54
October 2011 Claude Aguergaray with: V. I. Kruglov, J. D. Harvey e-mail : [email protected] Perturbations to self- similar propagation in optical amplifiers Physics Researc h 3

15.30 o4 c aguergaray

  • Upload
    nzip

  • View
    234

  • Download
    0

Embed Size (px)

DESCRIPTION

Research 3: C Aguergaray

Citation preview

Page 1: 15.30 o4 c aguergaray

October

2011

Claude Aguergaray

with: V. I. Kruglov, J. D. Harvey

e-mail : [email protected]

Perturbations to self-similar propagation in optical

amplifiers

Physics Resear

ch3

Page 2: 15.30 o4 c aguergaray

October

2011

Context

Sech input pulse in DSF Raman amplifier.

Self-

sim

ilar p

ulse

s

Parabolic pulses are generated asymptotically in the fiber amplifier independent of the shape or the noise properties on the input pulse, and possess linear chirp.

A class of solution of the nonlinear Schrödinger equation (NLSE) with gain.

Self-similar evolution of the pulse intensity and chirp the intensity profile retains its parabolic shape and resists the deleterious effects of optical wave breaking.

Page 3: 15.30 o4 c aguergaray

October

2011

Under normal dispersion, nonlinearity and gain

↪ ↪ Any input pulse will evolve asymptotically into a similariton with a parabolic intensity profile and positive linear chirp.

Context

Finot et al. OE 11 (2003).

Self-

sim

ilar p

ulse

s

Page 4: 15.30 o4 c aguergaray

October

2011

The Self-Similar dynamics :- Peak power- Temporal width increase exponentially with propagation length.- Spectral width

Context

Billet et al. OE 13 (2005)Dudley et al. Nature 3 (2007).

Self-

sim

ilar p

ulse

s

Page 5: 15.30 o4 c aguergaray

October

2011 The common point between Self-similar and the well-known technique of chirped-pulse

amplification (CPA) is that they aim at avoiding the pulse break-up due to excessive nonlinear phase shifts accumulated through the fiber.

- When CPA avoids nonlinearity by stretching the pulse before amplification,

- Self-similar amplifier actively exploits nonlinearity, (possibility of obtaining output pulses shorter than the initial input pulse).

For energies >µJ similariton amplifiers can be limited by the available gain-bandwidth. But they are a good alternative to more complex CPA systems below this limit.

Self-similar amplifiers have been demonstrated:- Several types of gain medium: ytterbium, erbium and Raman- For a broad range seed pulses in the range 180 fs–10 ps- Fiber lengths in the range 1.2 m to 5.3 km - Gains varying from 14 to 32 dB.

Self-similar pulse in amplifiers:

Self-

sim

ilar p

ulse

s

Context

Page 6: 15.30 o4 c aguergaray

October

2011

Amplification to the μJ level in an environmentally stable and polarization-maintaining configuration has been a demonstrated.

Self-similar pulse in amplifiers:

Compressed duration: 240 fsRepetition rate: 27 MHzAverage power: 21 WPeak power: 5 MW

Schreiber et al. OL 31 (2006).

Self-

sim

ilar p

ulse

s

Context

Page 7: 15.30 o4 c aguergaray

October

2011

Context

Billet et al. OE 13 (2005).

After 7m of propagation

All fiber compression stage by use of photonic bandgap optical fibre to replace bulk gratings lead to the realization of an all-fiber source delivering pulses in the 100 fs range at 1550nm.

Self-similar pulse in amplifiers:

FWHM136 fs

FROG measurement of compressed pulses

Self-

sim

ilar p

ulse

s

Page 8: 15.30 o4 c aguergaray

October

2011

Context

Self-similar pulse in lasers:

The net GVD of the cavity can be normal or anomalous.

- With large net anomalous GVD, soliton like pulses. These lasers (1st developed) have stringent limitation in energy (nJ) and pulse duration (ps) due to excessive nonlinear phase shift accumulated by the pulse.

To overcome this limitation researcher have developed laser cavity with dispersion map.

Self-

sim

ilar p

ulse

s

Page 9: 15.30 o4 c aguergaray

October

2011

Context

Self-similar pulse in lasers:

Aguergaray et al. OE 18 (2010).

- GVD ≈ 0, stretched-pulse operation occurs. The pulse energy can be an order of magnitude higher than in a soliton laser.

- GVD >> 0, higher pulse energies can be achieved directly from an oscillator.

Among these are the self-similar laser and the so-called chirped pulse oscillator (CPO). Pulse shaping in such a laser is based on spectral filtering of the chirped pulse, which cuts off the temporal wings of the pulse.

Laser output pulse energy: 21 nJ

Self-

sim

ilar p

ulse

s

Page 10: 15.30 o4 c aguergaray

October

2011

Motivation

Linearly chirped parabolic pulse is an asymptotic class of solution of the GNLSE with constant gain, normal dispersion and in the presence of non-linearities.

↪ ↪ Any input pulse with right energy will evolve into a similariton regardless of its shape and duration.

Until recently most theoretical descriptions of the self-similar (SS.) propagation had been done assuming that only low-order nonlinear effects and low-order dispersion effects dominate the pulse evolution.

But experiment have shown that SS. propagation can be perturbed by effect like:

- Gain bandwidth limitation,- Third-order dispersion,- Gain saturation.Se

lf-si

mila

r pul

ses

Page 11: 15.30 o4 c aguergaray

October

2011

Similariton propagation and break-up with third-order

dispersion influence

Page 12: 15.30 o4 c aguergaray

October

2011

Motivation

↪ ↪ Asymmetric temporal pulse shape,

↪ ↪ Peak shifted towards the edges of the pulse,

(direction depends on TOD sign).

Lead to generation of shock wave instability and pulse break-up.

TOD can have a detrimental effect on parabolic pulse propagation.

Sim

ilarit

on b

reak

-up

with

3

Page 13: 15.30 o4 c aguergaray

October

2011

Anterior work

z = 1 km z = 779 m

3 = 0.025 ps3/km3 = 0 ps3/km

Wabnitz and Finot have observed pulse break-up in Dispersion Decreasing Fibers (increased TOD influence with distance)

Wabnitz and Finot JOSA B 25 (2008).

10 % accuracy for pulse break-up critical distance.

z = 755 m

3 = 0.025 ps3/km

Sim

ilarit

on b

reak

-up

with

3

Page 14: 15.30 o4 c aguergaray

October

2011

Bale and Boscolo:

↪↪ Partial analytical description of the pulse before and at break-up,

↪↪ No theoretical prediction of critical distance.

z/zc = (a) 0.25, (b) 0.5,

(c) 0.75, (d) 1.

Bale and Boscolo, J. Opt. 12 (2010).

16% error between analytical and numerical simulations.

Sim

ilarit

on b

reak

-up

with

3

Anterior work

Page 15: 15.30 o4 c aguergaray

October

2011

Theoretical study.

Our Analytical model :1

2 2 60

2

21 1( , ) exp ( , )

2 3

g EU z gz Q z

Sim

ilarit

on b

reak

-up

with

3

Page 16: 15.30 o4 c aguergaray

October

2011

Theoretical study.

Numerical model :

(z) 3 controls pulse shape

(asymmetry of the pulse)

Critical parameter is given by

T2(z) = T3(z) (3 >0), orT1(z) = T2(z) (3 <0).

Condition gives critical length zc at which pulse breaks down.

Sim

ilarit

on b

reak

-up

with

3

Page 17: 15.30 o4 c aguergaray

October

2011

Theoretical study.

Yields to the critical distance parameter :

where

Sim

ilarit

on b

reak

-up

with

3

Page 18: 15.30 o4 c aguergaray

October

2011

Numerical results.

Analytical expression of critical distance :

2=0.13 ps2/m3=10-3 ps3/mg=2 m-1

=2.10-3 w-1m-1

Sim

ilarit

on b

reak

-up

with

3

Page 19: 15.30 o4 c aguergaray

October

2011

Numerical definition of pulse break-up :

↪ ↪ Pulse experiences growth of side peak under 3 influence

Numerical results.Si

mila

riton

bre

ak-u

p w

ith

3

Page 20: 15.30 o4 c aguergaray

October

2011

Numerical results.

Numerical simulations varying energy E0 :

3 = 0.96x10-3 ps3/m0.1pJ<E0<10pJ

lnc

<1% error for <1% error for critical length critical length

predictionprediction

33 0 3 4

0 52

10gE

3

2 2

g

Sim

ilarit

on b

reak

-up

with

3

Page 21: 15.30 o4 c aguergaray

October

2011

0 10E pJ3 3

3 10 /ps m

3 < 0 3 > 0

Numerical results.

Very good agreement between analytical prediction and numerical solution.

Sim

ilarit

on b

reak

-up

with

3

33 0 3 6

0 52

1.078 10gE

Page 22: 15.30 o4 c aguergaray

October

2011

Summary of β3 study

Novel analytical theory for propagating pulses in normal dispersion fiber amplifier with TOD.

Found the critical length zc at which the TOD generate pulse break-up for constant gain.

Shown numerically the limitations for input value providing a highly accurate analytical description of the quasi-similariton and the critical length: ≤ 10-4.

Critical distance zc does not depend on the sign of TOD.

Published in: Optics Letters / Vol. 35, No. 18, p. 3084 (2010)

Physical Review A / Vol. 84, No. 2, 023823 (2011)

Sim

ilarit

on b

reak

-up

with

3

Page 23: 15.30 o4 c aguergaray

October

2011

Motivation

Until recently most theoretical descriptions of the self-similar (SS.) propagation had been done assuming that only low-order nonlinear effects and low-order dispersion effects dominate the pulse evolution.

But experiment have shown that SS. propagation can be perturbed by effect like:

- Gain bandwidth limitation, - Third-order dispersion,- Gain saturation.

Self-

sim

ilar p

ulse

s

Page 24: 15.30 o4 c aguergaray

October

2011

Parabolic and Hyper-Gaussian similaritons

propagating in fiber with saturation effect.

Page 25: 15.30 o4 c aguergaray

October

2011

Gain saturation effect is important for the pulse evolution in normal fiber ring lasers.

It is negligible over the duration of a single pulse but cannot be neglected for a long pulse train since the amplifier gain will saturate over long time scale (> population relaxation time).

We use here the standard model equation for saturation effect obtained by averaging the gain dynamics in the presence of the pulse train.

Sim

ilarit

on w

ith g

ain

satu

ratio

n

0

1

ss

S

gg z

E z

E

Gain saturation model

Dependence of laser gain on the optical power at the steady state

Page 26: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

1

2

0

11 ,SS

S

g z g z dE

2 02

2 2z

g zi i

, , exp ,z t A z t z t

Analytical solution

The analytical solution is an exact asymptotical solution of the NLSE .

(From differential equation for the propagating pulses in optical amplifiers with an arbitrary gain function).

Slowly varying envelope:

Gain function:

Page 27: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

1

3 13

3ln

2s

p

13223

4 3sp

0g z

0

2

g

0 2

SS

E

g

0

SE

E

Analytical solution

3ln ln

Peak power:

Pulse duration:

Analytical solution with saturation effect takes the form:Dimensionless variables

Dimensionless parameters

1 22

1 2

2, 1p

u p

Page 28: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Parabolic similaritons

= 4000= 400

Simulation parameters: β2 = 0.02 ps2m-1 / = 2 10-5 W-1m-1 / g0 = 2 m-1.Input energy: E0 = 200 pJ 0 = 0.02.

Saturation energy: ES = 20 nJ S = 2.

Temporal profile and the chirp of the pulses for two different propagation distances:

Input parameters:

Page 29: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

= 600= 100

Input energy: E0 = 10 pJ 0 = 0.001.

Saturation energy: ES = 10 µJ S = 100.

Results for increased saturation energy:

Input parameters:

Parabolic similaritons

Page 30: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Hyper-Gaussian similaritons

For low amplification regimes, S < 0.3 (in our case ES < 3 nJ), the gain seen by the pulse goes very quickly to zero along the fiber.

The input pulses evolve into a different similariton regime with a linear chirp but a non parabolic shape. The pulse develops an Hyper-Gaussian shape.

It propagates through the fiber self-similarly with a linear chirp !

0

1

ss

S

gg z

z

Page 31: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Hyper Gaussian similaritons

2 4

, expE

pW W W

This function is a product of a Gaussian and a super-Gaussian therefore we named it Hyper-Gaussian pulse (HG pulse).

Two asymptotic non-linear attractors:

which route depends on S

Page 32: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Hyper-Gaussian similaritons

The spectral density of differs significantly parabolic shape.

HG pulses undergoe small spectral broadening due to weak non-linear effects what leads to a very smooth spectral shape (interest for fiber based amplification systems).

Linear chirp proving the self-similar aspect of the HG pulse propagation.

Page 33: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Hyper-Gaussian similaritons

Test for different input pulse shape:

All the pulses converge towards a HG shape pulse

with linear chirp !

HG pulse is a local asymptotic attractor.

Page 34: 15.30 o4 c aguergaray

October

2011

Sim

ilarit

on w

ith g

ain

satu

ratio

n

Our analytical solution for similariton pulses in a fiber amplifier with gain saturation allows an accurate predictions of the pulse temporal shape and chirp for a wide range of the saturation energy parameter.

A limit of s > 0.3 setting the lower boundary has been found

No upper limit on s… (computation time is restrictive).

A new local non-linear attractor leading to self-similar HG pulses has been identified.

Summary of ESAT study

Page 35: 15.30 o4 c aguergaray

October

2011

Conclusion

Two analytical solutions able to predict accurately the perturbations to the self-similar propagation caused by the TOD and the gain saturation.

Since low amplification is required for a pulse to evolve into a HG pulse,

Could be implemented in low energy laser systems delivering linearly chirped pulses.

Potential for pre-amplification stage of ultra-short pulse CPA systems to obtain linearly chirped

pulses with no spectral structure.

Page 36: 15.30 o4 c aguergaray

October

2011

Page 37: 15.30 o4 c aguergaray

October

2011

Page 38: 15.30 o4 c aguergaray

October

2011

Bric a brac

Page 39: 15.30 o4 c aguergaray

October

2011

0 expss

S

E zg z g

E

Page 40: 15.30 o4 c aguergaray

October

2011

Numerical simulations

The HG similaritons may form when:

- The energy E(z) of the pulse is a slowly growing function of distance,

- The peak power of the pulse is a constant or decreasing function of z.

Page 41: 15.30 o4 c aguergaray

October

2011

Overview

Motivations

Theoretical study

Numerical results

Conclusion

Page 42: 15.30 o4 c aguergaray

October

2011

Motivation

How to predict accurately the

critical distance and the pulse shape?

Sim

ilarit

on b

reak

-up

with

3

Page 43: 15.30 o4 c aguergaray

October

2011

Analytical pulse energy coincides with exact energy 0gzE z E e

Numerical results.

Page 44: 15.30 o4 c aguergaray

October

2011

OWN1 / Finot OSA OFC 2009

Context

Page 45: 15.30 o4 c aguergaray

October

2011

Theoretical study.

Renormalisation procedure : k

E zE z

z n

kz z

12 2 6

0

2

21 1( , ) exp ( , ) ( , )

2 3

g EU z gz Q z I z

Page 46: 15.30 o4 c aguergaray

October

2011

Motivation

However self-similar propagation is severally affected by Third Order Dispersion (TOD).

Novel features observed due to TOD in fiber amplifiers.

ΦNL accumulated in amplifier (SPM) compensated by TOD of

fiber stretcher + grating compressor

ΦNL = 0.4 π

ΦNL = 1.9 π

Zhou et al. (Wise) OE 13, 4869 (2005)

Grating stretcher and compressor best result.

ΦNL = 1.9 π

Page 47: 15.30 o4 c aguergaray

October

2011

Theoretical study.

No renormalisation With renormalisationprocedure applied

0 10E pJ3 3

3 10 /ps m

3 60 1.078 10

Page 48: 15.30 o4 c aguergaray

October

2011

Motivation

Novel features observed due to TOD in mode-locked lasers.

Logvin et al. OE 15, 985 (2007)

Page 49: 15.30 o4 c aguergaray

October

2011

Motivation

Net cavity GVD= 0.005 ps2

SMF and Yb fibers TOD (Negligible TOD)Similariton regime: symmetric pulse, top spectrum tilted.

PBF TOD = 500 fs3/mmCubicon-like features: asymmetric pulse, triangular shape spectrum.

PBF TOD = 1200 fs3/mm.Stretched Pulse regime: narrower pulse, broader spectrum with asymmetric sidebands.

Logvin et al. OE 15, 985 (2007)

Page 50: 15.30 o4 c aguergaray

October

2011 E0 3.5 3.5 8

0.029 0.1 0.1

Numerical results.

Numerical model :

with

Normalised variable

Page 51: 15.30 o4 c aguergaray

October

2011

Theoretical study.

Respecting < 10-4 condition :

52.44 10

0 1E 0.029

3

0E

Page 52: 15.30 o4 c aguergaray

October

2011

Theoretical study.

-8 -6 -4 -2 0 2 4 6 80.0

0.5

1.0

1.5

2.0

Ga

us

sia

n in

itia

l pu

lse

Normalised time

Pulse propagation in fibers with TOD is described by the following NLSE :

Analytical solution found using a first order perturbation theory :

Page 53: 15.30 o4 c aguergaray

October

2011

MotivationIn

tens

ity

Time [ps]

Dis

tanc

e [k

m]

Inte

nsity

Wavelength [nm]

Dis

tanc

e [k

m]

Latkin et al. OE 32, 331 (2007)

Page 54: 15.30 o4 c aguergaray

October

2011

Th

e A

ust

rali

an O

pti

cal S

ocie

ty