an introduction to nonlinear optics for accelerator
TRANSCRIPT
An Introduction to Nonlinear Optics for Accelerator Diagnostics
David Walsh
LA3NET, 1st Intl School on Laser Applications, GANIL, France,17 October 2012
What do we mean by nonlinear?
Linear response Nonlinear responses
x
f(x)
x
f(x)
Examples: Cheap stereos turned up too loud Electric guitar distortions (clipping)
Linear Optics
• Properties independent of light intensity – Linear refractive index – Snell’s Law – Linear absorption - Beer-Lambert Law
• Super position principle – Can look at inputs singularly and sum at the end
• Frequency of light is constant
– νin = νout
• Beams do not interact
– You can cross the beams with no effect – (not counting interference effects…)
Nonlinear Optics
• Refractive index and absorption can be a function of intensity – Two photon absorption – Self focussing
• Superposition no longer true
– Sum of responses to two inputs not the same as response to both inputs simultaneously
• Optical frequency can be changed
– Mixing processes – Harmonic generation
• Beams can interact
– Crossing the beams makes interesting things happen
Which Media are Nonlinear?
All of them!
• Solids
• Liquids
• Gasses
• Plasmas
• Vacuum
(via generation of vitual e- e+ pairs)
But it’s only observable if you pump them hard enough!
In general, lasers are required to observe nonlinear effects
The Birth of Nonlinear Optics
The field was kick started in 1961, by the first demonstration of second harmonic generation
This was made possible by the realisation of the ruby laser the previous year
Origin of the Nonlinear Response
The polarisation of atoms and molecules is affected by an incident EM wave
The potential gets very flat out at infinity, so the electron’s motion can easily go nonlinear!
Potential due to nucleus
Nucleus
Ener
gy
Position
Origin of the Nonlinear Response
Light wave
Incident light affects the polarisation of atoms/molecules throughout the medium, setting up a polarisation field.
This field reradiates EM waves, and in the linear regime is the origin of the refractive index.
Oscillating dipoles
Origin of the Nonlinear Response
Second Order Susceptibility Tensor
Third Order Susceptibility Tensor
E
P
E
P
E
P
Linear Susceptibility Tensor
Effects of the Nonlinearity
DC offset
Optical Rectification
Doubled frequency
Second Harmonic Generation
Effects of the Nonlinearity
Original frequency again!
Sum- and difference-frequency generation
Suppose there are two different-color beams present:
*
1 1 1 1
*
2 2 2 2exp( ) e exp( ) exp( ) x (( )) pE i t E i t E i t E i tt E
* *
1 2 1
* *
1 2 1 2 1 2 1 2
2 *
2
2 2
1 2
2 1 2
2
2
1 2
2 2
2
1 1 1
2
*
1
2
2 exp (
exp(2 ) exp(
2 exp (
exp(2 ) exp
) 2
) 2 exp
( )
( )
2
( 2
2
)
e )
)
x
2
p (
E i t E
E
i
E
t
E
E
t
E E
i t E E i
E i
i t E i
t
t E E
t
i t
E
Note also that, when i is negative inside the exp, the E in front has a *.
2nd-harmonic gen
2nd-harmonic gen
Sum-freq gen
Diff-freq gen
dc rectification
So:
Phasematching
• It seems that a there is potentially an awful lot going on
• Especially consider the cascading of processes that appears unavoidable
• How can we make practical use of these effect?
• Well, all the processes happen, but in general with low efficiencies
• We can drastically enhance efficiency via “phasematching”
Phasematching
Phasor Representation
( ) (2 )n n
2Frequency
Refr
active
index
Unfortunately, dispersion prevents this from happening!
The phase-matching condition for SHG:
Phasematching
Except in very rare cases near absorption features.
Birefringent Phasematching
z
x
y
nx
ny
nz
x
y
z
nx
ny
nz
The Index Ellipsoid
x The optic axis, is the direction in which both polarisations of light experience the same refractive index.
“Critical Phasematching”
nz
ny
k
z=optic axis
y
nx
ny
nz k
ne
y
z
Birefringent materials have different refractive indices for different polarizations. Ordinary and Extraordinary refractive indices can be different by up to 0.1 for SHG crystals.
( ) (2 )e on n 2Frequency R
efr
active index
ne depends on propagation angle, so we can tune for a given .
Some crystals have ne < no, so the opposite polarizations work.
We can now satisfy the phase-matching condition. Use the extraordinary polarization for and the ordinary for 2:
Birefringent Phasematching
Light created in real crystals
Input beam
Far from
phase-matching:
Closer to
phase-matching:
SHG crystal
Input beam
SHG crystal
Note that SH beam is brighter as phase-matching is achieved.
Output beam
Output beam
Walk-Off
The polarisation and electric fields are represented by vectors
2nd rank tensor
k
s E
D
Overall displacement
The Nonlinear Tensor
kjijkkjijki EEdEEP )2(0
)2(
0
Conventionally the susceptibility is replaced by the nonlinear tensor 2dijk
dijk is a 3x3x3 tensor, and maps all components of the vector E to vector P
Consider: d(i,j,k) describes the nonlinear coupling of fields Ea and Eb
along j and k into the polarisation along i
x,y,z=1,2,3
Note that the order of the E fields is not important d232 = d223
x
y
z
P Ea Eb
d232 describes
Simplifications of the Nonlinear Tensor Using the “piezoelectric contraction” the nonlinear response is now (order of fields not important)
Where elements j,k have been replaced with l
We can go one step further in lossless media, the Kleinman’s contraction
For given i,j,k then dijk = dikj = dkji = dkij = djik = djki
b
X
a
Y
b
Y
a
X
b
X
a
Z
b
Z
a
X
b
Y
a
Z
b
Z
a
Y
b
Z
a
Z
b
Y
a
Y
b
X
a
X
o
Z
Y
X
EEEE
EEEE
EEEE
EE
EE
EE
dddddd
dddddd
dddddd
P
P
P
363534333231
262524232221
161514131211
2
jk 11 22 33 23, 32 13, 31 12, 21
l 1 2 3 4 5 6
d11 d12 d13 d14 d15 d16
d16 d 22 d 23 d 24 d14 d12
d15 d 24 d 33 d 23 d13 d14
di,l =
Now there are only 10 independent values! The actual elements present depends on the crystal symmetry.
The Effective Nonlinear Coefficient
An example:
LBO crystal, E-fields along (1,1,0) and (1,-1,0)
)](,0,0[
2
1
11
0
0
0
11
11
...
.....
.....
2
2415
2
332415
24
15
ddP
ddd
d
d
P
P
P
o
O
Z
Y
X
normalising factor
Maxwell’s Equations
Evaluating the effect of an induced polarization on a wave requires solving Maxwells equations, with an induced polarization term. This gives rise to an extra term in the wave equation: This is the Inhomogeneous Wave Equation. The polarization is the driving term for a new solution to this equation.
2
2
02
2
0
2
tt
NL
PEE
(Normally = 0)
Coupled Wave Equations If we consider the interaction of 3 waves propagating in the z direction
Substituting the three waves to the inhomogeneous wave equation we get a set of 3 coupled wave equations
Ei
(1)(z, t) 1
2 (E1i
exp[i(1t k
1z)] + c.c.)
Ek(2 )
(z,t ) 12 (E2 k exp[i(2 t k2z)] + c.c.)
Ej(
3)(z,t) 1
2 (E3 j exp[i(3 t k3z)] + c.c.)
i,j,k are the wave polarisation, and can be x or y
dE1i
dz
i 1
cn1
dijk
' E3 jE
2k
* exp(ikz )
dE2 k
*
dz
i2
cn2
dkij'E1iE3 j
*exp(ikz )
dE3 j
dz i 3
cn3
djik
' E1iE
2 kexp(ikz )
Phase mismatch E is complex as it contains a static phase term
123
Manley-Rowe Relations dE1i
dz
i 1
cn1
dijk
' E3 jE
2k
* exp(ikz )
dE2 k
*
dz
i2
cn2
dkij'E1iE3 j
*exp(ikz )
dE3 j
dz i 3
cn3
djik
' E1iE
2 kexp(ikz )
E1
*n1 /1
E2n2 /2
E3n3 /3
n1d E1
2
1dz
n2d E2
2
2dz
n3d E3
2
3dz
idijk
'
cE1
*E2
*E3 exp(ikz)
I nco E
2
/2
1
1
dI1
dz
1
2
dI2
dz
1
3
dI3
dz
“Photon Splitting” Picture
ω3
ω2
ω1 “Parametric” – no absorption
Example Analysis of SHG Using the coupled wave equations we can now analyse second order processes.
All we need to do is apply appropriate assumptions and boundary conditions, and then solve the equations.
Quick example for SHG:
021 dz
dE
dz
dE
123 22
)exp(21
'
3
33 kziEEdcn
i
dz
dEjik
bkziaE )exp(3)exp(3 kzikai
dz
dE
kcn
EEda
jik
3
21
'
3 0@03 zEkcn
EEdab
jik
3
21
'
3
Assume negligible depletion
Trial solution
SHG Analysis Continued…
Signal grows quadratically
with length when phasematched
)exp(14
3
'
3 kzikcn
EdE
jik
*2
2
3
2
22'2
*
333 )exp(1)exp(1
2
4kzikzi
knc
IdEEI
jik
21
3 2
EEE
2
2
2
3
2
222'2
3
2
2sin
4
kz
kz
nc
zIdI
jik
Gain depends on wavelength Phasematching is important!
Quasiphasematching
sinc 2 ( kz / 2 )
converted field in crystal
Highest nonlinear coefficients don’t generally match with allowed phasematching conditions
Power oscillates between second harmonic and fundamental
What if we are phasematched say, here
Quasiphasematching
kLcoh
0 2 3dE1i
dz
i 1
cn1
dijk
' E3 jE
2k
* exp(ikz )
dE2 k
*
dz
i2
cn2
dkij'E1iE3 j
*exp(ikz )
dE3 j
dz i 3
cn3
djik
' E1iE
2 kexp(ikz )
iexp
iexp
iexp
1exp i
Change the sign of dijk?
Quasiphasematching
g 2lcoh 2
k kqpm k3 k2 k1 2
g
with domain reversal (quasi- phasematching)
no domain reversal (no phasematching)
con
ver
sion
g
z
This is possible in ferro-electric materials, e.g. lithium niobate, by applying a periodic high electric field which flips the material domains. Efficiency is improved by an order of magnitude!
Non-colinear phasematching The wave vector, k, is actually a vector property.
This allows us to have non-colinear waves and still achieve phasematching.
One application is the compensation of “walk-off”
In an appropriate material, it can also be arranged that multiple wavelengths phasematch with the same propagation angle, leading to a very large bandwidth for the nonlinear process.
3k
2k1k
If not compensation for walk-off, this has the effect of limiting the interaction length
Sum Frequency Generation 1
nonlinear medium
3
2
1
2
1
2sinc
22
0
3
321
2
21
2
3
3
Δkz
cnnn
zIIdI
ijk
In the general, low conversion efficiency case
Difference Frequency Generation
nonlinear medium 3
2
1
3
1
)(cosh)0()( 2
11 zIzI
)(sinh)0()( 2
1
212 zIzI
i.e. both waves experience gain at the expense of the “pump”
input
output
fields
z
2
21
2
3
2
21
cnn
Edijk
*Starting from low conversion conditions, it takes the quadratic form of the SFG solution
Optical Parametric Generation/Oscillation
• For DFG, both of the down-converted (lower frequency) waves experience gain
• Additionally, there exists spontaneously emitted photons (thermal, quantum noise)
• If the gain is sufficiently high, the few photons that are generated with the correct polarisation, frequency, and direction, will be amplified – i.e. you can get new wavelengths without the need to seed the process!
Parametric Gain
mirror mirror
Broadly tuneable laser-like sources
OPG
OPO
Pockel’s Effect
The second order nonlinearity can also be exploited to modulate the phase, amplitude or frequency of light by applying a suitable electric field
Consider the polarisation response:
And define the total susceptibility:
Now, the refractive index is:
Using aTaylor series approx.
This effect was observed >50 years before the invention of the laser!
P o((1)
E ( 2)
E2 ...)
E
P
o
n 1 (1
(1)
( 2 )E ...)
1 /2 (1
( 2 )
E
no2 )
1/2no
n no ( 2 )E
2no
Important Considerations
• Temperature range for phasematching
• Angular acceptance for phasematching
• Which element of deff to use
• Focussing trade-off
w, mm
x, mm
Intensity
Interaction Length
Walk-off
Ultrashort Pulses
kvg
Time
Starting at t=0 Say all beams overlap
3 pulses begin to separate Conversion still happens in overlap
Pulses are separate No more conversion Generated pulse “stretched”
Need to consider Group Velocities
Now, Some Applications
Photo Injector System
A short, intense, pulse of high energy photons is required at the photocathode in order to liberate electrons for acceleration.
Lasers can produce the synchronised, high intensity pulses, except they are at much longer wavelengths!
e.g. Ti:Sapphire 800nm, Nd 1064nm, Er 1550nm
The solution, as we know, is nonlinear optics.
What if we want UV light, as required by some photocathodes.
Try 3rd harmonic of Ti:Sapphire?
Third Harmonic Generation
Can be achieved via a 3rd order process but a very high pump field is required due to the weak susceptibility
Thankfully there is a better way – cascaded second order processes!
Femtosecond Ti:Sapphire
Second Harmonic Generation
800nm Sum Frequency Generation
400nm 266nm
Can be 50%+ efficient Also quite efficient
p
ppSHG
SFG
pp
SHG
33
111,
11111
Electron Bunch Length Diagnostics
• A number of techniques are being refined for the optical characterisation of the bunch temporal profile.
• Generally, this is achieved by shifting characteristics of the coulomb field onto an optical frequency carrier beam
• This optical signal is then characterised in order to infer properties of the electron bunch
The field of THz generation and detection by modelocked lasers is well developed, and as the coulomb field is similar in nature to a THz pulse, similar techniques are used.
Typically, THz detection is performed via the electro-optic effect.
Coulomb field of
relativistic bunch
probe laser
non-linear crystal
(electro-optic effect)
laser pulse
(linearly polarised) elliptically polarised
Refractive index modified by external (quasi)-DC electric field
intensity dependent
on ‘DC’ field strength
quasi-DC description ok if tlaser << time scale of EDC variations
(basis for Pockels cells, sampling electro-optic THz detection, ...)
N.B. Time-varying refractive index is a restricted approximation to the physics
( albeit a very useful and applicable formalism for majority of situations )
Physics of EO encoding ... standard description
Electron Bunch Length Diagnostics
Spectral Decoding
Spatial Encoding
Temporal Decoding
Spectral upconversion**
o Chirped optical input o Spectral readout o Use time-wavelength relationship
o Ultrashort optical input o Spatial readout (EO crystal) o Use time-space relationship
o Long pulse + ultrashort pulse gate o Spatial readout (cross-correlator crystal) o Use time-space relationship
o monochomatic optical input (long pulse) o Spectral readout o **Implicit time domain information only
(2)(;thz,opt)
opt + thz
convolution over all combinations of optical
and Coulomb frequencies
Electro-optic detection bandwidth
thz
opt
opt - thz
opt
description of EO detection as sum- and difference-frequency mixing
THz spectrum (complex)
propagation & nonlinear
efficiency
geometry dependent
(repeat for each principle axis)
optical probe spectrum (complex)
EO c
ryst
al
This is “Small signal” solution. High field effects c.f. Jamison Appl Phys B 91 241 (2008)
GRENOUILLE
• Grating Eliminated No-nonsense Observation of Ultrafast Laser Light E-fields…
• Spectrometer is replaced by a thick nonlinear crystal and a CCD – the phasematching condition now determines the angular spread
• Delay stage has been replaced by a Fresnel Biprism, thereby mapping the delay across the CCD and allowing single shot pulse characterisation
Thank you for listening