Optical Waveguide Power Divider

This invention relates to optical waveguides and more particularly to optical waveguide power dividers or couplers.

A power divider is an essential component of any amplitude modulator constructed with optical waveguides which operates by interfering two-phase shifted guided waves. Light passing through the two waveguides is controlled in such a manner that light may be shifted from one waveguide to another. Heretofore a 3-dB power transfer between optical waveguides has been achieved by providing synchronism between two optical waveguides over a critical interaction length, whose magnitude depends upon the coupling between the waveguides.

In prior art devices there are severe tolerance restrictions on both the smoothness and on their mean dimensions, i.e. length of the coupled guides. Such prior art devices have been set forth in an article "Improved Tolerance in Optical Directional Couplers," by M. G. F. Wilson et al., Electronics Letters, Vol. 9, #19, Sept. 20, 1973, pp 453-454.

This invention has been set forth in a published article, "Optical Modal Evolution 3-dB Coupler" by W. K. Burns, A. F. Milton, A. B. Lee and E. J. West, Applied Optics, Vol. 15, No. 4, April 1976, pp 1053-1065, which is incorporated herein by reference.

This invention makes use of an optical waveguide arrangement which comprises two paths for spatial power division into two equal parts. The device uses a single region of mode conversion followed by a region of adiabatic modal evolution in which no mode conversion occurs. These two regions are provided by an overlay with fast and slow tapers, which makes the device synchronous at a single point. Device length is non-critical. Synchronism at the desired point may be obtained by electro-optic or other index control.

FIG. 1 illustrates a cross-sectional view of an embodiment of the device.

FIG. 2 illustrates operation of the device with incident power from the opposite side.

FIG. 1 shows a cross-sectional view of the device. As shown, the device includes a substrate 10 upon which optical waveguide 11 is formed, waveguide 11 being of an optical material having an index of refraction which is greater than that of the substrate. A layer of optical material 12 having a thickness of 1/2 to 2.mu.m with an index of refraction which is the same as that of substrate 10 overlays optical waveguide 11 and an optical waveguide 13 is formed on the layer 12 of an optical material having a refractive index and thickness which is substantially the same as that of the waveguide 11. A triangular (cross-sectional) overlay 14 is formed over a portion of the waveguide 13. The overlay 14 has the same index of refraction as that of the layer 12 and is of sufficient thickness at its thickest part (at Z = 0) that the fields associated with guided modes in the structure do not extend beyond its upper boundary (1-2.mu.m). From its thickest part at Z = 0, the overlay 14 narrows down to zero thickness with a fast taper 14a as Z becomes more negative and with a slow taper 14b as Z becomes more positive. The physical slopes of these fast and slow tapers are defined qualitatively as follows: The fast taper 14a must be sufficiently steep that nearly complete mode conversion between normal modes of the structure occurs as these normal modes pass under it. The slow taper must be sufficiently shallow that essentially no mode conversion between normal modes of the structure occurs as they pass under it. This latter case is the case of adiabatic propagation. Quantitatively these slopes are defined by:

where K is the coupling constant between the guiding layers 11 and 13 and .gamma. is the transverse propagation constant in the overlay 14. Typical numbers for the taper slopes are 0.25 and 0.6 .times. 10.sup.-3 for the fast and slow tapers respectively. The design of the taper slopes is discussed in detail in "Optical Modal Evolution 3-dB Coupler" by W. K. Burns, et al. in Applied Optics 15, 1053 (1976). A sketch of the overall structure of the coupler device is shown by FIG. 4 in the published article. The relative dimensions and index-of-refraction of the various parts are shown on the sketch.

In operation with input from the negative Z side (FIG. 1). an input mode " a" is injected into arm 1 (layer 11) where it propagates to the beginning of the fast taper. As mode "a" passes through the region of the fast taper, mode conversion occurs and power is transferred from normal mode "a" to normal mode "b" . These modes are illustrated in FIG. 1 at or just after the point Z = 0 for clarity of the drawing. At the point Z = 0, the two coupled guides 11 and 13 have the same index and thickness, and are surrounded by cladding material of the same index. If they were well separated, i.e., if layer 12 were very thick they would support identical modes with the same velocity and they are said to be synchronous. Note that this is not the case when the overlay 14 is removed and the coupled guides are said to be non-synchronous. The purpose of the fast taper is to abruptly introduce synchronism and the purpose of the slow taper is to slowly remove it. It is known from the teachings of the Burns, et al., article referred to above that, if Eq. 1a is obeyed and the coupled structure is synchronous at Z = 0, 50% of the input power will transfer from mode "a" to mode "b" by mode conversion caused by the fast taper. Then, at position Z = 0, each normal mode will be excited with 1/2 the input power and power division between the normal modes will be achieved. Now as these modes propagate through the region of the slow taper 14b, no mode conversion occurs if Eq. 1b is obeyed. The modes evolve adiabatically and without power transfer between them until they attain the shapes shown in FIG. 1 in the non-synchronous region of the device. Mode "a" evolves so that it only exists in arm 1 (11) and mode "b" evolves so that it only exists in arm 2 (13). Power division between layers 11 and 13 is then achieved. The 50-50 power division is determined only by the requirement of synchronism at the point Z = 0 and by the conditions of Eq. 1. After synchronism has been removed, the two waveguide arms may be spatially separated to split the power into two different directions.

Operation of the power divider with input power from the opposite side is illustrated in FIG. 2. Input power could be injected into layer 11 or 13 but is shown in layer 13 (mode b) for purposes of illustration. After mode " b" is excited, it travels in the negative Z direction through the region of the slow taper and undergoes adiabatic modal evolution until it attains the shape shown in FIG. 2 at Z = 0. It then passes through the region of the fast taper and, as before, 50% of its power is converted to mode "a" shown. 50-50 power division has again been achieved between the modes "a" and "b" which are confined in the non-synchronous region to the layers 11 and 13, respectively. Thus spatial power division is achieved with input from either direction. The conditions for 50-50 power division in this case are the same as those outlined above.

This device is an improvement over power dividers of the mode interference type as described in the Wilson, et al., article referred to above in that here synchronism is only required at a point rather than over a critical coupling length. Also here the overall length of the device is not critical as long as Eq. 1 is obeyed, i.e., as long as the slow taper is sufficiently slow for adiabatic propagation through it.

Obviously many modifications and variations of the present invention are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended clains the invention may be practiced otherwise than as specifically described.