For purposes of validating our implementation we start with a comparison of phase propagation constants and attenuation levels. Tables 2.1 and 2.2 list values for angular mode numbers obtained with the present mode solver for two bend configurations adopted from Ref. [43], together with reference data from that source. We found an excellent overall agreement, for both the configurations with higher (Table 2.1) and lower refractive index contrast (Table 2.2).
|
|
The discussion of bent waveguides in Ref. [43] applies an alternative definition of the bend radius as the distance from the origin to the center of the core layer, which is related to the radius as introduced in Figure 2.1 by (hence the unusual values of bend radii in Tables 2.1, 2.2). Both definitions are meant as descriptions of the same physical configuration, i.e. both lead to the same angular field dependence (2.1), given in terms of the azimuthal mode numbers as determined by the dispersion equation (2.7). Via the relation , the different choices of the bend radius result in different values and for the propagation constant, and consequently in different values , and , for the phase and attenuation constants.
Still, for many applications one is interested in the variation of the phase constant and the attenuation with the curvature of the bend, expressed by the bend radius. Figure 2.2 shows corresponding plots for the configuration of Table 2.1, including values for the two different bend radius definitions. While on the scale of the figure the differences are not visible for the attenuation constants, the levels of the phase propagation constants differ indeed substantially for smaller bend radii. As expected, for low curvature the values of both and tend to the effective indices of straight slab waveguides with equivalent refractive index profile. For the present low contrast configuration, only minor differences between TE and TM polarization occur.
Certainly no physical reasoning should rely on the entirely arbitrary definition of the bend radius. This concerns e.g. statements about the growth or decay of phase propagation constants with (according to Figure 2.2 the sign of the slope can indeed differ), or discussions about the ``phase matching'' of bent waveguides and straight channels in coupler or microresonator configurations. Care must be taken that values for and or effective quantities like are used with the proper definition of taken into account.
With the present (semi) analytic solutions at hand, we have now a possibility to validate ``classical'' expressions for the variation of the bend attenuation with the bend radius. Beyond the high curvature region, Figure 2.2 shows a strict exponential decay of with respect to , as predicted by an approximate loss formula for symmetric bent slabs given in [93, Eq. 9.6-24]: