Supplementary Information

Transcription

1 Supplementary Information for Embedded Cavities and Waveguides in 3D Silicon Photonic Crystals by S. A. Rinne, F. García-Santamaría, and P. V. Braun, Nature Photonics, 2007 Figure S1. Figure S1 Detailed fabrication procedure 1 : 1. colloidal crystallization, 2. first atomic layer deposition of alumina, 3. TPP, 4. second atomic layer deposition of alumina (optional), 5. chemical vapor deposition of silicon, 6. reactive ion etching of top half-layer of silicon, 7a. wet etch of silica and alumina, OR 7b. removal of polymer by calcination, 8. wet etch of silica and alumina. S1

2 Figure S2. Artificial opals were deposited from silica microspheres grown via seeded Stöber growth processes, using either manual or continuous additions 2,3. Prior to crystal growth, silica microspheres were pre-heated at 600 ºC for no less than 48 hours. This prevented the artificial opals from cracking during post-processing at elevated temperatures 4. During pre-heating the microsphere diameter decreased 5% to 10% 5. The final diameters of the microspheres used here were ~725 nm or ~925 nm. The self-assembly procedure (Fig. S2) involved a modified vertical deposition method 6 adapted to incorporate some variations suggested by Norris et al. 7 Approximately 4 ml of a 1.5 vol.-% microsphere suspension in ethanol was dispensed into a 20 ml scintillation vial (Fisher) with a double-side polished silicon substrate (lateral dimensions ~30 x 12 mm 2, 500 μm thick). This vial was placed at an angle (about 45 ) in a round bottom cork holder on an aluminum block in an incubator (Fisher, Isotemp 125D). The temperature was set to 42.5 ºC and 45.5 ºC for the 725 nm and 925 nm colloids, respectively. Artificial opals grown in this fashion build rapidly to their maximum number of layers, plateau, and then become thinner due to sedimentation of microspheres from the growth solution. Typically, for high quality samples, as the thickness increases, the dimensions of the plateau with the maximum number of layers decreased. For high quality samples of ~30 layers, for example, the lateral dimensions of the plateau were approximately 0.5 x 7 mm 2. Features were written within these plateaus using TPP, assisted by in situ fluorescence confocal imaging 8. Figure S2 Cross-sectional schematic of the setup used to obtain colloidal crystals from large silica spheres. S2

3 Figure S3 We note that the appropriate temperature window used to grow high quality opals was small and occasionally needed to be experimentally adjusted. Increasing the temperature yielded thicker samples, however if the temperature was too high, the sample quality was very poor. Reflection and transmission spectroscopy were used to interrogate the thickness and quality of the samples (Fig. S3). Although most of the thin film opals obtained with this method showed excellent optical response, a significant number of samples were discarded. The high quality, thick samples used in this work contained up to 33 layers and exhibited absolute reflectivity peaks of ~80-85%, as grown; those with a maximum reflectivity below 70% were not used. Sample thickness varied significantly between samples. The thickest, high quality samples presented regions up to 33 layers, however most samples did not possess domains with more than 16 layers. Various environmental parameters, such as the humidity in the laboratory appeared at least in part responsible for these variations. Figure S3 Reflection and transmission microspectroscopy collected from the same region (~150 μm diameter spot) on a colloidal crystal grown via a modified vertical deposition setup from 725 nm spheres. Data was normalized to the reflectance from a silver mirror or the transmittance through a silicon wafer. S3

4 Figure S4 The thin, conformal layer of amorphous alumina was grown via ALD around the spheres to yield an interpenetrated array of spheres with enhanced mechanical stability. Alumina is mechanically and thermally stable and etches in hydrofluoric acid; further, it is transparent at 780 nm and it has a refractive index close to the colloids and monomer, to prevent heating or scattering during TPP. Since this is a sacrificial layer that will be etched at the end of the process, its thickness will determine the radius of the air spheres in the final inverse opal. This parameter can greatly alter the width of the cpbg 9,10. Figure S4a shows the dependence of the cpbg size with the radius of the hollow sphere. We find that for a radius ~0.36a, the gap size is maximized and yields the band structure 11 shown in Fig. S4b. Figure S4 (a) Gap width of a Si inverse opal as a function of the radius of the air spheres. The radius of the outer layer of the hollow Si spheres was 0.408a. (b) When the radius of the air spheres is 0.36a, the photonic band gap is maximized. S4

5 Figure S5 Occasionally, the desired alumina film thicknesses were obtained in two ALD growths, one before and one after TPP. The first ALD growth helps prevent cracking during TPP of large-area features and the growth after TPP may improve the edge resolution of air features in the final silicon-air inverse opals. Since the polymer shrinks upon polymerization, the second ALD step may fill gaps between the polymer and silica spheres at the periphery of the TPP features. This could prevent the gaps from being filled with silicon during the CVD step, possibly improving the edge resolution and minimum feature size. The accurate control over the degree of sphere interpenetration was confirmed by spectroscopy (Fig. S5). The peak position correlated well with band structure calculations which assumed one monolayer of Al 2 O 3 (~0.1 nm, n=1.60) grew per cycle. The target alumina layer thickness was chosen so as not to diminish the bandgap in the final silicon-air inverse opal. Figure S5 Reflection spectroscopy of a colloidal crystal as grown, and after 80 and 130 cycles of ALD on a sample assembled from 725 nm spheres. S5

6 Figure S6 It is essential that the features can withstand the elevated temperature required for silicon CVD, 325 C. Thermogravimetric analysis (TGA) revealed that our polymer does not lose significant weight when held at 325 C in an inert atmosphere (Fig. S6a). It is also important that the polymer retains optical transparency (Fig. S6b) if it will be present in the final structure, e.g. in Figure 4. Figure S6 (a) TGA results and (b) transmission spectra of poly(tmpta) samples (~1 mm thick) held at an elevated temperature for 3 h in N 2. The grey bar in (b) indicates the spectral range over which the near-ir micrographs in this paper were collected. TGA also showed that poly(tmpta) cleanly decomposed when heated to 500 C overnight in air, so to create air defects TPP features were removed via calcination. The polymer was removed after RIE, but before removal of the silica microspheres and alumina because the silicon-air inverse opal cracked upon heating to 500 C. Before silica removal, however, the TPP features could be isolated from the external environment, making their removal difficult. Thus, calcination is most appropriate for the removal of features that connect to the surface of the artificial opal. Debris was sometimes observed after removal of isolated, thin, or complicated features (Fig. 1). As others have demonstrated, it should also be possible to remove the polymer via oxygen plasma etching, after microsphere removal 12. S6

7 Figure S7 Previous recipes for Si CVD in opals involved operating above the decomposition temperature of our polymer (Fig. S6). Conventional low pressure CVD systems show negligible growth rates at that temperature 7. To enable reasonable Si deposition rates at 325 ºC, disilane rather than silane (SiH 4 ) was selected as the Si precursor given its lower decomposition temperature. Decomposition time controlled the thickness of the silicon film 10. Higher disilane pressures did not significantly increase the growth rate, though sometimes caused samples to lift off the substrate. Spectroscopy was correlated with band structure calculations to approximate the extent of silicon infilling (Fig. S7). Figure S7 (a) Reflection and (b) transmission spectroscopy taken from a Si/SiO 2 (microsphere diameter 925 nm) composite during and after silicon CVD. Data was normalized to a silver mirror or silicon wafer in reflection or transmission mode, respectively. As the Si filling fraction increased, the effective refractive index and photonic strength of the structure also increased, causing the peak wavelength to red-shift and increase in intensity. S7

8 Figure S8 After silicon CVD, reactive ion etching (RIE) was used to expose the silica colloids (Fig. S8a) enabling their removal via wet etching (see Methods, Wet Etch section) 7. Care must be exercised to avoid over-etching, since the first layer of silica colloids is not a perfect etch stop. When properly etched, a high-quality surface is achieved in the final, silicon inverse opal (Fig. S8b). RIE also removes the silicon overlayer from the top of TPP features that extend to the top surface of the photonic crystal. It is important for the TPP features not to protrude much past the surface, because the silicon that grows on the sides of the protruding portions is not removed during RIE, leaving silicon side-walls in the final structure (Fig. S8c). The surface geometry is also very important since the termination obtained after RIE prevents the deleterious effects from surface resonances we report in reference 13. Figure S8 (a) SEM of the top of a silicon-silica composite after RIE, taken at the edge of the RIE-exposed region. SEM of the silicon inverse opal obtained (b) without and (c) with a TPP feature that had protruded from the top of the sample. RIE removed the silicon from the top but not sides of the feature. S8

9 Figure S9 Using an ethanolic HF solution was important if the TPP features have not been removed, as it enables etching within the hydrophobic polymer regions. Complete etching required both the alumina ALD layer, and RIE removal of the silicon overlayer, which assist in creating an interpenetrated network of oxide and exposing the oxide microspheres at the surface, respectively. FIB cross-sections and spectroscopy revealed that wet etching only occurred in RIE-exposed regions and proceeded from the top-down (Fig. S9). The peak reflectance from the top of the PhC decreased by ~20-30% after RIE (Fig. S9d), suggesting that surface truncation plays a major role in coupling to photonic crystals 14,15. As expected, there was no change when measuring the sample from the back, before and after RIE. As previously discussed, the surface was masked to only RIE a small region (~1 x 2 mm 2 ), resulting in selective etching only in the region of interest of the sample. When large areas were etched, samples would crack and liftoff the substrate; limiting the etching area helped to mitigate these issues. Figure S9 (a) FIB cross-section of a partially etched silicon-silica composite in a RIE-exposed region. (b) FIB cross-section taken at the edge of an RIE-exposed region after HF etching (courtesy of Erik C. Nelson in our group.) Reflectance spectroscopy from the (c) top and (d) back of a photonic crystal assembled from 725 nm spheres after CVD, RIE, and HF etching. S9

10 Because etching proceeds from the top-down, it was possible to determine the etch extent by monitoring the reflectance spectra from the top and back of the sample (results were correlated with FIB). Spectra were normalized to a silver mirror. The reflectance from the silicon substrate was not accounted for, explaining the higher baseline in (c). S10

11 References 1. Pruzinsky, S. A. Two-photon polymerization of defects in photonic crystals. Thesis, Univ. of Illinois at Urbana-Champaign (2006). 2. Stober, W., Fink, A., and Bohn, E. Controlled growth of monodisperse silica spheres in the micron size range. J. Colloid Interface Sci. 26, (1968). 3. Arriagada, F. J. and Osseo-Asare, K. Synthesis of nanosize silica in a nonionic water-in-oil microemulsion: Effects of the water/surfactant molar ratio and ammonia concentration. J. Colloid Interface Sci. 211, (1999). 4. Chabanov, A. A., Jun, Y., and Norris, D. J. Avoiding cracks in self-assembled photonic band-gap crystals. Appl. Phys. Lett. 84, (2004). 5. García-Santamaría, F. et al. Refractive index properties of calcined silica submicrometer spheres. Langmuir 18, (2002). 6. Jiang, P., Bertone, J. F., Hwang, K. S., and Colvin, V. L. Single-crystal colloidal multilayers of controlled thickness. Chem. Mater. 11, (1999). 7. Vlasov, Y. A., Bo, X. Z., Sturm, J. C., and Norris, D. J. On-chip natural assembly of silicon photonic bandgap crystals. Nature 414, (2001). 8. Nelson E. C., and Braun, P. V. Registration and optical properties of embedded two-photon polymerized features within self-organized photonic crystals. arxiv: v1 (2007). 9. Busch, K. and John, S. Photonic band gap formation in certain self-organizing systems. Phys. Rev. E 58, (1998). 10. García-Santamaría, F. et al. Photonic band engineering in opals by growth of Si/Ge multilayer shells. Adv. Mater. 15, (2003). 11. Johnson, S. G. and Joannopoulos, J. D. Block-iterative frequency-domain methods for Maxwell's equations in a planewave basis. Opt. Express 8, (2001). 12. García-Santamaría, F. et al. Nanorobotic manipulation of microspheres for on-chip diamond architectures. Adv. Mater. 14, (2002). 13. García-Santamaría, F., Nelson E. C., and Braun, P. V. An optical surface resonance may render photonic crystals ineffective. Phys. Rev. B 76, (2007). 14. Hiller, J., Mendelsohn, J. D., and Rubner, M. F. Reversibly erasable nanoporous antireflection coatings from polyelectrolyte multilayers. Nat. Mater. 1, (2002). 15. Vlasov, Y. A. and McNab, S. J. Coupling into the slow light mode in slab-type photonic crystal waveguides. Opt. Lett. 31, (2006). S11