Tuning of InGaAsP planar photonic crystal nanocavities by local liquid crystal infiltration

H.H.J.E. Kicken

Research output: ThesisPhd Thesis 1 (Research TU/e / Graduation TU/e)

309 Downloads (Pure)


Future data-processing will increasingly employ photonic circuits in addition to conventional electronics. Photonic crystals (PC), a periodic arrangement of dielectrics, can influence the flow of light on the smallest scale, i.e. at or below the optical wavelength. Therefore PCs are a necessary tool for the ultimate miniaturization of photonic integrated circuits (PIC). The smallest PC-based building-blocks for PIC are wavelength-sized cavities which can be used as lasers, add-drop filters, and optical switches. The InP/InGaAsP /InP double heterostructure layer system is a standard platform for monolithic integration of active and passive components at the important telecom wavelength near 1.55 µm. In this thesis we have realized and characterized PC nanocavities in a hexagonal lattice of air holes. Extensive 2D and 3D Finite Difference Time Domain (FDTD) simulations were essential for design of the nanocavities. The nanocavities were fabricated by Inductively Coupled Plasma Reactive Ion Etching, after Electron Beam definition. In these high aspect ratio structures 4 new types of cavities were fabricated consisting of single unit cell sized defects: H0 (two shifted holes), H1 (single missing hole), H2 (7 missing holes) and the smallest possible ring cavity. The best quality factor found for these cavities is 300 for the ring cavity. In general, for applications, wavelength tunability is crucial. Wavelength tunability was demonstrated by infiltration of the holes of the PC with polymers and liquid crystals, shifting the spectral features of the devices over a wavelength range of 50 nm. Proof of principle of active tuning of the resonant frequency of an H1 cavity was performed using the temperature-dependence of the refractive index of infilled liquid crystals. In this experiment a tuning range of 7 nm was obtained. In addition, the same experiment was carried out on a band gap feature of a PC waveguide with consistent results. Ultimate design flexibility is achieved when the refractive index of an individual PC hole can be modified. We achieved this by a lithographic technique using Focussed Ion Beam milling. For the first time, additional etching and infiltration of individual holes near an H1 cavity was demonstrated. A blue-shift of the resonant frequency over a 40 nm wavelength range was shown as a result of the local additional etching, and an equal red-shift was observed for the local infiltration. The experimental results were consistent with Finite Difference Time Domain (FDTD) simulations. These tunable photonic nanocavities may ultimately be used as switch in optical logic circuitry. In chapter 2 the methods and materials used in this thesis were described. For numerical calculations the "CrystalWave" program is used, which uses the Plane Wave Expansion method for the calculations of bandstructures. Transmission spectra were calculated using the FDTD calculator. For the calculation of modal fields the Frequency Domain (FD) calculator was used. In general, parameters of a = 400 nm, r = 120 nm were used. For 2D-FDTD calculations an effective index neff of 3.25 is used, while for 3D-FDTD calculations refractive indices of 3.35 and 3.17 are used for InGaAsP and InP respectively. The samples were defined with Electron Beam Lithography (EBL), and etching using two types of hard masks. On the one hand 400 nm thick SiNx and on the other hand Cr/SiOx, which were opened by an CHF3 RIE process. Both the Cr opening process and deep etching process uses Cl2/O2 chemistry. The fabrication process produces PCs with a depth of at least 2.5 µm for holes with a diameter of 160 nm. For the Cr masking process a maximum hole depth of approximately 4.5 µm was obtained. The samples were optically characterized using transmission spectroscopy. The wide band spectra are obtained by measuring several PC structures with different lattice constants, so called lithographic tuning. As a result of the transmission measurements with approximately equally sized RidgeWaveGuides (RWGs) for coupling the light from free space to the PCs and vice versa, the transmission measurements are modulated by a large beat and fine Fabry-P´erot (FP) fringes. These features arise from the cavities formed by the mirror reflections at the RWG facets and the PC in between. Deeply etched PC devices may be integrated with membrane devices by conversion of the deeply etched devices to membranes by underetching. The prefabricated RWGs are still usable after the conversion. In chapter 3, The optical properties of W1 and W3 PC waveguides were explored for their suitability for subsequent infiltration experiments. Since for the infiltration experiments sharp features are required, the W3 waveguide was selected for further experiments, since it exhibits the Mini-StopBand (MSB) feature and shows a high transmission over a large bandwidth. PC waveguide bends were examined as reference for a local infiltration scheme, which is discussed further in chapter 6. A best estimate for the value of the propagation loss of the W3 PC waveguide is 110 dB/cm, consistent with other reports for InP/InGaAsP/InP. Chapter 4 discusses the optical properties of two classes of cavities in the deeply etched InP system, with a width of a single and three rows, which were fabricated and studied by transmission spectroscopy and analyzed in combination with FDTD calculations. The one row defect class includes the one missing row (FP1), H1 and H0 cavities. The three row defect class includes the three missing rows (FP3), H2 and ring cavities. For the single row defect class, H0, H1 and FP1 cavities have similar Q-factors, in the order of 50. The best result of 65 was obtained for the H1 cavity, which is comparable to results for membranes. For the three row defect class, the Q-factors of the FP3, H2 and ring cavity vary by one order of magnitude and mainly due to differences in sample quality. The best result of 300 was obtained for the ring cavity. In chapter 5, infiltration and subsequent tuning an H1 cavity and aW3 waveguide are discussed. An H1 cavity was infiltrated with LCs resulting in a shift of the resonance frequency of ??(a/??) = 0.009. A large increase in transmission of a factor of 5 was observed, this is attributed to improved coupling with the waveguides as the modal electric field profile is enlarged. The Q-factor of the cavities was reduced by 30% as a result of the infiltration. Subsequently, the orientation of the LCs in the PC holes was changed by variation of the temperature. A filling efficiency of 0.63 was found using 2D-FDTD calculations at temperatures above the clearing point. Upon crossing the clearing point an abrupt 7 nm shift was observed, consistent with the LC phase transition from nematic to isotropic. The nematic refractive index was found to be close to no. Furthermore, for a second sample an H1 cavity whose spectral signature was not observed before infiltration, was clearly detected after infiltration with BCB. The signal rise in this case was larger than an order of magnitude. Also, an H2 cavity was infilled with LC, with analogous results. The W3 PC waveguide described in section 3.2 was infiltrated with LCs. Before and after infiltration the MSB was observed, and was found red-shifted after infiltration by 30.3 nm. The orientation of the infiltrated LCs was subsequently changed by driving the temperature across the clearing temperature. At the transition temperature a further red-shift of 7 nm was observed. Analysis of the observed shift yields a filling efficiency of 80% and an effective refractive index of the LCs in the nematic state of 1.5, close to the ordinary index. The observed broadening of the MSB after infiltration is attributed to losses due to fabrication imperfections. In chapter 6, the local infiltration of a W3 PC waveguide bend and the local post-processing of six holes adjacent to an H1 cavity are described. In both cases a 400 nm SiNx mask layer was applied, which covers the PC holes, but does not fill the holes to a significant extent. The masking layer was locally thinned down using FIB milling. The holes beneath the thinned down region were opened using a CHF3 plasma in the case of the W3 PC waveguide and using wet chemical etching in the case of the H1 cavity. The CHF3 plasma process suffers from etching lag, causing the holes outside the thinned down region to be opened too fast, i.e. before the holes beneath the thinned down region are sufficiently open. Therefore, wet chemical etching is considered a better process. The six holes adjacent to the H1 cavity were post-processed by local digital etching and subsequent LC infiltration. This procedure resulted in a 40 nm blue shift of the resonant frequency with respect to the situation after fabrication for the digital etching. A 40 nm red-shift of the resonant frequency was obtained for the LC infiltration with respect to the situation after digital etching. A wide band transmission spectrum of all three conditions: after fabrication, after digital etching and after LC infilling showed that the airband edge has remained constant, proving that the local processing was indeed succesful, i.e. the holes beneath the thick part of the mask have not been affected by either the digital etching or been infiltrated with LC. 3D-FDTD calculations were performed for all three experimental conditions verifying the experimental results. This method is suited for mode selective tuning since PC holes may be individually selected for modification.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Applied Physics and Science Education
  • Salemink, Huub, Promotor
  • Fiore, Andrea, Promotor
  • van der Heijden, Rob, Copromotor
Award date7 Dec 2009
Place of PublicationEindhoven
Print ISBNs978-90-386-2073-2
Publication statusPublished - 2009


Dive into the research topics of 'Tuning of InGaAsP planar photonic crystal nanocavities by local liquid crystal infiltration'. Together they form a unique fingerprint.

Cite this