GB2397390A - Arrayed waveguide grating with flat spectral profile - Google Patents

Abstract

An Arrayed Waveguide Grating (AWG), such as used in optical communications for separation/combination of different optical wavelengths (demulti/multiplexing), has a free carrier or other loss inducing dopant dispersion effect to give a relatively flat spectral profile. The dopant may be phosphorus.

Description

FLAT OUTPUT RESPONSE ARRAYED WAV:EGUIDES GRATING (AWG) USING DOPING OR

DAMAGE TO ACHIEVE A SPF,ClFIED OUTPUT

PROFILE

Field of Invention

I'he present invention relates to a method of achieving flat spectral response in an Arrayed Waveguides Grating (AWG) such as used in optical communications for a variety of application, notably separation/combination of different optical wavelengths (demulti/multiplexing). This can be achieved by selectively introducing loss to the array waveguides. We give a particular example in a silicon AW(J by introducing the free carrier dispersion el'fect at the array waveguides section of the device, but the approach is also appropriate to other semiconductors, or to other materials when doping may cause loss via absorption, or the dopant could be introduced via ion implantation to cause loss due to material damage.

Background of Invention

The Arrayed Waveguides Grating (AWL) is used in Wavelength ivisi(-)n Multiplexing (WDM) optical communications systems. In such systems, difl'erent wavelengths are transmitted on a single optical fibrc, hence the need to multi/demultiplex multiple wavelengths has gained AWG significant popularity due to it low insertion loss, high stability high wavelength selectivity and good mass- producibility. The AWG has proven itself to be very flexible, being utilized in a number of configurations. For example, multiplexing, demultiplexing, add-drop multiplexing and N*N interconnects.

The generic schematic diagram of the AWG is shown in Figure 1 and the principle of operation is outlined as follows: An AWG comprises two slab regions (planar waveguides), which act as passive star couplers, and an array of ribs waveguides, of progressively increasing length. The increasing lengths typically introduce phase differences between adjacent waveguides of 2m at the centre wavelength. The first planar region is to excite the array of waveguides, and the second planar region is to allow multiple beam interference from the outputs of the array waveguides. If a single wavelength is introduced into one ol' the input waveguides, this wavelength is distributed to the array of rib waveguides via the first planar region. The purpose of the array rib waveguides is simply to add incremcutal phase shifts to each ray emerging from successive waveguides. This is achieved because each of the array waveguides is slightly longer than it neighbour. Therefore, the increasingly longer waveguides in the AWG introduce increasingly greater phase shifts. The difference in length between adjacent waveguides is kept constant, so that the phase shift 'added' to the each successively emerging ray of light is also constant. Hence our wavelength propagates through these waveguides, each ray emerging with the incremental phase shift due to length of the waveguide in question.

In the second planar region, the beams emerging from the array waveguides interfere to produce a pattern with a single principal peak that spatially coincides with one of' the output waveguides. If the second wavelength is introduced into the same input waveguides, it will also be distributed to the array of waveguides in the same way as z the first wavelength. It will pass through the array of rib waveguides, but will experience a different phase shift through these waveguides, due to the different slightly propagation constant associated with each wavelength. Once again the various components of the second wavelength will interfere in the second planar region resulting in another interference pattern. Due to the slightly different amount of incremental phase for the second wavelength, the interference pattern will have a peak that will occur at a different output waveguide, separating the two wavelengths.

Multiple wavelengths can be separated in this way, producing a wavelength demultiplexer, placing each wavelength at a different output waveguide. A typical response of such device is shown in Figure 2. Due to the narrow 'peak' of such response, it imposes a deficiency in WDM networking systems for the following reasons; Firstly, small fluctuation in the wavelength of a laser source causes a large loss. Secondly, the transmission loss of a conventional AWG increases monotonically around the centre wavelength of each channel. This places tight restrictions on the wavelength tolerance of laser diodes and requires accurate temperature control for both the AWG and the laser diode. Thirdly, due to the multiple uses of filters in a WDM ring/bus network, the cumulative passband width of each channel becomes much narrower than that of a single stage AWG filter. Hence, a flat spectral passband Wit]lin each channel becomes one of the most desirable characteristics of the AWC,.

Several methods have been proposed in designing a flat spectral response, however each has it own inadequacy. Okamoto et al [2] shows that by introducing excess lengths to the array waveguides, the field distribution follows a sin(x)/x function at the output of the array waveguides. However, the control of phase and transmission of individual waveguides poses great difficulties due to fabrication, temperature and radiation losses variation. He also proposed the used of parabolic waveguides horns at the input waveguides to flatten the spectral responses of the AWG multiplexer 13].

In U.. Pat No. 5,629,992, an alternative method of producing a flat spectral response is with the help of a multimode interference coupler (MMI) integrated at the entrance of the star coupler. Yet such technology increases the complexity of AWG and is not always compatible with some types of AWG, as it can only be implemented to one input port due to the widened input waveguide. For AWGs with more than one input waveguide, the device size has to be greatly enlarged to fit the MMI design.

The flattening of passband responses can also be achieved with help of multiple focusing points. The approach of such method is to tailor the focused image by varying the end positions of array waveguides so that superposing multiple images spatially, separates images at the output waveguides forming a desirable 'net' image.

Y. IIo. et al L41 showed that instead of placing all array waveguides on one grating circle, half of the array waveguides are placed on one grating circle and the other half on another. The two sets of array waveguides consequently form two spatially separated images. If the device is design such that two images are formed close enough to each other, the superimposition of the two fields will result in a relatively flat spectral response. IIowever, this approach increases the physical dimensions of the AWG and complexity of the fabrication process significantly.

Summary of the Invention

An AWG includes a pair of star couplers with at least one input/ output waveguide, interconnected by an array of waveguides, of varying length. The basic principle of achieving flat spectral responses is to obtain a Sinc function at the output of the array waveguides, point B (Figure 3). This field distribution is the Inverse Fourier Transform of the field at C; hence the field at C can be specified with an ideal rectangular function (i.e. a flat response). The field distribution across the array waveguides usually resembles a gaussian (see Figure 3) after propagation through the first star coupler. Therefore, in our invention we introduce free carriers (or other loss inducing dopant) selectively into the array waveguides, to convert the gaussian distribution to the Sinc function, by selectively absorbing light in the doped array waveguides.

With the introduction of such a methodology, we must control the dopants with a sufficient degree of accuracy, hence improving the transmission and phase variation of each individual waveguide.

Detailed Description of the Invention (Preferred embodiment) The Arrayed Waveguide Grating (AWG) according to the prior art is shown in Figure 1. When the device performs a demu]tiplexing function, different wavelengths are directed into the input waveguide and are separated at the output waveguides. leach of these wavelengths will be focussed at a predetermined output waveguide. In our preferred embodiment, we will be looking at a speci kc example in silicon- on- insulator (SOT). The waveguides are usually rib structures; these will have height and width ratio tailored to satisfy the singlemode condition [51 and should display minimum polarisation dependence The intensity field distributions produce at the interface between the input star coupler and the array waveguides is a bell-like or gaussian-like shape as shown in Figure 4 due to the dispersive property of' the singlemode input to singlemode output waveguides. When each of these rays propagates through the array waveguides, the array maintains a 2m phase difference between adjacent waveguides at the ccntre wavelength, hence allowing constructive interference at the output star coupler.

Figure 2 shows the modellcd output of an 8-channel AWG, but much larger devices are also possible. In order to obtain a flat spectral response, it is necessary to produce a sin(x)/x (Sine x) field intensity function across the output of the array waveguides.

['his is due to the fact that this field intensity is the Inverse Fourier Transform oi'thc field at the output waveguides. Consequently, to shape the field acquired in Figure 4 to the sin(x)/x function, we have selectively introduced loss into the array waveguides via l'ree carrier absorption. The free carriers can be introduced into the SOI waveguides using ion implantation for example.

According to Soref and Bennett 16], the refractive index and absorption changes in silicon, at communications wavelengths of 1.31lm and 1.551lm are given as:

-- -

At 1.3m An = Ane + Anh -[6 2*10-22 AN +6.0*10 (both) I (1) 6t = Able + A(Xh = 6.0 10 ANe + 4 0 10 ANh (2) At 1.551lm An = Ane + Anh =-[8.8*10 AN +8 5*10-18 (AN)08] (3) 6 = Ace! + Arch = 8.5 10. ANT + 6.0 * 10 18, (4) where: Ane-- change in refractive index resulting from change in free electron carrier concentrations Ant, = change in refractive index resulting lrom change in free hole carrier concentrations Ape = change in absorption resulting from change in lyre electron carrier concentrations Ear, = change in absorption resulting from change in free hole carrier concentrations The loss due to absorption can be specified in dBs as: Loss(dB) -- l Olog(exp(Aa*L)) (5) Where L is the active doping length of the waveguide.

The design methodology will be explained in conjunction with Figure 5. Each individual array waveguide is doped with a certain doping length and carrier concentration (either holes or electrons) based on the intensity difference between the conventional field distribution (Figure 5a) and the sin(x)/x field distribution across the output of the array waveguides (Figure fib). The necessary loss required in each array waveguide is determined by subtracting the Sinc function from the gaussian function, as shown in Figure 5c. In this embodiment, we will illustrate our theoretical analysis with a specific carrier concentration, for example, in this particular case, lel8cm3 Phosphorus, although other doping elements and/or concentrations may be used. Each arrayed waveguide will be doped with a particular length to induce the losses. However, because the absorption of light also results in a change in the refractive index of the material, the waveguides experience an unwanted change in phase. As a consequence, an additional path length is needed to compensate such changes.

Following expression (5), we can calculate the active Loping length and associate phase changes due to ion implantation using expression (6) and (7).

LaCjVe = In ( ] 0 ss/ I) / Aoc (6) A = 2.An.Lac,ive / Note that expression 6 is only an approximate expression for the change in phase.

Figure 6a shows a graph of the required doping length for each array waveguide and Figure 6b shows the-unwanted phase change induced as a result of the ion imp] antation.

In accordance with the present invcotion, it has been determined that in order to maintain a 2m phase difference between adjacent array waveguides at the centre vavelength, the phase change induced by the free carriers has to be compensated by an appropriate additional physical path length illustrated in Figure 6c. This is a consequence of the fact that in silicon, doping causes the refractive index to be reduced. Hence to restore phase, and additional path length must be introduced. If another dopant wore used that increased refractive index, or in other materials, if the refractive index is increased by doping, a shortening of the array waveguidcs would be necessary to restore the appropriate phase relationships. Hence, for the specific example discussed here, the appropriate geometrical layout of the array waveguides to include the compensating extra waveguide lengths is as shown in Figure 7.

Alternative embodiments In an alternative embodiment of the present invention, the use of different dopants and concentration levels resulted in a different geometry layout of the device. Two example designs using 9cl 7cm3 and 2el Scm3 Phosphorus are illustrated in Figure 8a and 8b respectively. The difference in concentration level of the dopants will introduce different absorption losses; hence the doping length, phase change and compensating optical path lengths will all be different. 'I'his will also be the case if different dopants are used.

It is important to note that, in our preferred embodiment, only communication wavelengths are discussed. However, we can employ this design metll-,dology to any other wavelengths because wavelength is a function of carrier refraction and absorption. That is, a change in wavelength will result in a dil'f:'erent doping length, phase change and optical path length. Thus the wavelength dependence of the dcmultiplexing function will result in output waveguides that are not evenly spaced.

Another important point to note is, by introducing absorption loss to each array waveguide, we can ultimately shape any kind of spectral response in taking tile Inverse Fourier Transform of the targeted response and subtract it from the conventional, "gaussian-like" response.

In our preferred embodiment, we encompassed our example with Silicon-on Insulator (SOT) because this material is of great interest in the field of optical communications.

. ' . , ' 1. I ' 1 However, our Design can oe appnea to any omer semlconauclor. we can also Include materials that are insulators such as silica (glass), or Lithium Niobate (LiNbO,) in which we can introduce absorption either through doping or crystal damage. In either case the doping can be performed by ion implantation, or some other doping mechanism, and the principles of the design procedures still hold.

The careful choice of a particular dopant can result in the elimination of polarization dependence. Since refractive index is changed by the introdueti-'n of free-earriers, this also results an effective index change of the TE and TM modes. thence by careful consideration of the change in effective index for each polarization, we can achieve a polarization independent device.

Although the present invention has been spceifieally described in a particular material, using free-earrier type absorption changes, at fixed concentration levels, it is clear that many variations can be made by someone skilled in the art within the.sccpe and spirit of the invention.

Rcfcrcaccs: [1] K. Olcamoto; Fundamentals of Optical Waveguides. Academic Press, ISBN ()- 12-525095-9. 2000 L2] K. ()kamoto, H. Yamada; Arrayed-waveguide grating multiplexer with flat spectral response. Optics I,etters. Vol 20. No 1. January 1995. 43-45.

[31 K. Okamoto, A. Sugita; Flat Spectral Response Arrayed-Waveguide Grating Multiplexer with Parabolic Waveguide Horns. Electronic Letters. 32. 1661 -1662. 1996.

[4] Y. P. Ho, Y. J. Chen; Flat Channel-Passband-Wavelength Multiplexing and Demultiplexing Devices by Multiple-Rowland-Circle Design. IEEE Photonics Technology Letters. Vol. 9. No 3. March 1997. 342-344.

51 R. A. Soref, J.Sachmidtchen, K. Petermann; Large Single Mode Rib Waveguides in (JeSi-Si and Si-on-SiO2, Journal of Quantum Electronic, 27, 1971 - 1974, 1991.

[6] Soref, R. A, Bennett, B.11.: Electroptical Effects in Silicon. IEEE J. QE-23 (1987), 123-129.

Claims (1)

1 -, . ,'. , r r r r r Patent Reference: GB 0227881.0 29'h November 2002

Claims 1. An arrayed waveguide grating, principally for use in wavelength division multiplexing in optimal communications systems, in which free carriers, or other loss-inducing dopant, converts the transmission intensity distribution at each output waveguide into a relatively Oat spectral profile.