Wavevector (Phase) Matching

k2 > k1

k  2 k ( )  k ( 2 )

I(2)  2 k vac ( ) n ( )  k vac ( 2 ) n ( 2 )

 2 k vac ( )[ n ( )  n ( 2 )]
- Need k small to get efficient conversion
- Problem – strong dispersion in refractive
z
index with frequency in visible and near IR

n = [n()-n(2)]  0 because of dispersion  linear optics problem

Solutions:
1. Birefringent media
n() 2. Quasi-phase-matching (QPM)
3. Waveguide solutions

 2
 Birefringent Phase-Matching: Uniaxial Crystals
Uniaxial Crystals - optically isotropic in the x-y plane
- z-axis is the “optic axis”
 no2 0 0 
- for k || z, any two orthogonal directions
 = 0 0 n 2 0
o are equivalent eigenmode axes

x

0 0 ne2  cannot phase-match for k || z k
no “ordinary” refractive index ne “extraordinary” refractive index z
y
in x-z plane, ne()
 x
 - Note: all orthogonal axes in x-y plane
E e k in x-z plane are equivalent for linear optics
- Eo always in x-y plane
- Ee always has z component
 ( 2)
- angle from x-axis important for d eff
y z

Convention: Eo along y-axis, no
no ( )ne ( )
ne ( ,  ) 
[ne2 ( ) cos 2 ( PM )  no2 ( ) sin 2 ( PM )]1 / 2
 Type I Phase-Match
1 fundamental eigenmode
+ve uniaxial  ne>no
1 harmonic eigenmode
+ve uniaxial oee no(2) = ne(,)
Harmonic (1 photon) Fundamental (need 2 identical photons)
k = 2ke() – ko(2) = 2kvac()[ne(,)-no(2]
x

k
Ee
y z
Eo
Critical phase-match 0<</2 Non-critical phase-match =/2
no(2) = ne(,) no(2) = ne()
n()
n()
ne ne
no
ne(,) no
 Because
 of optical isotropy in x-y plane
k for phase-matching lies on a cone
at an angle  from the z-axis

Eo (2 ) lies in x-y plane
  
Ee () is  to k and Eo (2)

( 2)
Note: d eff does depend on angle  from x-axis in x-y plane!!
no ( )ne ( )
no (2 )  ne ( , ) 
[ne2 ( ) cos 2 ( PM )  no2 ( ) sin 2 ( PM )]1 / 2
ne ( ) no2 ( )  no2 (2 )
insert cos ( PM )  1  sin ( PM )  sin(  PM ) 
2 2
no (2 ) no2 ( )  ne2 ( )

Range of phase-match frequencies limited by condition ne()  no(2)

Type I -ve uniaxial  no>ne
-ve uniaxial eoo no() = ne(,2)
Harmonic Fundamental

k = 2ko() – ke(2) = 2kvac()[no()-ne(,2]

Non-critical phase-match =/2 Critical phase-match 0<</2

n() n()
no ne()
no
ne
ne

no (2 )ne (2 ) ne (2 ) no2 (2 )  no2 ( )

no ( )  ne ( PM ,2 )   sin( PM ) 
ne (2 ) cos ( PM )  no (2 ) sin ( PM )
2 2 2 2 no ( ) no2 (2 )  ne2 (2 )
 Type II Phase-Match
2 fundamental eigenmodes
+ve uniaxial  ne > no
1 harmonic eigenmode
1
+ve uniaxial oeo no (2ω)  [no ( )  ne ( ,  )]
2
Harmonic (1 photon) Fundamentals, need 2
(orthogonally polarized) photons
k = ke() + ko() – ko(2) = kvac(){[ne(,)+no()] - 2no(2)}

n() ne
1
Range of [no ( )  ne ( ,  )]
2
1 no
 no ( )  [no ( )  ne ( )]
2
2ne ( ) no (2 ){no ( )  no (2 )}
sin(  PM ) 
2no (2 )  no ( ) no2 ( )  ne2 ( )
 Type II -ve uniaxial  no > ne

1
-ve uniaxial eoe ne ( ,2 )  [no ( )  ne ( ,  )]
2
Harmonic Fundamental

k = ke() + ko() – ke(2)= kvac()[ne(,) + no()] - kvac(2)ne(,2)

= kvac(){[ne(,) + no()] - 2ne(,2)}
Range of ne ( ,  )
 no ( )  ne ( )]
1
Range of [no ( )  ne ( ,  )]
2
1 n()
 no ( )  [no ( )  ne ( )] no
2 Unique 

1  no ( )ne ( ) 
 no ( ) ne
2  n 2 ( ) cos 2   n 2
( ) sin 2
 PM 
 e PM o 
no (2 )ne (2)

ne2 (2 ) cos 2  PM  no2 (2 ) sin 2  PM
 “Critical” Phase Match “Non-Critical” Phase Match

Type I eoo Z (optic) axis

 
PM k
ne(2,) k
no(2)
Poynting
vectors
no() = ne(2)
Curves are tangent

no()
Difference between the normals to
the curves represent spatial walk-off
between fundamental and harmonic
ne(,)
Reduces conversion efficiency

ne2 (2 , )  1 1 
linear optics  tan     2   sin( 2 PM )
 no (2 ) ne (2 ) 
2 2

n (2 )  no (2 )
for small birefringe nce  e sin( 2 PM )
no (2 )
 “Critical” Versus “Non-Critical” Phase Match
How precise must PM be? I(2)  sinc2[kL/2= /2]  4/2 0.5
e.g. Type I eoo (-ve uniaxial)

k  2k vac ( )[ no ( )  ne (2 , )]  2 [no ( )  ne (2 , )] no ( )  f ( )
c
   PM   PM  PM  angular detuning from phase - match (from maximum SHG)
n (2 , ) 1 ne2 (2 , )
ne (2 , PM   )  ne (2 , PM )  e  PM  PM   PM  PM
2

 2  2
k 2[no ( )  ne (2 , )]  n (2 ,  )
L L  L[no ( ) 0ne (2 ,  PM )  e  PM  PM ]
2 2c c 
ΔkL 
ne (2 ,  ) 
2    ( ) 1
Evaluating  PM  
2  
 4 L {no (2 )  ne (2 )} sin( 2 PM )
PM

Note key role of birefringence

I(2,)
PM (Half width at half
maximum) Usually quote the “full”
 acceptance angle = 2PM
 PM diverges as " non - critical phase - matching" (   / 2) is approached need next term
1/ 2
1  ne (2 ,  )
2  λ( ) 
|  2
in expansion   PM  
2  2 PM PM
 4 L[ no ( 2 )  ne ( 2 )] 

Small birefringence is an advantage in maintaining a useful angular bandwidth

Non-collinear Phase-Matching

We have discussed only collinear wavevector matching. However, clearly it is possible

to extend the wavelength range of birefringent phase-matching by tilting the beams.

Interaction limited to this region

Biggest disadvantage: Walk-off

 Quasi-Phase-Matching
( 2)
- direction of d ijk is periodically reversed along a ferroelectric crystal

Periodically poled LiNbO (PPLN): a is the “mark-space ratio”

3
 1a>0
z

x
2 p
(2)
 d333 ( x)   d333
( p)
exp[ i x]
p

PPLN
p’th Fourier component

dE3 (2 , x) k (2 )

  i vac  d 333 (2 ;  ,  )
( p)
dx 2n(2 ) p
2 p
 E32 ( , x) exp{i[2ke ( )  ke (2 )  ]x}


k = 2ke() – ke(2) + pK K  2 / 

Change phase-matching condition

by manufacturing different 
 Quasi-Phase-Matching: Properties (1)
 
 A – perfect phase match with k (2 )  2k ( )
E z
B – QPM with p=1
C- k  0
c/ 
k
x

n() n(2)

A modified form of
“non-critical” phase-match

vac ( ) vac ( )
 PM  | {[ ne (2 )  ne ( )]  [no (2 )  no (2 )]  }1 |.
4L 4
 Quasi-Phase-Matching: Properties (2)
( p)
The relative strengths of the Fourier components d eff depend on a.
sin( pa )
( p)
d 333  (2a  1)d 333 p  0 ( p)
d 333 2 d 333 | p | 1
p
k = 2ke() – ke(2) + pK
p0 Not useful since ke (2 )  ke ( )
p  1  2ke ( )  ke (2 )  K Not useful because ke (2 )  ke ( )

p  1  2ke ( )  ke (2 )  K Phase matching is possible

(1)
optimizing d 333  2d 333  [sin( pa )/p ] :
1 2
p  1 optimum for a   d 333
(1)
 d 333
2 
1 3 1
p  2 optimum for a  ,  d 333
( 2)
 d 333
4 4 
1 1 3 (3) 2
p  3 optimum for a  , ,  d333  d333
6 2 4 3
Higher order gratings can be used to extend phase-matching to
shorter wavelengths, although the nonlinearity does drop off, d333  2d333 / p
( p)
 State-of-the-art QPM LiNbO3

-fundamental and harmonic co-polarized

- d(2)eff  16 pm/V (p=1)
- samples up to 8 cms long
- conversion efficiency  1000%/W (waveguides)
- commercially available from many sources
- still some damage issues

Right-hand side picture shows blue,

green-yellow and red beams obtained
by doubling 0.82, 1.06 and 1.3 m
compact lasers in QPM LiNbO3
 Solutions to Type 1 SHG Coupled Wave Equations
e.g. Type I -first assume negligible fundamental depletion
 E() valid to 10% conversion
2 kL
 i
2 sinc kL E 2 (0,  )
E ( L,2 )  i ( 2)
Ld eff e
E() E(2) cn(2 ) 2
E(2) and E() are /2 out of phase at L=0!!!
1 2 2 | d eff
( 2) 2
| 2 kL 2
I ( L,2 )  n(2 )c 0 | E ( L,2 ) |  2
2
L2
sinc ( ) I (0,  )
2 n ( )n(2 )c  0 3 2

Large Conversion Efficiency (assume energy is conserved  Kleinman limit)

dE3 ( z,2 )  ~ ( 2) 2
i d eff E1 ( z,  )eikz
dz cn(2 )
dE1 ( z,  )  ~ ( 2)
i d eff E3 ( z,2 )E1* ( z,  )e ikz .
dz cn( )
Field Normalization
1 1
 1   1 
E1 ( z ,  )   n1 ( )c 0  I t (0) 1 ( z ) ei1 ( z ) E3 ( z ,2 )   n3 (2 )c 0  I t (0) 3 ( z )ei3 ( z )
 2   2 

Energy conservation : I total ( z )  I1 ( z ,  )  I 3 ( z ,  )  [ 12 ( z )  12 ( z )]I t (0) with 1 (0)  1, 3 (0)  0
  ~ ( 2) k
Defining : ~  d eff (-2ω; ω, ω) | E (0,  ) |   ~z s  ~   s  21 ( )  3 ( )
cn 
Normalized Coupling Constant Normalized Normalized “Global Phase”
Propagation Wavevector
Distance Detuning
Inserting into coupled wave equations and separating into real and imaginary equations
d d
1 ( )  1 ( ) 3 ( ) sin  3 ( )   12 ( ) sin 
d d
d d d d d 12 ( )
  s  2 1 ( )  3 ( ) 1 ( )  3 ( ) cos  3 ( )  cos 
d d d d d 3 ( )

d d d d 12 ( )
Substituting for 1 ( ) and 3 ( ) into     s  cos  [2 3 ( )  ]
d d d d 3 ( )
d cos  d
After some manipulation    s  ln[ 12 3 ]
d sin  d

Integrated by the method of the variation of the parameters

1
 cos [1  32 ( )]3 ( )  s32 ( )  C (constant independent of z )
2
s3 ( )
3 (0)  0  C  0  cos( )  
2[1  32 ( )]
 2
d  s 
 3 ( )   12 ( ) sin   Sgn[1  32 ( )][1  cos 2  ]1/ 2  Sgn[{1  32 ( )}2    32 ( )]1/ 2
d  2 
Sgn is determined by the sign of boundary (initial) condition sine( 21 (0)  3 (0))
The general solution is given in terms of Jacobi elliptic function sn(u 1 | u 4 )

12 ( )  1  u 2 sn 2 (u 1 | u 4 ) 32 ( )  u 2 sn 2 (u 1 | u 4 ) u 1  s / 4  1  (s / 4) 2

Solutions simplify for s=0,

i.e. on phase-match
z
3 ( z )  tanh( )
 pg
z
1 ( z )  sech ( )
 pg
cn
 pg  ~
d eff
( 2)
| E (0,  ) |

The conversion efficiency saturates at unity (as expected)

 Note the different shape of the harmonic
Δs0 Δs=0.2 response compared to low depletion case

k2 > k1

I(2)

Δs=0.2
z

The main (Δk=0) peak with increasing

input which means that the tuning bandwidth
becomes progressively narrower.
The side-lobes become progressively narrower
and their peaks shift to smaller ΔkL.

~  3 (solid black line);~  1.5 (dashed black line); ~  0.75 (red dashed line);
~  0.25 (solid blue line, curve multiplied by factor of 4).
 Solutions to Type 2 SHG Coupled Wave Equations
d 2 ~ ( 2)
E1() E3 ( z ,2 )  i d eff E1 ( z ,  )E2 ( z ,  )eikz
dz n3 (2 )c
 2
d  ~ ( 2)
E3(2) E1 ( z ,  )  i d eff E3 ( z ,2 )E2* ( z ,  )e ikz
 dz n1 ( )c
d  ~ ( 2)
E2() E2 ( z,  )  i d eff E3 ( z ,2 )E1* ( z ,  )e ikz .
dz n2 ( )c
Normalizations
1
 1 
Ei ( z ,  )   ni (i )c 0  I t (0) i ( z ) eii ( z ) I t (0)  I1 ( z ,  )  I 2 ( z ,  )  I 3 ( z ,  ).
 2i 
~ 4 3 k
~  d ( 2) It   ~z s  ~   s  1 ( )  2 ( )  3 ( )
eff
 0c n1 ( )n2 ( )n3 (2 )
3 

Note that for these normalizations : i 1 i2 ( z )  1

3

Physically useful solutions are given in terms of the photon fluxes N(), i.e. photons/unit area

N i ( )  I t (0) i2 ( ) / 

Simple analytical solutions can only be given for the case Δs=0
 Type 2 SHG: Phase-Matched
Solutions for Type II SHG N 3 ( )  N 2 (0) sn 2 ( 1 (0) | m)
for s  0 and N1 (0)  N 2 (0) N 2 ( )  N 2 (0)  N 2 (0) sn 2 ( 1 (0) | m)

 (0) N1 ( )  N1 (0)  N 2 (0) sn 2 ( 1 (0) | m)

m 2
1(0) 2
1 (0)   22 (0)
define  
12 (0)   22 (0)
2 K [(1   ) /(1   )]   0.33
L period 
1  N1 ( )
K  elliptic function
as   0, L period  

N 3 ( )
1. No asymptotic final state
2. All intensities are periodic
with distance N2 ( )
3. Oscillation period depends
on input intensities