Chapter 9

Second Harmonic Generation

To mathematically model and analyze second harmonic generation. To calculate the conversion coefficient To qualitatively summarize the method of designing second harmonic generation.

The optical polarization of dielectric crystals is due mostly to the outer, loosely bound valence electrons that are displaced by the optical field. Denoting the electron deviation from the equilibrium position by x and the density of electrons by N, the polarization p is given by p (t ) = Nex (t ) In symmetric crystals the potential energy of an electron must reflect the crystal symmetry, so that, using a one-dimensional analog, it can be written as m 2 2 m 4 V (t ) = 0 x + Bx + .. (9.1) 2 4 2 Where 0 and B are constants and m is the electron mass. Because of symmetry V(x) contains only even power of x, so V(-x)=V(x). The restoring force on an electron is V 2 = m 0 x mBx 3 + ... (9.2) F = x And is zero at the equilibrium position x=0. The linear polarization of the crystals in which the polarization is proportional to the electric field is accounted by the first term in Equation (9.1). To see this, consider a low frequency electric field E(t) that is, a field whose Fourier components are at frequencies small compared to 0. The excursion x(t) caused by this fields found by equating the total force on the electron to zero 2 eE (t ) m 0 x(t ) = 0 So that e x(t ) = E (t ) (9.3) 2 m 0 This resulting in an asymmetric crystal in which the condition V(x)=V(-x) is no longer fulfilled, the potential function can contain odd powers of x and thus m 2 2 m V (t ) = 0 x + Dx 3 + .. (9.4) 2 3 Which corresponds to a restoring force on the electron V ( x) 2 = ( m 0 x + mDx 2 + ...) (9.5) F = x

An examination of (9.5) reveals that a positive excursion (x>0) results in a large restoring force, assuming D > 0, than does the same excursion in the opposite direction. It follows immediately that if the electron force on the electron is positive (E<0), the induced polarization is smaller than when the field direction is reversed. This situation is depicted in Figure 9.1.

Next consider an alternating electric field at an (optical) frequency to the crystal. In a linear crystal the induced polarization will be proportional, at any moment, to the field, resulting in a polarization oscillating at as shown in Figure 9-2(a). In a nonlinear crystal we can use Figure 9-1(b) to obtain the induced polarization corresponding to a given field and then plot it (vertically) as in Figure 9.2(b). The result is a polarization wave in which the stiffer restoring force at x>0 results in positive peaks (b), which are smaller than the negative ones (b). A Fourier analysis of the nonlinear polarization wave in Figure 9-2(b) shows that it contains the second harmonic of as well as an average (dc) term. The average, fundamental, and second-harmonic components are plotted in Figure 9.3. To relate the nonlinear polarization formally to the inducing field, we use Equation (9.5) for the restoring force and take the driving electric field as E() cos t. The equation of motion of the electron F=mx is then d 2 x(t ) dx(t ) eE ( ) jt 2 2 + + x ( t ) + Dx ( t ) = (e + e jt ) (9.6) 0 dt 2 dt 2 where we account for the losses by a frictional force - mx. An inspection of (9.6) shows that the term Dx2 gives rise to a component oscillating at 2 , so we assume the solution for x(t) in the form 1 (q1e jt + q 2 e j 2t + c.c.) (9.7) 2 where c.c stands for complex conjugate. Substituting the last expression into (9.6) gives j 2 (q1e jt + c.c.) 2 2 (q 2 e j 2t + c.c) + (q1e jt c.c.) 2 2 x(t ) =

2 (9.8) * q q D 2 2 j t 2 4 jt e + (q1 e + q2 + 1 1 + 2q1 q 2 e 3 jt 4 2 * q q eE ( ) jt * j t e + 2q1 q 2 + 2 2 + c.c. = (e + c.c.) 2 2m If (9.8) is to be valid for all times t, the coefficients of ejt and ej2t on both sides of the equation must be equal. Equating first the coefficients of ejt, assuming that 2 |Dq2| << [( [( 0 2 ) 2 + 2 2 ]1 / 2 , gives

+ j (q 2 e j 2t c.c) +

02

(q1e jt + q 2 e j 2t + c.c.)

1 eE ( ) (9.9) q1 = 2 m ( 0 2 ) + j The polarization at is related to the electronic deviation at by

 Ne (q1e jt + c.c.) = 0 [ ( ) E ( ) e jt + c.c.] (9.10) 2 2 Where () is thus the linear susceptibility. By using (9.9) in (9.10) and solving for (), we obtain
P ( ) (t ) =

Ne 2 ( ) = (9.11) 2 m 0 [( 0 2 ) + j ] We now proceed to solve for the amplitude q2 of the electric motion at 2. Equating the coefficients of e2jt on both sides Of (9.8) leads to 1 2 q 2 (4 2 + 0 + 2 j ) = Dq12 2 And, after substituting the solution (9.9) for q1, we obtain De 2 ( E ( ) ) 2 (9.12) q2 = 2 2 2 ) + j ] 2 ( 0 4 2 + 2 j ) 2m 2 [( 0 In a manner similar to (9.10), the nonlinear polarization at 2 is

1 Ne (q 2 e 2 jt + c.c.) = {d ( 2 ) [ E ( ) ] 2 e 2 jt + c.c.} (9.13) 2 2 The second of equation (9.13) defines the normal optical coefficient d(2). If we denote the complex amplitude of the polarization as P(2) we have, from (9.13), P ( 2 ) (t ) = P ( 2 ) (t ) = and 1 ( 2 ) 2 j t (P e + c.c.) 2 (9.14)

P ( 2 ) (t ) = d ( 2 ) E ( ) E ( ) That is, d(2) is the ratio of the (complex) amplitude of the polarization at 2 to the square of the fundamental amplitude. Substituting (9.12) for q2 in (9.13), then solving for d(2) , results in d ( 2 ) = DNe 3 2 2 2 ) + j ] 2 ( 0 4 2 + 2 j ) 2m 2 [( 0 (9.15)

Using (9.11) we can rewrite (9.15) as

3 (9.16) mD[ ( ) ] 2 ( 2 ) 0 2N 2e3 Equation (9.16) is important since it relates the nonlinear optical coefficient d to the linear optical susceptibilities and to the inharmonic coefficient D. Estimates based on this relation are quite successful in predicting the size of the coefficient d in a large variety of crystals. Relation (9.14) is scalar. In actual crystals we must consider the symmetry so that the second harmonic polarization along, say, the x direction, is related to the electric field at by a third rank tensor dijk.

d ( 2 ) =

( 2 ) ( ) ( ) ( 2 ) ( ) ( ) ( 2 ) ( ) ( ) Px( 2 ) = d xxx E x E x + d xyy E y E y + d xzz Ez Ez ( 2 ) ( ) ( ) ( 2 ) ( ) ( ) ( 2 ) ( ) ( ) + 2d xzy E z E y + 2d xzx E z E x + 2d xxy Ex Ey

(9.17)

Crystals are usually divided into two groups, depending on whether the crystal structure remains unchanged upon inversion or not. Crystals belonging to the first group are called centrosymmetric, whereas crystals of the second are called noncentrosymmetric. In crystals possessing an inversion symmetry, all the nonlinear ( 2 ) must be zero and there is no second-harmonic generation. optical coefficients d ijk

The conversion efficiency from to 2 is very important parameter in the second harmonic generation: P2 3 / 2 2 d 2 l 2 sin 2 [k / 2] P = 2( ) (9.18) SHG = P 0 ( k / 2 ) 2 A n3 where P /A is the intensity of the fundamental beam (A-the cross section of the beam);

= is the wave impedance in ; l the thickness of the crystal; d-the 0

nonlinear optical coefficient; - the frequency of the fundamental frequency in Hz=1/s. The main peak of second harmonic generation occurs at k=0 which means that 2 2 k = k ( 2 ) 2k ( ) = [n n ] = 0 (9.19) c Since sinc(x=0)/x=0=1, Equation (9.18) becomes

P2 2 d 2 l 2 P = 2( ) 3 / 2 (9.20) P 0 A n3 However, (9.18) and (9.20) hold only when input power P of fundamental wave is assumed to be undepleted (the low conversion efficiency approximation). Generally speaking, the conversion efficiency from to 2 should be expressed as:

max SHG =

SHG =

P2 2 d 2 l 2 sin( kl / 2) P 1 / 2 = tanh 2 [2( ) 3 / 2 ( ) ] (9.21) 0 P kl / 2 A n3 e x ex ) e x + e x

(note: tanh 2 x =

The crystal length is called coherence length, lc, when the second-harmonic power becomes maximum value. According to (9.18) and (9.19), a prerequisite for efficient second-harmonic generation is that k=0 or, using (9.19), k ( 2 ) = 2k ( ) (9.22) If k0, the second-harmonic power generated at some plane, say z1, having propagated to some other plane (z2). This results in the interference described by the factor sin 2 (k / 2) . (9.23) ( k / 2 ) 2 The main peak and the first zero of this spatial interference pattern are separated by the so-called coherence length 2 2 lc = = ( 2 ) = (9.24) ( ) 2 k k 2k 2 | n n | where is the free-space wavelength of the fundamental beam. If we take a typical value of =1m and n2-n=10-2, we get lc50m The coherence length lc is thus a measure of the maximum crystal length that is useful in producing the second harmonic power. Under ordinary circumstances it may be no larger than 10-2 cm. this is because the index of refraction n normally increases with so k is given by (9.19). The index of refraction ne and no for the crystal KH2PO4 at different wavelength is listed in Table 9.2

If ne 2< no , there exists an angle m at which ne 2()= no ,so if the fundamental beam (at ) is launched along m as an ordinary ray, the second-harmonic beam will be generated along the same direction as an ordinary ray. The situation is illustrated by Figure (9.4). The anglem is determined by the intersection between the sphere (shown as a circle in the figure) corresponding to the index surface of the ordinary beam at , and the index surface of the extraordinary ray ne 2(). The angle m , which defines a cone, for negative uniaxial crystal that is, crystals in which ne < no is that satisfying ne 2()=no or, m is satisfying sin 2 m =
2 2 2 (no ) (no ) 2 2 2 2 (ne ) (no )

(9.25)

To get (9.25), we need to know that (see Figure 9.4) 1 cos 2 sin 2 = + (9.26) 2 ne2 ( ) no ne2 or 1 1 cos 2 sin 2 = = + 2 2 2 2 2 no ( ) ne2 ( ) (no ) (ne ) (9.27)

Figure 9.7 and 9.8 show how can we get a second harmonic generation by focus lenses for increasing the intensity of fundamental lasing.