POLARISATION COMPENSATED OPTICAL WDM-DEVICE
Field of the Invention
[0001] The present invention is directed toward optical communications, and more particularly toward the reduction of polarization-dependent wavelength shifts in optical multiplexers /demultiplexers.
Background of the Invention
[0002] Dense wavelength division multiplexing (DWDM) enables multiple channels at distinct wavelengths within a given wavelength band to be sent over a single mode fiber, thus greatly expanding the volume of data that can be transmitted per optical fiber. The wavelength of each channel is selected so that the channels do not interfere with each other and the transmission losses to the fiber are minimized. Typical DWDM allows up to 40 channels to be simultaneously transmitted by a fiber.
[0003] DWDM requires two conceptually symmetric devices: a multiplexer and a demultiplexer. A multiplexer takes multiple beams or channels of light, each at a discrete wavelength and from a discrete source and combines the channels into a single multi-channel or polychromatic beam. The input typically is a linear array of waveguides such as a linear array of optical fibers, a linear array of laser diodes, or some other optical source. The output is typically a single waveguide such as an optical fiber. A demultiplexer spatially separates a polychromatic beam into separate channels according to wavelength. Input is typically a single input fiber and the output is typically a linear array of waveguides such as optical fibers or a linear array of photodetectors.
[0004] Two types of integrated wavelength multi/ demultiplexers that have been widely investigated are grating-on-a-chip spectrometers and phased-array devices, known as phasars. Grating-based devices utilize a series of parallel grooves to diffract different wavelengths of light at different angles such that multiple wavelengths are separated from one another in different directions of
propagation or combined to form a single beam when incident from multiple angles. Grating-based devices require high quality, deeply etched grating facets. The optical loss of the device depends critically on the verticality and smoothness of the grating facets.
[0005] Phased-array (PHASAR) devices, also called arrayed waveguide gratings (AWG), direct an input multiwavelength signal through an input waveguide. The signal is spread out in a coupler, and enters into multiple curved waveguides. The waveguides have slightly different lengths, so that the light takes different times to pass through each waveguide, resulting in interference and diffraction when the light paths are combined in an output coupler. As a result, each wavelength emerges on the other side of the coupler at a different angle and the various wavelengths can thus be separated. As phased- array bases devices are realized in conventional waveguide technology and do not require the vertical etching step need in grating-based devices, they appear to be more robust and fabrication tolerant.
[0006] One of the major drawbacks in wavelength multi/ demultiplexers is the polarization sensitivity of the device. Since an optical signal propagating through an optical fiber has an indeterminate polarization state, the switching /routing devices must be substantially polarization insensitive. In phasar-based devices, polarization independence is achieved if both the orthogonal TE (transverse electric) and TM (transverse magnetic) fundamental modes propagate in the arrayed waveguide section with the same propagation constants, and thus the wavelengths of the corresponding modes (measured in the waveguides) are identical. However, planar waveguides usually have different propagation constants for TE and TM waveguide modes. For wavelength multiplexers/demultiplexers, this difference in propagation constants results in a wavelength shift in the spectral response peak or the passband of each wavelength channel. This wavelength shift is sensitive to the design of the planar waveguide, and can be as large as 3 nm. As WDM systems
are being designed towards smaller and smaller channel spacing (from 1.6 nm to 0.8 nm or even less in the future), even a small polarization dependent wavelength shift (e.g. 0.3-0.4 nm) is of concern.
[0007] Several techniques can be used to reduce the polarization dependent wavelength shift. One method proposed, involves the insertion of a half-wave plate (λ/2) in the middle of the waveguide array (H. Takahashi et al., Opt. Lett. Vol. 17, 499, 1992). In this way, light entering the array in the TE-polarized state will be rotated by the half-wave plate and travel through the second half of the array in the TM-polarized state; similarly, TM-polarized light will traverse half the array in the TE-state. As a consequence TE- and TM-polarized input signals will experience the same phase transfer regardless of the birefringence properties of the waveguides. This method is only applicable to waveguide structures with a small NA.
[0008] Other methods proposed include dispersion matching with adjacent diffraction orders (M. Zirngibl et al., Electron. Lett. Vol. 29, 201, 1992), use of a special layer structure with low birefringence (H. Bissessur et al., Electron. Lett. Vol. 30, 336, 1994), insertion of a waveguide section with a different birefringence in the phased array (M. Zirngibl et al., Electron. Lett. Vol.31, 1662, 1995 and U.S. Patent No. 5,623,571 issued April 22, 1997 to Chou et al.), or addition of a polarization splitter at the input of the AWG (M. K. Smit and C. van Dam, IEEE Journ. of Select. Top. in Quant. Electr. Vol 5, 236, 1996). These techniques suffer from drawbacks ranging from fabrication difficulties to poor performance, or they are limited to specific structures, materials, or operating conditions.
[0009] A method of etching a compensating region in slab waveguides has been described (J. -J. He et at., IEEE Photon. Tech. Lett. Vol. 11, 224, 1999). This is a particularly attractive, easy-to-fabricate device, but it requires deep etching of the materials and devices with large polarization dependent wavelength shifts, which in turn degrades device performance due to losses in such deeply etched compensators.
Summary of the Invention
[0010] In one aspect, the present invention provides a method for compensating polarization dependent wavelength shift in phasar-based devices. The method consists of overlayer deposition or etching of compensating regions, preferably asymmetrically arranged and prism-shaped, in input and output couplers of the device, respectively. The compensating regions have effective TE/TM mode refractive indices different from those of the original slab waveguide, and thus provide the polarization dependent properties that cancel the polarization dependent wavelength shift of the original uncompensated device. The present invention also provides, in one aspect, an optical device employing two compensating regions.
[0011] The invention is based on the compensation of polarization dependent wavelength shift by two compensating regions located in input and output couplers of a phasar-based device, respectively. The use of an asymmetric compensation scheme eliminating total internal reflection problems is also novel.
[0012] By using two compensators etched in both input and output couplers, respectively, etch depth required for eliminating the polarization dependent wavelength shift is significantly reduced, as is the extra insertion loss penalty. The overall loss is reduced as a result of a better mode matching between such shallower compensators and the coupler slab waveguide. The splitting the compensator also reduces Fresnel reflection loss at the effective index step at the compensator/ coupler boundary.
[0013] Other features and advantages of the invention will become apparent from the following description and from the claims.
Brief Description of the Drawings
[0014] The invention will now be explained by way of example only and with reference to the attached drawing in which:
FIG. 1 shows a schematic diagram of an arrayed waveguide according to the
prior art;
FIG. 2 shows an arrayed waveguide with two asymmetric compensating regions, in accordance with one embodiment of the invention;
FIG.3 is a graph showing the reflectivity of the compensator boundary as the light passes from the higher to the lower effective index slab, in accordance with one embodiment of the invention;
FIG.4 shows input wavelengths joining the input coupler slab, in accordance with one embodiment of the invention;
FIG. 5 shows arrayed waveguides joining the output coupler, in accordance with one embodiment of the invention;
FIG. 6 shows the TM and TE spectra for channel #4 of the compensated device, in accordance with one embodiment of the invention; and
FIG. 7 shows the spectrum of a polarization compensated AWG, in accordance with one embodiment of the invention.
Detailed Description of the Preferred Embodiments
[0015] Referring to FIG. 1 depicting an AWG as is known in the prior art.
The light signals received from waveguide 103, diffracts in the input coupler slab region 105 and is coupled to multiple grating arms (or waveguides) 109. The light travels through the grating arms 109 from the input coupler slab region 105 to the output coupler slab region 107. The length differences between neighboring arms 109 is a constant ΔL. This length difference translates into a regular delay of the light, or phase difference. As a result, each wavelength emerges on the other side of the output coupler slab region 107 at a different angle. The output waveguides 111 collecting the light from the output coupler slab region 107 are spaced to collect the desired focused wavelengths.
[0016] Focusing is obtained by choosing the length difference ΔL between adjacent waveguides equal to an integer number of wavelengths, measured inside the array waveguides
ΔL = mλc/Ng = mc/Ngf
in which m is the order of the phased array, λc is the central wavelength in vacuo, f is the central frequency in vacuo, and N is the effective refractive index of the waveguide mode.
[0017] However, waveguide birefringence, i.e., a difference in propagation constants between the TM and TE modes (ΔN=Ntm-Nte), will result in a shift Δf for the spectral responses with respect to each other, which is called the polarization dispersion, such that
Af « f- N,e -N ,m m
where Νte and Ntø are the effective refractive waveguide indices for TE and TM polarization, and Nfe is the group index of the waveguide TE mode. In practical phasar-based devices, the condition for polarization independent operation Δf ~ 0 is rarely satisfied due to material and waveguide birefringence. Thus, the routing properties depend not only on the signal wavelength λ but also on its state of polarization, leading to this undesirable polarization sensitivity.
[0018] The polarization compensation technique disclosed is this invention is based on creating two asymmetric prism-shaped compensating regions in the input and output coupler slab regions, respectively, with the effective indices of the compensating regions (Nc te and Nc/im for TE and TM modes, respectively) different than the effective indices (Ns,te and Ns/tm) of the original slab waveguide. Such a difference in the effective indices can be created by overlayer deposition or etching an appropriately shaped region in the coupler. The strength of the compensator can be adjusted by selecting the etch depth or the thickness and /or refractive index of the overlayer, and thus the induced change in the effective refractive indices of the compensating regions.
[0019] In accordance with an aspect of the invention, an arrayed waveguide with two compensator regions is shown schematically in FIG.2. The light signals received from waveguides 203, diffract in the input coupler 204, first in the non-etched input slab waveguide 205, which is the higher effective index region and then pass through into the compensator portion 207, the lower effective index (etched) region. After passing through the input coupler 204, the light is coupled to multiple grating arms (or waveguides) 209. The light travels through the grating arms 209 from the input coupler 204 to the output coupler 210. There it passes first through the non-etched slab region 213 and then into compensator region 211, where it is focused onto the various output waveguides 215.
[0020] The shape of the compensating regions can be determined as follows. In order to eliminate the polarization dependence, it is sufficient to assure that the wave fronts corresponding to both TM and TE slab modes have the same tilt in the output coupler, thus converging to the same position at the focal line. This will be satisfied if the difference between the total optical phase (accumulated as the light propagates from the beginning to the end of the AWG) through the waveguide number i of the phased array and the total phase through the waveguide k of the phased array (z and k being arbitrarily chosen) is identical for both TM and TE polarizations. It can be shown that this condition of constant phase difference for the two orthogonal polarizations yields the following formula defining the compensator shape:
, j 1 imδλ a j — aQ = ■
2 δns - δnc
wherein d4 is the distance between the end of the array waveguide i and the point where the compensator boundary intersects a line joining the end of waveguide i and the beginning of the central output waveguide. The locus of these intersection points defines the boundary between the compensating and non- compensating parts of the output coupler; d0 = dj (i =0); m is the AWG order; δλ =
λte - λta is the wavelength shift between the TE and TM spectra to be compensated; δns = ns/te - nS tm and δnc = nc/te - nc/tm is the effective index birefringence of the slab waveguide and the compensating region, respectively.
[0021] As explained above, this compensator formula assumes splitting the compensator in two parts, one in the input coupler and one in the output coupler. Given the symmetry of the AWG device and the phase conditions assumed when deducing the formula, the two parts of the compensator should be mirror images of each other along the symmetry axis of the AWG device. However, such symmetric design can result in total internal reflection (TIR) on one of the compensator-slab boundaries, degrading device performance. In the case when the compensating regions are etched, the effective refractive index of the compensator region is lower than that of the non-etched region of the combiner. Such design can result, in the input coupler, in total internal reflection (TIR) for some rays refracted at the slab /compensator boundary, as they pass from a higher effective index (non-etched) to a lower index (etched) regions. It should be noted that in contrast to the situation in the input coupler, in the output coupler light propagates first through the lower index and then through the higher index regions. Consequently, TIR cannot arise in the output coupler when etched compensators are used. In the case when compensator is created by deposition of an over-layer, similar arguments applies, but because in such case the effective refractive index of the compensating region is higher than that of the original slab waveguide, TIR may occur in the output coupler while TIR cannot arise in the input coupler.
[0022] Fig. 3 shows the intensity reflectance versus the angle of incidence assuming 0.4 mm deep etched compensator in 1.5 mm thick Si slab waveguide. It is observed that the reflectivity significantly increases for angles of incidence above 80°, with TIR occurring at ~83.1°.
[0023] To avoid possible TIR in one of the couplers, the compensator etching or the overlayer depostion are reversed in the coupler in which TIR may
occur, so that in the actual device light propagates first through the lower index and then through the higher index region of the latter. To maintain the same polarization compensation however, the compensator region in that particular coupler must also be reflected about the symmetry axis of the latter. This results in asymmetric compensator regions shown in FIG. 2.
Example
[0024] The following example illustrates the effectiveness of the novel technique even when compensating very large polarization dependent wavelength shifts that are typical for compact silicon-on-insulator (SOI) phasar- based devices.
[0025] AWG demultiplexers were fabricated on SOI substrates with a 1.5 μm thick Si layer of (100) orientation on a 1 μm thick Si02 layer. FIG.4 and FIG. 5 show SEM images of different parts of the fabricated device. Polarization dependent wavelength shift of a non-compensated device was δλ = λte - λta = 2.22 nm. This large red shift of the TE spectrum with respect to the TM one was reduced to δλ = 1.78 nm upon the first 0.124 μm deep etch step and further reduced to 0.54 nm by etching compensator additional 0.218 μm.
[0026] By further trimming, polarization dispersion was reduced down to
0.04 nm, as shown in FIG. 7 where TE and TM spectra are depicted for a channel near the AWG central wavelength after full-compensation. No crosstalk performance degradation was observed for a fully compensated AWG. It was found that compensator induces a significant red shift of ~5 nm in the peak wavelength. This red-shift is expected because of shifting the focus position for a particular wavelength at the output coupler focal line is induced by refraction at the compensator/coupler boundary. An adjustment in AWG design can easily be made to account for this shift. It was also found that overcompensation of the demultiplexer is possible by etching the compensator an additional 0.22 μm, yielding δλ = -3.65 nm. The spectrum of the compensated device is shown in
FIG. 7.