LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 510.84 1^6r no l80-l70 cop. 4 The person charging this material is re- sponsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for disciplinary action and may result in dismissal from the University. UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN NOV 15 M 1973 L161 — O-1096 Digitized by the Internet Archive in 2013 http://archive.org/details/novelelectroopti170ujhe DIGITAL COMPUTER LABORATORY 10.170 :op.3 UNIVERSITY OF ILLINOIS URBANA, ILLINOIS REPORT NO. 170 NOVEL ELECTRO- OPTICAL LIGHT MODULATORS by Gab or Kalman Ujhelyi October 26, 196k (This work was submitted in partial fulfillment of the requirements for the Degree of Doctor of Philosophy in Electrical Engineering, September, 1964, and was supported in part by the Office of Naval Research under contract Nonr-l834(l5).) DIGITAL COMPUTER LABORATORY UNIVERSITY OF ILLINOIS URBANA, ILLINOIS REPORT NO. 170 NOVEL ELECTRO-OPTICAL LIGHT MODULATORS by Gabor Kalman Ujhelyi October 26, I96U (This work was submitted in partial fulfillment of the requirements for the Degree of Doctor of Philosophy in Electrical Engineering, September, 196^, and was supported in part by the Office of Naval Research under contract Nonr-l83Ml5)-) ACKNOWLEDGEMENT The author wishes to express his appreciation to Professor W, J. Poppelbaum for his counsel and guidance in the preparation of this dissertation. The author is also deeply indebted to his colleague Dr. Sergio T. Ribeiro for his friendship, encouragement, many helpful ideas and for his collaboration in the experimental part of this work. Sincere thanks are also due to Professors D. Holshouser and 0. Gaddy who provided valuable advice, space and equipment for the experiments. 111 TABLE OF CONTENTS CHAPTER 1 CHAPTER 2 INTRODUCTION THE THEORY OF ELECTRICALLY INDUCED BIREFRINGENCE CHAPTER 3 CHAPTER 4 CHAPTER 5, CHAPTER 6. BIBLIOGRAPHY 2.1 2.2 2.3 2.4 Birefringence in anisotropic materials- The derivation of the electro-optic effects. The index ellipsoid of some important electro- optic materials. 2.3.1. Linear electro-optic effect in uniaxial materials. 2.3.2. Linear electro-optic effect in cubic materials. 2.3.3. The quadratic electro-optic effect. Material properties of some electro-optic materials INTENSITY MODULATION BY CONTROLLED PARTIAL REFLECTIONS 3.1. The Fresnel relations for the reflected amplitudes. 3.2. The dependence of the reflected intensity on the index of refraction n^ . 2 3.3. The modulation M as a function of the modulating electric field E . 3.4. The physical realization of the intensity modulator. 3.5. Beam intensity modulation. Effects of eollimation and diffraction. 3.6. Velocity matching condition for high speed operation , 3.7. Dimensional tolerance requirements. 3.8. Optimization of modulator dimensions. 3.9. The choice of the operating temperature. 3.10. Modulating bandwidth limitations. BEAM DIRECTION MODULATION BY CONTROLLED INH0M0GEKEIT7 4.1. The theory of wave propagation in an inhomogeneous medium . 4.2. Beam deflection by means of an electro-optical device. 4.2.1. Beam deflection by the linear electro-optic effect . 4.2.2. Beam deflection using the quadratic electro- optic effect. 4.3. Beam distortion effects. EXPERIMENTAL RESULTS 5.1. Experiments with the light intensity modulators. 5.2. Experiments with the beam bending device. CONCLUSIONS 4 4 17 21 21 22 22 24 25 25 29 34 37 40 44 47 50 52 54 57 57 62 63 69 70 77 78 80 82 85 iv NOVEL ELECTRO-OPTICAL LIGHT MODULATORS Gabor Kalman Ujhelyi, Ph.D. Department of Electrical Engineering University of Illinois, 1964 Since the availability of a coherent, well collimated and powerful light beam, there is a great demand for efficient, broadband light modulators. To the already well developed family of Kerr-cell type devices we propose to add two novel electro-optical devices: a direct intensity modulator and a light beam deflecting device. To provide a firm basis for the analysis of these devices, the theoretical background of the electro-optically induced birefringence is examined in detail . Starting from the anisotropic material equations and Maxwell's differential equations, an analytical derivation of this birefringence is given. Further- more it is shown how the electro-optical effect influences the material equations hence its effect on the birefringence is derived. The experimentally determined electro-optic effect is tabulated for the most important materials . The new intensity modulator operates on the principle of electro-optically controlled partial reflections. The light beam to be modulated traverses a transparent isotropic material of index of refraction n = n and it is incident on the electro-optic material with n = n (n > n ) . The reflected amplitude (hence the intensity'.) depends on the index of refraction n , which in turn is controlled by the modulating electric field through the electro-optic effect . A detailed study of the theoretical and practical aspects of this intensity modulator reveals that its performance is in many ways comparable to the most sophisticated of the Kerr-cell type devices. A systematic and quantitative procedure is presented for the optimized design of a broadband intensity vi modulator. The modulating frequency of the device is fundamentally limited only by the dispersion of material parameters. The beam deflecting device uses the fact that a variation of the index of refraction n causes a light beam to change direction of propagation. The desired variation of n (hence the deflection of the beam) can be achieved by placing an electro-optical material in a suitable electric field distribution. This field distribution is determined for the case of minimal beam distortion for linear and quadratic electro-optic materials. The experimental results obtained with three devices support the theoretical predictions. Intensity modulation up to 50% and a deflection of ,003 radian were observed in the laboratory using nitrobenzene as the electro- optic material . CHAPTER 1 INTRODUCTION The modulation of a light beam is by no means a new idea: navies used light signals as means of communications ever since ancient times. The development of the intensity modulator can be traced through the progress in the measuring techniques used for the experimental determination of the velocity of light. The rotating wheel of Fizeau (1849), the rotating mirror of Foucault (1860) and the electro-optical shutter of Kerr (1874), Karolus and Mittelstaedt (1928) represent the major advances in this develop- ment . The use of these devices, however, was restricted mainly to the field of experimental physics and to high speed photography in particular. (2) The invention of the new light source, the laser , increased the importance of the light modulators manifold. This new device provides a coherent, highly stable, powerful and well collimated beam of electro- 14 magnetic radiation with a frequency of the order of 10 cps . Much work has been done in the recent years trying to utilize the information carrying capabilities of this beam. A number of modulating systems were designed with a modulating bandwidth of several kilomegacycles . However, all of these electro-optical modulators are basically phase modulators in that the modulating fields cause a difference between the phase velocities of the two component waves in the medium and hence influence the polarization of the output beam. The phase modulation may be converted to intensity modulation by passing the output beam through a polarization analyzer. (These are the Kerr-cell type modulators.) The purpose of this dissertation is to analyze the operating character- istics of two novel electro-optical light modulators: 1.) An amplitude or intensity modulator which does not operate on the Kerr principle described above, but obtains intensity modulation directly from controlled partial reflections on the interface between an isotropic and an electro-optic material. 2.) A device which deflects a light beam by the inhomogeneity induced by a modulating electric field through the electro-optic effect . Our intensity modulator compares favorably with the Kerr-cell type devices on several features, due mainly to the fact that the modulation is accomplished at the boundary surface rather than in the volume of the electro-optic material. Some of these points are: 1.) The modulated light beam does not travel in the usually lossy electro-optic material, but stays within the nearly loss- leys isotropic medium. 2.) A very thin layer (several wavelengths) of the electro- optic material is sufficient for the modulation process and consequently the modulating power i b Ln this material is \ tly reduced. 3.) The device does not have a limiting modulating frequency correj'r ling to the cutoff frequency of ;1 i Kerr-cell type devices, (occurring when the microwav n dulating field varies appreciably across the cross section of the optical beam.) These and other points will be ei d the text. 3 The organization of the remainder of the dissertation is as follows: in Chapter 2 the theory of the electrically induced birefringence is derived from basic principles and the experimentally measured electro-optical effects are given for the most important materials. Chapter 3 presents a detailed analysis of the intensity modulator and provides a systematic and quantitative procedure for the design of an optimal modulator. Chapter 4 analyses the operational characteristics of the beam deflector device. Chapter 5 contains the results of experiments designed to test the performance of the two devices. In Chapter 6 we conclude with a tabulation of the performance expected from modulators made of presently available materials. CHAPTER 2 THE THEORY OF ELECTRICALLY INDUCED BIREFRINGENCE 2.1, Birefringence in Anisotropic Materials . The theory of electromagnetic waves is based on two quite separate founda- tions: Maxwell's differential equations provide the first, and the material equations give the second. Whereas in isotropic media the vector quantities in the material equations are related to each other by scalars, for aniso- tropic materials these scalars become tensors, and the velocity of propagation becomes a function of the direction of polarization and the direction of propagation. In this section we wish to study the propagation of an electro- magnetic wave in an electrically anisotropic, non-conducting, (transparent), non-magnetic (|i = \i ) , source-free and homogeneous material . Let us assume the simplest relation which can account for the anisotropic behavior, namely that each component of the dielectric displacement vector D is linearly related to the components of the electric field intensity vector E: D. = 6. . E. (2.1) where € is the dielectric constant tensor. (Note that we define D. as the component of D in the x. coordinate axis direction.) We can show that e. . is a symmetric tensor. This is a consequence of the definitions of the electric energy density w , the magnetic energy density w , the Poynting e m (11) vector P, and the continuity equation for the electromagnetic energy. These definitions are w = - D • E = - e,.E. E, e 2- - 2 ijij w = : i B • H (2.2) m 2 — - P = E x H and the continuity equation is 5 dw dw - div P = — (w + w ) = — - + — - (2.3) — dt e m dt dt The left hand side of (2.3) can be computed from (2.2) using a well known vector identity, yielding - div P = E curl H - H curl E (2.4) Substituting for curl E and curl H from Maxwell's equation we have dD dB - div P = E — + H — (2.5) — — dt — dt Comparing (2.3) and (2.5) we find that dw dD dt — dt On the other hand from (2.2) we obtain dw dE dD dt 2 dt — — 2 — dt — dt Equations (2.6) and (2.7) must be equal, hence dD dE E — = D — (2 .8) - dt - dt which in tensor form reads dE . dE . £, . E. —J- = €.. -r- 1 E. (2.9) lj l dt ji dt l dE. and since E. -~— = 0, we must have l dt €. . = G.. (2.10) which result shows that €. . is a symmetric tensor indeed. This fact allows us to eliminate the off-diagonal elements of e , ij or equivalently, choose the coordinate system x. in such a manner that e 1 ij becomes a diagonal tensor. Let us call the diagonal elements e the ii principal dielectric constants, and let e. . = €. . Then we can write 11 1 D. = e.E. (2.11) Equation (2.11) calls our attention to the fact that if the E of the wave coincides with one of the coordinate axes x. , the material behaves formally as an isotropic material with dielectric constant e. . For these particular waves we can use the definition of the propagation (phase) velocity and the index of refraction as given for isotropic media, and write 1 v. = (2.12) 1 F& and O O 1 o where c is the velocity of light in vacuum, e*.T is the relative principal *i dielectric constant and n. is the principal refractive index. The remainder of this section will be devoted to the study of a general electromagnetic (TEM) wave traversing the anisotropic medium. It will be shown that for any given direction of propagation there are at least two directions of allowed polarization, and that these directions, and the corres- ponding propagation velocities can be obtained from the knowledge of the principal refractive indices n. . l Let us find which plane waves can propagate in the anisotropic medium in the direction of N unit vector. In tensor notation the possible solutions can be described by 7 x.N. E. = E . exp [ ju(t ) ] (2.14) J oj ° v where u> is the angular velocity in radians per second, v is the phase velocity (propagation velocity) in the direction of N. For a rectangular coordinate system Maxwell's equations in tensor form read 8 H. 5 D . e S X. i 5 t e kl J 5e J 5 X. l So B k 8t 8H * 5 D. l = 8 X . (2.15) (2.16) and (2.17) OB. i = (2.18) 8 X i ik All quantities are expressed in the MKS system of units. The tensor e *• A U ( 13 > is defined by ijk . e =0 if i = j , orj=k, or l = k =+1 if ijk are in even permutation (2.19) (1, 2, 3; 2, 3, 1; 3, 1, 2) =-1 if ijk are in odd permutation (1, 3, 2; 3, 2, 1; 2, 1, 3) Substituting (2.14) into (2.16) we find S h x.N. -\x — -^ = - j - [ e klJ E , N. ].exp [ JO) (t - - J - i ) ] (2-20) integrating this equation with respect to time (from - oo to t and assuming that H k = for t = - <*> ) H k = ^ [ e klJ E^N. ] exp [ jifct ~ ^ ) ] (2.21) from which we can conclude that H • E = and also H • N = 0, implying that both E and N are normal to H. Substituting (2.21) into (2.15) gives 6T = - j ,"7 l eJlk (eklJ E oj N i> N i 1 ex » t ^ (t - ^ < 2 - 22 > Integrating again with respect to time gives x N Dj = [e Jik (e kij E QJ N.) N. ] -^ exp [ ju>(t - -^)] (2.23) which can also be written in the form D. = - ^-2 [e jik (e kiJ E.N.) N. ] (2.24) J H Q V 2 J i i v Expanding this tensor equation and noticing that (since N is a unit vector) 3 Z N. 2 = 1 (2.25) i=l X equation (2.24) can be simplified to the form D. = -^— §- [E. - (E.N.) N. ] (2.26) J U. v 2 j li j From this equation we see that since D is a linear combination of vectors E and N, D must be in the plane of E and N, and since E and N are normal to H, (2.21), D must be normal to H. It is also seen from (2.26) that D • N = 0, implying that D is normal to N also. Figure 2.1 illustrates the relative positions of these vectors. D, E, N and P vectors are in a plane normal to B and H vectors. The vector u is defined as a unit vector in the x. — — —x. 1 l coordinate axis direction. Figure 2.1. Relative orientation of vectors E, D, H, B, N, and P. It will be more convenient to use the inverse of the 6. tensor. Since £. 1 l is a diagonal tensor, its inverse is simply the reciprocal of the elements £. , i.e., hence p. -[e.]" 1 -^ 1 L 1 J f . *- 1 E. = p.D. l 'l l (2.27) (2.28) where (3. is called the impermeability tensor. With this definition (2.26) becomes D. = B.D. - (E.N.)N. J .. .,2 J J 1 1 J We can rearrange this equation to the form (2.29) D.(p. = Ul v 2 ) - (E.N.)N. j K j To l i j (2.30) and divide both sides by (p. - Kx. v 2 ) ^ D = J (E.N.)N. <3 -/V*> (2.31) 10 i We have noted before that D • N = 0, hence 3 3 (E.N.)N. 3 3 N 2 1=1 J=l K J O 1=1 D =l p g o 3 Since Z (E.N.) = for anisotropic media, we can conclude that i=l N. 2 E ^ V 2 =° (2.33) Let us define a. as the principal phase velocity by a j = L- (2 - 34) Dividing (2.33) by |i and substituting (2.34) yields 3 N. 2 £ — - 2 = (2.35) i=l i We can transform this equation into an equivalent form by defining the index of refraction n as n = - . (2.36) v and using the definition of n. (2.13) we obtain i * > 3 £ - t- (2.37) i = 1 n 2 - eV n i Equations (2.35) or (2.37) give two values for v 2 (or n 2 ) for a given N and G f . tensor. Equation (2.35) can be expanded to read v 4 - (n x 2 (a 2 2 + a 2 3 ) + N 2 2 (a 2 1 + a 2 3 ) + N^Ca^ + a 2 2 ) ] v 2 + + N W 2 3 + N V 2 i a2 3 + N YV 2 3 = ° (2 - 38) 11 which has the solutions V2 = I [ N V a % + a Y + "r.<»\ + a*.,) + N^ 3 fc.» + a^) ± ± ^ N4 l 2+N 3 (a2 l a2 2 )2 - 2 l N V a Y a2 3 )N V a Y a V + ♦ » ! 1 »V*V 1 V ( "V" ! 2 )€ VV aI i-»V (>! i-"V] (2.39) The two solutions of (2.39), (v 2 ) 1 and (v 2 )" give the two phase velocities and therefore the two indices of refraction. Let us consider equations (2.34) and (2.39) in more detail. If we let N = u , then N = 1, N = N = and (2.39) has the two solutions (a 2 ) and X.. J. a O 2* (a 2 ) , which correspond to p and p respectively, I (2.34) J . Thus we find that for this case there are two plane waves which can propagate through the medium: one is polarized in the u direction, the other in the u direction, 2 3 and they propagate with v' = a and v" = a respectively. Notice that we found the directions of polarization by finding the directions for which the value of p corresponds to the solutions of (2.39), namely to v' and v". This method can be generalized by defining v as v 2 = -&- (2.40) ^ o where v is the phase velocity of the wave in the N direction and p* relates the components of D and E in the plane normal to N. We shall show that as the plane of polarization is rotated around the N vector, p* describes an ellipse, whose major and minor axes correspond to v f and v " and therefore the only two permitted planes of polarization will be determined by the major and the minor axes of the ellipse together with N. This derivation provides 12 an analytical proof for the well know geometrical method for the construction of the two permitted indices of refraction from the index ellipsoid. The coordinate system x ± was chosen in such a manner as to coincide with the directions of the principal dielectric constants of the anisotropic material. Since we want to investigate general directions, (N) , we have to set up a rotated coordinate system in which vectors D, H and N are in the direction of the axes y. . Fig. 2.2. t = angle between (x y A plane and y axis cp = angle of (x y ) plane and x axis 6 - angle between x and y The rotated coordinate system as related to the original axes The vectors D, H and N will have the following form in the new coordinate system D = D u y ~y i ± N = N u y.3 -tJ H - H • u y 2 -: Since p is a second rank tensor* we can write (2.41) Since V 3 D 2 = 2w > for D. f 0, the left hand side of this equation is a positive definite quadratic form, hence p. are the principal semi-axes of an ellipse and also 6. is a second rank tensor^ ^ 13 ^ J 13 8y ± &y i 0*. . = ' * • ft , (2.42) The desired rotation can be done by = 1 = cosO cos cp cos ■■\- sincp sin? (2.43) = 1 =-cos9 costp sin? - sincp cos? 8 x l 6y 2 5 x l 5 y 3 6 X l 5 y l 8 X 2 8 y 2 5 X 2 6 y 3 5 X 2 5 y-L 5 X 3 5 y 2 5 X 3 B y 3 8 X 3 = 1 = cos cp sinG = 1 = cosG sincp cos? + cos cp sin? = l OQ = cos cp ccffe? - sincp sin? cos© = 1 = sincp sin0 = 1 = - sinG cos? = 1 = sine sin? = 1 33 = cosG The impermeability tensor 0*. . relates vectors E and D expressed in the y. coordinate system. Specifically E = B* D (2.44) y I U y l since D = D =0. (Refer to (2.41)). y 2 y 3 Let us compute the maxima arid minima of B* (?) as ? varies from to X K n Using (2.42) and (2.43) we can express 3* in terms of cp , 9 and Y • P*ll = ^lA + l2 21^2 + l2 3lP 3 (2.45) = [ (cos 2 9 cos 2 cp ) p 1 + (cos 2 0sin 2 cp ) p + (sin 2 0)p ] cos 2 ^ + + [cosO • si g 2Cp (P 2 -P 1 ) ] sin2^ + + [ (sin 2 cp )p + (cos 2 cp )p ] sin 2 ^ which can be written as P *ll = A i cos2 * + A 2 sin2 ^ + A 3 sin 2 ^ (2.46) where A. depends only on 3 . , and cp . Equation (2.46) is recognized as the equation for a rotated ellipse; the direction of the axes (maxima and minima) can be obtained from 5 P*11 — - = = -2A 1 cos^ sin^+ 2A cos2 Y + 2A sin Y cos Y (2.47) Ol -L Z 6 which has the solution 1 A 2 ¥ • = -= arctan — — - (2.48) * A l A 3 As we have expected, (2.48) indicates that the extrema are jt/2 apart: y » » = y ' + it/2 . Substituting (2.48) into (2.46) and rearranging terms we find that ?*-,-,( ¥*) = - [ p (sin 2 epsin 2 0+eos 2 0) + p (cos 2 cp sin 2 + cos 2 0) + XX £ X ^ + 3 3 sin 2 + l/(3 1 -3 2 ) 2 (cos 2 0cos 2 cp +sin 2 cp ) 2 + 2(p 1 ~P 2 )(P 3 -p 2 ) . __ _ — 1 • sin 2 0(cos 2 0cos 2 cp -sin 2 cp ) + (p -p ) a sin 4 9 ] (2.49) Let us now transform (2.39) into the new coordinate system y. . This can be accomplished by letting 15 H ± -l la < 2 -50) Performing this substitution into (2.39) and rearranging terms we obtain v 2 = - [a 2 (sin 2 cp sin 2 + cos 2 9) + a 2 (cos 2 cp sin 2 9 + cos 2 9) + a 2 „sin 2 9 + a X £ O — - 1/ ( a2 1 ~ a2 2^ 2 ^ cos2ecos2c P + sin2( P ) 2 + 2(a 2 - a 2 )(a 2 -a 2 ) • • sin 2 9(cos 2 9 cos 2 cp - sin 2 cp ) + (a 2 -a 2 ) 2 sin 4 9 ] (2.51) Substituting a. from (2.34) into (2.51) we find that the resulting equation is identical to (2.49). Hence we can write p* , ( r) =ii (v«) 2 11 ° (2.52) P*ll ( r) = ^o (v " )2 Since 8* ( ^) is an ellipse, and we find that the extrema of (3* correspond to the two permitted propagation velocities, we conclude that Y * and ¥*', or the axes of the ellipse are the only permitted directions of polarization. We find that (3* , (^ ) is an ellipse for any 9 and cp , which implies that 8* , is an ellipse for any orientation of the (y, y~) plane. Since the only surface, which is cut by any diametial plane in an ellipse is the ellipsoid, we can consider that B* . is the radius vector of an ellipsoid. We have considered (3 and v in these calculations as the basic variables, but e n , and n can be used as well. From (2.44) we have 11 D .. = \r — E = €* . E_j (2.53) A p*u yi n y? hence 1 e* 11 8* p n 16 and from (2.40) we can substitute for P*, , and obtain e *n m = pS-TY) = |T> =€ o n2 (2 ' 54) 11 o Since P* , can be considered as a radius vector of an ellipsoid, we can conclude from (2.54) that n is also a radius vector of an ellipsoid. (Note that p* ^ 0.) A diametral plane normal to N, (the y_ y plane) cuts both the p* and -L z the n ellipsoids in ellipses. Because of equation (2.54). the extrema of these ellipses occur for the same ^ ' and ^ ", i. e. and 11 'O Equation (2.55) indicates that n ellipsoid may be also used for the determination of the permitted directions of polarization. This ellipsoid is called the index ellipsoid, but it is also known as the ellipsoid of wave normals, or the optical indicatrix. The equation of the index ellipsoid is = 1 (2.56) The construction of n' and n" from the given N and the index ellipsoid is illustrated in Fig. 2.3. The two permitted indices of refraction for the case illustrated in Fig. 2.3 are (n')" 2 = COS l 9± + Sin29 i (2.57) 1 * 2 . x 2 * 2 o 1 2 3 »'l + " 2 2 + „ 2 n 3 n 3 n 2 it n = n 2 i x t 3 17 Fig. 2.3. Illustration of the use of the index ellipsoid. Equations (2.54) and (2.55) show that tensors €. , p. and the index ellipsoid contain identical information about the properties of the anisotropic medium. 2.2, The derivation of the electro-optic effects . The anisotropy mentioned in the previous section can be attributed to the crystal structure, or in some materials it can be induced by electric fields and stresses. These induced anisotropic properties can be expressed as the derivatives of one of the thermodynamic potentials. The suitable thermodynamic potential is the internal energy U which can be written in terms of strain (5. .), electric displacement (D) and entropy (S) as independent variables, ant stress (T) , electric field (E) and temperature (0) as dependent variables . The differential of U can be given by dU = T. . d (S . .) + E d D + 9 dS (2.58) ij ij mm Let us assume that for our purposes adiabatic (dS = 0) and constant stress (dS . . = 0) conditions prevail. In this case (2.58) simplifies to 18 dU = E d D (2.59) and we can develop E in terms of D m m m 1 5 m 1 5 3 E m E m = FF °n + 21 65-85 D n D p + 3! JT f^t^ D rPp D q + '" (2 ' 60) n n p H 5 d 8d 6d m P n p q The coefficients of this expansion can be written as 5 2 D s E m ^ = a D mn n 5 D &D m n 5 3 D = R 2E Sd Sd n p = S D &D m n P b mnp (2.61) (2.62) and 5 4 U 5 E m = Q (2.63) mnpq 5d&d&dSd 5d5d&d mnpq n p q Thus (2.60) can be rewritten as 1 1 E=aD + -c DD+-C DDD +... (2.64) m mn n 2 mnp n p 6 mnpq n p q and E=(a +77C D+-C DD)D +... (2.65) m mn 2 mnp p 6 mnpq p q n Our present aim is to show the interaction between the modulating electric field and the optical impermeability tensor 3 Let us assume that the D and E vectors consist of components at the modulating frequency w' and at the light frequency w' v . E. -!• e jU,t 4- E". e JW,,t (2.66) 11 l and . ., M^ D. = D 1 . e J + D e J J J J Substituting (2.66) into (2.65) yields 19 E = U + [^S£ + JH^L (D . e J w 't + D - e jw"t ] e Jto't + D „ jco"t \ m \ mn u 2 6 q q p 1 / • (D« e^' 1 + D" e^"*) (2.67) Performing the multiplication and collecting the terms of approximately co frequency we obtain b E" = a D" + (D* D" + D' D" ) -^- + m mn n p n n p 2 c + m " Pq (D" D' D* + D' D" D* + D' D' D" ) (2.68) 6 npq npq npq and considering that X Z (D' D" + D" D' )] = 2 Z. \ Z. < D ' D " >1 , I ., p n p n J . L *— , p n -J n=l p=l n=l p=l (2.69) equation (2.68) can be further simplified to " = [a + b D' + \ c D' D' I D" (2.70) m L mn mnp p 2 mnpq p q J n E The effective optical impermeability tensor is then c 6 = a + b D' + m " Pq D' D' (2.71) mn mn mnp p 2 p q We desire to express the interaction in terms of the electric field intensity E' , rather than in terms of D' . ; hence from (2.1) we can write i i D' = c' E' (2.72) The substitution of (2.72) into (2.71) yields B = a + b F' E' + i- c <=' J E' E' (2.73) mn mn mnp ps s 2 mnpq <- ps °qt s t Let us define new tensors 8'' , r and p as mn mns imnst S" = a the optical impermeability tensor K mn mn ^ J r = £ b c' the linear electro-optic tensor (2.74) mns o mnp '-ps n - f — c c d the quadratic electro-optic tensor Nmnst o 2 mnpq c ps c qt 20 Thus (2.73) becomes 8 =8" + £ _1 r E' + O £ _1 E' K' (2.75) mn mn o mns s mst s t J m This is the general expression for the electro-optic effect . In the m m 2 MKS system the units of r is — and that of p is — . Using the J m s V Ymst y2 results of the previous section we can express the effect of the modulating fields on the index ellipsoid. From (2.54) we can write 8 = * 2 (2.78) m £ n £ o m Expressing n 2 from (2.78) and substituting for 6 we obtain m m (2.79) m <- /„ti ^ -1 „, _ -1 £ (8" + £ -i r E' +8 O E' E' ) o m o ms s o >mst s t Let us define n by om n 2 = ^ I,, (2.80) om £ n & «, o m hence (2.79) becomes n 2 om n 2 = m 1+n 2 (r E' + O E' E' ) om ms s >mst s t [ 1-n 2 (r E' + P E' E' ) + . . . (2.81) om u om ms s >mst s t 21 Taking the square root of both sides we obtain r n * i n = n 1 2E ( r E' + P E' E' ) + ... (2.82) m om L 2 ms s ^ mst s t J Thus the change in the principal index n is m An = - - n 3 (r E' +P E' E' ) (2.83) m 2 om ms s »mst s t 2 .3 The index ellipsoid of some important electro-optic materials . In the previous section we have derived the general relationship (2.81) between the modulating electric field and the principal refractive indices of the material . For some media the significant elements of the tensors r and Q have been measured, but for most materials the available data is ms >mst (17), (18), (19), (20) incomplete . 2.3.1. Linear electro-optic effect in uniaxial materials. Among other materials, the important electro-optic crystals KDP and ADP (with a symmetry D ), belong to this group. If an electric field E is applied to this material along its optical X 3 axis x , equation (2.81) becomes n 2 1+n 2 r E o 63 x 3 (2.84) n 2 1-n 2 r E o 63 x„ and The material constants n , n , r are given in Table 2.1. Thus the o' e' 63 index ellipsoid in normal form is <~i- + r «<, E ) x 2 n + ( 4- " r._ E ) x 2 + K- x 2 = 1 (2.85) n 63 x„ 1 n" 4 63 x„ 2 n 3 o 3 o 3 e 22 Although we are going to discuss this later, we should mention here the fact that KDP and ADP are ferroelectric materials, and that both r 63 and € s are very strongly temperature dependent. 2.3.2. Linear electro-optic effect in cubic materials. For cubic materials with symmetry T , to which cuprous chloride belongs, the principal indices are given by (the electric field is in the x direction again) 1+n 2 r„,E o 41 x 3 n 2 iT5 — (2.86) 1-n^ r._E* o 41 x o n 2 = n 2 and the index ellipsoid is given by ■r )X 2 n + (-2 o ° o «j o ( ^" + r 41 E x ' x2 l + ^ r 41 E :X y 2 + *>- X % " * <2 - 87) For cubic materials the orientation of the index ellipsoid depends on the orientation of the electric field. 2.3.3. The quadratic electro-optic effect . All electro-optic substances which have a center of symmetry in their molecular structure exhibit this quadratic effect; for this case tensor r = n ms For cubic materials of symmetry 0, the number of independent p tensor elements are reduced to three. Furthermore, if the field is applied in the direction of an edge of the cube, e.g. the x axis, the principal indices are given by 23 n 2 = n 2 = ^!° (2.88) Z 1+n 2 P E 2 O 2 x 3 and n 2 .2 n 3 " 1+n 2 Pl E 2 X 3 and the index ellipsoid is given by ^ + P 2 E V x2 1 + < J"" + P 2 E \ > x2 2 + ( ^" + Pl E2 x > x2 3 " * (2 - 89) o 3 o 3 o 3 It can be demonstrated that this quadratic effect gives rise to the Kerr effect. Let us consider the difference <°\ - n V - " a o t i + ,/ p ,e» - 1 ; / Pl E* ' < 2 - 90) o p 2 x„ o M l x 3 3 - „2 also n 2 [1 - n 2 p E 2 - 1 + n 2 p E 2 ] o o^2x o K lx = n ( Pl-P2>:"-^3 (n 2 x - n 2 3 ) = (n x + n 3 > (^ - n g ) (2.91) = 2 n (n - n ) O 1 o hence ^ • < n l " V " \ n 'o ( Pl - (21) The empirical law of the Kerr effect states that An = K\ E 2 (2.93) X 3/ where K is the Kerr constant, \ is the wavelength of light in vacuum and An is the difference in indices as defined by (2-92). Comparing (2.92) and (2.93) we find that K\ = | n 3 Q (p 1 - p 2 ) (2.94) 2 .4 Material properties of some electro-optic materials (MKS units are used.) Material Units Temp. (^ - (p 2 ) r K 41 -16 m 2 , rt -12 m 10 v* 10 V 10 63 -12 m KDP ** ** ADP CuCl ZnS BaTiO SrTi0 3 K TaO nitro- benzene 120 148 393 33 173 — -- 293 — — 173 OQQ -- — zy«3 293 — 6 293 — 2 .4 393 23 -- 403 6.5 -- 77 0.31 -- 293 0.02 -- 4 4.3 -- 77 0.1 — 300 0.0124 3.0 1 .5095 60 6.8 1 .5095 20 — 1 .5246 16 5.5 1 .5246 14 — 1 .93 8.3 — 2 .368 10.25 — 2 .4 6000 2 .4 4500 — 2 .409 300 — 2 .409 300 — 2 .5 4400 — 2 .5 800 1 .55 34 T is the Curie temperature discussed in Section 3.9 c **n = 1.468 for KDP and n = 1.4792 for ADP e e CHAPTER 3 INTENSITY MODULATION BY CONTROLLED PARTIAL REFLECTIONS 3.1. The FresnaJl relations for the reflected amplitudes . Let us consider the problem of a plane wave incident on a plane boundary between two media n and n . If both materials are anisotropic, then there are four resultant waves in general: two reflected inton,, and two refracted into material n . The solution for the intensities, phase velocities, directions of propagation and directions of polarization can be obtained from the boundary conditions on the E, D, H, B vectors and from the index ellipsoids of the two materials. The analytical solution of this general problem is quite difficult; the geometrical constructions involve fourth order surfaces and are very lengthy. Fortunately, a few simple, realizable assumptions reduce the complexity of the problem to an acceptable level . Let us assume that both materials are homogeneous, non-conducting, non- magnetic, and that material n is isotropic and material n is electrically anisotropic of the type treated in Chapter 2. Let us further assume that the plane of incidence and the plane of boundary coincide with the planes of symmetry of the index ellipsoid of medium n . Notice that this is a very significant simplification, since now both the normal and the parallel (with respect to the plane of incidence) components of the incident optical electric field E are in the direction of permitted polarization for material ^ ; hence both of these waves have only single refracted and reflected waves. In Fig. 3.1 E and E are the normal and parallel components (with respect to the plane of incidence) of the incident optical field 25 26 anisotropic material n (xz) plane is the boundary (xy) plane is the plane of incidence Fig. 3.1. Plane wave is incident on an anisotropic medium. E • N. , N . , N and 0., 0^, are the unit wave normals and the angles — — i — t — r i t r & of the incident, transmitted (or refracted) and reflected waves respectively The principal refractive indices of medium n„ are n„ , n^ and n„ , in the 2 2x 2y 2z direction of coordinate axes x,y,z. Let us further assume that medium n n is denser than medium n , i.e. n i >/n 2 (3.1) so that there exists a critical angle 0. for which 0, = n/2 . Let us con- ic t sider the case when is slightly below the critical angle 0. . The re- fracted wave propagates nearly in the direction of the x axis. i.e. (3.2) The diametral plane normal to u cuts the index ellipsoid of medium —x N = u — t — x n in an ellipse with semi -axes n and n , therefore planes (x, y) and (x, z) are the allowed planes of polarization, and the refractive indices are n„ and n^ in these planes respectively. If the incident wave is 2y 2z polarized in the direction normal to the plane of incidence, then 27 trace of ellipsoid n, Figure 3.2. The two allowed polarizations for N = u . E=E =Ee JTi i — — n n — z (3.3) where r-Ni T. = OJ (t - — 1 v. ) = CJ [t - — (x sine. + y cosG.)] (3.4) The magnetic field intensity vector H can be obtained from Thus H = n n \| — (E cosG.e J i u - E sinG. e "* u ) lll[-i ui — x u i — y (3.5) If we define R and T as the complex amplitudes of the reflected and n n transmitted waves, we have for the transmitted fields E^ = T e J t u — t n — z (3.6) where G o " jT t " jT t H = n„ 1 — (T cos9 e u - T sine e u ) -t 2z y u n t -x n t -y i o (3.7) T -«[t - (x sine + y cos6 )] C X c (3.8) and for the reflected fields E = R e — r n u — z (3.9) 28 i € o "^r " jT r -r = n i V 7T (R n COs9 r e H x " R n sine e u ) (3.10) where "l T = oj [ t (x sinG + y cos9 ) ] (3.11) The boundary conditions at y = demand that the tangential components of E and H be continuous. Hence we must have E + E = E - -r -t and (3.12) H*u + H • u =H - u — —x — r —x — t —x (The normal components of B and D are continuous across the boundary automatically.) Substituting the electric and magnetic vectors into (3.12) and using the fact that cosG = - cos9., we obtain r i E + R = T n n n and n, cosG, (E - R ) = n cosG T (3.13) l.i n n 2z t n We can solve (3.13) for the reflected amplitude R n n n cos9 . - n„ cos9__ R = -^ j± 25 * E (3.14) n n^cos9. + n cos9 J n 1 i 2z t Let us define r as the relative reflected amplitude; then we have n R n. cos9. - n cos9^ r - — = zr ZT (3.15) n E n, cos9. + n„ cos9^ n 1 i 2z t In an analogous manner we can derive the relative reflected amplitude of the parallel polarized wave and obtain 29 R n cos0. - n, cos9 jp 2y 1 1 t r p ! E n cos9. + n, cos0^ (3.16) P 2y l 1 t Equations similar to (3.15) and (3. 16) were derived by Frcsncl in 1823 and by tradition bear his name. The angles 0. and 9 are related through the law of refraction n i sinG = — sine , n > n_ (3.17) t n 1 1 2 where n = n for the normal component and n = n for the parallel component . The critical angle of incidence is different for the two waves in general, since n. p n. . Let us denote by G and 0^ the critical angles for the nor- 2z ' 2y n t mal and parallel components respectively. Then we have (3.18, ii n and sin0 = n n l sin0 P 2yo n l (3.19) 3.2. The dependence of the reflected intensity on the index of refraction n . If the anisotropy of material n is due to the electro-optic effect, then the principal refractive indices n , n and n depend on the modulating field E . In this section we want to examine how the changes in these indices m 6 effect the relative reflected amplitudes and the relative reflected intensities (I and I ) of the light waves. The intensities and the amplitudes are n p related by I = (r ) 2 n (3.20) I - (r ) 2 P P Let us denote the changes in indices by An [ refer to (2.83)] n 2i = n 2io + ^21 (3.21) 30 where i ranges through x, y, z and n are the principal refractive indices ZIO with E = 0. — m As we have noted previously, our interest lies in cases where the incident angle is close to the critical angles given by (3.18) and (3.19). We shall find it convenient to introduce the variables Ot and 3, where and a = e - 9. n n i B = rt/2 - 9^ n tn (3.22) for the normal component, and a 9-9. P i 8 = ^2 - 9 P tp (3.23) for the parallel polarized wave. »-x Fig. 3.3. Illustration of the new angular variables for the parallel component . Substituting (3.21), (3.22) into the expression (3.15) for r we obtain 31 n_cos(9 - a ) - (n +An ) cos(n/2 - ft ) 1 n n 2zo 2z K n r n " n.cos(9 - a ) + (li +An ) cos(jt/2 - ft ) (3.24) 1 n n 2zo 2z *n which can be written into the form 1 - A sinft r n = 1 + A n sin S n (3 ' 25) n "n where (n„ + An„ ) 2zo 2z . . A = 7E — 7TT (3.26) n n, cos (9 -a ) 1 n n Using (3.18) we can rewrite equation (3.26) as An a 2 A = tanG (1 + ) (l - tan a tan9 + . . .) (1 -2 + . . .) (3.27) n n n_ n n 2 2zo which yields after multiplication An A = tanG [1 + ( - tan a tanG ) + ...] (3.28) n n n_ n n 2zo The second and higher order terms were neglected, since a and An are -2 smaller than 10 for cases under consideration. The term sinft can be computed from equation (3.17) using the new variables (3.21) and (3.22); then we can write n l sine = cosS = t sin(0 - a ) (3.29) tn ^n n„ +An„ n n 2zo 2zo hence An a 2 , cosft = (1 — + ...)(1 - tana catane + ,,.)(1 ^+ ...) (3.30) "n n„ n n 2 ' 2zo 2z = 1 - ( + tana cotan9 ) + . . . n_ n n 2zo .2* n An. Then cos ft is approximately equal to cos 2 R = 1 ■- 2 ( — — + tan a cotan 9 ) (3.31) K n n„ n n 2zo 32 and from the identity sin 2 B + cos 2 B = 1 we obtain n K n An sinB = [ 2(- n n 2z 2zo + tan a cotan ) ] n (3.32) The reflected relative intensity I is given by I = (r ) 2 n n 1 - A sin 6 n ^_n 1 + A sin ft n n (3.33) = 1 - 4 A sinB + 8 A 2 sin 2 B - . . . n K n n K n Substituting (3.28) and (3.32) into this equation and retaining only the first order term we obtain An An I =1-4 tanG (1 + - tana tanG ) [ 2( + tana cot anG )] n n n_ n n L n n J 2zo 2zo (3.34) Let us now define the relative intensity modulation M by M =1-1 (3.35) For the normal component this becomes M and n An 1 M = 4 tanG (1 + — — . _ . rr> , 2z _ . -, 2 fo ._. n n n n tana tanG ) [2( + tana cotanG )] (3.3b) 2zo n n n„ n n 2zo We can derive analogous results for the parallel component. We can write the expression (3.16) for r in terms of the new variables as P (n„ + An„ ) cos(G - a ) - ncos(fi/2 - 6 ) 2yo 2y^ p P 1 P r p (n„ + An„ ) cos(G - a ) + ncos(it/2 - 6 ) 2yo 2y p p 1 K p (3.37) Let us define A as P 33 n i A P ' (n. + An )cos(9 - a ) (3,38) 2yP _2y P P Then (3.37) can be rewritten as 1 - A sin|3 r = A P . P (3.39) p 1 + A sine ' o K p Using similar arguments as in the derivation of A , we obtain n t an9 An A = . 2 P [ 1 - ( £ + tana tan9 ) ] p sir8 n_ P P P 2yo (3.40) The term sinB can be formally obtained from (3.32) by replacing n by n„ and 9 by 9 . Thus J 2y n J p ^2 - sinB = [ 2( £ + tanQ; cotan9 ) ]2 (3.41) P n 2yo P P P The quantities I and M for the parallel component are given by I =1-4 A- sinB + 8 A 2 sin 2 B ~ ... (3.42) P P P P P and M s 4A sinB - 8 A 2 sin 2 S + ... (3.43) P P P P P Substituting A and sinB from (3.40) and (3.41) we have P P tan9 An An — I =1-4 . 2 P [ 1 - ^ - tanatan9 )] [ 2( + tana cotan9 ) ] P sm 2 9 n_ P n n p p P 2yo 2yo (3.44) and , tan9 An An - M - 4 J 9 P (1 ^ - tan a tan9 ) [ 2(- — * + "tanacotan9 ) ] p sin 2 9 n„ p n_ o p 2yo 2yo (3.45) 3^ We should remark here that An and a, can be negative as well as positive, but real modulation occurs only if the quantity under the square root sign in (3.36) and (3.45) is positive. -3 Since for most electro-optic materials An<10 and a is chosen to be small, ( this point will be discussed later), we can write ^z \ M = 4 tanG [ 2( + tana cotanG )] (3.46) 2zo n and 1 tanG An - M = 4 . J [2( y - + tana cotanG ) ] 2 (3.47) p sin^G n p p p 2yo The last two equations will be used in subsequent calculations. 3.3. Modulation M as a function of the insulating electric field E m . The electro-optic effect was given specifically for three important crystal groups in section 2.3. Considering then crystals as medium n , we shall obtain M and M as functions of E and a. Two cases should be n p — m considered separately: 1. when E has only a normal component (i. e. E = E u ) and 2. when E has only a parallel component (E - E u ). — m z — z — m * *■ «- —roy—y Let us consider first the use of a linear uniaxial material. For the first case we have x_ = x, x. = y, x^ = z and E = E . Thus we 1 2 3 -m z can write n = n + \ n 3 r__ E (3.48) 2y o 2 o 63 z and n„ - n 2z e Substituting these quantities into (3.46) and (3.47) we find ± 35 M = 4 tanG [2(tanQl cotanG )1 (3.49) n n L n n J and tanG M = 4 - P E r i -.1 • 2 Q L 2 (o n r c< 3 E + tana cotanG ) 2 (3.50) nn^G L 2o63z p p J In a similar manner we can determine M (E , a ) and M (E , oc ) for the n -m n p — m p other crystal groups. These results are summarized in Table 3.1. It is apparent from this table that several media exhibit negative An. For these cases modulation can be achieved only if ol is chosen such a way that (— + tanacotanG) > (3.51) n o The apparent electro-optic constant of a quadratic material can be in- creased by the use of a bias electric field E . Let us consider An of o y a cubic quadratic material if E = E + e , and the E field is in the y — m -o -m — m direction, (where e is the magnitude of the modulating field) . Then m An = - \ n 3 O (E + e ) 2 (3.52) y 2 o >1 o m = " h n3 9, (E 2 + 2E e + e 2 ) 2o)l o om m Now if E > > e we can neglect e 2 in (3.52) and write o m m An = (- J n 3 Q E 2 ) + (- i- n 3 0, 2E e ) (3.53) y2o>lo2o>lom Since E is a static field, we can consider the first term of (3.53) as o part of n and we can say that the effective change An v will be y o An' = - n 3 o E e (3.54) y o > 1 o m Thus in this case the quadratic material behaves as a linear material with electro-optic constant ( p,E ). N 1 o 36 cd G CD N G co CO 0) i— i o .a ^ od rH O CO •H o C/J • rl r< -h o u ft ft Z-, +j Jh M < U IS1 m g w ^ as S [Q .* rl a g ft G as S s s S ft ft s s s 0) >> >> >> !>> C5 rH rH rH rH * «, G G g G G G O o S a a 05 a Ol rH CDl -H 01 rH 01 r^ 01 rH 01 rH ■H c 1 a G 1 G g \ g G I G G | G G [ G a tn N) N >> w W N N >> 00 r-i w W CM >» CO > 0) O IN CI G G G G G G c a> g CD! rH 01 rH 01 <-{ 01 r-\ Ol rH 01 rH •H g 1 g c 1 g g 1 g G 1 G G I G G 1 G r/1 >> SJ >> H i>> fr) w CO H w H SI CD rH tH CM > N >> N >> W W w W W W w O gg •H rH l-H •P rt Ct f-i s-\ O rH o as ■H •H aS "3 M •ri CS /^ M -H u h M X CD CM rH £1 1) 5 M ,.Q "O /-s CD a) G Q 3 G M 3 aS Si •4-> •H -H ^ O -rH H O 3 O d G rJ ^i V O" v^ S £> bD G •H -P aS r-< 3 XJ o s CO 3 •H tH aS > rH O «H /-s t* ^ ft • S CO >— ' rH •o «H G aj aS +-> /-s d CO ■d -^ G •H blD G •H • S-i (0 aS -p fu G ft CD ft G aS O ft W S G O O O •H 4J t3 CJ rH G CD 3 -H ft n ) 38 As indicated in Fig. 3.4, a is the separation between the two parallel reflecting surfaces, L is the length of the device and b is the height of the device (in the z direction.) The number of reflections m is given by m = — cotanB. (3.55) a i (Note that the length L should be chosen so that all of the beam is reflected on the last reflection.) In section 3.2 we have derived I and I for a single reflection. n p The reflected relative intensity I after m identical partial reflections m is given by I = (i) m = (1 - M) m (3.56) m Raising the expression for I to the m^ n power yields (I ) m = I = 1 - 4m A sin 6 + 8 m 2 A 2 sin 2 6 - ... (3.57) n mn n n n n and similarly for (I ) P (I ) m = I = 1 - 4m A sinB + 8 2 A 2 sin B - ... (3.58) p mp p K p m p K p Let us define the resulting modulation by M m M =1-1 (3.59) m m Neglecting the higher order terms in (3.57) and (3.58) we have the M mn and M mp M = 4 m A sin 8 = m M /0 cn v mn n K n n (3.60) M = 4 m A sin 8 = m M mp P P P Thus we can conclude that the modulation increases m-fold due to the multiple reflections. 39 We should point out that the intensity modulation can be realized in a dual way also. Let us suppose that the denser anisotropic material is n , and medium n_ is the isotropic, less dense material. (Refer to Fig. 3.4). Then internal total reflection can occur in the anisotropic material n , and controlled partial reflections are possible. The analysis of this case is slightly more involved, since here the index of refraction for the parallel polarized wave does not coincide with the principal index of refraction n n . The results for this case can be summarized by M* and ly mn M* mp "^lz 2 M* = 4 m tan 0* [ 2( + tana* cotan0 f ) 1 (3.61) mn n n, n n J lzo and 1 tane 1 - M* = 4 m . * P L2(-^?- + tana 4 cotan0' ) 1 (3.62) mp sin^G n r , p o J p 1 where a* , a' , 0' and are defined similarly as the unprimed variables, n p n p and n ? is computed from cos 2 0„ sin 2 0_ (3.63) (n^) 2 cos 2 p + sin 2 P lx 2 iy In the following sections we are going to concentrate on the analyis of the first type of modulator, but most of these results can be easily ex- tended to the second kind of modulators described by (3.61) and (3.62). It is also interesting to note that if a =0, the modulation M as n mn given by (3.60) is independent of material n . Substituting m from (3.55) we obtain 1 An 2 M = 4 - (2 — ^ ) (3.64) mn a n„ 2zo 40 Thus as long as n, > n_, and the angle of incidence is 0. =9 , the 12 in actual value of index of refraction n. does not influence the modulation M 1 mn 3.5. Beam intensity modulation. Effects of collimation and diffraction . Up to this point we have considered the case of a single ray, or in other words, a plane wave incident on an infinite plane boundary. Evidently this is a simplified version of a beam incident on a crystal surface. We can, however, extend our results to include this more general problem by considering the beam as an infinite number of superimposed plane waves (rays), then computing the modulation for each ray by the derived formulas and finally summing all these modulated rays to obtain the output beam intensity. Suppose that either because of lack of collimation or diffraction effects the beam intensity is distributed between the limiting incidence angles OL and ot . Since we are going to consider relative intensities, the input absolute intensity is unity. If w(a) is the distribution function, then it has to satisfy the relation J w(a) da = 1 (3.65) at! Note that w (a) = for a outside the interval a < a< CL. Let us define the relative beam modulation M and the relative beam intensity I by ■ / w(a) M(a,An)da (3.66) a 2 M °2 I -.; i - m = f wta) I(a,An)dC( (3.67) °1 41 From these equations we can compute M(An) and I (An) by using the formulas for M and M developed in section 3.4. One can consider various forms of mn mp w(0!) , but since we are looking for an estimate of the performance of a modulator, a pessimistic approximation can give us a sufficient answer. Thus we can assume that the beam intensity if uniformly distributed over a, i.e. w.. id) = constant. In the case of diffraction the actual beam is concentrated more in the center; for a parallel slit the intensity of the first lobe is distributed as 1 sin 2 y 2 V ~" jt y s w„(a) = - (3.68) where , 2* , a i +a 2 y = ( ) a - Jt b i 1/2 Z = r „.,° : , _. x (3.85) We may require Z to be some convenient impedance, (say 50 ohms), and we can solve (3.85) for b . b=Z ol yU Q w(w-a + E^a) £o (3.86) As we have noted before, these velocity matching conditions are effective only for frequencies below f , (3.80). As the modulation frequency is in- creased, the modulation becomes less effective. 3 . 7 Dimensional tolerance requirements ■ In order to determine the tolerances that must be met in the manufacture of a modulator, we have to examine the influence of inexact dimensions on the performance of the device. On Fig. 3.4 we have labeled the input and output windows A and C repsectively , and surfaces B and D are called the reflecting surfaces . Evidently all these surfaces should be optically flat, since any irregularity causes the deterioration of the collimation of the beam. The angular position of surfaces A and C are not critical, but these should be normal to the beam in order to reduce the reflection losses . A lack of parallelism in the reflecting surfaces B and D degrades the modulator performance in two ways: the angles of incidence for the successive i+8 reflections will vary, and in the case of high frequency modulation, the modu- lating and the light waves progressively slip out of synchronism. Let us assume that surfaces B and D are normal to the xy plane and the angle between them is 1r. Using the notation defined in section 3.2 we th find that the angle of incidence at the i reflection is a. = a, + (i - 1) r (3.87) where > 1, we can approximate (3.88) by a + (m-l)f M r M(a' , An)da' (3.89) QL Integral (3.89) is similar to (3.70), hence we can use the result of that integration. Then we obtain for M" M mn 8 fim tan 2 ' n 2z . n 2zo + (a + (m - 1) ipjcotanG - (- 2zo + a, cotanG ) 1 n From this equation we can conclude that the effect of Op is equivalent to an increase in the spread of the beam. From the microwave modulation point of view we find that nr- changes the effective optical phase velocity v^ v , thereby ox causing a mismatch between v and v ox mx 49 Fig. 3.7. Notation for calculating the effect of Y* . If O^is small, (/*-<< 1) , we can write for the change of the optical path after m reflections n AL = - ma c^rr Substituting (3.55) for m we obtain at. L Q a T T "l AL = - - cotanG t rr n, = - a n = -^ D^ (3.101) — 1 n b and b > D,_ (3.102) — b In order to avoid diffraction effects, we require that equation (3.78) be satisfied, hence 2 L < L = x- 5 (3.103) J2 " A n 2zo UNIVERSITY OF ILLINOIS LIBRARV If the microwave losses are appreciable in material n (and we assume that w and a are such that most of the microwave energy is carried in medium n ),then at some distance L we have E mz (L 3 ) = \ E mz (0) = E mZ (0) eX *> <" V 3 ) (3 ' 104) hence where ln2 L 3 =— (3.105) m % = — f^\ tan 8 1 (3.106) ^•m The quantity tan 5 is the loss tangent of material n . The optimum modulator dimensions can now be found by choosing a = a , height b from (3.86) consistent with (3.102) and length L from L = min (L , L 2 , L 3 ) (3.107) The thickness of the electro-optical material n is such that the n n interface does not disturb the optical fields at interface £ 1 o n l n , and that the light rays totally reflected on interface n h do not reach interface n, 1 n 2 again. 3.9. The choice of the operating temperature . It is well known that the electro-optic and the dielectric constants of the electro-optic materials are temperature dependent. In ferroelectric materials (BaTiOg, SrTiO , KDP etc.) both the electro-optic coefficients and (23) the microwave losses increase sharply near the Curie temperature. We would thus like to determine whether there exists a temperature which maximizes the figure of merit F. Repeating the expression (3.99) for F we have 53 r An 8L(1 - a si n L) F = n 2zo 2z n„ >- 2zo tr — w(w-a + £' a) ^o 1 1/4 m 1/2 (3.108) Assuming that the properties QI , , £.' , , n, of the isotropic material n do si 1 1 1 not vary rapidly with temperature, we can say that the temperature dependence of F is due to the variation of the properties of anisotropic material n , namely to the variation of n„ and n J 2z 2zo It should be noted that n_ , to the extent to which it is determined by 2zo J the electronic polarizability, should be almost independent of temperature. Thus, only n remains in (3.108) as a temperature dependent variable. Since the electrically induced birefringence arises fundamentally from a change of polarization, we expect that the temperature coefficient of the linear electro-optic follows that of the dielectric constant . In fact Pockels original formulation of the electro-optic effect was in terms of constants e. . relating the constants of the index ellipsoid to the induced polarization relation between Pockels e. . and our r. . is given by r . . = 6 ( 61 - 1.) e- . (24) The (3.109) Thus for material with high «£. we expect r , to be proportional to £? , and to the extent that e. . is temperature independent, to have the same temperature dependence as £* . A similar argument for the quadratic electro-optic effect suggests that 0. should be proportional to (g.') 2 . From (3.108) it is clear that F is maximum if An„ is at its maximum. n 2z By the above argument the maxima of An and e' occur for the same temperature For ferroelectric materials such as SrTiO , KDP etc, in which g r is dominated by a (T-T ) temperature dependence, the modulation is optimal near the Curie temperature T . It should be stressed again that although the losses increase sharply in medium n at the Curie temperature, this has little effect on driving power requirements, since material n has a very small volume . 3.10. Modulating bandwidth limitations. Since both the modulating and the optical signals propagate as TEM- like waves, it might be thought at first that the bandwidth of the device would be limited only by dispersion in the material parameters of materials n and n . For some materials this would imply a microwave bandwidth of more than 10 KMC. A more careful analysis reveals however that the bandwidth of the device is also limited by: 1. The microwave impedance of the structure. 2. The velocity synchronism condition is not satisfied for some higher frequencies. 3. The field intensity at the n n interfaces is reduced at higher J I £ frequencies . The first of these factors, the microwave impedance, has a practical significance in that it is difficult to achieve a broadband match between the modulator and the driving transmission line. There are however a number of standard microwave techniques which help to overcome this difficulty. Note : not the modulating frequency ' 55 The second factor, the variation of the microwave phase velocity v mx with frequency, was discussed in some length in section 3.6. The only way this problem may be eliminated is to choose material n in such a way that V€« ^1 n 2o (3.110) in which case w = a, and the synchronism condition is satisfied for all frequencies. This might not be possible however, for lack of suitable mater- ial n . As we have mentioned in section 3.6, for high frequencies the modulating field is essentially contained in medium n , with a cosinusoidal field dis- tribution in the y direction. Fig. 3.8 illustrates this case. y- 4) \ \ \ \ \ \ n o n i A E — rn // n 2 n \ o \ \ \ \ \ \ conductor plates Fig. 3.8. Representative microwave field distribution for very high frequencies. This difficulty can be overcome at the expense of increased modulating power requirement by the addition of layers of n. material between media n. 1 2 and n . The resulting sandwich type structure is illustrated in Fig. 3.9. . interface D y. & n 2 V / xx_ / / n. interface B Fig. 3.9. Modulator designed for improved high frequency response. It is interesting to note that the Kerr-cell type devices have a fundamental cutoff frequency due to the variation of the modulating field, (hence the phase velocity) across the cross section of the light beam. Only approximate velocity matching is possible for these devices. On the other hand, our intensity modulator does not have a similar cutoff frequency. Since the modulation occurs only at a boundary surface, exact velocity matching is possible and theoretically the only frequency limit is due to the dispersion of material parameters. CHAPTER IV BEAM DIRECTION MODULATION BY CONTROLLED INHOMOGENEITY 4.1. The theory of wave propagation in an inhomogeneous medium . It is generally very difficult if not impossible to obtain an exact solution for Maxwell's equations in an inhomogeneous material. In many cases however, an approximate solution is satisfactory. If for instance we can assume that in a region the spatial variation of the fields and the material constants are negligible within an optical wavelength, we can obtain an approximate solution which is analogous to the plane wave in a homogeneous medium. (This approximation is analogous to the WKB-method used in quantum mechanics.) This solution forms the basis of the geometrical optics. The reason why this approximation holds for a great number of problems is that most light beams are much wider than the wavelength, and since the approximation holds for the interior of the beam, only a small part (the periphery) of the beam will be erroneously analyzed. In the following analysis we intend to show that the path of a light beam can be determined from the material equations of the medium. Consider the general harmonic fields at a point r , v - i cot E(r.,t) = E (r) e J ~°~ . , (4.1) , x - i cot H(r ,t) = H (r) e J — o — in a non-conducting, non-magnetic, isotropic medium. It is well known that for a homogeneous medium the complex vectors E and H are given by E - E • e J c n( ^ * - } ^o -1 e ° (4.2) .CO , . H = H . e J c n( ^ * Y -> — o —1 57 58 where E , H are constant vectors, n = |e' and s is a unit vector specifying the direction of propagation. (a • r) = constant represents a plane in which the phase of the wave is constant. Similarly, it can be shown, that for an inhomogeneous material E and H will be solutions of Maxwell ^s equations (within the "geometrical optics" approximation) if E and H are of the form — o — o E =E l( r) e j c nS( *> -o -1 - (4.3) „ „ /„x J - nS(r) H = H n (r) e c - — o —1 where S(r) is the "optical path of the wave" and S(r) = constant represents the geometrical wave-fronts. Substituting (4.3) into Maxwell's equations and neglecting terms in accordance with the "geometrical optics" approximation, we find (4.4) (grad S(r) x H ) + c€ E =0 (grad S(r) x E ) - e|AH =0 E • grad S(r) =0 H • grad S(r) =0 From these equations we can obtain a condition on S(r) . Using the first two equations of (4.4) we find e n grad S x (grad S x E n ) + which can be written as ^o E, = (4.5) grad S . (grad S • E ) - (grad S) 2 • E + n 2 E^ = (4.6) The first, term of this equation vanishes (eq. 4.4) and since (4.6) must hold for all E, 59 [grad S(r)] 2 = n 2 (r) (4.7) This equation is known as the eikonal equation and it is the basic equation of geometrical optics. From equation (4.4) we see that both E and H are normal to grad S(r) ; we also know that the direction of the energy propagation, or the direction of the Poynting vector is also normal to both E and H . Hence it follows that the direction of the Poynting vector is the same as that of grad S(r) . Thus we find that the energy propagates along the normal trajectories of S(r) = constant surfaces. The geometrical light rays are then these orthogonal trajectories . Let us define the unit vector N along the ray dr(s) N = — (4.8) — ds where s is the arc length along the ray and r(s) is the position vector of a point on the ray. We have noted that grad S is in the direction of the ray and from the eikonal equation (4.7) we also see that write grad S = n. Therefore we can N = grad S = grad S (4 9) — | grad S| n hence from (4.8) and (4.9) it follows that dr(s) ds = grad S »r(s) (4.10) We now want to arrive at an equation where r(s) is given by n(r) only Let us differentiate (4.10) with respect to ds . 60 d d - (s) d Ts (n -^r } ■ d^ Lgrads(r(s))] (4.11) dr = (div grad S) * (ds ) . / ■> \ /grad S x (from 4.9) = div (grad S) • (- ) n = — grad [ (grad S) 2 ] = — grad [ n 2 ] from (4.7) 2n hence dr(s) — (n— - ) = grad n (4.12) ds ds This is the vector form of the differential equations for the light rays. (4.12) indicates that the rays through the medium can be computed from the knowledge of n(x, y, z) and the initial direction and position of the ray. For a homogeneous medium n = n and thus (4.12) reduces to dr(s) %- (n '— ) = (4.13) ds o ds hence , 9 d^r - — = (4.14) ds which is satisfied by r = as + Jb (4.15) where a and b are constant vectors. Thus (4.15) confirms the fact that the light rays are straight lines in a homogeneous medium. A second example can be the case when n depends only on the distance R from a fixed point 0, (let R = £J) n =n(R) (4.16) 61 In this case (4.12) becomes dr r dn — (n — ) = grad n = — — (4.17) ds ds R dr This equation is satisfied by 1 x ns = const. (4.18) This equation indicates that the path will be in a plane including the origin of the spherically symmetrical system. Another important parameter of the ray, the curvature vector K can also be determined from n(r) . Let us define >£ as the unit normal vector at a point r(s) on the ray and O as the radius of curvature at that point . Then K will be given by K = - V (4.19) I V n > n Fig. 4.1. Illustration of the bending of a ray in an inhomogeneous medium . K also can be given in terms of r(s) dr(s) d 2 r(s) K = T- < "3 > = Hg (4-20) — ds ds ds 62 dr We should note here that from the definitions of \T and — it — ds follows that these vectors are normal to each other. Rewriting (4.12) into an expanded form yields dr d 2 r Hi * di + n dT = grad n < 4 - 21 > and multiplying both sides scalarly by K gives Id d - 1 d2 - ^d^ ( ^-d^ )+( ^ )n dT- = ^ gradn.y (4.22) We find that the first term of this equation vanishes; (by the above remark.) Then since the second term of (4.22) equals to n (k I 2 , we have n | K | 2 = - grad n • V* (4.23) and this can be rewritten as K • K = (- grad n) ' (- v) (4.24) - - n ^ - since from (4.19) \k\ = — , we can write K | = | = ^ grad n • Y (4.25) hence K = (- grad n • ^) ' V (4.26) This equation can also be used for the computation of the equation of the rays through the medium. 4.2. Beam deflection by means of an electro-optic device . As we have seen in the previous section, the curvature of a light ray is directly related to the variation of the index of refraction in the medium We have also seen in Chapter 2 that the index of refraction in an electro- optic material depends on the imposed electric field. Let us then consider 63 the case when a non-uniform electric field is imposed on an electro -optic material and a beam of light is propagated through this medium. Based on our experience with the inhomogeneous , isotropic material treated in the last section, we expect that the beam will not propagate through this medium in a straight line. Since the electro-optic material in a non-uniform field behaves as an inhomogeneous, anisotropic medium, the results of the last section do not apply directly. In fact we cannot even define a ray for this case, since a unique trajectory for the propagation of optical energy cannot be found in general . If however, the polarization of the input ray is such that only a single wave propagates in the medium, we can use the previously derived results. As we have seen in Chapter 2, there are two major groups of electro-optical materials* in the first the linear, in the second the quadratic electro- optic effect dominates. It will be seen later that it is convenient to treat these groups separately. 4.2.1. Beam deflection by linear electro-optic effect . Let us consider the case when a slab of electro-optic material n is placed in a non-uniform electric field. The optical axis of the material n is oriented in the direction of the z axis. The light beam propagates approximately in the +y direction. Fig. 4.2 illustrates the situation. Let us assume that by some means, (later we shall go into more detail), we have generated a modulating electric field E in material n n . Let us further —ml assume that the y component of E , E is equal to zero, and that the other ' -m —my 6k z II n 1 / beam j* K (ths »-x y 'Qz? *o n Fig. 4.2. Light beam incident on an electro optic material, components of E do not depend on y. Also, E is chosen in such a way that in the neighborhood of the (xy) plane we have E (x.z) » E (x, z) for z = (4.28) mz ". mx ' The principal indices of refraction (see Section 2.3) are given by n. = n n _ (1 + - n 2 r E ) lx 10 2 10 mz n. = n (1 - - n 2 r E ) ly 10 2 10 mz (4.29) n n = n_ lz le where r equals r__ for uniaxial materials (KDP) and r = v. for cubic materials bo 41 (Cu CI) . The wave normal is in the y direction approximately and therefore the two allowed directions of polarization are x and z. From (4.29) it is evident that the optical field E will be influenced by E only if — o — m E = E • u (4.30) — o ox —x where u is the unit vector in the x coordinate direction. —x Under these conditions, (4 .28) - (4 .30) material n will behave as an "isotropic" (only a single wave propagates!), inhomogeneous material with 65 index of refraction n' n = n_ (E ), and we are in the position to apply the 1 lx mz rr J results of Section 4.1. From the differential equation for the ray (4.12) and the equation giving n 1 = n , (4.29), we can theoretically compute the ray for any function n (E ). A simple and very useful solution of (4.12) is when the ray describes a circular arc of radius R. In this case the radius of curvature will be constant and Q = R. Equation (4.25) gives in terms of n and v , hence our requirement is - = (— r— grad n' ) •: V = const. = - (4.32) ? V l X " R As we will see later the direction of grad n and V are nearly parallel . _3 Furthermore the changes in n' are expected to be less than 10 * and thus we can approximate n' by n . We can write then I = I grad n' I ■ — (4.33) R I 1 ' n o The value of grad n' I can be computed from (4.29) h [ l gradn'^ = - n^ (1 + - n^rE^) 1 ox 31 — X I"Vt! < 4 - 3 «> since n' = n n . 1 lx Equating (4.33) and (4.34) we find that 1 ! 2 mz (A r>^\ — - — n „r — v (4.35) R 2 10 x 3 x From this equation we can obtain the required E by integrating with respect to x E = — K — • x (4.36) mz Rn^ r 66 where the integration constant was chosen to be zero in order to reduce the field requirement . Thus if E is of the form shown in (4.36), the rays close to the (xvl mz i ■" plane (z = 0) describe circular arcs. We can compute the angular deflection from Fig. 4.3. *■ ^T Xi ~y Fig. 4.3. Illustration of a ray bent in a circular arc. The calculation of the angle of deflection can be performed as follows For a given modulating field distribution, E = Cx, we can compute R from mz (4.36). Also, from Fig. 4.3 we can write hence 9 s = 2(x. - x ) 1 o LC(n* )r 2(x. - x ) 1 o R 10' (4.36a) (4.37) We can also compute the maximum possible deflection with a given electro optic material by substituting the linear variation of n into equation (4.33) | = grad n'^x) ± n'fx.) - n'(x ) 11 1 o 10 n io (x i *o> £n'(x. )-An' (x ) 11 1 o n io (x r x o ) (4.37a) 67 Assume that the device is designed so that n',(x ) = n, _ and n', (x, ) = 1 o 10 11 n' (E *,) , where E _ is the break-down field strength of the material. Then 1 mB mB An'., = n',(x,) - n* (x ) and from (3.46a) and (3.47a) we have 1 max 11 1 o =i/2 1 max (4.38) max U 1 n io Thus we have derived the remarkable result that the maximum achievable deflection depends only on the material parameters: n_ n , r and the break- down strength. E _ . We can show that a suitable electric field can be mB generated by the potential field V , if V X > y * Z) = - Axz • A = Rn 2 r (4.39) Since all second derivatives of V equal to zero, V satisfies Laplace's m m equation, and thus the field obtained from V will be static field:.. The m components of E are obtained from — m E = - grad V = A(z u + x u ) (4.40) — m m —x — z The equipotential surfaces, hence the suitable electrode surfaces are given by - V z = r^- , for all 0< y < L. (4.41) A X The orthogonal curves to these equipotential surfaces give the direction of E , or the field lines of E . The equations of these lines are — m -m z = / x a - x 2 ' (4.42) o The constant I E lines will be concentric circles as given by 1 — m ' IE I 2 IE I 2 R 2 n* r 2 2 9 ~ m —ml 10 , N x 2 + y 2 = '- ' = (4=43) A 2 4 68 rH * CO co at r! o o •H Q w ■H •H u a. ex H Eh a> a a p d at Sh +j w < <5 N m 03 at s *■ c co C •H -H -H X 4-> (1 10 u 10 >. CD X rH (S !h u at & -H at O rH T3 g e •h at W| ID +J Q) OS ft Ifi ^ N N S X M E H s 5 M rH M M CO ^ w W CD r4 rH CM X i Fh o 0^ 0^ C i o O N CM N IN N CM N CM c a a c 1 + i 1 X o o o c c c c c c a X s N X N s s w N M (M w 1 O N 1 1 o rH u o cy 0^ G N CM N CM N est C c ". ■ i 1 IS) o o c c c c c c a •a o X N X N X N s e s s E e E H w w w W H O r-{ •H at U h ■P •H at O at at x cd -a •H CD o u t-i at C cm rH £} fi T3 M •h -a xj •H -H Q M 3 -H H M .Q at O G J O hJ ►H 3 3 £> o o- bfi C •H -P O 0) rH •H +-> O at h «H CD U O CD S -a CD rH at 69 4.2.2. Beam deflection using quadratic electro-optic effect . We turn our attention now to the case when material n is replaced by a quadratic electro-optic material n (Fig. 4.1). For these materials the most significant interaction between E and E occurs when these two vectors ° — m — o are parallel. In terms of E and n„ this can be stated as -m 2 n = n_ (E ) 2x 2x mx n = n (E ) 2z 2z mz (4.44) As we have done in the previous case, we require that E (x, 0) = E ' u (4.45) — m mz — z Then we choose the polarization of E such that — o E = E . u (4.46) — o oz — z Note that the major axis of the index ellipsoid for n is always in the direction of the modulating field E . In the above case(z^O) we — m have assumed that E is in the z direction, hence n„ is the major axis of n„ — m 2z J 2 ellipsoid. For the above case then the material n behaves as an "isotropic", inhomogeneous medium, and therefore we can use the results of Section 4.1 again with n' = n . From Section 2.3 we can obtain the relation for n 2z n» = n„ ~ n (1 - ^ n 2 o _ E 2 ) (4.47) 2 2z 2o 2 2o U mz In order to deflect the beam in a circular arc again, we have to require that - = — I grad n* I = const. = ^ (4.48) D I O 70 From (4.47) we can compute grad n', ()E, grad n' = - - n 3 Q 2E — r 2 2 2o ^ 1 mz ^ x hence n R 9 2E <) E mz 2o \ 1 mz b x (4.49) (4.50) Integration with respect to x gives n — (x - x ) = - — n 3 P , R o 2 2o 1 1 hence E should be of the form mz mz = (k?t— ) f^ 7 mz x 2o 1 (4.51) (4.52) The electric modulating field is then 1 E = ( — m Rn 2 — )2 [ >o 1 L -z r* 2 x - x u + ¥ x - x u — x o — z -> (4.53) The field E as given by (4.53) satisfies (4.52), curl E = 0, but div E = — m -m — m except for z = 0. Thus we may conclude that E is not a static field, and it cannot exist without a charge distribution source. 4.3. Beam distortion effects . In the previous section we have considered the path of a very narrow beam through the inhomogeneous material. Since in practical experiments the beam diameter is not negligible, we must examine what happens to these wider beams as they pass through the same material. There are also several other factors, which have to be considered in the case of a practical beam deflecting device. Some of the important questions are: 71 1 . How do the small angular variations in the input beam effect the output beam? 2. What are the consequencies of the "constant radius of curvature" bending? 3. What is the influence of the fact that grad n does not have uniform magnitude and direction for the different parts of the beam? 4. How do unwanted variations of n due to strains and temperature variations effect the output beam? These effects will be treated qualitatively in general, but simple quantitative results are given also whenever it is feasible. The main effect of the lack of collimation (or small angular derivations in the input) is that the width of the beam increases as it passes through the device. The diameter D 1 at the output is given by D' = D,_ + QL (4.54) b b where D is the beam diameter at y = 0, a is the angle between the extreme rays in the beam and L is the length of the device. If (Qli) is comparable to D we can conclude that: 1. the device must be designed larger in order to accommodate the larger beam D', and 2. the larger beam (D' ) will be more b b distorted . The effect of non-parallel, but constant radius bending is a minor effect in practical cases. Assuming that two parallel rays enter at y = (input face) D distance apart, and that R is the constant radius of curvature of both rays, then at y = L (output face) the angle between these two deflected rays will be y- 72 T D b . L R 2 (4.55) and the distance between the rays will be D" J b D " b = D b • cose = \ h - w = \ (4 - 56) for L « R (which is the case in practice) . -5 Since normally R » D , / > is a very small angle ( *— 10 ) . The most serious of all these factors is that our approximation of treating the material as an isotropic" material breaks down for points where z = 0. The approximation gets worse as we increase z , hence it is the worst for the outer portions of beam. This problem is less important for linear, uniaxial materials, (KDP, ADP etc.) than for cubic media. For KDP the change in n due to E occurs only normal to the optic axis (z direction*.), hence only the z component E of E influences n., . Since E is inde- mz — m lx mz pendent of z (4.36) all rays will have the same radius of curvature, independent of (x , z ) coordinates on the input face ( y = ) . For cubic materials the case is much more complicated, since the permitted directions of E in the medium depend on the directions of E . As we have — o — m seen (4.42) the lines of E describe a hyperbola; the direction of E at -m J -m any point is given by the slope of this hyperbola, which is given by x fx 2 - x 2 o ' f(x) = , - X (4.57) Figure 4.4 illustrates this case. The index ellipsoid for material n has its major axis always in the direction of E . At point P., there are two directions of permitted optical —ml vectors E in the direction of E and E , normal to E . The index of — oa - m — ob — m refraction for these waves are written as n and ru respectively. From 73 V = const., equi potential surface , E field lines x b., Fig. 4.4a. Illustration of E field for the linear case — m b. Components of E vector. Section 2.3.2 we find that n is independent of E a -in but n depends on E — m Namely — m (4.58) n = n a o Since E increases in the radial direction, (Fig. 4.4), grad n -m b will be also in this direction (approximately), and the ray going through P will be bent in a plane which includes P, and the y axis, and not in a plane parallel to the xy plane. We also find that component E is not deflected — oa at all, since n = n . Thus we expect that portions of the beam which are a o further away from the xy plane (z = 0) are considerably distorted. Figure 4.5 gives the approximate shape of the deflected beam at the output face of the device, x and x are the positions of the centers of the input and output beams respectively. Actually some part of the beam 7^ output projection of an extreme ray x Fig. 4.5. Illustration of beam distortion for linear, cubic materials. corresponding to E vectors is smeared , between the two positions x and — oa c o x . The angle o between the extreme rays on the output beam can be given by (4.59) D, D^ (x n - x ) <^= ^ . e = -* -i 2 x x L o o V x R o From (4.59) we observe that for a desired deflection 9 we can improve the collimation of the output beam by reducing D and increasing x . The case of the quadratic cubic materials is even more complicated, since E is a more complicated function of the coordinates x, z. Arguments given for the linear cubic material hold approximately for this case also, thus there is a portion of the beam which is not bent and the other component is bent toward the x direction. The major difference is that for quadratic materials the roles of n and n are interchanged. Component E and n cl D Oa Si will be influenced (refer to Fig. 4.4/b) by | E I , and E and n will not be changed much by E , since we expect O » O (10) The constant curves are also circles, but these are not concentric, as was the case for linear media. For z — I E j is given by m E I 2 = A 2 f (x - x) + — -ml L o x -: which has the normal form r \SJ f + z 2 = E 2 ^n 2A* J 75 (4.60) (4.61) Figure 4.6 illustrates the equal E lines; the gradient of n will be approximately normal to these curves. output beam / projection of an extreme ray Fig. 4.6. Illustration of beam distortion for quadratic, cubic materials. From these arguments we may conclude that the distortion is greater for quadratic materials than for linear media. The stress and temperature effects are also important, since the index of refraction depends strongly on the temperature. In all probability, any unwanted, random gradient of n will scatter the beam, and it might obscure the bending due to the electro-optical effect . The stresses might be due to 76 uneven crystal growth; the temperature gradient can be due to uneven conduction of the heat generated by dielectric losses . This factor is very stronly influenced by the design and the manufacturing techniques of the device. The modulating material must be chosen such that even at the very high field strengths required for operation, the temperature of the medium does not rise significantly. ( i. e . the conductivity must be negligable. ) CHAPTER 5 EXPERIMENTAL RESULTS The main purpose of the experimental part of this work is to show that the modulation effects predicted by theory actually occur in practical devices. Due to this objective and to the fact that our manufacturing and test facilities were limited, only three simple devices were constructed and tested experi- C 25) — f 97^ mentally. The results of these experiments can be called satis- factory, since there is a good agreement between theoretical predictions and experimental measurements. Nitrobenzene was chosen as the electro-optic material for a number of reasons: it is readily available, and it has the highest electro-optic con- stant among liquids. The use of a liquid electro-optic material also simplified the construction of the devices considerably: it was relatively simple to obtain an optical quality interface between the isotropic material and nitro- benzene. Glass was used as the isotropic medium, since it was available with various indices of refraction n. The modulating fields were generated by gold plated electrodes immersed into the nitrobenzene liquid. (Refer to Fig. 5.1). The modulating signal was supplied by a magnetron pulser capable of delivering up to 10 KV pulses of duration 1-10 M-sec, with a repetition rate of 60 pulses per second. A He Ne gas laser emitting at \ = 6328 A provided the input light beam and an RCA photomultiplier detected the intensity modulation of the output beam. The positioning of the device was accomplished by the accurate turn- tables and other mechanisms of a surveying instrument . 77 78 5.1. Experiments with the light intensity modulators . The intensity modulators described in Section 3.4 were tested in the experimental apparatus described above. The Type 1 modulator, studied in detail in Chapter 3, consisted of a denser isotropic and a less dense, electro- optic material, (glass n n =1.65 and nitrobenzene n = 1.55 respectively.) (Refer to Fig. 5.1). The dimensions of the device were L = 70 mm and a = 1 mm. The device was operated in the normal modulation (M ) mode, since both the nm modulating and the optical fields were in the z direction. The other constants of the device were measured to be = 70° and m = 28 . The effective spread n -3 of the beam was estimated ( 0L- - Cd, ) = 5.10 radians from the properties of the output beam. The fact that only 22% intensity modulation was produced by this device was attributed to this large spread of the beam. Figure 5.1 shows the construction of this device. output beam ■ electrode. Fig. 5.1. Illustration of Type 1 intensity modulator 79 Equation (3.73) and families of curves similar to Fig. 3.5 were used for the calculation of the theoretically predicted modulation. From Fig. 5.2 we can see that the experimental measurements agree reasonably well with the theoretical results. M inn O.2.. 0.1- / experimental 6 W Fig. 5.2. Comparison of theoretical and experimental results for the Type 1 intensity modulator. We should point out here that the beam spread can be reduced to below -4 10 radians by careful manufacture of the device and by the use of a suit- able laser source. The intensity modulation for this case should be much larger (by an order of magnitude) than our observations. A second light intensity modulator, the dual of the one described above, was also constructed and tested. In thisfType 2)modulator the electro- optical material nitrobenzene has the higher density, and the light beam travels in this medium. The constants of the device were measured to be L = 65 mm, a = 1.1 mm, m = 8, 9 = 78 . A maximum intensity modulation n 80 fi v of 55% (M = 0.55) was observed for a modulating field of E = 9.10 — . mn —mm The beam was better collimated in this case and thus the observed modulation effect was also higher. The main reason for using pulsed modulating field was to separate the electro-optic modulation effects from slower, but significant temperature gradient effects. The index of refraction of the nitrobenzene depends strongly on the temperature and thus static experiments were: not suitable to show the electro-optical modulation without extensive purification of the nitrobenzene . 5 .2 Experiments with the beam bending device . It should again be emphasized that the objective of these experiments was to prove the feasibility of these modulation methods, rather than to develop a practical device. A simple beam bending device was constructed and tested using the apparatus described above. (Refer to Fig. 5.3) the non- uniform electric field was obtained by the use of the fringe field of the rectangular electrodes. This field distribution evidently does not satisfy the conditions (grad n = const.) developed in Chapter 4 for constant radius bending, but we expect that the peak grad n was high (since the electric field fell off rapidly in the x direction, (Fig. 5.3).) light beam field lines electrode Fig. 5.3. Illustration of the beam bending device 81 We expected (and observed) considerable beam distortion due to non-constant grad n for different portions of the beam. The experiments consisted of sending a narrow light beam close to the edges of the electrodes through the nitrobenzene and observing the distribution of the beam by a small pinhole on the face of photomultiplier mask. As the modulating field was turned on, we found that the beam distribution shifted toward the electrodes . The de- flected distribution (which was considerably distorted) was mapped by the pinhole arrangement for several values of the modulating field. A maximum deflection of 5 mm was observed at a distance of 1500 mm from the output face of the device, which corresponds to an angle of deflection 9 = 0.0033 radians. Again, a short (1 Usee) modulation pulse was used in order to reduce the additional distortion caused by the temperature gradients in the nitrobenzene. The length of this device was 70 mm and the separation between the electrodes was 0.5 mm. The maximum modulation occurred for 2 KV, and for higher voltages only the distortion increased. This was expected since for these voltages the deflected portion of the beam travelled outside of the active fringe field region, hence no further deflection was possible. A more sophisticated experiment was performed independently by V . J. Fowler, et (28) al . ' who used KDP as the electro-optic material and obtained a deflection of = 9 or 9 - 0.0026 radian at a modulating voltage of 10.4 KV . CHAPTER 6 CONCLUSIONS From the detailed analysis of the modulators and from the corresponding experimental results we may conclude that these modulation schemes are indeed feasible and that practical modulators operating on these principles can be constructed. The underlying theory of these modulators has been established and some of the important practical problems have been considered in sufficient detail, so that a realistic design procedure could be presented. As we have pointed out earlier, the intensity modulator treated in this work compares favorably with the existing light modulators dm some aspects. The innovation of this device may be summarized by saying that the modulation process has been changed from a volume modulation to a boundary modulation, in that instead of changing the properties of a volume of a modulating medium, we modulate only the boundary conditions. This results in a reduction of the active volume, namely we need to change the properties of a thin layer (only several optical wavelength thick) of material only. This fact should result in reduced power absorption in the device. Furthermore, whereas some properties of the modulating materials cause serious problems to the (basically) phase modulator devices, our modulator is relatively insensitive to these effects. The random strain which exists in some electro-optic materials (CuCl, Sr Ti ) seriously degrades the performance of these modulators ' , since the light beam actually travels through these materials. On the other hand, the light beam in our device travels in an isotropic material only, and since the effects of small variations in n due to the random stresses are not cumulative, the performance of the device will not be affected seriously. 82 83 Finally, let us compare the various available electro-optic materials as the modulating materials in our modulators. For M mode of operation mn * we found in Chapter 3, (3.64), that the modulation is unaffected by the choice of the index of refraction n of the isotropic material. Let us assume that we construct a device with dimensions L = 100 mm, a = 1 mm, b = 1mm and we specify n only to the extent that n > n . Then from equation (3.64) and Table 3.1 we can compute the modulating field intensities E* necessary to produce 50% modulation (M^ =0.5) for a well collimated * beam.( The modulating voltage V = E m b is also given in Table 6.1. ) We can also compute the maximum field intensity E** needed to deflect -2 the beam in the beam bending device by = 10 radian. (Constant radius bending is assumed, equation (4.38) is used in computations.) The results of these computations are given in Table 6.1. From this table we can conclude that intensity modulators operating with relatively low voltages (less than 100 volts) can be constructed from available electro- optic materials. The advantage of the quadratic materials for beam bending is also apparent from this table. Qk Intensity Modulation CM = 0.5) mn Beam Bending (9 = 0.01) Materials Temperature E * m V * m E ** m units 10 s . - m V m KDP 293 ADP 293 Cu CI 293 ZnS 293 BaTiO 393 403 SrTiO 77 293 KTiO 4 77 nitrobenzene 293 1.06 106 6.85 • 1.29 129 8.3 • 0.70 70 4.5 • 1.16 116 7.4 • 0.109 10.9 8.65 • 0.222 22.2 16.4 . 0.93 93 74 3.65 365 29ft 0.241 24 19.3 • 1.58 7 .9.9 158 799 126 P180 10 10 10 10 10 4 10 10 t 10 10 io' 10 Table 6.1 Comparison of the electro-optic materials. 85 BIBLIOGRAPHY 1. E. Bergstrand, Encyclopedia of Physics, edt . S. Flugge (Berlin, Springer) V. 24, (1956) . 2. A. L. Schawlow and C.H. Townes, Infrared and optical masers, Phys . Rev., Vol. 112, pp. 1940-1949, Dec. 1958. 3. P. Kaminow, Microwave modulation of light by the electro-optic effect. NEREM Record, V. 3, p. 117, Nov. 1961. 4. D. F. Holshouser, H. von Foerster and G. L. Clark, Microwave modulation of light using the Kerr effect, J. Opt. Soc. Am. V. 51, p. 1360, Dec. 1961. 5. R. H. Blumenthal, Design of a microwave-frequency light modulator, Proc. IRE, V. 50, p. 452, April 1962. 6. S. E. Harris, et . al . , Modulation and Direct Demodulation of Coherent and Incoherent Light at a Microwave Frequency. Applied Physics Letters, V. 1, Oct. 1962, p. 37. 7. Rignod, W. W. and Kaminow, I. P., Wide-Band Microwave Light Modulation, Proc. IEEE 51, Jan. 1963, p. 137. 8. White, R. M. and Enderby, Electro-Optical Modulators Employing Inter- mittent Interaction, Proc. IEEE, 51, Jan. 63, p. 214. 9. Peters, C. J., Gigacycle Bandwidth Coherent Light Travelling-Wave Phase Modulator, Proc. IEEE, Vol. 51, Jan. 1963, p. 147. 10. M. DiDomenico, Jr., and L. K. Anderson, Broadband Electro-Optic Traveling Wave Light Modulators. B.S .T.J . .V.42, pp. 2621-2678, Nov. 1963 . 11. Born, M. and Wolf, E., Principles of Optics, Pergamon Press, New York, 1959. 12. Jenkins, F. A. and White, H. E., Fundamentals of Optics, McGraw-Hill, New York 1957, 3rd. ed . 13. C. E. Springer, Tensor and vector analysis, The Ronald Press Company, New York, 1962. 14. Mason, W. P., Optical Properties and Electro-Optic and Photoelastic Effects in Crystals Expressed in Tensor Form, B.S. T.J. 29, April 1950, p. 161. 15. W. P. Mason, Piezoelectric Crystals and Their Application, to Ultrasonics , Van Nostrand Co., New York. 86 16. Max Born and Ken Thang, Dynamic Theory of Crystal Lattices, Clavendon Press, Oxford, 1954. 17. R. 0*B. Carpenter, The electro-optic effect in uniaxial crystals of the dihydrogen phosphate type. J. Opt. Soc . Am. Vol. 40, p. 225, April 1950. 18. I. P. Kaminow, Microwave modulation of the electro-optic effect in KH PO , Phys. Rev. Letters, Vol. 6, p. 528, May 1961. 19. Kaminow, I. P. and Harding, G. 0., Complex Dielectric Constant of KH PO at 9.2 Gc/sec, Phys. Rev., 129, Feb. 1963, p. 1562. 20. B. H. Billings, The electro-optical effect in uniaxial crystals at the type XH PO . ( Theoretical) J . Opt. Soc. Am., V. 39, p. 797, Oct. 1949. 21. Kerr, J., On a New Relation Between Electricity and Light. Philosophical Magazine (London, England) 4th Series, V. 1, p. 337, 1875. 22. Kaminow, I. P. and Liu, J., Propagation characteristics of Partially Loaded Two-Conductor Transmission Line for Broad Band Light Modulators, Proc. IEEE, 51, Jan. 1963, p. 133. 23. Landamer, R., Loss Tangent in a Ferroelectric Above an Order Disorder Transition. Bull. Am. Phys. Soc. 8, Jan. 1963, p. 61. 24. Pockels, F., Lehrbuch der Kristalloptik, B. Teubner, Leipzig, 1906. 25. Technical Progress Report, Section III, Digital Computer Lab., U. of I., Urbana, Illinois, July, 1963. 26. Technical Progress Report, Section III, Digital Computer Lab., U. of I., Urbana, Illinois, August, 1968. 27. Technical Progress Report, Section III, Digital Computer Lab., U. of I., Urbana, Illinois, September, 1963. 28. V. J. Fowler, et . al . , Electro-Optic Light Beam Deflector, IEEE Proc, V. 52, N. 2, p. 193, February 1964. 29. G. K. Ujhelyi, and S. T. Ribeiro, An Electro-Optical Light Intensity Modulator, Proc. IEEE, 52, July, 1964. p. 845. UNCLASSIFIED 87 Security Classification DOCUMENT CONTROL DATA - R&D (Security classification of titlt. body ol abstract and indexing annotation must 60 entered when the overall report ia claaailied) I. ORIGINATING ACTIVITY (Corporate author) Digital Computer Laboratory University of Illinois Urbana, Illinois 2a. REPORT SECURITY CLASSIFICATION UNCLASSIFIED 26 CROUP 3. REPORT TITLE NOVEL ELECTRO-OPTICAL LIGHT MODULATORS 4. DESCRIPTIVE NOTES (Type of report and Inclusive dates) Technical Report 5 AUTHORfS,) (Last name, first name, initial) Ujhelyi, Gabor K. 6- REPORT DATE October 26, 1964 la- TOTAL NO. OF PACES 9 2 lb. NO. OF REFS 29 8a. CONTRACT OR GRANT NO. Nonr 183M15) 9a. ORIGINATOR'S REPORT NUMBERfS.) 6. PROJECT NO. d. Report No. 170 96. OTHER REPORT t*0(S) (Any other numbers that may be assigned this report) None 10. AVAILABILITY/LIMITATION NOTICES Requesters may obtain copies of this report from DDC. Additional copies may be obtained from the Digital Computer Laboratory, University of Illinois, Urbana. 11. SUPPLEMENTARY NOTES None 12. SPONSORING MILITARY ACTIVITY Office of Naval Research 230 North Michigan Avenue Chicago 1. Illinois 13- ABSTRACT Since the availability of a coherent, well collimated and powerful light beam, there is a great demand for efficient, broadband light modulators. To the already well developed family of Kerr-cell type devices we propose to add two novel electro-optical devices: a direct intensity modulator and a light beam deflecting device. To provide a firm basis for the analysis of these devices, the theoretical background of the electro-optically induced birefringence is examined in detail. Starting . from the anisotropic material equations and Maxwell's differential equations, an analytical derivation of this birefringence is given. Furthermore it is shown how the electro-optical effect influences the material equations, hence its effect on the birefringence is derived. The experimentally determined electro-optic effect is tabulated for the most important materials. The new intensity modulator operates on the principle of electro-optically controlled partial reflections. The light beam to be modulated traverses a trans- parent isotropic material of index of refraction n = n-j_ and it is incident on the electro-optic material with n = rv> (n-^ > ru). The reflected amplitude (hence the intensity!) depends on the index of refraction ru, which in turn is controlled by the modulating electric field through the electro-optic effect. A detailed study of the theoretical and practical aspects of this intensity modulator reveals that its performance is in many ways comparable to the most sophisticated of the (Continued on following page) DD FORM 1 JAN 64 1473 UNCLASSIFIED Security Classification 88 13. ABSTRACT (Continued) Kerr- cell type devices. A systematic and quantitative procedure is presented for the optimized design of a broadband intensity modulator. The modulating frequency of the device is fundamentally limited only by the dispersion of material parameters. The beam deflecting device uses the fact that a variation of the index of refraction n causes a light beam to change direction of propagation. The desired variation of n (hence the deflection of the beam) can be achieved by placing an electro-optical material in a suitable electric field distribution. This field distribution is determined for the case of minimal beam distortion for linear and quadratic electro-optic materials. The experimental results obtained with three devices support the theoretical predictions. Intensity modulation up to 50 per cent and a deflection of .003 radian were observed in the laboratory using nitrobenzene as the electro- optic material. UNCLASSIFIED 89 Security Classification 14- KEY WORDS LINK A ROLE WT LINK B LINK C "electro-optical devices" "light intensity modulator" "light beam deflector" INSTRUCTIONS \. ORIGINATING ACTIVITY: Enter the name and address of the contractor, subcontractor, grantee. Department of De- fense activity or other organization (corporate author) issuing the report. 2a. REPORT SECURITY CLASSIFICATION: Enter the over- all security classification of the report. Indicate whether "Restricted Data" is included. Marking is to be in accord- ance with appropriate security regulations. 26. GROUP: Automatic downgrading is specified in DoD Di- rective 5200. 10 and Armed Forces Industrial Manual. Enter the group number. Also, when applicable, show that optional markings have been used for Group 3 and Group 4 as author- ized. 3. REPORT TITLE: Enter the complete report title in all capital letters. Titles in all cases should be unclassified. If a meaningful title cannot be selected without classifica- tion, show title classification in all capitals in parenthesis immediately following the title. 4. DESCRIPTIVE NOTES: If appropriate, enter the type of report, e.g., interim, progress, summary, annual, or final. Give the inclusive dates when a specific reporting period is covered. 5. AUTHOR(S): Enter the name(s) of authors) as shown on or in the report. Enter last name, first name, middle initial. If military, show rank and branch of service. The name of the principal author is an absolute minimum requirement. 6. REPORT DATE: Enter the date of the report as day, month, year, or month, year. If more than one date appears on the report, use date of publication. 7a. TOTAL NUMBER OF PAGES: The total page count should follow normal pagination procedures, i. e. , enter the number of pages containing information. 76. NUMBER OF REFERENCES Enter the total number of references cited in the report. 8a. CONTRACT OR GRANT NUMBER: If appropriate, enter the applicable number of the contract or grant under which the report was written. 86, 8c, & 8cf. PROJECT NUMBER: Enter the appropriate military department identification, such as project number, subproject number, system numbers, task number, etc. 9a. ORIGINATOR'S REPORT NUMBER(S): Enter the offi- cial report number by which the document will be identified and controlled by the originating activity. This number must be unique to this report. 96. OTHER REPORT NUMBER(S): If the report has been assigned any other report numbers (either by the originator or by the sponsor), also enter this number(s). 10. AVAILABILITY/LIMITATION NOTICES: Enter any lim- itations on further dissemination of the report, other than those imposed by security classification, using standard statements such as: (1) "Qualified requesters may obtain copies of this report from DDC " (2) "Foreign announcement and dissemination of this report by DDC is not authorized. " (3) "U. S. Government agencies may obtain copies of this report directly from DDC. Other qualified DDC users shall request through (4) "U. S. military agencies may obtain copies of this report directly from DDC Other qualified users shall request through (5) "All distribution of this report is controlled. Qual- ified DDC users shall request through If the report has been furnished to the Office of Technical Services, Department of Commerce, for sale to the public, indi- cate this fact and enter the price, if known. 1L SUPPLEMENTARY NOTES: Use for additional explana- tory notes. 12. SPONSORING MILITARY ACTIVITY: Enter the name of the departmental project office or laboratory sponsoring (pay- ing (or) the research and development. Include address. 13- ABSTRACT: Enter an abstract giving a brief and factual summary of the document indicative of the report, even though it may also appear elsewhere in the body of the technical re- port. If additional space is required, a continuation sheet shall be attached. It is highly desirable that the abstract of classified reports be unclassified. Each paragraph of the abstract shall end with an indication of the military security classification of the in- formation in the paragraph, represented as (TS). (S), (C), or (U). There is no limitation on the length of the abstract. How- ever, the suggested length is from 150 to 225 words. 14. KEY WORDS: Key words are technically meaningful terms or short phrases that characterize a report and may be used as index entries for cataloging the report. Key words must be selected so that no security classification is required. Identi- fiers, such as equipment model designation, trade name, military project code name, geographic location, may be used as key words but will be followed by an indication of technical con- text. The assignment of links, roles, and weights is optional. UNCLASSIFIED Security Classification UNIVERSITY OF ILLINOIS-URBANA 510 84 IL6R no C002 no 160 170(1964 Report/ 3 0112 088398166