You are on page 1of 12

The effects of focusing in the threefrequency parametric upconverter

Shekhar Guha and Joel Falk



Citation: Journal of Applied Physics 51, 50 (1980); doi: 10.1063/1.327353
View online: http://dx.doi.org/10.1063/1.327353
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/51/1?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in
Frequency upconverted lasing of nanocrystal quantum dots in microbeads
Appl. Phys. Lett. 95, 183109 (2009); 10.1063/1.3242352

Secondary echoes in three-frequency nuclear quadrupole resonance of spin-1 nuclei
J. Chem. Phys. 118, 5071 (2003); 10.1063/1.1545442

Erratum: Irreducible tensors and selection rules for threefrequency absorption [J. Chem. Phys. 8 3, 2009 (1985)]
J. Chem. Phys. 84, 2901 (1986); 10.1063/1.450859

Irreducible tensors and selection rules for threefrequency absorption
J. Chem. Phys. 83, 2009 (1985); 10.1063/1.449343

THREEFREQUENCY HETERODYNE SYSTEM FOR ACQUISITION AND TRACKING OF RADAR AND
COMMUNICATIONS SIGNALS
Appl. Phys. Lett. 15, 420 (1969); 10.1063/1.1652885


[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
The effects of focusing in the three-frequency parametric upconverter
Shekhar Guha and Joel Falk
Department of Electrical Engineering, University of Pittsburgh, Pittsburgh, Pennsylvania 15261
(Received 22 January 1979; accepted for publication 11 May 1979)
This paper extends the theory of parametric upconversion to describe the mixing of two focused
optical beams with unequal confocal parameters. The results of this paper have greatest
applicability when focusing of one input beam is limited by physical constraints, e.g., damage.
The effects of phase matching, walk-off, and diffraction are treated herein. The analysis shows
that if the beams have fixed confocal parameters b
l
and b
2
, optimum mixing generally occurs with
b
2
=1=b
l
If both confocal parameters can be arbitrarily chosen, maximum efficiency occurs with
b
l
= b
2
= (crystal length)/2.84 in the absence of Poynting vector walk-off, and with b
l
=l=b
2
in the
presence of walk-off. For weak focusing upconversion efficiency is written in a closed form which
shows explicitly the effect of walk-off.
PACS numbers: 42.65.Bp, 42.60.He
I. INTRODUCTION
Optical parametric upconversion has received wide at-
tention as a means to sensitive ir detection. I Upconversion
mixes a strong laser pump of radian frequency WI with a
weak ir idler W
2
to produce a signal output of W3 = WI + W
2
.
The power generated at the sum frequency depends on the
input power and on focusing of both the pump and idler
electric fields.
In spite of the large amount of literature available on
upconversion, the choice of optimum beam sizes in the up-
converter remains uncertain. The basic question is this:
What should be the relative sizes of the pump and idler
beams? Should the beam spot size of the infrared wave be
chosen equal to the pump-beam size or does optimum focus-
ing dictate that the ir and the visible beams have equal confo-
cal parameters? The latter assumption implies that the visi-
ble pump laser be focused to an area / A.
z
times the ir
input. The resolution of this problem is of particular impor-
tance if practical considerations limit the minimum beam
size of either pump or ir beam. For example, in many upcon-
verter systems the pump-beam size must be kept large in
order to avoid crystal damage.
2
In these upconverters is it
beneficial to focus the ir more tightly than the pump?
It may be argued that simple physical intuition dictates
that equal beam sizes provide optimum focusing since that
configuration provides maximum overlap of the two beams.
On the other hand, there are two pieces of evidence that lead
one to believe that the choice of equal beam sizes is not
optimum.
(1) A quasiplane-wave analysis
3
of the upconverter pre-
dicts an efficiency proportional to + where WOI
and W
02
are the 1 Ie pump and idler Gaussian beam sizes. We
note that for a particular choice of W
OI
' conversion efficiency
is not maximized at W
02
= W
OI
but increases monotonically
as W
02
decreases.
(2) A second piece of evidence supporting the conjec-
ture that equal beam sizes do not represent optimum focus-
ing comes from the literature on optimum focusing for other
nonlinear optical processes. Boyd and Kleinman' justify the
fact that for resonant parametric downconversion and para-
metric oscillation optimum focusing is achieved with equal
confocal parameters and not with equal beam sizes. Experi-
mental investigations of frequency mixing have sometimes
implicitly assumed the validity of the equal confocal param-
eter assumption.
l
The analysis of focusing in any nonlinear optical pro-
cess is essentially the finding of a solution to the Maxwell
wave equation driven by a nonlinear polarization. For the
upconverSIOn process,
v2E _ n
2
a
2
E = 41T p(t) , (1)
c
2
at 2 c
2
at 2
where P and E are the polarization and electrical field at the
signal frequency. The polarization is produced by a mixing
of pump and idler fields. The size of these input beams as well
as the phase synchronism between signal polarization and
electric field determine the efficiency of the interaction.
We have employed two approaches to solving Eq. (1)
for arbitrary input beam sizes. The first approach involves
finding a Green's function for the upconversion system. A
Green's function G (r,r') describes the electric field produced
at point r due to an impulse of radiation at point r'. Once a
Green's function is found the signal wavelength electric field
due to an arbitrary polarization may be written
E(r) = f P(r')G(r,r')dr'. (2)
This approach has been used by Kleinman, Ashkin, and
Boyd to analyze focusing in second-harmonic generation
(SHG).'
The second technique analyzes the problem in the
Fourier or spatial frequency space. The three-dimensional
Fourier transform [P (K)] of the driving polarization [P (r)] is
calculated. The electric field, E (K), produced by each spatial
frequency is found and the total electric field is written
E(r) = fE(K)d3K.
(3)
Bjorkholm as well as Kleinman have used this Fourier ap-
proach to treat SHG.'7
Both approaches to the calculation of the signal electric
50 J. Appl. Phys. 51(1), January 1980 0021-8979/80/010050-11 $01.1 0 1980 American Institute of Physics 50
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
field give identical results for E (r). The Green's function
method of analysis for upconversion is treated in the main
portion of this paper. The Fourier technique is discussed in
the Appendix.
The power Pproduced by the electric field can be found
from the relation:
(4)
(n is refractive index, the integration extends over the entire
cross section). It is the behavior of this expression as a func-
tion of phase mismatch and focusing that forms the principal
results of this paper.
II. GREEN'S FUNCTION ANALYSIS
We consider type-I three-frequency upconversion in a
noncentrosymmetric crystal. The pump (lUI) and idler (lUz)
electric fields are ordinary waves and the signal (lU
2
) is ex-
traordinary. The two input beams have TEMoo modes. The
upconversion medium is a crystal oflength I and second-
order nonlinearity d. We include the effects of double refrac-
tion and phase mismatch in the crystal. Absorption at all
wavelengths is neglected. The field of the pump or idler beam
of power focused to a waist W
Oj
at the center of the crystal
is given by'
where
7
j
= 2(z and IEoj 12 = 16P
j
2 (6)
] nj CW
Oj
h
j
is the confocal parameter of the Gaussian beam and is
related to its spot size wO
j
by the relation h
j
= . The
refractive index at Wj is nj and k
j
= nj lU
j
/ C The pump and
the idler fields are designated by subscripts 1 and 2 in each of
their parameters. The nonlinear polarization they give rise to
is
(7)
We consider only one component of the nonlinear po-
larization and omit the vector notation.
This polarization drives the signal field E3 which satis-
fies the inhomogeneous wave equation
2 J2E3 47r a
2
\7E3 - --- = --P
NL
(8)
c
2
at
2
C
Z
at
2
The time dependence of P
NL
is given by exp - i(lU
l
+ lUz)t,
i.e., exp( - ilU
3
t). We cast this equation in an integral form,
i.e.,
lU
Z
J E
3
(x,y,z,t) = 7 PNL(x',y',z',t')G(r,t;r't')d
3
r'dt'. (9)
For a homogeneous isotropic medium the disturbance pro-
duced by a source point r' at time t ' proceeds as
I r' - r'l ( ') 0
--'---'- - t - t = ,
c/n
3
(10)
51 J. Appl. Phys., Vol. 51, No.1, January 1980
and the time-dependent Green's function is given by8
,,8[ir-r'I(c/n
3
)-I-(t-t')] (11)
G(r,t;r ,t ) = .
Ir - r'l
This gives
where the time dependence exp( - lU
J
t) has been sup-
pressed. For an anisotropic medium the disturbance pro-
duced by the source at r' does not propagate as a spherically
symmetric function as in Eq. (10). If the anisotropy is small,
the wave surface at any instant is R = constant where
R2=XZ+y2+Z2, (13)
X = x - pz - (x' - pz'),
Y=y-y', (14)
Z=z-z',
and where p is the Poynting-vector walk-off angle.
6
For a
medium of small anisotropy we may, therefore, replace
I r - r' I by R and obtain
(15)
where
(16)
and where r' = (x' ,y' ,z') denotes a source point inside the
crystal. The assumption that P
NL
is approximately a plane
wave is implicit here so that inside the crystal the energy flow
is along thep line (Fig. I). Let us denote the coordinates at
the exit surface of the crystal by x" ,y", and I. If the medium
beyond the exit surface is air (n = 1) and if we assume nor-
mal incidence upon the interface, the field in air just outside
the exit surface (at z = 1+ ) is
2n
E3(X",y",l + ) = --3-
E3
(x",y",l_).
n) + 1
(17)
x 'y 'z'

( z - z')
p
R = constant
(X_Xl)
FIG. I. Energy flow in the upconvener. Source points are indicated by
(x',y,z'). Observation points are (x,y,z). The wave surface is given by
R = canst. The Poynting vector at the signal wavelength is along the p line.
S. Guha and J. Falk 51
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
Outside the crystal P NL = 0 and the wave equation becomes
2 1 a2E3
\J E3 - -;- -- = 0, (18)
c- at 2
the solution to which is
E ( )
- W3 f f exp(i
w
3/3 Ic)
1 x,y,z - --
- 21Tic /3
XE
3
(x",y",1 + )dx" dy", (19)
where
/3
2
= (X_X")2+ (y_y")2+ (Z_/)2. (20)
Using Eqs. (14)-(17) this becomes
2rriw3 f 1
El (x,y,z) = -- P
NL
(r')Ge(r,r') d -,',
cn)
(21)
where
2n3
-----
n3 + 1
X f f dx" dy" G (r',r") /3 Ic)
(22)
denotes the external Green's function.
The approximations used to compute G
e
are (following
Ref. 6) as follows:
(1) P
NL
(r') is nearly a plane wave, so that its divergence
angle is small and only paraxial rays need be considered: If
R " is the value of R evaluated at x" ,y",1, we may write
" , 1 (x" _X')2+ (y" _y')2
R + - (23)
2 (/- z')
(2) Outside the crystal we consider the far field, i.e., z
very large, so that
I (X-X")2+ (y_y")2
/3c:::::(z -I) + - . (24)
2 (z -I)
Replacing R " and /3 in the denominators by their leading
terms and using
f
OC exp( - iax) exp( - ibx
2
) dx = lI2 exp( ia
2
) , (25)
- oc lb 4b
we obtain
G (r,r')= n3
w
3 exp{(iw3 Ic)[n3(l-z')+z-/]}
e n,+12rric l-z'+n3(z-/)
[(
iW3n3) {[x-x'-p(/-z')P+ (y_y')2}]
Xexp -- .
2c 1 - z' + n
1
(z - I)
(26)
This external Green's function G
e
is identical to that in
Ref. 6 [Eq. (3.45)]. It is only in the form of P
NL
that upcon-
version differs from SHG.
For upconversion P
NL
is given by
P (
') _ dEoJE02
NL r -
(1 + i7';)(1 + i7';)
(
X,2 + '2)
Xexp[i(k
l
+ k
2
)z'] exp - e'y B(z') ,
(27)
where
52 J. Appl. Phys., Vol. 51, No.1, January 1980
(28)
= (z' -
b
2
2'
III
-=-+-,
C' e;
(29)
e; = w61 (1 + i7';),
(30)
e; = W62 (1 + i7';),
B (z') = 1 for z' < I,
= 0 for z' > I. (31)
Equations (26) and (27) can be substituted into Eq. (21), and
the integrations over x' andy' can be evaluated, yielding the
field outside the crystal
E
3
(x,y,z)
= A2 i'dz'
o z - Z3
exp[(iw
3
n
1
Id '){[x - p(l- z')] 2 + y2}]
X - d' , (32)
where
Lik = kl + kl - k3 ,
d' = 2c[l-z' + n
3
(z -I)] + iw
3
n
3
C', (33)
2n 3 rrcw61 W62 k J k2
A2 = ----- dEol E02 exp(ii) ,
n, + 1 c
2
i(k
l
+k2)
W3
dJ= -(n31+z-/),
c
lib,
Z3 = - +--
- 2 2
= + +b
2
k
J ).
2 2 kJ + k2
(34)
(35)
(36)
The field at position (x,y,z) outside the crystal is a function of
x, y, and z. We square E3 and integrate over the whole x-y
plane to obtain the output power,
where
(40)
where
S. Guha and J_ Falk 52
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
and
I
SI= -,
hI
I
S2= -,
h
2
SIS2(1 + k)
S) = S2 + kSI '
SI + kS2
S4= ---
1 + k
SI + kS2
a=
(1 + k)SISz'
(ZI - ZD(Z2 - Z3*)
/= --------------
(zl- A I)(Z2- A f)+C
I
(41)
(42)
(43)
Evaluation ofEq. (37) shows that P3 is independent of x,y,z.
Thez independence mathematically follows from a cancella-
tion ofthe variable z during the x,y integration and is expect-
ed since (outside the crystal) the total power flow through
any z = const surface must be the same. The far-field ap-
proximation [Eq. (24)] is consequently unnecessary. If we do
not desire an explicit expression for E (x,y,z), we need not
propagate the field from the output surface of the crystal.
Our expression for P
3
reduces to that obtained by Boyd and
Kleinman
4
and Bjorkholm7 for SHG when we set kl = k2
and b I = h
2
The variation of signal power P
3
with focusing
is implicit in the function h =!I Is
4
This function is analo-
gous to the function h developed to describe focusing effects
in SHG.
4
III. COMPUTER RESULTS
The dependence of P
3
on ilk, B, k, SI' and Sz is con-
tained in the function h [Eq. (39)]. For a given experimental
situation the choice of A I , A
2
, and the upconversion crystal
andB=!pB/(k
l
+k2)]I12. The val-
ues of S I = 1/ k I and S 2 = II kl are determined by
the focusing of the input lasers. Mismatch ilkl is adjusted by
controlling the nonlinear crystal's temperature or angular
orientation.
We have numerically evaluated the expression for h,
performing the integrations specified in Eq. (40) on a digital
computer. For given values of SI' S2' k, and B, ilk is varied
to maximize h (and hence signal power P3)' The maximum
value of h is denoted by
(44)
If k = 1 and SI = S2 our upconversion analysis becomes a
treatment of SHG and the numerical evaluation of Eq. (37)
53 J. Appl. Phys., Vol. 51, No.1, January 1980
reduces to the focusing results of Boyd and Kleinman (Ref.
4, Fig. 2). For a general experimental situation (B and k fixed
and S I generally not equal to S 2) we can consider S I a param-
eter and compute the value of h
m
as a function of S2' Figures
2-9 show the variation of h
m
with focusing for various values
of SI' S2' and k. We note that the absolute maximum value
of h
m
occurs at B = 0, k = 1, and Sl = S2 = 2.84. This point
occurs with (ilkl)m = 3.25.
If both SI and Sl can be arbitrarily chosen the maxi-
mum value of h m can be determined for given values of k and
B. This value of h
m
is denoted by
h
mm
= h [(ilkl)m,Slm,S2m,k,B] .
Figure 10 shows the values of S I and S 2 that maximize h m In
the absence of Poynting vector walk-off(B = 0) or if
AI = A2 (k = 1), the conversion efficiency is maximized
with Sim = S2m' For all other values of Band k the maxi-
mum efficiency occurs with Sim =!=-S2m'
Figure 11 shows the value of h
mm
as a function of walk
off (B ) and k. For k = 1 these results are identical to those
given by Boyd and Kleinman. Boyd and Kleinman have
shown that if S I = S 2 the analytical description of upconver-
sion reduces to a mathematical treatment of second-har-
monic generation. In the SHG case they determine the value
of Sm = SI = S2 that maximizes second-harmonic output.
They suggest that upconversion efficiency is maximized by
this choice of focusing parameter for both the signal and the
idler!
The present analysis shows that the choice of
SI = S2 = Sm maximizes upconversion efficiency only if
k = 1 or B = O. The first case corresponds to second-har-
monic generation and the second is upconversion in the ab-
sence of walk-off. Figure 10 shows that in the presence of
walk-offupconversion efficiency is maximized with
Slm*"S2m'
The qualitative behavior of these curves may be viewed
1.0,--------------------,
0.1
k :: 1.0
'I:: 0.01
8::0
2
4
8
0.00 I '----'-_....I..-..LJL..L._..L----.l.---.l.-LL---"""--.-L-...l-I:::O
0.01 0.1 1.0 10.0
FIG. 2. The function h
m
(proportional to signal power) as a function of idler
focusing. 51 = O.o!. k = 1.0.
S. Guha and J. Falk 53
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
1.0
k =
1.0
, = 0.1
I
8=0
0.1 2
4
hrn
8
16
O. 00 I
0.01 0.1 1.0 10.0
FIG. 3. The function h
m
(proportional to signal power) as a function of idler
focusing. SI = 0. \, k = I.
as a competition among three effects; the enhancement of
nonlinear polarization due to focusing, the reduction due to
walk off, and phase matching. As the idler (liJ
2
) or pump (liJ I )
is more strongly focused, i.e., as 52 or 51 increases:
(I) The signal polarization increases due to the increase
in idler and pump electric fields (P NL = dE I E
2
). This leads
to an increase in the signal electric field.
(2) The signal electric field decreases because small
idler or pump beams imply a large angular spread in the
signal polarization. The upconversion process is limited by
10 ,----------=::::===::::----"1
0.1
k = 1.0
'I = 1.0
8=0
2
4
0.00\ L-----l._---'----'---'-'_--L_-L...-'-.J......I..._-'-_..L..--L-L.J
001 0.1 10 10.0
FIG. 4. The function h
m
(proportional to signal power) as a function of idler
focusing. S I = 1.0, k = I.
54 J. Appl. Phys., Vol. 51, No.1, January 1980
10
k = 1.0
8=0
'I = 10.0
2
0.1
4
hrn
8
16
o. a a I "'---lL_...L.--L..l.--'-----'L....----'-......L...L.L_--'-------'--'-..L.I
0.01 0.1 1.0 10.0
FI G. 5. The function h m (proportional to signal power) as a function of idler
focusing. SI = 10, k = I.
phase mismatch in its ability to use all angular components
of the signal polarization.
(3) The signal beam size decreases and limits on the
upconversion efficiency are imposed by the Poynting vector
walk-off at the signal wavelength. This walk-off effectively
limits the crystal length useful for upconversion.
We first consider the competition among these effects in
the absence ofwalk-6ff. For a given 51' (large or small) the
conversion efficiency h
m
increases with 52 until 52 is near
unity. This increase of h
m
is a consequence of the increased
10 ,-----------------
0.1
k = 4.0
= 0.01
8=0
2
4
8
16
a 00 I L---'-_-'---'-LL_-'------'----'L..L..L----'._--'----'--.LJ
0.01 0.1 1.0 100
FI G. 6. The function h m (proportional to signal power) as a function of idler
focusing. 5, = 0.0\, k = 4.
S. Guha and J. Falk 54
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
l.0
k = 4.0
= 0.1
B=O
2
4
O.t
8
h
m
16
a .001 L--..L_--L-...L.l.....l.-__
001 0.1 1.0 10.0

FIG. 7. The function h", (proportional to signal power) as a function of idler
focusing. 5r = 0.1, k = 4.
pump and idler field strength and hence signal polarization.
As t 2 is increased to about unity the role of phase matching
becomes important. When S2 = 1, the idler spreads to (2)112
times its waist size by the time it reaches the crystal's surface.
For S2 I upconversion efficiency drops because the upcon-
version process cannot phase match the extreme rays of the
idler beam. The precise peak of the upconversion efficiency
occurs when the gain in nonlinear polarization is just bal-
anced by the loss due to ineffective phase matching. Figures
2-9 show that the peak efficiency always occurs with S 2 of
order unity.
We consider the value of S2 necessary to maximize con-
1.0,-------------=-"""':---.
k = 4.0 8 = 0
= 1.0
0.1
o 00 I '--''-'-_--'---'-.L..L_--'--_'--"-I.....L-_'-----'---'-....L-'
0.01 0.1 1.0 10.0
FIG. 8. The function h
m
(proportional to signal power) as a function ofidler
focusing. 5, = 1.0, k = 4.
55 J. AppL Phys., Vol. 51. No. 1. January 1980
1.0
k =
4.0 B:O
= 100
0.1
2
4
h
m
8
16
0.01

0.01 0.1 1.0 10.0

FIG. 9. The function h", (proportional to signal power) as a function of idler
focusing. s! = 10. k =. 4.
version efficiency h",. Figures 2-9 show that the optimum
value of t2 increases as SJ increases toward unity. This fact
has a simple physical interpretation. A small value of S \ im-
plies that the pump acts essentially as a plane wave, i.e., as a
nearly collinear light source. A high value (> 1) of S 1 implies
a pump rich in off-angle radiation. (We have seen that the
peak value of h
m
always occurs withs
2
near unity, i.e., where
the idler beam is rich in divergent rays.)
lfboth S 1 and 52 are large (but less than an off-
angle idler ray can combine with an off-angle pump ray for
near-perfect phasematching. However, if 5 J is small, few off-
angle rays are available. Hence, if S 1 is large the phase
matching of a greater proportion of a diverging pump beam
is possible than if 51 is small. Consequently, the value of S2
that produces a maximum of h
m
increases with 51 until SI is
approximately unity. If S 1 and S2 are both much greater than
unity, the angular spreads of both beams are very large and
only a small proportion of the signal polarization can be
phase matched.
3.0r-------

w
"
c
o
E
;"J 1.0
o 10 2.0 3.0
B
4.0
k=20
10
5
5
10
20
5.0 60
FIG. 10. The values of 51m and 52", as a function of Band k.
S. Guha and J. Falk 55
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
1.2
E 08
1
E I
: 1
04l
k=20
L
10

o 10 20 3.0 40 5.0 60
B
In the presence of walk -off CB :f 0) the qualitative expla-
nation given above becomes slightly more complicated. Here
again the competition between increased nonlinear polariza-
tion due to focusing and decreased efficiency due to phase
mismatch occurs. However, the peaks of the individual
curves shown in Figs. 2-9 are also greatly influenced by the
finite useful crystal length determined by walk-off. In most
cases, walk-off rather than phase-matching limits the value
of 52 that maximizes efficiency (h
m
). From Figs. 2-9 we see
that the values of 52 required to maximize efficiency de-
crease with increasing walk-off (B), and are always less than
those useful without walk-off. The fact that phase matching
does not limit the useful maximum values of 51 or 52 can be
seen from Figs. 15-20 which show the rather minimal effect
that wide deviations fromJk = 0 have upon the efficiency of
the upconversion process when walk-off is large.
Figures 12-14 show the computation of h
m
for three
upconverters that have been previously experimentally in-
10.0
k = 6.6
, = 0.72
I
1.0
8=0
h
m
0.01
0.1 1.0 10.0 100

FIG. 12. Focusing in a LiNb0
3
-argon laser upconverter. This upconverter
can be phased matched without walk off. The pump confocal parameter b is
6.94 cm, / = 5 cm, 5, = 0.72. This corresponds to a pump-beam size of 50
j.lm. (Ref. 13) The quasiplane wave solution for these values is shown for
comparison (Ref. 3) A, = 0.SI45j.lm, A, = 3.39 pm.
56 J. AppL Phys., Vol. 51. NO.1. January 1980
10.0.------------------.
1.0
k = 17.7
, = 0.3
I
o. I _ _L_ _ __'_ _ _'__...l.._L_ _ __L_ _ __'__.L_.L_J
0.1 1.0

100
FIG. 13. Focusing in a AsGaS, upconverter (Ref. 2). The pump focusing
parameter 5, = 0.3, B = 0, A, = 0.598 pm, A, = 10.6 pm.
vestigated. 2, 11.13 Note that in general the maximum value of
h
m
occurs neither when 51 = 52 nor when W
OI
= W
02
(52 = kS1)' In the 90 phase-matched upconverter treated
in Fig. 12 the results of a quasi plane wave analysis
3
are
shown for comparison.
The variation of h with phase mismatch 11k! can be
computed from Eq. (40). Figures 15-20 show this variation
for strongly and weakly focused upconverters with and with-
out walk-off. In the absence of walk-off (B = 0), and with
weak focusing <51,S2<1), these curves follow a
1.0
k = 4.8
, = 0.01
I
B = 22.3
0.1
h
m
0.01
O. 00 I
0.01 0.1 1.0 10.0
FIG. 14. Focusing in a lithium iodate, ruby-laser upconverter (Ref. II). The
pump focusing parameter is;-, = 0.0 I. Phase matching without walk off is
impossible (B = 22.3, / = 5 em). A, = 0.6943j.lm, A, = 3.39j.lm.
S. Guha and J. Falk 56
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
I
12
!
r
NO "t
-c I
T
I
/
I,
J
o
___ --'--- __ ._
-32.0 - 16.0 0.0 160 32.0
6k
FIG. 15. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = 0. The curve with lowest peak h corre-
sponds to B = 8. 5. = om 51 = 0.01 B = 0,2,4,8. Figures 15 and 16
<5. = 52) are valid for any value of k (Ref. 4).
OBOf-
f\
i I "
0.60
L
I I
L:. I I \1
Q40- ! \
l

-32.0 - 16.0 0.0 16.0 32.0
6kR
FIG. 16. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = 0. The curve with lowest peak h corre-
sponds to B = 8. 5. = 1.0 52 = LO B = 0,2,4,8. Figures 15 and 16
<5. = 52) are valid for any value of k (Ref. 4).
24r
r-
I
16L
N
0
-c
O.B
01
0.0
6kP
16.0
-------'--
32.0
FIG. 17. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = O. The curve with lowest peak h corre-
sponds to B = 8. k = 1.0 51 = om 52 = 1.0 B = 0,2,4,8.
57 J. AppL Phys., Vol. 51, No.1, January 1980
1.2
32.0
FIG. 18. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = O. The curve with lowest peak h corre-
sponds to B = 8. k = LO 5. = 0.01 5, = 10 B = 0,2,4,8.
i
I

O.OB
Ie '
I
\\
006
-c
OOT
002r
o Ooc.
II
\
,J \\\<"
.. ===-
-320 - 16.0 0.0 16.0
6k2
FIG. 19. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = O. The curve with lowest peak h corre-
sponds to B = 8. k = 10 51 = om 51 = LO B = 0,2,4,8.
I
4 Bl
:-
I
3.2
N
0
-c
1.6
\.
.. -
O -
-320 -160 00 160 32.0
6k
FIG. 20. The effect of phase matching on upconversion. The curve with
highest peak h corresponds to B = O. The curve with lowest peak h corre-
sponds to B = 8. k = 10 51 = 0.01 52 = 10 B = 0,2,4,8.
S. Guha and J. Falk 57
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
1.0
Cl.
E
E 0.5
Cl
o
__
o 2 4 6 8 10 12
P
FIG. 21. The function I,opl>m"m'(P) [Eq. (54)]. The function, valid for loose
focusing, describes the decrease in conversion efficiency due to walk-off.
p=2B(kff, +ff,)/(l+k).
[(sin!Llkl )/(!Llkl)] 2 variation. For very strong focusing (51
or t 2 ? I 0) and/or in the presence of walk -off (B=I=O) the
upconverter's tolerance for phase mismatch (Llkl) is much
greater than with t1 and t2 small and B = 0. In the case of
strong focusing this increase comes about because of the
availability of off-angle polarization rays that can be phase
matched although Llkl, the collinear phase mismatch, is far
from zero. When walk-off is large the substantial Llkl toler-
ance is due to the fact that only a small portion of the crystal
length (the aperture length, la) contributes in a coherent way
to the build up of the signal's electric field. The phase mis-
match developed over this length does limit the interaction.
However, since implies a large Llkl tolerance
as illustrated in Figs. 15-20.
The large Llkl tolerance for B O has important impli-
cations for the operation of upconverters. The upconverter's
angular as well as its spectral acceptance is determined by
this quantity. The large upconversion bandwidths reported
by Gurski for the Lil0
3
upconverter are at least in part due
to this large Llkl tolerance caused by walk-off. 12
We note that Figs. IS and 16 are independent of k. This
is consistent with Ref. 4 where Boyd and Kleinman show
that if tl = t2 then h is independent of k.
A. Limiting case: weak focusing with double refraction
It is well known that for loose focusing in the absence of
double refraction, upconversion efficiency varies as
l!(w61 + W62 ).3 However, heretofore no similar expression
has been derived for similar focusing in the presence of dou-
ble refraction.
We consider the quasi plane wave case, i.e., both beams
very large (51 ,t2 small) but we do not neglect double refrac-
tion. The starting point for our analysis is the integral equa-
tions for the signal electric field, Eqs. (19), (24), and (27). We
consider both input fields to behave as collimated Gaussian
beams, i.e.,
58 J. Appl. Phys., Vol. 51, No.1, January 1980
so that
(45)
Then,
E, = A I exp(i )2CC1T (' exp(iLlkz') dz
Jo d'
X exp[ ){[X - p(1 - Z/)]2 + y2}] , (46)
where
liJ
3
dE
01
E
02
2n3
c
2
n3 + I
(47)
Equation (46) can be further simplified by noting that in the
far field (z>/), d'c:=2cn
3
z - CliJcn3C is independent ofz'.
This gives
P3 = J J I E3 12 dx dy
where
[= t t dz, dZ
2
exp[iLlkl(z, -Z2)]
Jo Jo
X [ - p2(ZI - Z2)2] ,
p= pi
V2w,
and W, is the signal beam waist defined by
I I I
-2 =-2-+-2-'
W, W
OI
W
02
For no walk off, p goes to zero and [becomes
sin2QLlkl)
(1kl)2
When walk-off is present
(1T)II2{
[= -- exp( - q2)[(p + iq) erf(p + iq)
2p2
(49)
(50)
(51)
+ (p - iq) erf(p - iq) - 2q erf(q)] + _1_ (52)
(1T)I12
X [exp( - p2 - iLlkl) + exp( - p2 + iLlkl) - 2]} ,
where q = Llkl /2p.
We have noted earlier that for loose focusing the opti-
mum value of Llkl is close to zero. Making use of this fact
here we obtain
[(optimum) (P) = [(1T)1I2p erfp + exp( - p2) - 1] .
p2
(53)
This is identical to the function G (t,O) defined in Ref. 4 [Eq.
(2.40)]. For nonzero walk-off the conversion efficiency is
reduced by this factor. The variation of [(optimum) (p) withp
is shown in Fig. 21.
Figure 21 can be used to define an aperture length la due
to double refraction. If we allow upconversion efficiency to
be reduced 50% by walk-off this means [ (p) = 0.5. This oc-
S. Guha and J. Falk 58
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
curs at p = 2.7S which gives
fa = 3.89w/ p.
ACKNOWLEDGMENT
The authors thank Y.C. See for many helpful discus-
sions of this work and for aid in the computer analysis. The
work reported in this paper was supported by NASA God-
dard Space Flight Center.
APPENDIX: CALCULATION OF SIGNAL ELECTRIC
FIELD BY FOURIER ANALYSIS
In the main body of this paper we used a Green's func-
tion analysis to derive an expression [Eq. (32)] for the upcon-
verter's signal field as a function of several parameters.
The field produced by the nonlinear polarization can
also be found by an alternative method. The polarization
P (r) is broken up into its Fourier components
P(K)= -1-fp(r)eXp(-K.r)d3r. (AI)
(2Jr)3
Each Fourier component radiates into an electric field.
Kleinman
6
has studied this problem in connection with sec-
ond-harmonic generation. He shows that the electric field
produced at upconverted frequency by each of these plane
waves is given by
(A2)
where
g(x) = f exp( - xp) dp, (A3)
and the mismatch .JK is a function of K and is
K2 +K2
.JK=K
z
-(kl +k2 -.Jk)+pKx + x y. (A4)
2(k
l
+ k
2
)
The total electric field produced is then obtained by integrat-
ing over all K
E
3
(x,y,z) = f E
K
(r)d
3
K.
(AS)
PK , the Fourier amplitude distribution of the nonlinear
polarization, is found directly from Eq. (A I) and (27), if in
(27) we drop the slab function B (z') following the argument
in Ref. 6.
JrdEol E02 kl k2
P K = ----:.--=-
(2Jr)
3
(k
l
+ k
2
)
Xexp[iaz
3
+b(b
31
+b
32
)]Jo(2s)
for K
z
< Kc , (A6)
P K = 0 otherwise,
where
59
a = kl + k2 - K
z
,
K2 +K2
b = x y
2(k
l
+ k
2
) ,
b
3
- b
l
b ----
31 - 2 '
J. Appl. Phys., Vol. 51, No.1. January 1980
(A7)
(A8)
Substituting Eqs. (A6) and (A2) into Eq. (AS) we find the
electric field inside the crystal to be
Ek)=B' f dp
Xexp[ip(k
1
+ k2 - .JK)z]I
K
/
K
/
Ko
' (A9)
where
B
' = /z 2JrUJ3 rdEo1Eo2
---------exp[i(k
1
+ k2)Z3] '
n,c (2Jr)3 (k
1
+ k
2
)
(AW)
(All)
and I K ., I K ,., and I
K
, are integrals over K
x
' Ky' and K
z
,
respectively. These integrals are
I K , = ( . . (K;+K;) dKz
Jk, + k, - K,.> -'--"--,-"-
2(k, + k,)
xexp{ - iK
z
[Z3 - z(l - p)]}J
o
(2s), (AI3)
The integrals can be evaluated to yield
a(kl + k
2
) - DK}

a
where
a b
31
b
32
D = ----.:-- + -----
2(k
l
+ k
2
) 2(k
1
+ k2)a '
a= -i[Z3 -z(1-p)],
Hence,
E3(r)
(A 14)
(AIS)
(AI6)
(AI8)
=B'Jr t dp exp[a(k1 +k2)] exp[ip(k
1
+k2 -.Jk)z]
Jo a
S. Guha and J. Falk 59
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39
(x _ ppZ)2 + y2
Xexp- (AI9)
4c
where
(A20)
We are interested in the electric field at the output surface on
the crystal, z = l. If we perform the substitutionp = 1 - z'/l
we can easily show
a = - i(z, - z')
(A21)
where C' is given by
C' 2 + 2 (A22)
wol(l +ir;) w
02
(1 +ir;)
Ifwe assume kl + k2 ~ k , and note Eq. (33), then
. 2e(l - z') - iW3 n3 C'
4c = I -------=--=---
and we obtain
id'
(A23)
4B 'W3 n31T
E3(r) = l exp[i(k
l
+k2)(l-z3)]exp(-iJ.kl)J
(A24)
60 J. AppL Phys., Vol. 51, No.1, January 1980
J = i'dZ' exp(iJ. ~ )
o Z, -z
exp [(iw
3
n3 /d '){[x - p(/ - Z')]2 + y2}]
X - - - - - - - - - - - - - - ~ - - - - - - - - ~ ~
d'
(A25)
Equation (A24) calculated by a Fourier analysis is the
electric field at the output surface of the upconversion crys-
tal. This expression is identical to that [Eq. (32)] derived
from a Green's function approach.
'D.k Kleinman and G.D. Boyd. J. Appl. Phys. 40, 546 (1969).
'w. Jantz and P. Koidl, Appl. Phys. Lett. 31, 99 (1977).
'G.D. Boyd and A. Ashkin, Phys. Rev. 146, 187 (1966).
4G.D. Boyd and D.A. Kleinman, J. Appl. Phys. 39, 3597 (1968).
'G.D. Boyd, Thomas J. Bridges, and E. Gardner Burkhardt, IEEE J.
Quantum Electron. QE-4, 515 (1968).
"D.A. Kleinman, A. Ashkin, and G.D. Boyd, Phys. Rev. 145, 338 (1966) .
'J.E. Bjorkholm, Phys. Rev. 142, 126 (1966).
'J.D. Jackson Classical Electrodynamics, 2nd ed., (Wiley, New York, 1975,
p.224.
"M. Born and E. Wolf, Principles o/Optics, 3rd ed. (Pergamon, Oxford,
1965).
"'M.M. Abbas, T. Kostiuk, K.W. Ogilvie, Appl. Opt. 15,961 (1976).
liT. Gurski, Appl. Phys. Lett. 23, 273 (1973).
"T.R. Gurski, H.W. Epps, and S.P. Maran, Appl. Opt. 17, 1238 (1978).
"J. Falk and Y.c. See, Appl. Phys. Lett. 32, 100 (1978).
S. Guha and J. Falk 60
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
164.125.41.50 On: Sun, 02 Nov 2014 10:55:39