Transcript Lecture 3 Ultra-short pulse parametric devices

Lecture 3
Ultra-short pulse parametric devices
David Hanna
Optoelectronics Research Centre
University of Southampton
Lectures at Friedrich Schiller University, Jena
July/August 2006
Lecture Outline
• General features and attractions of ultrashort pulse parametric
devices
• Synchronously Pumped OPOs (SPOPOs): general considerations
• Specific examples of SPOPO performance
• Optical Parametric Amplifiers (OPA), Optical Parametric Chirped
Pulse Amplifiers (OPCPA) & Optical Parametric Generators (OPG)
• Carrier Envelope Phase considerations
Attractions of parametric processes in the ultrashort pulse
regime
•
High gain; damage intensity behaves ~1/(pulse duration)½
•
Broad gain bandwidth
•
Wavelength flexibility (eg different from Ti:Sapphire!)
•
Reduced ASE, reduced background, good contrast
•
High Quantum efficiency
•
Low thermal effects
•
Good beam quality
•
Scalability
Some disadvantages of parametric processes
•
Small aperture dimensions available
•
No energy storage
•
Synchronisation requirements
•
High pump brightness required
Some general features of ultra-short pulse
parametric devices
•
High gain and wide bandwidth can be obtained in a single pass of a
parametric amplifier: lasers require regenerative amplification
•
For the shortest pulses, ensure a large enough gain-bandwidth + good
temporal overlap between the interacting waves over the NL medium
Short crystal length can ensure the above, but places limits on the
achievable gain
•
Alternative ways to increase the gain bandwidth include:
near-degenerate operation
non-collinear phase-matching
•
Double refraction effects are reduced for shorter crystals
•
Non-collinear phase-matching can contribute to group-velocity-matching
Dependence of double-refraction effects on crystal length
For a given double-refraction walk-off angle ρ, and beam diameter D, the effect of
walk-off in a crystal of length is insignificant if
ρL/D << 1
For confocal focussing, 2πw02n/λ = L, i.e., D = 2w0 = [2Lλ/nπ]½
so;
ρL/D = ρ[πnL/2λ]½
Hence, for shorter crystals, as required for shorter pulses,
confocal focussing is less compromised by double refraction
10x shorter pulse →10x shorter Xtal → tolerate√10x greater ρ value
Synchronously-pumped OPO
>
>
Mode-locked pump:
pulse separation
matches round trip
of OPO
N.L.Xtal
>
>
Signal and idler
output pulse train
>

OPO gain corresponds to the peak power of the pump pulse

Crystal length must be short enough so that group velocity
dispersion does not separate pump, signal and idler pulses in the crystal.
SPOPO pump requirement versus crystal length
If length L is determined by the allowable Group Delay Difference,
then, L  T
and if confocal focussing is used,
then, gain  LP = LE/T  E
Hence, threshold is specified by an energy, independent of pulse duration,
& for a given repetition rate,
threshold average power is then independent of pulse duration.
But Self Phase Modulation is more problematic for shorter pulses, since
effect of SPM ( fractional spectral broadening)  IL  PL/L  E/T
(T,P,E,I are, respectively, pump pulse duration, power, energy, intensity)
Some Attractions of SPOPOs

Low threshold average power (amenable to diode pumping)

Power scalable, eg via fibre-pumped SPOPOs

Very wide tuning

Synchronised outputs at two wavelengths
(e.g. for CARS)

Very high gain possible, can oscillate even with
very high idler loss

Very high efficiency,
e.g. makes the tandem OPO practical
SPOPO facts and figures

Average output power
> 20 W

Shortest pulses
13 fs

Tuning range
0.45 – 9.7 micron

Efficiency
(diode  laser  OPO)
25%
Slope efficiency
>100% (170% observed)

Crystal length constraint for a SPOPO
 Require enough signal gain bandwidth for a
signal pulse duration ~ pump pulse duration T
0.4 x 2 / T 
L

L v g12  v g11
Tv g 2 v g 1
v g1  v g 2
(away from degeneracy)
Use higher order terms in
Taylor expansion if the vg
are nearly equal
 Require signal (& idler) pulse not to walk away from pump pulse
L / vg 3  L / vg 2  T
L
Tv g 3 v g 2
vg3  vg 2
Signal case
Typical resonator arrangement for SPOPO
How to tune a QPM OPO
Angle tuning may not be an option, so:

Fixed pump; tune crystal temperature (fine tune)
change grating period (coarse tune)

Tune pump wavelength

Fixed pump; tune across gain-bandwidth via intra-cavity filter, or
diffraction grating reflector.
SPOPO slope efficiency of > 100%
L.Lefort, et al., Optics Communications Vol.152 pp.55-58 (1998)
Order of magnitude pulse compression in a PPLN
SPOPO
4ps pump, 250fs signal,
20mm PPLN
~100fs/mm pump/signal
Group delay difference
Lefort et al. Opt Letts, 24(1),28,1999
Other features of SPOPO

Cavity length change can change signal wavelength:
not a good technique for tuning as pulse characteristics will change

Oscillation tolerates cavity length changes of many pulse widths.
Stabilise cavity length via stabilising the output frequency

Tuning through the gain profile can lead to higher
order transverse modes of the signal

Tuning elements involving angular dispersion, eg grating, produce tilted
pulses

In QPM materials, many additional outputs may be
seen (2ωs, 2ωi, ωs+ωp, ωi+ωp).
PPLN SPOPO with feedback via diffraction grating
Tilted signal pulse
is ‘cleaned up’ in
PPLN amplifier
before exiting the
cavity
Hanna et al J Phys D Appl Phys,34,2440, (2001)
Tilted pulses produced by diffraction grating
From Hanna et al.J Phys D, Appl Phys., 34,2440, (2001)
CdSe tandem-pumped SPOPO
M.A.Watson, M.V.O'Connor, D.P.Shepherd, D.C.Hanna Optics Letters 28 (20) pp.1957 (2003)
CdSe SPOPO
Non-critical (θ = 90o ) type-II
phase-matching curves in CdSe, for
pump-wavelength tuning. The pump
wavelength range has been limited
at the long end to the signal range
from the pump OPO and at the short
end by twice the band gap
wavelength, where two-photon
absorption would become
significant. Inset: diamonds indicate
experimental idler tuning points.
M.A.Watson, M.V.O'Connor, D.P.Shepherd, D.C.Hanna Optics Letters 28 (20) pp.1957 (2003)
Infrared absorption edge of Lithium Niobate
Sato et al Appl. Optics 38, 2560, 1999
SPOPO with idler absorption
Signal gain, if small, is
For large αL this is

(1)
 2 L2 2exp(L / 2)  1  L/(L / 2) 2
 2 L2 (4 / L)
i.e. threshold is increased by αL/4
Lowenthal IEEE JQE, 34, 1356 (1998)
Lefort et al APL, 73 (12), 1610 (1998)
Watson et al Opt.Letts 27 (23), 2106 (2002)

SPOPO with idler absorption (2)
Photon conversion efficiency to idler output:
2
8 s  i D (1  D ) d eff
Ip
I i / i

I p /p
 2 (1  R )n 3 0 c 3
(D is pump depletion, R is signal round-trip loss)
Output idler power is that generated in last extinction
length of the crystal
Strategy for efficient idler generation:
Increase Ip until D~ 0.5 and make R as small as
possible (eg use ring resonator).
But avoid excessive (damaging) signal intensity
M.A.Watson et al. A.P.L.73 (12), 2108,(2002)
SPOPO with idler absorption
(3)
M.A.Watson et al, Optics Letters Vol.27(23) pp.2106-8 (2002)
SPOPO pumped by femtosecond mode-locked
fibre laser
O’Connor et al Opt Letts., 27 (12), 1052, (2002)
High power femtosecond fibre feedback SPOPO
19W av o/p@
1450nm,
7.8W @3570nm
Südmeyer et al. Opt Letts. 29, 1111, (2004)
Fibre feedback SPOPO: insensitivity of output
power to resonator length changes
Südmeyer et al. Opt Letts., 29,1111,(2004)
Femtosecond (down to 13fs) visible OPO
via non-collinear phase-matching in BBO
Gale et al. JOSA B, 15, 792, (1998)
Coupled NL equations for signal & idler in the pump pulse frame
Gale et al. JOSA B 15, 792, (1998)
Non-collinearly phase-matched femtosecond OPA with a
2000cm-1 bandwidth
Shirakawa and Kobayashi Appl. Phys. Letts., 72(2),147, 1998
Matching of group velocities by spatial walk-off in collinear
three-wave interaction with tilted pulses
Danielius et al., Opt. Letts., 21, 13, 973, (1996)
Pulse-front matched OPA for sub-10-fs pulse generation…
Shirakawa et al. Opt. Letts., 23,16,1292, (1998)
Visible pulse compression to 4fs by OPA +programmable
dispersion control
Prism P3 imparts tilt
(angular dispersion)
to the SH (ie pump)
beam
Baltuska et al., Opt Letts., 27,306, (2002)
Visible compression to 4fs by OPA+ programmable
dispersion control
Dashed curve is for
monochromatic pump.
Inset shows spectrum of
SH used as pump
Baltuska et al., Opt. Letts., 27, 306, (2002)
Yet more OPA designs…
• OPCPA +multiple pumps, at different wavelengths, to increase the
gain bandwidth.
Wang et al., Opt Commun., 237,169, (2004)
• Use of chirped broadband pump + operation near degeneracy.
Limpert et al., Opt. Express, 13, 19, 7386, (2005)
• Ultrabroadband (octave-spanning) OPCPA, using angularly
dispersed signal
Arisholm et al., Opt. Express, 12, 518, (2004)
Efficiency-enhanced soliton OPA
• Pump, signal and idler are mutually trapped in a spatial soliton
• This requires a phase-mismatch whose ideal value depends on the
mix of pump, signal and idler powers
• These powers evolve through the amplifier, hence ideally one needs
a longitudinally varying phase-mismatch through the medium
• SOLUTION: Use aperiodic QPM medium
Rodriguez et al JOSA B,19, 1396, (2002)
Tandem-chirped OPA grating design for simultaneous
control of group delay and gain control
• Chirped grating 1 produces idler with frequency-dependent group delay
• Idler from grating 1 acts as signal for grating 2, hence idler from 2 has
frequency of original signal
• Grating 2 compensates group delay dispersion of grating 1
Charbonneau-Lefort et al., Opt. Letts., 30,634,(2005)
Cavity-enhanced OPCPA
• Cavity acts as a reservoir and amplifier for the pump
• Long pump pulse avoids cavity dispersion issues
• Need to minimise optical Kerr effect in cavity
Ilday & Kärtner, Opt . Letts.,31, 637, (2006)
Generation of few cycle terawatt light pulses via OPCPA
CEP stabilised pulses from TiS oscillator maintain their CEP in OPA + compressor
Witte et al., Opt. Express, 13, 4903, (2005)
Carrier Envelope Phase (CEP)
• Carrier phase offset between
carrier peak and envelope peak
can vary from pulse to pulse
• This has significant effects in
high field experiments using
few-cycle pulses
Brabec and Krausz Rev. Mod. Phys., 72,545,2000
Self-stabilisation of CEP via parametric processes
In an OPA, with signal only as input, the phase relation, φp-φs-φi = -π/2 ,
applies through the medium if Δk = 0
If the signal is derived from the pump, eg as in generation of supercontinuum,
signal and pump have the same phase behaviour.
So, using the pump to amplify this signal in an OPA leads to a CEP stable
idler even if the pump is not CEP stable.
If this CEP stable idler does not have the desired power it can be used as the
input signal to a second amplifier, OPA2
Since this amplified signal has its phase preserved in OPA2 one now has a
high power pulse that is CEP stable
Baltuska et al., Phys Rev Letts. 88, 133901, (2002)
Generation of high energy self-phase-stabilised
pulses via DFG and OPA
DFG between spectral components of the supercontinuum produced in
the fibre gives a CEP stable pulse whose stability is maintained in OPAs
Manzoni et al. Opt Letts., 31, 963, (2006)
Concluding remarks
•
OPAs are widely seen as a preferred alternative to TiS for
amplification of ultrashort pulses to high powers
• Much needs to be done to establish power-scaling limits of OPOs,
and OPAs.
• Designs for OPAs are numerous and new proposals keep
appearing. Not yet a mature field; work is in progress.
• Different circumstances, e.g. pulse energy, duration, wavelength,
call for different designs. Not a case of “one size fits all”
• Numerical calculations need to include transverse effects. Planewave models are ignoring vital aspects