot lasers with hydrogenated amorphous silicon waveguides on silicon. Optics Express 2008; 16:5136-5140.

25/01/2014

15


29


30

Schematic view of InP grown from SiO2 trench on Si through Ge seeding. The bottom picture of the rrepresents the created atomic steps on Ge

for avoiding APB’s.From: G. Wang et al., J. Electrochem. Soc.

2011; 158: H645-H650.

25/01/2014

16


31

Lattice matched growth

Liebich et al, Appl. Phys. Lett. 99, 071109 (2011)

Monolithic approaches on Silicon

32

Si-Ge-Sn on Silicon

Yijie Huo, Strained Ge and GeSn band engineering for Si photonic integrated circuits, Ph.D. Dissertation, 2010; Stanford University, USA. (http://snow.stanford.edu/thesis/Huo.pdf)

R.A.Soref, C.H.Perry. Predicted band gap of the new semiconductor SiGeSn. J. Appl. Phys. 1991; 69:539-541.

25/01/2014

17

Monolithic approaches on Silicon

33

Monolithic approaches on SiliconIII-Sb on Si

34

• AlSb buffer layer on Si => absence of vertically penetrating threading or screw dislocations

• N.B.!!! Large mismatch of 13%!• AlSb grown on Si initially consist of a crystalline AlSb QD ensemble

• =>coalesces into bulk material where the strain is reduced by crystallographic undulations, which relieve strain

Ref.: G.Balakrishnan et al., Appl. Phys. Lett. 2005; 86: 034105.

25/01/2014

18

Monolithic approaches on SiliconIII-Sb on Si

35

J.R.Reboul et al., APL, 99,121113(2011)

CW operation of 2 µm laser @ RT Pulsed operation of 1.55 µm laser @ 90 K and RT

L. Cerutti et al., IEEE Photon. Technol. Lett., 22, 533 – 535, 2010.

Selective Area and Epitaxial Lateral Over Growth( S. Naritsuka et al., Jpn. J. Appl. Phys., 34(1995) L1432-35)

SiO

2m

askPrinciple of ELOG

InPELOG

ELOG InP on Si from nano-openings

InPELOG

SiO2

mask

ELOG InP on Si from μ-openings

• Y.T Sun and S. Lourdudoss, Proc. of SPIE, Vol. 4997 (2003)

• F. Olsson et al., Journal of AppliedPhysics (2008).

• C. Junesand et al., 22nd Int. conf. on Indium Phosphide and Related Materials, 2010, Kagawa, Japan, 22nd IPRM Conf. Proc., 232-235.

• W. Metaferia et al., Journal of Crystal Growth, vol. 332, pp. 27-33 (2011).

36

25/01/2014

19

Heteroepitaxy: Advantages of ELOG

• Overcomes problems associated with 8% lattice mismatch of InP on Si: misfitdislocations

• Potential for high quality InP on SiGrow large size InP on Si wafers for volume productioneven for InP based devices

• Enables monolithic growth of InP lasers on siliconNo costly InP substrates

• Si/SiO2 waveguide as the mask in ELOGCoupling of Si/SiO2 waveguides with InP devices feasible

Patents:1. S. Lourdudoss and F. Olsson, Semiconductor heterostructures and manufacturing thereof, US patent 2012.2. C. Junesand and S. Lourdudoss, Active photonic device, US Patent 2012.

37

Planar defect studies of ELOG InP/Si

1. Carl Junesand et al., Optical Materials Express, Vol. 3, Issue 11, pp. 1960‐1973 (2013) http://dx.doi.org/10.1364/OME.3.0019602. Carl Junesand et al., Mater.Express,Vol.4(1), 41‐53, 2014, doi:10.1166/mex.2014.1140

38

On InP (ref)

On Si

25/01/2014

20

Si

InP seed

ELOG InP

Photoluminescence of multi quantum welllaser structure

On InP(ref)

On Si

Big step Towards Integration Of InP laser on Silicon Collaboration: Intel (USA) and John Bowers, UCSB

Multi quantum well laser on Si fabricated in Electrum Lab

1. Z. Wang et al., Materials Science and Eng. B, 177(17), 1551‐1557 (2012), http://dx.doi.org/10.1016/j.mseb.2011.12.006.2. H. Kataria et al., Semicond. Sci. Technol. 28, 094008, 2013; doi:10.1088/0268‐1242/28/9/0940083. H. Kataria et al., IEEE JSTQE, 20(4), 8201407, 2014; doi:10.1109/JSTQE.2013.2294453

dx.doi.org/10.1021/nl402815v, Nano Lett. 14, 37‐43 (2014)

25/01/2014

21

Multi quantum wells of InP/InGaAsPon silicon

41

Himanshu Kataria et al. IEEE JSTQE, 20(4), 8201407, 2014; DOI: 10.1109/JSTQE.2013.2294453.

42

25/01/2014

22

KIC InnoEnergy | Boosting Innovation for Sustainable Energy | Speaker: Yanting Sun 43

- Si subcell for 1.0eV band photon

absorption: CPV efficiency >50%

- Tandem solar cell on Si: cost

saving: 70%

High efficiency tandem solar cell on silicon substrate with mass

production compatibility

PI: Yanting Sun

Si substrate

Si subcell (1.1 eV)

Subcell 1.4eVSubcell 1.8eV

New start-up company TANDEM SUN AB

Direct interface of InP on Si

Picture removed (not yet

published)

Hybrid bonded QD lasers on Si

K.Tanabe et al, III-V/Si hybrid photonic devices by direct fusion bonding, Scientific Reports, 2012.

44

25/01/2014

23

Growth of QDs

Control of SITE, SIZE, SHAPE and DENSITY

Direct growth of QDs on the substrate= requires high resolution

patterning

Pyramidal frustum > QDs

Patterning- NIL, OL…

PF

QD

On InP: Appl. Phys. Lett. 91, 243106 (2007), Appl. Phys. Lett. 94, 143103 (2009)

45

Growth of InP Nanopyramidal Frusta (NPF)

Aixtron HVPE reactor @ KTH

Nanoimprinted holes on InP substrate

Growth conditionGrowth T=5900C, Time 2.5 min, V/III=10, Sulphur doped InP

Circular hole openings in SiO2 on InP(seed)/Si

D=300 nm and S= 500 nm ( main focus of thispresentation)

Soft UV-Nanoimprint Lithography

46

25/01/2014

24

D=120nm, S=180nm

D=200 nm,S=300nm

D=300nm,S=500nm

Growth of InP NPF

InP NPF grown from different diameter and spacing of hole openings in the oxide mask 47

RT-Cathodoluminescence Studies

CL measurement done at Laboratory of Thin Film Physics, Linköping University, 581 83 Linköping, Sweden (Galia Pozina and Lars Hultman )

NPF show brighter compared to the InP seed layer Defect filtering involved epitaxial necking effect and reduced area growth The CL spectra from NPF is 15 times stronger, blue shifted and broader

CL-spectra of NPF on Si0.5 μm0.5 μm

SEM Panchromatic CL

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.80

5

10

15

20

25

30

CL

-Int

ensi

ty (

a.u)

Photon energy (eV)

NPF InP(seed)/Si

(b)

SEM (left) and (P-CL) images of the NPF on Silicon substrate

48

25/01/2014

25

InP on Si

1. W. Metaferia et al., Phys. Status Solidi C, 1–4 (2012) / DOI 10.1002/pssc.2011006782. Position paper on nanophotonics and nanoelectronics, Nanonewsletter, no. 24, Dec. 20113. W. Metaferia et al., RSC CrystEngComm, Accepted for publication (2014)

SEM micrographs of InP nanopyramids grown on InP precoated Si substrate

SEM and Panchromatic CL images of InPnanopyramids grown on InP precoated Si

substrate

49

Site controlled quantum dots on Si on Nanoimprinted patterns

Collaboration: KTH + ORC, Tampere University of Technology + Linköping University

50

HeteroepitaxialMaterials Systems

Advantage(s)Operating

wavelength of light sources

Challenges CMOS compatibility

Suitability for large volume production

Lattice matched/graded materials on Si

Thin buffer layer on Si for facile integration with Si

Good thermal dissipation

So far < 1.2 µm.

Localised growth/Growth on non‐planar silicon wafers for integration.

Lasing wavelength beyond 1.3 µm

Feasible for silicon photonics

Electronics –photonics integration not yet available

Good

III‐Sb on Si Also useful for long wavelength detectors


1.5 µm and beyond

Localised growth/Growth on non‐planar silicon wafers for integration

Control of 90o

dislocations Thin buffer layers for

facile integration with Si

‐do‐

Good

QDs on Si Low lasing threshold and high temperature stability


Mostly ~1 µm; recently ~1.3 µm

Thin buffer layers for facile integration with Si

CW laser operation at RT and above at wavelengths >1.3 µm

‐do‐Good

GeSiSn on Si Lattice and thermal matching feasible for laser, detector and modulator structures.

Growth at relatively low temperature


Potential for 1.5 µm

Materials issues on adequate Sn solubility in Si‐Ge for bandgap engineering


Electronics –photonics integration feasible

Good

S.Lourdudoss, Current Opinion in Solid State and Materials Science, 16, 2(2012)91-99

25/01/2014

26

51

HeteroepitaxialMaterials Systems Advantage(s)

Operating wavelength of light sources

Challenges CMOS compatibilitySuitability for large volume production

ELOG of InP Perfect template for growing lattice matched InGaAsP for lasers, detectors and modulators

Waveguide structure can be used as defect filtering mask and at the same time for coupling light

Adequate thermal dissipation

1.3 µm and 1.5 µm

Necessity of good seed layer.

Complete elimination of coalescence defects

Device design topography for ELOG on certain specific titled angle of the mask openings



Good

SAH of InP in Si trenches

Perfect template for growing lattice matched InGaAsP for lasers, detectors and modulators

Ideal for integration with silicon devices at same horizontal level


1.3 µm and 1.5 µm

Enlarging the growth area by lateral growth above the trench level

Complete elimination of APBs



Good

Bonding Flexible Thin buffer layer and

good integration strategy with Si photonic circuits

Any desirable wavelength

Very large scale integration

Uniformity, reproducibility and reliability in wafer and die bonding



Electronics –photonics integration feasible

Moderate

S.Lourdudoss, Current Opinion in Solid State and Materials Science, 16, 2(2012)91-99

Epiclarus AB

• Active 2012‐

• Co‐founders:– Carl Junesand, CEO

– Bo Hammarlund, VP marketing and sales

– Sebastian Lourdudoss, senior scientist

• Innovation and Product:– Semi‐insulating regrowth for photonic integration and Quantum Cascade Lasers (~ 15 µm deep mesas)

– Templates of III‐V on Si

– InP on Si for Si Photonics

52

25/01/2014

27

Tandem Sun AB

• Tandem Sun AB‒ Active: 2013‐‒ KTH Innovation‒ EIT‐KIC InnoEnergy partner

• Innovation and product‒ High efficiency multi‐junction solar cell on silicon substrate

(SiMJSC)• Founders

‒ Professor Sebastian Lourdudoss‒ Dr. Yanting Sun

• Technology‒ Coherent III‐V/Si heterojunction‒ PCT/SE2013/050355

53

Summary and Conclusions

• Need for heterogeneous integration of InP on Si felt by big actors

• Monolithic integration is more flexible

• QW growth on Si ‐ promising

• Site controlled growth of InP based QDsfeasible even on Si

• Need for more technological developmenttowards monolithic integration with respectto hybrid integration

54

25/01/2014

28

Acknowledgements• Wondwosen Metaferia , Himanshu Kataria• Dr. Carl Junesand, Dr. Fredrik Olsson• Dr. Yanting Sun• Prof. Lech Wosinski, Fei Lou and

Dr. Zhechao Wang (Univ. Ghent), Prof. Lars Thylén

• Prof. Tapio Niemi and Prof. Mircea Guina, ORC, Tampere, Finland

• Dr. Galia Pozina and Prof. Lars Hultman, Linköping University• Prof. John Bowers, UCSB• Dr. Hyundai Park and Dr. Hai‐Feng Liu, Intel

55

Heteroepitaxy and selective area heteroepitaxy Lourdudoss/Menu...InP seed ELOG InP Photoluminescenceof multi quantum well laser structure On InP(ref) On Si Big step Towards Integration

Documents