Sunteți pe pagina 1din 13

Home

Search

Collections

Journals

About

Contact us

My IOPscience

Leaky waves in planar optical waveguides

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 1975 Nouvelle Revue d'Optique 6 273 (http://iopscience.iop.org/0335-7368/6/5/304) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 67.207.12.4 The article was downloaded on 10/12/2011 at 01:40

Please note that terms and conditions apply.

Nouv. Rev. Optique, 1975, t. 6, no 5, pp. 213-284

DEPARWYTOF ELECTRICAL ENGINEERING AND ELECTROPHYSICS POLYTECHNIC INSTITUTE OF NEW YORK


333 Jay St., Brooklyn, N . Y . 11201 f U . S . A . )

LEAKY WAVES IN PLANAR OPTICAL WAVEGUIDES (*I


bv
I KEYWORDS:
Leaky waves Integrated optics
MOTS CLES :

T. TAMIR

Ondes de fuite Optique integree

Ondes de fuite dans les guides optiques plans.


RkSUME : On presente les proprietts de base des ondes de fuite ainsi que leurs applications, presents et futures, en optique integree. Tout dabord on montre, quen plus des ondes de surface, les champs excitts par des sources reelles disposees le long dune couche mince comportent aussi ces ondes de fuite. Les champs peuvent se decrire en termes dondes planes inhomogenes qui se reflechissent sur les faces de la couche mince en rayonnant de lenergie vers les milieux exterieurs. On etudie ensuite les champs dondes de fuite gentrb par des systemes multicouches ou g r i o diques et on discute leur r61e dans les dispositifs de couplage. Pour terminer on presente certninc dispositifs a coefficient de fuite

variable permettant d Obtenlr un grand nombre de rranstormations faisceaux. SUMMARY : The basic properties, applications and future potential uses of leaky waves in integrated optics are presented. It is first noted that, in addition to surface waves, the fields excited by realistic sources along thin films also include leaky waves. These fields can be described by inhomogeneous plane waves that bounce between the film boundaries and radiate energy into the exterior open regions. Leaky wave fields on multilayered and periodic media are examined and their role in beam and waveguide couplers is discussed. Finally, the construction of structures having variable leakage is described and their capabilities in providing a wide range of beam transformations are outlined.

1.

- INTRODUCTION

The subject of waves guided by open conhgurations was studied in acoustics [l] and electromagnetics [2] already towards the end of the last century. These waves played a most important role in microwave technology, which employed extensively waves guided either by closed structures (e.g., modes propagating inside metallic waveguides) or by open configurations (e.g., surface and leaky waves on long antennas [3]). In optics, the utilization of these waves came relatively later, with surface waves having been considered first along fibers [ 4 , 51 and then in planar films [6, 71 ; leaky waves were thereafter examined in the context of beam-to-surface-wave couplers [8-111, in lateral. beam shifts [lo, 111 and fn other related optical processes [12] which occur in the presence of thin dielectric films, as well as in fibers having circular cross-section [ 131. Currently, both surface and leaky waves fulfil important functions in the area of integrated optics [ 141, where their application increases
(*) Invited paper presented at the Colloquium on (( the Optic of Guided Waves D, (Paris, April 8-1 1,1975) under the title (( leaky-wave

field configurations in optical waveguides D.

so rapidly that it may rival their use in microwave technology. Because of their earlier utilization and broader application, surface waves are familiar to physicists and engineers specializing in optics. By now, discussions of their properties can be found in many recent articles and text-books, of which references [14-17] represent only a few of the latest publications on this subject. In contrast, leaky waves are less well understood, because their theory is more complicated and their application is more restricted than that of surface waves. The aim of this paper is therefore to review the basic theory of leaky waves, to present their most significant applications in the area of integrated optics and to discuss some of their features which can possibly lead to future developments. Readers interested in additional material concerning leaky waves should consult references [I 1-19] for further details. Although leaky waves have been investigated also in circular geometries 1131, we shall discuss here leaky waves in two-dimensickd planar configurations only, because the latter are conceptually simpler. The extension to three-dimensiona1 non-planar situations can thereafter be carried out: such a generalization
~

274

T. TAMIR

Nouv. Rev. Optique, 1975, t. 6, no 5

involves greater mathematical complexities but no new physical aspects. We start with some basic considerations in Section 2 and mention that leaky waves appear as complex solutions of boundaryvalue problems involving sources. We then construct leaky wave fields in Section 3 by using simple arguments involving inhomogeneous plane waves. The properties of these fields are discussed in Section 4 which considers both multi-layered and periodic configurations. Practical applications of leaky waves are presented in Section 5, where beam and waveguide couplers are discussed. The lateral displacement of beams incident on leaky wave structures is next treated in Section 6, which uses simple power flux arguments to explain a rather complex phenomenon. Finally, Section 7 discusses leaky wave structures having variable characteristics and speculates on the future potentialities of these structures for application to integrated-optics devices.
2. -FIELDS

In general, the solution of F for given s is not easy to find. One therefore often seeks, at least as a first step, solutions of F in the absence of sources, i.e., s = 0, in whioh case Eq. (1) reduces to the homogeneous Helmholtz equation (2)

(Vz + k2) F

=0,

I N PLANAR STRUCTURES

To put leaky waves into proper perspective, we consider first the basic electromagnetic problem which is posed when solving for the optical field in planar configurations. Such configurations usually consist of a stratified multi-layered structure having an upper planar boundary exposed to a semi-infinite medium, which is usually air. As shown in figure 1, we take this upper boundary so as to coincide with the xy plane. To have an optical field, it is necessary that a source be present ; this may be either a laser beam, an aperture illumination or any other form of radiant energy, which is described schematically by the three arrows in figure 1 .
2.1. The field excited by sources
In the presence of a given monochromatic source having a time variation exp(- io?) implied but suppressed, the optical field excited by the source will be found by solving the inhomogeneous Helmholtz equation
(1)

subject to the same boundary conditions as before. In this case, the solutions are in the form of guided waves which can be obtained by solving a transverseresonance relation [la or by using other well known methods. However, these guided waves cannot actually appear unless a source s is present. Phrased differently, the foregoing statements imply that guided waves, as obtained by considering Eq. (2) only, represent fields which can potentially exist provided that a suitable source is brought in so as to excite such fields. To better understand what fields actually appear along a planar structure, it is therefore necessary to examine the excitation problem, viz. the solution of F in Eq. (l), with s # 0, as discussed below. 2.2. The longitudinal representation of the fields Because the geometry shown in figure 1 is invariant with respect to both x and y, this solution can be written, for all points z > 0, in the form
m

(3)

F = J-mf(k,) exp[i(k, x

+ k, 41 dk,

which is basically a Fourier transform for F with respect to k, and (4) k;

+ k,Z = k 2 ,

(V2 + k2) F

= S,

where F is any scalar component of the electromagbetic field, k = 2 n/A is the propagation factor in free space and s = s(x, z ) is the source distribution, which is assumed known; to restrict the discussion to two-dimensional fields, we also assume that s(x, z ) is independent of y , i.e., a/ay = 0. The solution for F = F(x, z) must, of course, satisfy field continuity conditions at the boundaries, which may include boundaries additional to that shown at z = 0 in figure 1. A radiation condition stipulating that power is outgoing at infinity is also needed.

where k, is selected so that it is positive real or positive imaginary for k real, i.e., only lossless media are examined. The extension to lossy media is straightforward [19] but is not considered here. It is evident that F in Eq. (3) consists of a continuous infinite spectrum of waves having an amplitude density f(k,) exp(ik,x) and varying as exp(ik,z) along z, i.e., they progress with propagation factor k, and travel in a direction normal to the boundary z = 0. Because this spectrum of waves is given by the continuous real variable k,, which is itself associated with the longitudinal coordinate x, the representation of F in Eq. (4) is often referred to as a longitudinal representation [ 19, 201. The disadvantage of this representation is that, although relatively simple, it does not show explicitly any surface-wave fields or any other guided fidds.

Source

i
Planar boundary

2.3. The transverse representation of fields


To explicitly obtain the surface waves excited by actual sources, it is necessary to modify the expression for F in Eq. (3). By using a contour deformation [19], the integral along the real axis in the complex k, plane can be transformed along a path that yields an

\! I '
7/////////

///?//I// +Y. /

FIG. 1. - Geometry for planar Configurations.

Nouv. Rev. Optique, 1975, t. 6, no 5

T. TAMIR

275

integration along all real values of k, (instead of real k,). In the process, a number of pole singularities are captured between the original (real kJ and the deformed (real k,) integration contours, thus obtaining

J-m

be accounted for in the first (continuous spectrum) term of Eq. (5), in which case this term can no longer be neglected. To deal with such a possibility, it is then convenient to transform Eq. (3) by using a contour deformation which is different from that used in arriving at Eq. (5). This different deformation is often obtained by using a steepest-descent path, as discussed in references [19-221. For the present purpose, we shall observe only that, as a result of .such a deformation, the field F can be written as

where

+ 2 xi C f(kxn) exp[i(km x + kzn z)]


n

Now, F in Eq. (5) consists of two terms, of which the first represents a continuous spectrum of waves having an amplitude density g(z, k,) and varying as exp(ik, x) along x, i.e., these waves progress with propagation factor k, and travel parallel to the boundary z = 0. The second term of F in Eq. (5) stands for the summation of residues around poles located at k,, in the complex k, plane (or at k,, in the complex k, plane). Each of the term in the summation represents a guided wave, usually a surface wave for which k , is positive imaginary and k , is real, so that propagation is again along x. Because F in Eq. ( 5 ) is given in terms of a spectrum with respect to k,, which is associated with the transverse variable z, this representation is referred to as a transverse representation [19, 201. The advantage of the transverse representation is that it exhibits surface waves explicitly in the second term of Eq. (5). These waves form a discrete rather than continuous spectrum. As f(k,) is known and the value of k , has presumably been obtained by solving the source-free problem of Eq. (2), the surface waves in the presence of the given source are readily found. In most practiced cases, only a few such waves (often, only one) are significant. If, in addition, the first (continuous spectrum) term of F is known to be relatively small, the excitation problem is then easily solved by neglecting all field contributions except for the few significant surface-waveterms.
2.4. The angular representation of field

The foregoing transverse representation is very useful in those cases where the sources excite primarily surface waves. We recall that these surface waves are characterized by real values of k, = k , , and by positive imaginary values of k, = k,,, because the) propagate as exp(ik, x) parallel to the boundary and they decay as exp( - I k , I z) in a direction normal to that boundary. Thus, surface waves account for energy flow along a polar-angle direction given by 8 = k 900 in figure 1. However, in certain circumstances, a considerable amont of optical energy may travel along an angular direction 8 which is different from 900. In that case, the transverse representation is no longer too helpful because a large energy would

where C is a suitable contour in the complex w plane, with w being related to k, by an angular transformation k, = k sin w . As obtained in Eq. (7), F has a form similar to that in Eq. (5) because it again includes a continuous spectrum and a discrete spectrum, the latter being the result of capturing poles between the original integration path along the real k, axis and the deformed contour C. However, there is a significant difference in'that now the discrete spectrum contains not only surface-wave fields (with k, real), but also leaky wave fields (with k,, complex). The advantage of F in Eq. (7) over F in Eq. ( 5 ) is that the former expression may exhibit a continuous spectrum which is small even though the complete field contains a considerable amount of energy flow at an angle 8 # 900. Thus, this energy flow has been relocated into the second (discrete-spectrum) term of Eq.' (7) and is accounted by one or more complex values of k , in the summation. Obviously, Eq. (3), (5) and (7) are all alternative representations for the same quantity F. Depending on the specific problem at hand, one of these representations is more suitable than the others because, by using discrete terms in a rapidly converging summation, it puts into evidence that portion of the field that may have a greater physical significance. We should note that the third representation of F in Eq. (7), which may be referred to as the angular representation, is the most general because it includes the longitudinal representation (for 8 = 0) and the transverse representation (for 8 = 900) as special cases. However, as the angular representation involves , , , it is pertinent to discuss complex values of k , = k the properties of fields characterized by such complex propagation factors.
3. - FORMATION OF LEAKY-WAVE FIELDS

Although leaky waves, like surface waves, can be viewed as modes, i.e., resonant solutions of boundaryvalue problems [17-191, we prefer to present here a more intuitive description in terms of inhomogeneous plane waves. These plane waves are therefore discussed first and they are then used to construct leaky waves by using simple superposition arguments.

276

T. TAMIR

Nouv. Rev. Optique, 1975, t. 6, no 5

3.1. Inhomogeneous plane waves


In an unbounded medium, solutions of the homogeneous Helmholtz equation (2) take the form

(8)

F, = C exp[i(k, x

+ k, z ) ] ,

where C is a constant, k, and k , are related by the dispersion formula (4) and the subscript p in F, signifies that we are now dealing with a particular solution of Eq. (2) rather than with a complete solution as was the case for F in Eq. (3). Hence now k, and k, take on special values and are no longer continuous variables. For real k, and k,, we recall that F, in Eq. (8) represents a homogeneous plane wave. However, neither k, nor k, need be real to satisfy Eq. (2); they must only be related by Eq. (4). If we let these parameters be complex, we can write
(9) (10)

FIG.3. - Reyection and refraction of a plane inhomogeneous wave at the interface between two different media.

+ ia, k, = b + i a .
k,
=

fl

We then find [18] that F, still describes a plane wave, which is, however, no longer homogeneous. As shown in figure 2, this wave now has plane equiphase
L

power transmission at the interface is predicated solely by the field continuity conditions at the z = 0 boundary, reflection and refraction are determined by the same Fresnel coefficient as in the case of homogeneous plane waves, except that now k, and k, assume complex values. In particular, the reflection coefficient for perpendicular polarization is given by
(1 1)

k,, r(kx) = k,,

+ k,,

k,, - b , - b, b, b,

+ i(a, - a,) + + i(a, + a,)

'

where the subscripts 1 and 2 in the above quantities refer to the two different media. Because now all of the parameters a,, b,, a, and b, are non-zero, it is easily seen that I r(k,) I is always smaller than unity, so that no total reflection can occur for 8, # 0. Thus, unlike homogeneous waves, some energy must always be transmitted into the second medium in the case of inhomogeneous plane waves.

FIG.2. - Field of a plane inhomogeneous wave. The solid lines denote equiamplitude contours ; the density of these lines suggests the magnitude of the,fieldintensity. The dashed line indicate equi-phase contours, whereas the arrows denote the direction of power,flow.

3.3. Guidance of an inhomogeneous wave by a layer


Consider next an inhomogeneous plane wave which is incident inside a dielectric layer of thickness t and refractive index n, as shown in figure 4.For simpficity, we assume that the lower boundary is a perfect mirror with reflection coefficient yo = & 1. Also, we may assume that the reflection coefficient r1 at the upper boundary is sufficiently large so that the wave can

and equi-amplitude contours that are orthogonal to each other. The power flux progresses parallel to the amplitude contour at an angle Bo = tan-' (fllb) and the amplitude decays along the equi-phase contours. We recall that this decay is not due to absorption losses. In fact, for a = b = 0, we get the picture of a surface-wave field which is a special case of an inhomogeneous plane wave.

3.2. Reflection and refraction of plane inhomogeneous waves


If a plane inhomogeneous wave is incident at the interface between two media with different refraction indices n, and n,, it undergoes reflection and refraction in a manner similar to that of a homogeneous plane wave. This is indicated in figure 3 wherein a wave is incident from below. Because

FIG.4. - Inhomogeneous fielak guided by a planar layer. The field refracted into the upper (air) region constitutes a leaky wave which radiates at the leaky-wave angle Bo.

Nouv. Rev. Optique, 1975, t. 6, no 5

T. TAMIR
4.

277

bounce many times between the two boundaries before it loses most of its energy by refraction. Under the above conditions, the field inside the layer will progress for a considerable distance along x before its magnitude decreases significantly. In particular, we recall that, for surface waves, a self-consistent field is established by means of a homogeneous plane wave bouncing between the two boundaries of the layer if we satisfy the transverse-resonance condition (12)

OF LEAKY-WAVE FIELDS In microwave structures, a wide range of configurations are known [3] to support leaky-waves. In optics, only two categories of such structures have been studied : multi-layered media and periodic (grating) configurations. These categories are discussed below.
4.1. Multi-layered media In general, any structure that guides surface waves can also support leaky-waves. It can thus be shown that leaky-waves exist on a single dielectric layer placed in air [3], as well as on a thin film deposited on a substrate [12, 181. However, these leaky-waves do not usually satisfy the slow leakage condition (13) and therefore they are not very important. On the other hand, if the layer is a metal so that its permittivity at optical frequencies is well described by a plasma medium, strongly guided leaky-waves may occur for TM modes [23, 241, which have found applications in studies of surface properties.

- VARIETIES

ro r, exp(2 ik,, t,)

1,

where k,, denotes the phase-shift factor along z inside the layer. This self-consistency relation holds also for inhomogeneous plane waves provided we let the quantities r,, r, and k,, become complex. In fact, as for surface waves, the transverse-resonance condition (12) yields those values of k,,.that produce constructive interference between the various bouncing fields, thus setting up a mode which is guided along the layer. Consequently, the same zig-zag mechanism of a bouncing field, which is used to describe a surface wave, can also explain the more complex guided wave described schematically in figure 4. Because now the zig-zag travelling wave is an inhomogeneous plane wave, the present field differs from that of a surface wave in two important respects :
a) The field in the exterior (air) region travels along an oblique direction given by Bo, in contrast to a surface wave for which the flux travels parallel to the boundary. This happens because refraction is always non-zero in the case of inhomogeneous waves. b) Due to the power flow at the angle Bo, energy leaks continuously out of the layer region and therefore the total field must decay as it propagates along x. This decay is consistent with the fact that the propagation factor k, was assumed complex in Eq. (9), with a denoting the attenuation due to energy leakage.

FIG. 5 .

- Leaky

wave supported by a four-media layered structure.

Because complex fields guided by layers always account for leakage out of the layer, such fields have been labelled (( leaky waves D in the literature. Obviously the energy leaked out represents radiation which appears in the far field as a beam oriented at an angle 8, with respect to the planar structure. For surface waves, i.e., for real values of k,, only a few solutions of the transverse-resonance relation (12) exist. In contrast, an infinite set of solutions of Eq. (12) exist for leaky waves, i.e., complex values ia. However, only a few of these satisfy of k, = p the strong guidance condition

(1 3)

a < p ,

which implies that leakage per wavelength is very small, so that the wave extends over a long distance along the layer. If a is comparable to /3 or larger, the energy leaks out fast and the wave is only weakly guided. For practical purposes, only strongly guided leaky waves satisfying condition (1 3) are important. We shall therefore assume henceforth that only this kind of leaky waves is being considered.

In recent integrated-optics work, the largest use of leaky-waves has been in the context of four-media configurations, as shown in figure 5. In the absence of the top layer with refractive index np, this configuration can guide a surface wave along a thin film with refractive index nf deposited on a substrate having refractive index n,, provided nf > n, > n, [14-171 ; the medium above the film is usually air, so that n, = 1. If the top layer is brought in and its refractive index satisfies n p > n p the surface wave is modified into a leaky-wave, as indicated in figure 5. By taking a coupled-mode approach, the wave leaked into the top denser medium is often explained [8-10, 141 in terms of energy crossing through the evanescent field in the air gap' by means of a tunnel effect. A characteristic feature of leaky-waves in multilayered dielectric media is that the horizontal component of the radiation field in the open (exterior) regions is oriented along the same direction as that of the energy flow inside the principal (G surfacewave ))) layer. For the case shown in figure 5, both of these components flow along the + x direction. Hence the field in this case is referred to as a forward leaky-wave. Exceptions to this may occur only if one (or more) of the dielectric media is replaced by a metal [23], but we shall not discuss such cases here.

4.2. Dielectric gratings


Leaky-waves along periodic structures were first discussed by Hessel and Oliner [25] in the context of Wood's anomalies on metallic diffraction gratings.

278

T. TAMIR

Nouv. Rev. Optique, 1975, t. 6, no 5 From Eq. (14)-(16), it is clear that k,, may be real for some v < 0 even if a is zero, provided only d is small enough. These real values of v correspond to grating orders for which energy will radiate or leak away from the structure. Because of this leakage, k,, must then have a non-zero value of U if the grating is present; this a again expresses the decay of the fields in the same manner as for multi-layered media. As frequency increases, the first grating order occurs for v = - 1, in which case energy leakage appears as shown in figure 6a. As indicated therein, this leakage has a horizontal component oriented along - x whereas the energy of the wave bouncing inside the layer flows along + x. Hence, this kind of field is referred to as a backward leaky wave. With increasing frequency this leakage swings clockwise and becomes a forward leaky wave as was the case in figure 5. However, when this occurs, another grating order (either v = - 2 or v = 1) usually appears in the backward direction, as shown in figure 6b. In this case, the total leaky wave field accounts for two outgoing beams in each of the two open (substrate and air) region. This is in contrast to the situation in multi-layered media where only one beam in each region can occur for every mode. Thus, leaky waves on dielectric gratings are more versatile than those on multi-layered media. In the former case, several beams may appear in each open region and their character may be of either the forward or backward type.
4.3. Effect of asymmetry of grating profile

Most recently, they have been investigated in studies of dielectric gratings for integrated-optics applications [I 1, 261. Although the characteristics presented here also apply to metallic, gratings, we shall restrict our discussion to leaky-waves in dielectric gratings.

-X

FIG. 6 . - Leaky waves supported by dielectric gratings : a) Single leakage with backward characteristics ; b) Multiple leakage with backward and forward beams.

Unlike the situation in uniform layers, a wave bouncing inside a dielectric grating does not encounter a simple planar boundary on both sides. Thus, as shown in figure 6, the upper boundary is non-uniform and characterized by a periodicity d. Because of this periodicity, energy is scattered into the upper and lower media. If a surface wave propagating as exp(i&,x) can exist when the grating is absent, this wave will be modified by the grating to a field having the Floquet form
W

(14) where

F, =
,=-U3

a,@) exp(ik,,

(15) k,, = B,

2 VR + iu. = Bo + 7 + icr

v = 0, & 1, & 2,

...

Obviously, here 8, refers to the propagation factor along x of the v-th space harmonic. When the grating acts as a small perturbation, Bo is close to B, of the surface wave. In the two open (air and substrate) regions, propagation along z is given by (16) k,, = (kj2 - k,J
2
112
3

where kj denotes the plane-wave propagation factor in the a x ( j = a) on the substrate ( j = s) medium.

Some typical shapes of dielectric gratings are shown in figure 7. In most cases, the grating profile is symmetric about a reflection plane parallel to the yz axis. However, by suitable fabrication methods, the profile can assume an asymmetric form, as is the case in figure 7b. When this happens, the leaky waves supported by the structure may exhibit interesting directional effects, which are described below. Because of the profile asymmetry, a wave travelling along the + x direction encounters a structure which is different from that seen by a wave travelling along the - x direction. Thus, as shown in figure 8, a wave incident from the left may produce a field such that most of the energy is scattered into the substrate; on the other hand, if the wave is incident from the right, the energy is scattered mostly into the air. By taking a rigorous model for an asymmetric grating, Peng and Tamir have shown [271 that a judicious choice of the profile parameters may concentrate all of the radiation into one of the two open regions; conversely, the radiation appears entirely in the other region if the incidence direction is reversed. This directional discrimination is analogous to the blazing of a plane wave incident at an angle on a diffraction grating. In that case, changing the incidence angle 8 to - 8 also produces a different field configuration. However, the blazing effect described here deals with the incidence of a surface or leaky wave along the structure rather than with a plane wave incident obliquely onto the structure.

Nouv. Rev. Optique, 1975, t. 6, no 5

T. TAMIR

279

5. -LEAKY WAVES IN BEAM AND WAVEGUIDE COUPLERS


I

na
-X

So far, the widest application of leaky waves in optics has been in the context of beam .couplers for integrated optics. To discuss these, consider a thin film, as shown in figure 9 4 which supports a surface wave to the left of the x = 0 plane and a leaky wave to the right of the x = 0 plane. A surface wave incident from the left is partly or fully converted into a leaky wave, which then radiates energy away into the open region (or regions) in the form of an outgoing beam at the leaky-wave angle'8,. To achieve the change of regime between a surface wave and a leaky wave, we can use either a prism as shown in figure 9b, or a grating as shown in figure 9c. In the former case, we need to satisfy the condition n p > nf > n, > n, already discussed in conjunction with figure 5 , the prism providing now the denser medium with refractive index np. In the latter case, the grating may provide one or more beams in each open region, as already discussed in connection with figure 6 . In the case of prim couplers, the leakage into the substrate is very close to the grating angle (e N 90) and its intensity is then negligible. However, the beams leaked out by grating couplers appear in both the air and substrate regions. Thus, if energy is to be

+X

(d1

FIG. 7. - Several varieties of dielectric gratings. Note that the grating in b) has an asymmetric profile.

( a ) Incidence olong + x direction

\
-X

/
( b ) I n c i d e n c e along - x direction

FIG.8. - Directional discrimination m the field supported by an f ~7ux lines suggests the intensity asymmetric grating. The number o of the field m each region. Note that most o f the energy in the top case is leaked into the substrate whereas most o f the energy in the bottom case is leaked into the air region.

FIG. 9. - Leaky-wave interpretation of beam couplers : a) Basic structure that converts an incoming surface wave into an outgoing leaky-wave beam; b) Implementation o f beam coupler by using a prism ; c) Implementation of beam coupler by using a grating.

280

T. TAMIR

Nouv. Rev. Optique, 1975, t. 6, no 5


couplers, because it is separated from the linear waveguide by a gap. Thus, if a mode is incident from the right in the linear guide, leakage starts to occur whed the gap region is reached. If the wide guide is oriented at the correct leakage angle, the energy is coupled into this guide in a manner which is analogous to that shown in figure 9b. Leaky waves also play significant roles in the operation of the Lummer-Gehreke interferometer [ 121 and in enhancing the power carried along fibers [13].
6. - BEAM DISPLACEMENT AT LEAKY-WAVE STRUCTURES

transferred from a surface wave into a single outgoing beam, the grating coupler is generally less efficient. However, by suitably designing the grating and, in particular, by choosing a specific profile so as to enhance the blazing effect discussed in Section 4.3, it is possible to increase the efficiency of grating couplers to be nearly equal to that of prism couplers. Of course, by using reciprocity arguments, it is evident that a beam incident on the configurations shown in figure 9 can be converted into a surface wave. This reciprocal function can be visualized by simply reversing the direction of the arrows in figure 9. It is thus seen that the operations of prism and grating couplers can be explained by means of a unified approach in terms of leaky-wave fields. The reader interested in applying these concepts to the design of actual devices is referred to chapter 3 of reference [14].

d
Immersion fluid "Prism"

A closer inspection of the behavior of beams incident on leaky-wave structures reveals an interesting phenomenon, which is similar to the GoosHaenchen shift [29]. However, leaky-wave structures can produce a displacement of the reflected beam which is several orders of magnitude larger than the Goos-Haenchen shift. A simple heuristic explanation of the beam displacement due to leaky-waves can be given by examining figure 11, which describes the profile of an outgoing beam in an output coupler. Except for the smooth rounding off of the left-hand edge, this beam exhibits the expected exponentially decaying shape of the leaky-wave amplitude. The rounding-off of the left edge is due to the weak diffraction that prevents an abrupt discontinuity in the beam profile [14, 181. If the power flux of the entire beam is considered, we find that its center of gravity moves along the thick flux line shown in figure 11. Because the beam field is identical to the leaky-wave in the region x > 0, it is evident that energy travels horizontally for a distance I = I' B before being completely leaked off into the upper medium. It can be shown from simple arguments that I is closely equal to the inverse of the leakage parameter a.

FIG. 10. - Waveguide couplers using leaky waves : a) Junction between a planar waveguide and a fiber ; b) Junction between a planar waveguide and a linear waveguide.

Although of less wide application, leaky waves have played a role also in coupling waveguides of different shapes, two examples of which are shown in figure 10. In the first scheme, which was reported by Kersten [28], a high-index fluid is used to provide leakage from a planar waveguide. By then introducing the end of a fiber into the liquid and aligning the fiber along the leakage angle e,, energy can be transferred from the waveguide to the fiber, or vice-versa. The second waveguide coupler, shown in figure lob, is the analog of the prism coupler reduced to a planar geometry. In this case, the beam is replaced by a surface wave travelling along a broad waveguide, which is coupled to a narrow linear waveguide (of the rib or strip variety) via a slanted termination. This termination plays the role of the prism in beam

-X

,Surface-waveregime

-Leakywave regime

FIG. 1 1. - Leaky-wave beam produced by an output beam coupler. The thick .flux line indicates the center of gravity of power .flow.

Consider now figure 12 where we assume that the x axis represents the upper boundary of a structure which can support a leaky-wave having leakage at the angle 6,. In this figure, the trajectory Z ' B D in figure 12a represents the same flux as that of figure 1 1.

Nouv. Rev. Optique, 1975, t. 6 , no 5


z

T. TAMIR

28 1

section, we find [l 1, 181 that it can be described as a superposition of two fields, of which one has leakywave characteristics. Because this leaky-wave portion behaves in the form suggested in figure 12, the total beam field undergoes a displacement. Although the net reflected beam may exhibit a profile that is often distorted when compared with the incident beam, the flux displacement is quite evident and, under proper conditions, the beam shift may be nearly as large as the beam width.

FIG. 12. - Displacement of power flux at leaky-wave structures : a) Outgoing leaky-wave .flux I' BD and its reciprocal incident flux IAD' ; b) An incident leaky-wave,%.xIAD' musf ultimately continue as the outgoing leaky-waveflux I' BD ;c ) Flux displacement AB in the case of forward leakage; d) Flux displacement AB in the case of backward leakage.

FIG.13. - Beam shift X at leaky-wave structures

: a) Forward

By reciprocity, if a beam with a leaky-wave profile is incident on the structure, it will follow the path IAD'. The same path IAD' is repeated in figure 12b, except that it has been reflected through the yz plane. However, AD' represents energy flux flowing horizontally and we recall that such a horizontal flux, when described by I ' B , must ultimately leak into the upper region along BD. It therefore follows that the horizontal flux AD' generated by the incident beam IA must continue along a path Z'B that radiates away along BD. This is described in figure 12c which simply connects the two points D' and 1'. Hence, the beam IA re-emerges as a reflected beam BD only after it has undergone the displacement AB. The construction in figures l l a - l l c refers to a beam incident at the leakage angle on a leaky-wave structure of the forward type and the beam displacement AB in figure 12c is therefore a forward one. A similar displacement occurs for incidence on structures supporting a leaky-wave of the backward variety, such as was shown in figure 6a. The construction for this case is given in figure 124 where it is noted that the incident flux IA must intersect the reflected flux BD at some point above the structure. In practice, incident beams seldom have leakywave profiles of the form shown in figure 11. However, if we analyze a realistic beam such as a laser beam with Gaussian intensity distribution in its cross-

shift, and b) Backward Shift.

As a result, beams incident on leaky-wave structures are displaced as shown in figure 13, where the shift of the forward type is shown in figure 13a and the backward type is indicated in figure 13b. The former beam-shift variety occurs and has been observed [30] in the case of multi-layered media. The backward shift may occur in the case of periodic structures, but its experimental verification has not been reported yet. In fact, the backward shifted beam shown in figure 13b occurs when only a single grating order (for v = - 1) radiates into the upper region. If more than a single grating order radiates, diffracted beams appear and each one of them is laterally shifted. The situation is then predicted [ l l ] to be as indicated in figure 14, where a specular and a diffracted beam are shown. It is noted that each one of the scattered beams is displaced by a different distance X , and that some beams undergo a forward shift (e.g., the specular beam in fig. 14) whereas others are shifted backwards (e.g., the diffracted beam in fig. 14). Although beam shifts along dielectric gratings have not found any applications yet, they may provide the scope for very interesting basic studies, with the backward shift serving as an open question for experimental verification.

282

T. TAMIR

N o w . Rev. Optique, 1975, t. 6, no 5

Film
I I
I

>h,,:
Rim
(b)

Substrate

FIG. 14. - Lateral displacement of beams scattered by a periodic structure.

7 . - STRUCI'URES WITH VARIABLE LEAKY-WAVE CHARACTERISTICS


In all of the foregoing discussions, we have assumed that the structures considered were characterized by constant values of /3 and E . However, structures having variable leaky-wave characteristics may have considerable potential uses in the future. In particular, if the structures vary (( slowly )) with x, they could be characterized by continuously varying parameters p(x) and CL(X), thus leading to leaky-wave properties of a generalized nature. In the case of multi-layered media, the variation in a(x) and p(x) may be achieved by means of a layer whose thickness varies with x, as suggested in figure 15a. Such a variation has already been investigated by Harris and Shubert [31] and by Ulrich [32] for the purpose of increasing the efficiency of beam couplers of the prism variety. In that case, the practical implementation follows the scheme outlined in figure 1%. Here the air gap is made variable in height by manipulating the pressure produced by the clamp that holds the prism and the film waveguide together. Saad et al. [33] extended this approach to situations having other variations, but their theoretical considerations have not been verified by experimental studies. Because of their backward-wave characteristics and multiple-beam capabilities, dielectric gratings with variable characteristics have even greater potential applications. Their variation can be achieved as shown in figure 16, where it is noted that either the grating height re or the periodicity d, or both, could be changing with x. It is thus conceivable, for example, that a leaky-wave supported by such gratings would change from the forward to the backward variety as it progresses along the x direction. ,Other such variable features can be introduced and a wide range of field changes may thus be achieved. By suitably controlling the phase variation p ( x ) and the leakage parameter a(x), it would thus be

FIG.15. - Layered structures with variable leakage : a) Decreasing air gap produces a value of a(x) that increases with x ; b) Practical implementation of the variable air gap in aprism coupler for improving the coupling efficiency.

possible to realize leaky-wave structures that perform complex functions within small dimensions of a thin-film structure. In particular, incident beams may be transformed into a plethora of various fields. We recall that the beam couplers discussed in Section 5 convert a beam into a surface wave. Other beam conversions and transformations could be achieved by structures with variable leaky-wave characteristics. Some of these possibilities are shown in figure 17. Although most of these beam transformations must yet be explored, it is easy to imagine what kind of variation is conceptually needed to implement most of them. In particular, the beam-splitting operation

tg = tg(x)

d = d(x)

--X

tg= tg(x)

and

d(x)

FIG.16. - Dielectric gratings with variable leakage.

N o w . Rev. Optique, 1975, t. 6, no 5

T. TAMIR

283

[ 21 LECHER (E.). - Eine Studie uber elektrische Resonanzer-

FIG. 17. - Beam functions realizable by means of variable leakywave structures.

shown in the bottom right corner has been shown [33] to be achievable by using a structure similar to that depicted in figure 1%. I~ conclusion, it can be noted that leaky-waves have already played a significant role in integratedoptics applications. By extending the concepts already developed to configurations having variable leakage characteristics, it is expected that leaky-waves may bring about an even greater and more diversified use for controlling light beams by means of miniature and reliable structures.

***
REFERENCES
[ 11 Lord RAYLEIGH. - On Waves Propagated Along the Plane Surface of an Elastic Solid, proc. London Math. soc., 1885, 17, 4-11. Also in Scientific Papers, vol. 2, Cambridge Univ. Press, London, 1900, pp. 441-447.

scheinungen, Ann. Phys. Chem., 1890, 41, 850. - Ed. - Antenna Theory, Part 2, McGraw Hill, New York, 1969, pp. 151-348. [ 41 SNITZER (E.), OSTERBERG (H.), - Observed Dielectric Waveguide Modes in the Visible Spectrum, J. Opt. Soc. Amer., 1961, 51, 499-505. [ 51 KAPANY (N. S.), BURKE (J. J.). - Fiber Optics : 1X. Waveguide Effects, J . Opt. Soc. Amer., 1961, 51, 1067-1078. [ 61 Y ~ (A.), N LEITE (R. C.). - Dielectric Waveguide Mode of Light Propagation in p-n Junctions, Appl. Phys. Lett., 1963, 2, 55-58. [ 71 OSTERBERG (H.), SMITH (L. W.). - Transmission of Optical Energy Along Surfaces : Part 11, Inhomogeneous Media, J. Opt. Soc. Amer., 1964, 54, 1078-1084. [ 81 HARRIS (J. H.), SHUBERT (R.), POLKY (J. N.). - Beam Coupling to Films, J. Opt. Soc. Amer., 1970, 60, 1007-1016. [ 91 TIEN(P. K.), ULRICH (R.). - Theory of Prism-Film Coupler and Thin-Film Light Guides, J. Opt. Soc. Amer., 1970, 60, 1325-1337. [lo] MIDWINTER (J. E.). - Evanescent Field Coupling Into a Thin-Film Waveguide, ZEEE J. Quantum Electronics, 1970, QE-6, 583. [ll] TAMIR(T.), BERTONI (H. L.).- Lateral Displacement of Optical Beams at Multilayered and Periodic Structures, J. Opt. Soc. Amer., 1971, 61, 1397-1413. [12] ULRICH (R.), PRETTL(W.). - Planar Leaky Light-Guides and Couplers, Appl. Phys., 1973, I , 55-68. [13] SNYDER (A. W.). - Leaky-Ray Theory of Optical Waveguides of Circular Cross-Section, Appl. Phys., 1974, 4, 273-298. [14] TAMIR(T.), Editor. - Integrated Optics, Springer-Verlag, Heidelberg, 1975. (N. S.), BURKE (J. J.). - Optical Waveguides, Aca[15] KAPANY demic Press, New York, 1972, pp. 1-89. [16] TAMIR (T.). - Inhomogeneous Wave Types at Planar Interfaces : 11. Surface Waves, Optik, 1973, 37,204-228. [17] MARCUSE (D.). - Theory of Dielectric Optical Waveguides, Academic Press, New York, 1974. [18] TAMIR (T.). - Inhomogeneous Wave Types at Planar Interfaces : 111. Leaky Waves, Optik, 1973, 38,269-297. [19] TAMIR (T.), OLINER (A. A.). Guided Complex Waves, Proc. Z.E.E.E., 1963, 110, 310. [20] SHEVCHENKO (V. V.). - Continuous Transitions in Open Waveguides, Golem Press, Boulder, Colorado, 1971. ~211 B~~~~~~~~ (L. M.). - waves in ~ ~ Media, ~ &ade~ mic Press, New York, 1960, pp. 245-261. [22] TYRAS (G.). - Radiation and Propagation of Waves, Academic Press, New York, 1969, pp. 118-125. [23] TAMIR (T.), OLINER (A. A.). - The Spectrum of Electromagnetic Waves Guided by a Plasma Layer, Proc. ZEEE, 1963, 51, 317-332. [24] OTTO (A.). - Streustrahlung von Silber durch Anregung von Oberffachenplasmaschwingungen, 2. Physik, 1969, 224, 65-73. [251 H B S ~ (A.), OLINER (A. A.). - A new Theory of Woods Anomalies on Optical Gratings, Appl. Optics, 1965, 4, 1275-1297. [26] PENG (S.T.), TAMIR (T.), BERTONI (H. L.). - Theory of Periodic Dielectric Waveguides, IEEE Trans. Microwave Theory Tech., 1975, MTT-23, 123-133. [27] PENG(S. T.), TAMIR (T.). - Directional Blazing of Waves Guided by Asymmetrical Dielectric Gratings, Optics Comm. 1974, 11, 405-409. [28] KERSTEN (F. T.). - Coupling Between Slab-Waveguide and Glass Fibers, Digest of Tech. Papers, Topical Meeting on Integrated Optics, New Orleans, La., 1974, WB5-1-4. [29] HOROWITZ (B. R.), TAMIR (T.). - Unified Theory of Total Reflection Phenomena at a Dielectric Interface, Appl. Physics, 1973, I , 31-38. 130i M1DWINTER (J. E.), ZERNIKE (F.1. - Experimental Studies of Evanescent Wave Coupling into a Thin-Film Waveguide, Appl. Phys. Lett., 1970, 16, 198-200.
[ 3 1 COLLIN (R. E.), ZUCKER (F. J.).
~

284

T. TAMIR

Nouv. Rev. Optique, 1975, t. 6, no 5


ACKNOWLEDGMENT. - The author wishes to express his appreciation to Prof. C. I ~ ~ B E R for T facilitating his participation at the f Guided Waves )) as well as for encouColloquium on (( the Optics o raging the presentation and subsequent preparation of this paper. This work has been supported by the U.S. OfJice of Naval Research, under Contract No N00014-75-C-0421.

[31] HARRIS (J. A.), SHUBERT (R.). - Variable Tunneling Exci; tation of Optical Surface Waves, IEEE Trans. Microwave Theory Tech., 1971, MTT-19, 269-276. [32] ULRICH(R.). - Optimum Excitation of Surface Waves, J. Opt. Soc. Amer., 1971, 61, 1467-1476. [33] SAAD(A.), BERTONI (H. L . ) , T m (T.). - Beam Scattering by Nonuniform Leaky-Wave Structures, Proc. ZEEE, 1974, 62, 1552-1561.

(Manuserit re( U le 27 mai 1975.)

S-ar putea să vă placă și