Virtual Photonics: from fundamental science to industrial applications

1
Virtual Photonicsfrom fundamental science to
industrial applications

• Gian-Luca Oppo
• Department of Physics, University of Strathclyde,
  Glasgow

Supported by SHEFC, EPSRC Long term collaboration
with SGI Special thanks to Andrew Scroggie, John
Cowen, Ivan Rabbiosi, John Mc Sloy, Graeme
Harkness, Willie Firth, David Hutchings, John
Arnold, Iain Wallace Department of EEE,
Glasgow University
2
Photonics

• Integration of optical and electronic
  technologies
• It employs light to process and transfer data
  (e.g. lasers)
• Main applications communications, optical
  storage, CD-DVD players, sensors, components,
  technology for medicine and science etc.
• Present market world-wide over 20 billions U
• Worldwide commercial laser revenue in 2002 4.5
  billions U (down from 5.6 in 2001 and a
  staggering 8.8 in 2000).
• Growth expected in 2003.
• Present status on the stock market not that
  healthy because of recent downturns. However,
  photonics companies with solid product lines are
  expected to come back to profit soon.

3
Photonics in Scotland

• Employs more than 5000 people at present
• Steady and fast growth until 2002.
• Turnover over 600 million
• Well established companies as well as a large
  number of new and innovative companies born in
  the last 10 years
• Small - medium and few large size companies
• Many spin-offs from Scottish Universities
• Good industry-academia partnerships
• Major support from Scottish Enterprise

4
The PDSS Project

• Photonics Device Simulation Services
• Collaboration between Strathclyde, EEE at Glasgow
  University and several Scottish Photonic
  Industries
• These Industries have no expertise or personnel
  for HPC
• Available software is not flexible to accommodate
  RD

HPC Facilities for PDSS at Strathclyde

• 12 parallel processors SGI ONYX 2 Virtual
  Reality T.
• 16 parallel processors SGI Origin 300 (installed
  in 2001)
• Tendering for a new multi-processor facility
  (Altix, hopefully)

5
Virtual Photonics at Strathclyde
VIRTUAL DEVICES
BIOPHOTONICS
SEMI CONDUCTORPHYSICS
SENSORS ULTRASONIC
ULTRA-SHORT, HIGH POWER LASER PULSES
SCIENTIFIC APPLICATIONS
6
PDSS Industrial Partners

• Intense Photonics
• Kelvin Nanotechnology
• Coherent Scotland
• British AErospace
• Terahertz Photonics
• Kymata - Alcatel
• Institutes of Photonics and Biophotonics
• Thales
• Essient (?)
• Tsunami (Ireland)

7
Overview

• What is Virtual Photonics (VP) ?
• Cheap assessment of devices ahead of
  fabrication
• Few Examples 1) Short Pulse generation in OPO
• 2) Ultra-short Pulses in Surface Emitting Lasers
• 3) Optical Memories with Spatial Solitons
• 4) Fundamental Science Optically Trapping,
  Rotating and Releasing Atoms and Cells (Optical
  Tweezers)
• Why High Performance Computing (HPC)
• Why High Performance Visualisation (HPV-VAN)

8
Virtual Photonics (VP)

• Mathematical Modelling
• Set-up of appropriate Partial Differential
  equations
• A) Description of light propagating through
  materials (Maxwell equations materials
  properties guiding)
• B) Description of propagation through optical
  elements, cavities and diffractive objects
• C) Description of light outputs in time and space
• Differences with Academic Approach
• Numerical codes, High Performance Computing (HPC)
  and tests of accuracy

9
Short Pulses in Optical Parametric Oscillators
(OPO)
Mathematics Interaction of light beams with
nonlinear crystals and optical elements.
PERIODICALLY PULSED
10
Short Pulses in OPOMathematical Model
phase matching
absorption
dispersion
Stochastic, nonlinear partial differential
equations
11
Short Pulses in OPO Output Pulses
Without Absorber Long and messy output pulses
With Absorber Short and clean output pulses

• More realistic devices
• Curved Mirrors
• Cavity Diffraction
• Apertures
• Polarizers

• Dispersion Compensation
• QPM materials
• Thermal effects
• Asynchronous pumping

12
Virtual Laboratory
At Strathclyde MSci in Photonics MSci in
Physics with Visual Simulations
13
Short Pulses in VECSEL
Main Aims

• To construct comprehensive computer models of
  Passively Mode-locked Vertical External Cavity
  Surface Emitting Laser (VECSEL) systems
• To provide a complete theoretical analysis

Why use simulations and models?

• To save time and money
• To visualise optical systems (virtual prototypes)
• To guide experimental work

input parameters ? outputs ? clearer picture of
the system
14
The Vertical External Cavity Surface Emitting
Laser (VECSEL)

• Very high output powers
• Semiconductor gain medium
• Low cost, easy to protect

15
850nm VECSEL wafer structure
Cap (InGaP)
Confinement Window
(AlGaAs)
GaAs QWs
Gain Region
/
(17
)
l
2

AlGaAs
/
l
2
30 Layer Bragg
Mirror
(AlAs/AlGaAs)
GaAs Substrate
0.5mm

16
The Numerical VECSEL Model
Laser cavity including Gain Element with
population dynamics Saturable absorber with
population dynamics Mirror BW 75-300
nm Optional Diffraction between gain element and
saturable absorber Pump shape
Output
Curved Mirror
Flat Mirror
Pump beam
Absorber
Gain Element
17
Numerical implementation of the model

• Models run over PDSS web interface
• Uses SMP computers with up to 16 parallel
  processors
• Huge speed up for results (from one month to
  three days)

Tail off
Close to linear speed up
18
Running the codes over the Web
Wide Range of Input Parameters
Forms and data are accessible via the web through
PHP
19
Results
Pulse Optimisation by changing Device
Characteristics
Final Pulse after 20000 round trips
Space
Time
20
Short pulse optimisation in solid state lasers

• Inclusion of diffraction
• Inclusion of apertures
• Inclusion of curved mirrors
• Inclusion of walk-off
• Inclusion of temperature profiles
• Optimisation of pulse shape in both space and time

21
Virtual Devices
Optical Material for info coding

• Numerical simulations of Laser systems and other
  Photonics devices
• High level of accuracy of numerical models
• Parallel implementation on HPC machines
• Possible reliable comparison with experiments
• Virtual Prototypes

Laser
22
Photonic Crystals
23
Spatial Solitons Useful for Optical Memories and
Photonics IT. Observed in semiconductor
micro-resonators, VCSEL, liquid crystals
Nonlinearity
Diffraction
External Driving
Cavity Losses
24
Numerical Analysis
To determine peculiar solutions of the system of
equations we developed codes based on three
numerical methods

• Newton Method - to find spatial structures
• Arnoldi Method - to determine their stability
  via the diagonalization of very large matrices
• Split-Step Integration - using alternating
  Runge-Kutta and FFT methods

Both Newton and Arnoldi Methods are extremely
computationally intensive and have been
parallelised with OpenMP and MPI on a SGI Origin
300 with 16 parallel processors. Integration is
much faster than on cluster machines with 60-80
Pentium processors or other SMP machines.
25
Spatial Solitons for Optical Memories
26
Scientific ApplicationTrapping, Rotating and
Cooling Atoms
E
A
0
(2)
c
A
1
27
Optically Trapping, Rotating and Releasing Atoms
Slow atoms remain trapped while fast ones leave
the beam (Cooling)
28
HPC and HPV

• Joining HPC from below. Constant development and
  testing of codes. Need of fast and reliable
  parallelisation.
• SMP configurations allow for easy parallel
  programming and superb utilisation of large RAM
  and Disk space.
• 16 SGI R14000 can easily beat 70 P4 in a
  cluster.
• Once the codes are tested, companies need easy
  access to optimise the design of the photonics
  devices.
• Need to transfer large number of data and
  results.
• We used to transfer files but they were not
  analysed.
• Transfer of images, movies and data
  visualisation at user convenience (VAN).

29
Handling Complex Data in a Virtual Environment
BIOPHOTONICS

• Optical detection
• Removal of noise (filtering)
• Data decoding reconstruction
• High dimensional correlations
• Comparison with data from virtual models
  prototypes
• Three, four higher dimensional renderings
• Data coding and long distance transmission

SEMI CONDUCTORPHYSICS
ULTRA-SHORT, HIGH POWER LASER PULSES
30
Conclusions

• Virtual Photonics with HPC is a reality
• Parallel computing allows for the study of real
  devices with few approximations
• Requires continuous test and update of codes
• Fewer publications gt Confidentiality
• Distribution of result data in 3D plots and
  animations (VAN)
• Applications in Research and Industrial RD
• Preparation of future generation of skilful
  employees
• Thanks to Stuart Wilson and SGI-UK

31
VP Projects in PDSS

• Beam Propagation and Finite Difference Time
  Domain
• Semiconductor Lasers and passive OE Devices
• Harmonic Sub-harmonic frequency generators
• Photonic Crystals Couplers
• Temporal and Spatial Solitons
• Optimisation of Short Pulses in VECSEL VCSEL
• Multi-mode emission of ridge waveguide
  semiconductor lasers

32
High Performance ComputingParallel Speedup
Saturation
33
(No Transcript)
34
0.05
0.07
0.10
0.12
35
The VECSEL
Light emitted perpendicular to the structure
plane

• High modulation speeds
• Low divergence circular beam
• Small active volume ?Low operating current

36
Model Equations
Gain Element Equations
E - Electric Field p - Population Inversion ? s-
Linewidth Enhancement Factor
po - Equilibrium Population Inversion ?s - Gain
Spontaneous Emission time ISAT - Saturation
Intensity (gain)
Saturable Absorber Equations
q - Ground state Population ? r- Linewidth
Enhancement Factor
qo - Equilibrium ground state Population ?r -
Relaxation Time ISAT - Saturation Intensity
(absorber)