A compact passive polarization converter optimized foractive-passive integration on InP/InGaAsP

L.M. Augustin, J.J.G.M. van der Tol, E.-J. Geluk, M.K. SmitCOBRA Research Institute, Technische Universiteit Eindhoven

Postbus 513, 5600 MB Eindhoven, The Netherlandsl.m.augustin@tue.nl

Abstract: An improved design for an integrated polar-ization converter is presented. The device is speciallysuited for monolithic integration with active and pas-sive components on InP/InGaAsP. A novel simplifiedfabrication process is shown. Measured polarizationconversion > 97% agrees well with simulations.

IntroductionPolarization handling is of ever greater importance inmodern telecommunications networks. First of all a lotof components in the network are highly polarizationdependent, furthermore polarization mode dispersioncan degrade the transmission in an optical fiber. On theother hand, the polarization can be employed in e.g.polarization multiplexing, and polarization based fil-tering. In all these cases, polarization converters (PC)are key-elements.Passive polarization converters that are able to be in-tegrated with both active and passive components arepreferred. Polarization conversion can be obtainedby periodically loaded waveguide sections [1, 2, 3],by using integrated bends [4], or by using a singlewaveguide section with a vertical and a slanted side-wall [5, 6, 7].One problem of these designs is the difficulty tointegrate them with passive and specifically activecomponents. The polarization converters and activecomponents (such as semiconductor optical amplifiers(SOAs) and phaseshifters [8]) can be made in thesame layerstack, but the optimal thickness of the top-cladding differs by more than 1 µm. For polariza-tion converters a large birefringence is needed: hencea thin topcladding is preferred (typically smaller than300 nm); for SOAs and electro-optic phaseshifters, athick topcladding (typically 1500 nm) is needed belowthe contact to avoid optical losses in the contactlayer.This is clarified in Fig. 1: an SOA, phaseshifter anda conventional polarization converter are shown. Crit-ical lithographical definition is not possible with theseheight differences which complicates processing andintegration of these devices.A new design is proposed as is shown in the rightmostpicture in Fig. 1. This new proposed design uses thesame topcladding as an active device and can be eas-ily integrated with the other components. The bire-fringence with the thicker cladding is only slightly re-duced with respect to the thinner cladding. Hence theincrease in length for this design is less than 40µm.

Fig. 1: Cross section of active devices (SOA, phase-shifter), a conventional PC and the new design.

DesignThe layer stack consists of an InP substrate, a 500 nmInGaAsP[Q(1.25µm)] waveguide layer and a 1500 nmInP topcladding. Simulations with the commercialwaveguide solver FIMMWAVE predict an optimalconversion of TE to TM and vice versa larger than 99%for a width of 0.94 µm and a length of 141 µm. A con-version above 95% is expected for a width deviation of100 nm.Deviations from the calculated width and length are ex-pected, because the refractive indices used in the sim-ulations are not accurately known for these materials.The total device contains an asymmetric waveguideas the converter section, 1.2 µm wide deep input andoutput waveguides, coupled to 3 µm wide shallowwaveguides via 75 µm long tapers (Fig. 2). At theinterface of the input waveguides and the polarizationconverter section a small ridge is present, perpendicu-lar to the waveguide. This is needed to prevent etchingof the input waveguide in the wet etch for the slantedsidewall.

Fig. 2: Overview of the polarization converter includ-ing in- and output waveguides and tapers.

WA3

FabricationThe polarization converters have strict tolerances, sothe converter sections will be defined using Elec-tron Beam Lithography (EBL). EBL is not suited towrite large circuits, therefore all other waveguides willbe defined using standard optical lithography. TheEBL written parts have to be aligned to the opticalwaveguides. This can be done by the writefields withalignment marks as shown in Fig. 3.

Fig. 3: Optical waveguides and EBL writefields in-cluding EBL alignment marks.

The processing scheme of the polarization converter isshown in Fig. 4.

a. First the waveguides and writefields for the EBL,including alignment marks (Fig. 3), are definedoptically into a Silicon Nitride (SiNx) mask. Nextthe polarization converters are defined in Ti on topof the SiNx using EBL and a lift-off process.

b. Before etching any waveguide, the second EBLstep is done to open the straight side of the PC

c. Now the nitride at the straight side of PC isopened. The shallow waveguides and the straightside of the PC are etched with CH4/H2 ReactiveIon Etching (RIE).

d. All shallow waveguides are covered with resist,the PC area is opened with a non-critical opti-cal lithography step. The SiNx at the sloped sideof the PC can be openend and RIE etched until300 nm above the waveguide layer, while etchingthis side, the straight side is etched even deeper,well below the waveguide layer.

e. Silicon Nitride is deposited on the whole sam-ple and the shallow waveguides are again cov-ered with resist. The SiNx is etched back usingCHF3 RIE. Because of the directional etching, theetched sidewalls stay covered with SiNx, whichserves as a mask for wet etching.Br2-Methanol is used to etch the slope. Thisetchant etches both InP and InGaAsP with an an-gle of 54o with respect to the surface.

f. Finally all the nitride is removed using an HF so-lution.

(a) Define waveguides and write-fields with optical lithography.Define PC in Ti with EBL.

(b) Open straight side of PC.

(c) Etch straight side of PC andall standard waveguides.

(d) Cover all standardwaveguides. Open slopedside of PC and RIE etch till300 nm above waveguide layer.

(e) Cover sidewalls by Silicon-Nitride. Br2-Methanol is used toetch the slope.

(f) Remove all nitride.

Fig. 4: Processing scheme of the integrated Polariza-tion converter.

The fabricated converters are shown in Fig. 5. Fromthese figures it is clear that there is an underetch at theshallow side (the sloped side), but there is no underetchat the deep side. This is most probably caused by stressin the masking material during the wet-etching. The in-fluence of this additional underetch on the conversionperformance of the device will be very small as can beunderstood from Fig. 6. The field is located in the In-GaAsP layer, below the sloped side. The width of thepolarization converter at the position of the waveguide-layer is not strongly affected and thus the tilt of themodes is still correct. However, because the top of thefieldprofile is influenced, the coupling losses will beincreased with respect to the design.

MeasurementsThe polarization converters are measured using asetup as shown in Fig. 7. The device is excited usingan EDFA with a bandpass filter set to 1555 nm asa source. The filter has a 2 nm bandwidth, largeenough to prevent Fabry-Perot resonances. A polarizerat the input of the chip is used to select the inputpolarization. At the output another polarizer selectsthe polarization that is measured using the photodiodeand the lock-in amplifier.

WA3

Fig. 5: SEM photographs of the integrated Polarizationconverter. Top: transition of a straight waveguide to thepolarization converter. Bottom: reverse transition.

Fig. 6: Field profile of a mode in the polarization con-verter, the structure is indicated by the solid line, thedashed line shows the etched topcladding caused bythe underetch.

Polarizer

EDFA BPF PD

DUT Polarizer PinholeChopper

Lock-inamplifier TIA

Fig. 7: Setup used for characterization of the polariza-tion converter. BPF: Bandpass filter, PD: Photodiode,TIA: Trans impedance amplifier

The power in both polarizations at the output is mea-sured for both TE and TM polarized light at the in-put. The conversion is defined as the fraction of theconverted polarization in the output power. The con-version C for the two polarizations (i, j = TE, TM) isdetermined from the following equations:

Pi j = αiα j C P j ; i , j (1)Pi j = αiα j (1 −C) P j ; i = j (2)

where P j is the input power with polarization j; Pi j isthe output power in polarization i when the input polar-ization is j; αi, j are the losses for the two polarizationsin the input and output waveguides.By solving these equations, the propagation losses inthe input and the output waveguides for the two po-larizations are eliminated and the conversion C is ob-tained:

C =√

x1 +√

x(3)

wherex = Pi jP ji

PiiP j j; i , j (4)

This conversion is plotted as a function of width andlength in Fig. 8.

0.75 0.8 0.85 0.9 0.95 1 1.050

0.2

0.4

0.6

0.8

1

Width [µm]

Con

vers

ion

measurementssimulation

0 100 200 300 400 5000

0.2

0.4

0.6

0.8

1

Length [µm]

Con

vers

ion

measurementsfitsimulation

Fig. 8: Measured conversion as a function ofwidth(top) and length(bottom).

The maximum conversion from TE to TM and viceversa occurs at 131 µm length, corresponding to the

WA3

half beat length between the modes of the convertersection, and back to zero at the full beat length(262 µm). The maximum conversion, for this deviceis 97.5%.The large scattering on the measured values is causedby the underetch, this not uniform throughout thewhole device. Hence a non-uniform width is obtainedalong the converter section.The losses of the polarization converter including inputand output tapers are measured to be 2.4±0.3 dB for TEand 2.6±0.3 dB for TM. This value is higher than sim-ulated (less than 0.5 dB), most probably caused by thenon-optimal coupling of the waveguide and the con-verter because of the underetch. This can be optimizedby using a more controlled wet etch.

ConclusionA new type of polarization converter is shown, spe-cially suited for easy integration with other active andpassive devices. The device is fabricated and mea-sured. The polarization converter shows a maximumconversion larger than 97%.

AcknowledgmentsThis research is carried out in the framework of theIST-MUFINS project. The European Commission isgratefully acknowledged for financial support. TheCentre for Integrated Photonics in Ipswich, UK iskindly acknowledged for supplying the wafer for thisexperiment.

References[1] J.J.G.M. van der Tol et al, “Realization of a short

integrated optic passive polarization converter,”IEEE Photon. Technol. Lett., vol. 7, pp. 893–895,Aug. 1995.

[2] Y. Shani et al, “Polarization rotation in asymmetricperiodic loaded rib waveguides,” Appl. Phys. Lett.,vol. 59, pp. 1278–1280, Sept. 1991.

[3] H. Heidrich et al, “Passive mode converter witha periodically tilted InP/GaInAsP rib waveguide,”IEEE Photon. Technol. Lett., vol. 4, pp. 34–36,Jan. 1992.

[4] C. van Dam et al, “Novel compact polarizationconvertors based on ultra short bends,” IEEE Pho-ton. Technol. Lett., vol. 8, pp. 1346–1348, Oct.1996.

[5] F.H. Groen et al, “Compact polarisation con-verter on InP/InGaAsP using an asymmetricalwaveguide,” in Proc. 11th Eur. Conf. on Int. Opt.(ECIO ’03), pp. 141–144, Prague, Czech Repub-lic, April 2–4 2003.

[6] H. El-Refaei et al, “Slanted-rib waveguideInGaAsP-InP polarization converters,” J. Light-wave Technol., vol. 22, pp. 1352–1357, May 2004.

[7] U. Khalique et al, “Ultrashort polarization con-verter on InP/InGaAsP fabricated by optical litho-graphy,” in Technical Digest Integr. Photon. Res.and Apps. (IPRA ’05), p. IWA3, San Diego, USA,Apr. 11–Apr. 13 2005.

[8] J.J.M. Binsma et al, “InP-based photonic integra-tion technology,” in Technical Digest Integr. Pho-ton. Res. (IPR ’04), p. IFB1, San Francisco, USA,Jun. 30–Jul. 2 2004. Invited paper.

WA3