Circular grating resonators as nano-photonic modulators

Circular Grating Resonators as Nano-Photonic Modulators

Nikolaj Molla, Sophie Schonenbergera, Thilo Stoferlea, Thorsten Wahlbrinkb, Jens Boltenb,Thomas Mollenhauerb, Christian Moormannb, Rainer F. Mahrta, and Bert J. Offreina

a IBM Research, Zurich Research Laboratory, Saumerstrasse 4, 8803 Ruschlikon, Switzerland

b Advanced Microelectronic Center Aachen (AMICA), AMO GmbH, Otto-Blumenthal-Strasse25, 52074 Aachen, Germany

ABSTRACT

Circular grating resonators could lead to the development of very advanced silicon-on-insulator (SOI) basednano-photonic devices clearly beyond state of the art in terms of functionality, size, speed, cost, and integrationdensity. The photonic devices based on the circular grating resonators are computationally designed and studiedin their functionality using finite-difference time-domain (FDTD) method. A wide variety of critical quantitiessuch as transmission and reflection, resonant modes, resonant frequencies, and field patterns are calculated.Due to their computational size some of these calculations have to be performed on a supercomputer like amassive parallel Blue Gene machine. Using the computational design parameters the devices are fabricated onSOI substrates consisting of a buried oxide layer and a 340-nm-thick device layer. The devices are defined byelectron-beam lithography and the pattern transfer is achieved in a inductively coupled reactive-ion etch process.Then the devices are characterized by coupling light in from a tunable laser with a lensed fiber. As predictedthe measured transmission spectra exhibit a wide range of different type of resonances with Q-factors over 1000which compares very well with the computations.

1. INTRODUCTION

For future applications, such as chip-to-chip or on-chip optical interconnects, a high physical integration densityis necessary. Exploiting integrated micro-cavities could lead to such highly dense and novel optical devices.1–4

This paper reports on a specific type of integrated micro-cavities, namely circular grating resonators (CGR).The design of such a CGR is illustrated in Fig. 1. It consists of concentric rings of high index material surroundinga center disc which acts as a defect. The circular grating around the center disc acts as a Bragg mirror andimposes a photonic band gap, i.e. stop band, which prevents lateral leakage of light from the cavity. Theresonance frequency which has to lie inside the band gap is then mainly determined by the radius of the centraldefect. The waveguides which penetrate the circular grating act as in-coupling and out-coupling waveguides.The penetration of the circular grating improves the coupling from the waveguides to the defect in the center.Light which not propagates from the defect into the direction of the waveguides will be reflected back.

Even at relatively low refractive index contrasts CGRs offer full two-dimensional light confinement. Theycan have a very small footprint of a few micrometers, which essentially corresponds to one of the smallest opticalresonators. Because the volume in which the light is confined is so small, also the refractive index change whichhas to be induced to switch can be very small. Therefore, it is expected that the use of CGRs will enablehigh-speed electro-optical modulators to be built. All in all, this makes CGRs a very promising candidate forfuture integrated photonic devices.

2. DESIGN OF THE CIRCULAR GRATING RESONATORS

To achieve a small cavity volume, which is imperative for high switching speeds, we optimize the geometricparameters of the circular grating to maximize the band gap. A larger band gap leads to a larger reflectivityR of the circular grating for the same number of gratings. Therefore, the light decays faster inside the circulargrating and decreases the spatial extension of the cavity mode.

The band gap of the circular grating can be approximated by the corresponding linear grating which inprinciple is just the cross-section of the circular grating. The band structure of this linear grating is computed

Invited Paper

Silicon Photonics III, edited by Joel A. Kubby, Graham T. Reed Proc. of SPIE Vol. 6898, 68980B, (2008) · 0277-786X/08/$18 · doi: 10.1117/12.766290

Proc. of SPIE Vol. 6898 68980B-12008 SPIE Digital Library -- Subscriber Archive Copy

SiO2

Si

Figure 1. A CGR consisting of concentric rings of high index material with in-coupling and out-coupling waveguides.

using a simulation tool MIT Photonic Bands (MPB),5 which is based on a frequency-domain technique. Itperforms a direct computation of the eigenstates and eigenvalues of the full Maxwell’s equations using a planewave basis.

Fig. 2 shows the cross section of the linear grating. The duty cycle D = q/a is the length of the air sectionrelative to the period of the circular gratings, i.e., the lattice constant a. Together with the height h of the Sistructure, the duty-cyle D, and the lattice constant a they are the three free parameters for the mirror structure.These parameters are varied to obtain the largest possible band gap centered around the telecom wavelength of1550 nm. Here we focus only on the TM polarization.

The results of our calculations are shown in Fig. 3. There is a maximum band gap of 12% (band-gap widthdivided by center wavelength) for a thickness of approx. 320 nm. For thicknesses larger than 420 nm, even largerband gaps can be achieved.

Because SOI wafers with a Silicon thickness of 340 nm are easily available, we chose this height for our devicedesign. Furthermore, for this thickness the band gap is close to the local maximum in Fig. 3. In Fig. 4 the bandgap and the lattice constant are plotted as a function of the duty cycle for a Silicon thickness of 340 nm. The

z

r

Si

SiO2SiO2

h

q

a

AirAir

Figure 2. The cross section of the linear grating with a height h, a lattice constant a, and a air gap q. The duty cycle isgiven by D = q/a.

Proc. of SPIE Vol. 6898 68980B-2

340nm

0

0.1

0.2

0.3

0.4ba

nd g

ap

0 100 200 300 400 500 600 700 800

height h (nm)

Figure 3. The band gap of the linear grating centered around the telecom wavelength of 1550 nm as a function of the Sithickness h. Each of the circles represents one configuration with a certain duty cycle D = q/a, lattice constant a andheight h.

linear grating which maximizes the band gap for a Si thickness of 340 nm has a duty cycle of 0.14 and a latticeconstant of 350 nm. For the fabrication of the devices we choose little different parameters which are easier tofabricate: a duty cycle of 0.3 and a lattice constant of 400 nm.

To achieve the target specification of switching on a picosecond timescale, we designed a CGR with a quality-factor Q of a few thousands. This gives us a target reflectivity R of the circular gratings of approx. 85%, whichcorresponds to approx. 6 to 8 circular gratings around the defect.

3. CHARACTERIZATION OF THE FABRICATED DEVICES

The CGR devices are fabricated on SOI substrates consisting of a 2-µm-thick buried oxide layer and a 340-nm-thick device layer. The devices have been defined by electron-beam lithography (EBL) using HydrogenSilsesquioxane (HSQ) as negative tone resist. Then pattern transfer has been achieved in a HBr-chemistry-basedinductively coupled reactive-ion etch process. A two-step process has been used. First, in the main etch step,pure HBr has been used. The etching time for this step is set to stop at a residual top-Si thickness of 20 nm.Then the etch conditions were changed to the second or overetch step. For this part a mixture of HBr and O2

is used. The overetch step exhibits an excellent selectivity between the top-Si and the buried oxide. The entireprocess flow, including both lithography and etching, has been optimized to minimize the surface roughness ofthe CGRs.

Proc. of SPIE Vol. 6898 68980B-3

0 0.2 0.4 0.6 0.8 1duty cycle D

band

gap

0 0.2 0.4 0.6 0.8 1200

300

400

500

600

latti

ce c

onst

ant a

(nm

)

lattice constant

band gap

0

0.05

0.1

0.15

0.2

Figure 4. The band gap and the lattice constant are plotted as a function of the duty cycle for a Silicon thickness of340 nm.

A scanning electron microscope (SEM) of a CGR device is shown in Fig. 5. We characterize these CGRdevices by coupling light in from a tunable laser with a lensed fiber. The output light then is collected withanother lensed fiber. The transmission spectrum measured is shown in Fig. 6. For each defect radius thetransmission spectrum shows a couple sharp resonances with quality factors of a few thousands and a contraston/off (corresponding to peak height) of approx. 10 to 15 dB. Varying the defect radius shifts these resonancesin the wavelength in regularly fashion. Some other features of the transmission spectra of the CGRs howeverremain the same when varying the defect radius. These features are not due to transmission through the defect.

Looking at the polarization dependency of the device we note different transmission spectra for TE and TM.As expected, this dependency vanishes outside the band gap. To obtain the absolute transmission of the device,the transmission spectrum must be normalized to the adjacent straight waveguides. When measuring differentstraight waveguides, we observe that most of them have qualitatively the same spectral shape of the transmission.The absolute transmission however varies around 10 dB due largely uncontrolled input coupling or variations inthe waveguides. Thus, we inherit the uncertainty to the absolute transmission. Depending on whether if we usethe straight waveguide above or below the circular grating we obtain a total transmission between 8% and 60%.Using tapers at the end of the waveguides will improve the coupling and result in more reliable results.

Because variations on the order of a few nanometers can significantly affect the device performance, weevaluate the reproducibility of the fabrication process from chip to chip by measuring the same CGR deviceon different chips. The spectrum shape and relative positions of the resonant features are nicely reproduced.However, we observe a 5-nm shift in absolute wavelength for all resonances, which is about 0.3% of the wavelength.For such a shift, the radius of central defect size has to be changed by 3 nm. This agrees very well with theexpected tolerances of the fabrication process.

Proc. of SPIE Vol. 6898 68980B-4

2pm2 µm

Figure 5. A SEM of a CGR device fabricated in Si.

4. CONCLUSIONS

We designed circular grating resonator (CGR) devices computationally and studied their functionality. Using thecomputational design parameters the CGR devices were fabricated in SOI. The transmission spectra measuredexhibit a wide range of different type of resonances with Q-factors higher than 1000. CGRs can have a verysmall footprint of a few micrometers, which essentially corresponds to one of the smallest optical resonators andmakes CGRs a very promising candidate for future integrated photonic devices.

Acknowledgments

The authors would like to acknowledge the support by the EU-Project Circles of Light.

REFERENCES1. J. Niehusmann, A. Vorckel, P. H. Bolivar, T. Wahlbrink, W. Henschel, and H. Kurz, “Ultrahigh-quality-factor

silicon-on-insulator microring resonator,” Opt. Lett. 29, pp. 2861–2863, Dec. 2004.2. V. R. Almeida, C. A. Barrios, R. R. Panepucci, M. Lipson, M. A. Foster, D. G. Ouzounov, and A. L. Gaeta,

“All-optical switching on a silicon chip,” Opt. Lett. 29(24), pp. 2867–2869, 2004.3. V. R. Almeida, C. A. Barrios, R. R. Panepucci, and M. Lipson, “All-optical control of light on a silicon chip,”

Nature 431, pp. 1081–1084, Oct. 2004.

Proc. of SPIE Vol. 6898 68980B-5

1520 1540 1560 1580 1600

wavelength λ (nm)

tran

smis

sion

1500 1620

830 nm

820 nm

810 nm

800 nm

rin

Figure 6. Measured transmission spectra of CGR devices with defect radii from 770 nm of 800 nm.

4. A. Gondarenko, S. Preble, J. Robinson, L. Chen, H. Lipson, and M. Lipson, “Spontaneous emergence ofperiodic patterns in a biologically inspired simulation of photonic structures,” Phys. Rev. Lett. 96(14),pp. 143904–1–143904–4, 2006.

5. S. G. Johnson and J. D. Joannopoulos, “Block-iterative frequency-domain methods for maxwell’s equationsin a planewave basis,” Opt. Exp. 8(3), p. 173, 2001.

Proc. of SPIE Vol. 6898 68980B-6