Journal of Materials Chemistry

Thin film photonic crystals: synthesis and characterisation

Martyn A. McLachlan^ab, Nigel P. Johnson^b, Richard M. De La Rue^b and David W. McComb^a
aDepartment of Chemistry, University of Glasgow, Glasgow, UK G12 8QQ. E-mail: martynmc@chem.gla.ac.uk; Fax: 01413304888; Tel: 01413306022
^bDepartment of Electronics & Electrical Engineering, University of Glasgow, Glasgow, UK G12 8QQ

The results of an investigation of the major factors that influence colloidal self-assembly of thin film photonic crystals are reported. The effect of temperature, relative humidity, sphere diameter, colloidal concentration and substrate angle were investigated: the results establish clearly that temperature is the most critical factor. Quantitative analysis of the results using Design of Experiments methodology has identified the optimum conditions for the growth of large area, low defect density thin film photonic crystals.

Since the concept of a three-dimensional (3-D) photonic crystal (PC) was first proposed by Yablonovitch and John¹ there has been tremendous activity across a wide range of scientific disciplines in an attempt to synthesise and investigate the structure–property relationship in this new class of functional materials.² One of the key properties of a PC is the presence of a photonic band gap (PBG). The PBG arises due to periodic modulation of the dielectric function in at least one crystallographic direction.³ If the modulation of the dielectric function (or the refractive index) occurs in all directions in the crystal a 3-D PBG will result.⁴

Current fibre optic telecommunications operate at wavelengths of 1.3 µm and below for short-haul⁵ and around 1.5 µm⁶ for long-haul applications. Photonic materials that operate at these wavelengths and can be utilised in switching/filtering applications are commercially available.⁵ However, it is predicted that in the future, short-range communication technologies will operate at wavelengths in the visible and near infrared regions and there is a need to develop PCs that operate in this regime.

The search for 3-D PCs that could potentially exhibit a PBG at visible wavelengths focussed initially on the gemstone opal. Opal is a crystalloid composed of a face centred cubic (fcc) array of mono-sized, hydrated, amorphous silica spheres.⁷ Natural opal exhibits two characteristics that makes it unsuitable for photonic crystal applications; firstly, it is polycrystalline and secondly, impurities are almost always present in the interstices between the silica spheres. Potentially both of these problems can be overcome by the formation of synthetic opal in clean, controlled laboratory conditions. There are many techniques used in the formation of synthetic opal from colloidal dispersions of mono-sized spheres including sedimentation, centrifugation, sedimentation electrophoresis, and controlled vertical drying.⁸

Controlled vertical drying, also known as colloidal self-assembly, shows great potential as a method for the formation of thin film 3-D PCs and has excited much interest from researchers interested in fabrication of PBG devices. The technique has been extensively utilised in the last few years by a large number of groups. In essence the meniscus of a colloidal solution moves across a substrate, depositing colloidal spheres at the evaporation front where they can undergo spontaneous self-assembly. In some cases the substrate is mechanically withdrawn out of the vessel containing the colloid.

Although colloidal self-assembly is widely used for the formation of thin film photonic crystals, the factors influencing the growth process and the optimum conditions for growth of large crystals with low defect densities do not appear to have been systematically investigated. The use of colloids of different materials, sizes and dispersions, as well as the use of numerous characterisation methods also makes it difficult to obtain a clear understanding of the growth process. It can be anticipated that many of the factors influencing the growth process will not be independent, and understanding the interaction between these variables is critical. The information contained within the current literature often focuses on the variation of parameters associated with the colloid, usually sphere diameter and concentration^9–11 and characterisation of the resultant thin films is carried out by a wide range of techniques. Consequently, comparison of the results presented by different research groups is often problematic.^9,12 In addition, the lack of a consistent measure of film quality (e.g. defect density or domain size) for these materials makes it difficult to fully utilise the results that have been reported in the literature.

In this contribution, the results of a systematic study of the factors influencing the controlled vertical drying technique are reported. The factors chosen for investigation in this study were temperature, relative humidity, sphere diameter, colloidal concentration and substrate angle. The colloidal crystals reported in this work were assembled from polystyrene spheres in aqueous solution and have been characterised using a range of microscopy and spectroscopy techniques.

Styrene (bp 145–146 °C, 99%, Aldrich) was distilled at reduced vacuum to remove traces of the inhibitor 4-tert-butylcatechol (bp 285–286 °C). Thermal self-initiation during transit and storage is prevented by the addition of a small quantity of inhibitor (10–15 ppm) by the manufacturer.¹³ The risk of thermal self-initiation is minimised during the removal of the initiator by distilling at reduced pressure, and the use of a dark environment reduces the risk of light-initiated polymerisation. After distillation the monomer is stored in the refrigerator, in amber bottles and under a nitrogen atmosphere, until required.

Colloidal suspensions of polystyrene spheres were synthesised by a suspension polymerisation using a free radical initiator (potassium persulfate, K₂S₂O₈, 99%, Aldrich) and the inhibitor-free styrene.¹⁴ The polymerisation was carried out in aqueous solution at 80°C for 24 h under a nitrogen atmosphere. The reaction mixture was agitated using a twin paddled overhead stirrer operating at 350–450 rpm. Colloidal suspensions of polystyrene spheres with a mean diameter in the range 200–750 nm and exhibiting low polydispersity have been synthesised by this method. Variation of the mean diameter can be achieved in various ways; alteration of reagent concentrations, variation of stirring speed and altering reaction time.¹⁵ In this work the concentration of the monomer was varied in the range 0.3–0.7 mol l^–1 and the concentration of the initiator, which was small, was kept constant. This produced colloidal suspensions with a volume fraction (v.f.) of solids in the range 3–6%.

The diameter of the colloidal spheres was determined using a transmission electron microscope (TEM – JEOL 1200EX). A droplet of the diluted suspension was placed on a carbon coated copper grid and the solvent was allowed to evaporate. The TEM magnification was calibrated using a patterned (2160 lines mm^–1) grid, which was measured at a range of nominal magnification values to obtain a plot of nominal vs. actual magnification. For each colloid produced, TEM images were recorded and printed. At least 200 spheres were measured manually and their absolute diameters were determined by reference to the calibration plot. The diameter and polydispersity of the colloids used in this study are detailed in Table 1.

The colloidal crystals were grown on glass substrates (10 × 60 mm) cleaned by sequential washing (5 min each) in Opticlear, acetone, methanol and de-ionised water, and finally nitrogen blow-dried prior to use. The colloid suspensions were placed in an ultrasonic bath for 10–15 min to ensure that any aggregates were broken up before use. The glass substrates were placed into cylindrical vessels of volume 12 cm³ and a known volume of the colloid added. The vessels were then placed into a temperature stable incubator until the growth was complete.

In the course of this study, growth temperatures between 20 and 70 °C were investigated. The incubators used during this study were allowed a minimum of 24 h to stabilise prior to any samples being introduced; temperature drift during the growth cycles was controlled to within ±0.2 °C. Several methods to vary the angle of the substrate relative to the meniscus were investigated. The method selected was to attach the substrate to the base of the evaporation vessel using silicone adhesive, which was cured with the substrate at the desired angle. Although a wide angular range was investigated, only the results obtained from substrate angles of 90 and 75° will be presented in this report.

The final parameter explored during this work was the effect of relative humidity (RH). RH can be defined as the amount of water vapour in the air divided by the maximum amount of water vapour the air can hold at a given temperature and is usually expressed as a percentage. In some of the initial experiments growth of colloidal crystals was carried out with no control of RH. Subsequent measurements revealed that in these cases the RH values remained between 10 and 20% at each temperature. In order to achieve and maintain higher values of RH, a water tank was placed in the incubator, and air was blown through a perforated rubber tube placed in the bottom of the tank. Using this method it was possible control and maintain the RH in the incubator to within ±2% of the desired value for the duration of the growth cycle. The temperature and RH values were monitored and recorded using a data logger (Dicksonware TM125) placed inside the incubator for the duration of the growth. The real time output from the data logger allowed the conditions within the incubator to be closely monitored and adjusted if necessary. The size of sample grown was kept constant (10 × 40 mm) regardless of the growth temperatures used. Although this resulted in a significant variation of the growth times at different temperatures, it allowed the same areas on different samples to be analysed. The growth times ranged from over 40 days to less than 4 days depending on the experimental conditions. The volume fraction of solid in the as-synthesised polystyrene colloid was equal to or greater than that utilised in the experiments. If dilution was necessary, the same distilled, de-ionised water used as the solvent for the original polymerisation was used.

The thin films were analysed using a range of complementary techniques in order to obtain information on the microscopic and macroscopic structure and properties. The principal techniques used were optical microscopy, reflectance spectroscopy and scanning electron microscopy (SEM). Three positions at the centre of each sample, each separated by 10 mm, were analysed using optical microscopy and reflectance spectroscopy. In the case of SEM analysis, each of these points was sub-divided into five smaller regions, and each was analysed. A schematic diagram of the analysis positions is shown in Fig. 1.

	Fig. 1 Schematic illustration of analysis positions on each thin film produced – the overall area covered by the thin film was maintained at 400 mm2.

The thin films were observed using a Leica optical microscope (INM20) equipped with a CCD camera. Three areas, approximately 400 × 300 µm, in the middle of each position (Fig. 1) were imaged. Typical examples of the images obtained are shown in Fig. 2. The colour change of the reflected light is due to a change in sphere size, which is a direct observation of these thin films acting as photonic crystals. It is immediately apparent that there are a large number of cracks in the thin film and that the number of cracks is not the same in all specimens.

	Fig. 2 Optical microscope images of thin films fabricated from the three sphere diameters indicated in the captions.

Each image was digitally processed and analysed to obtain the crack density, which is defined as the area of a sample covered by cracks as a percentage of the total area. This factor was used as a quantitative measure of macrostructural quality and facilitates comparison of different samples. The crack density can be influenced by the vacuum and by electron beam heating in the SEM. The use of optical microscopy methods allows evaluation of the crack density immediately following the solvent removal and without any additional perturbation.

A white light source, filtered to remove second harmonic was passed through a single grating monochromator (Jobin Yvon THR1000), beam chopper and focused to obtain a 1 mm² spot of monochromatic light on the sample. The reflected light was collected and focussed onto a photomultiplier tube connected to a lock-in amplifier. The intensity of the reflected light was collected over the desired wavelength range (300–900 nm). The samples were mounted on a goniometer allowing the angle of incidence of the light on the sample to be varied. As the wavelength of the incident light approaches the stop-band wavelength it can no longer transmitted through the samples and is reflected. A photonic stop-band exists when the refractive index contrast is insufficient to allow the existence of a full PBG. By altering the angle of incidence of the light relative to the sample surface normal the wavelength at which the maximum in the reflection peak occurs will shift (Fig. 3(a)). By performing angular resolved studies it is possible to manipulate the reflectance spectra to give detailed information about the thin films. In particular, a plot of the wavelength of the reflection maxima squared against the corresponding incidence angles should produce a straight line with a y-intercept equal to n²_eff, where n_eff is the effective refractive index, and a slope of 1/(4d²), where d is the (111) interplanar spacing (Fig. 3(b)). This is based on the combination of Bragg's law (eqn. (1)) with Snells' law (eqn. (2)) to obtain eqn. (3) below and assumes that the sample surface normal is aligned along a <111> direction.^16,17

n = 2dsin

(1)

n1sin1 = n2sin2

(2)

= 2d111(n2eff – sin2)1/2

(3)

	Fig. 3 (a) Reflectance spectra thin films formed from 230 nm polystyrene spheres at the incidence angles shown (relative to surface normal). (b) The linear plots discussed in the text for all three-sphere diameters. The difference in the slope of each of the lines is attributed to the difference in diameter.

Increasing the angle of incidence relative to the surface normal shifts the position of the reflectance peak to shorter wavelength. The samples could also have been measured in transmission mode, where the wavelengths unable to propagate the crystal would appear as a dip in the transmission spectrum. However, reflectance is the preferred technique, as additional sample preparation is not required, and provided that there are no appreciable absorptions, the same information is obtained.

Each region analysed by optical microscopy and reflectance spectroscopy was sub-divided into five smaller areas for analysis by SEM. The imaging was carried out using either a conventional SEM (Hitachi S3000) or a field emission gun (FEG) SEM (Hitachi S4700). An image was recorded at each corner and an additional image was taken at the centre of each of the original regions (Fig. 1). Micrometer control of the SEM sample stage enabled the distance between each image to be accurately measured. A two-dimensional Fast Fourier Transform (FFT) of each image was calculated in order to evaluate the homogeneity of the growth directions both spatially within a given thin film and between different specimens. The FFTs were also utilised to assess the long-range order of the thin films.^9,18

Some samples were also cleaved to permit viewing of the stacked internal planes and to facilitate evaluation of the sample thickness. The analysis of the cross-sectional images confirmed that a <111> direction was parallel to the sample normal and enabled microscopic point and line defects to be observed. A typical cleaved sample is shown in Fig. 4. Very rarely a surface plane other than a {111} type is observed, this has been reported by other authors¹⁹ and the appearance of such surfaces does not appear to be associated with any of the growth parameters investigated.

	Fig. 4 A cleaved sample of thin film, grown from 230 nm diameter spheres at 25 °C such images confirm that the <111> direction is perpendicular to the substrate.

We have carried out a systematic study into the effects of factors considered important in the colloidal self-assembly of thin-film photonic crystals. The effects associated with each variable and also the results obtained overall are discussed in turn for each factor. Using a DoE methodology²⁰ a quantitative analysis for the effect of four growth parameters on domain size is presented. This report focuses on the macroscopic faults and defects in these thin films. Classical point and extended defects are not discussed in this contribution.

The growth temperature used has been found to be the most significant parameter studied. The growth temperature effects the thin film formed in a number of ways. The first effect observed is an increase in domain size, with increasing temperature, measured by optical microscopy (Fig. 5). At 25 °C the domains are in general less than 50 µm in any direction. This increases to 50–100 µm at 45 °C and at 65 °C domains of length 250–300 µm are obtained.

	Fig. 5 Increasing growth temperature leads to a direct increase on the size of domain produced, as shown in the optical micrographs. SEM images (inset) confirm that smaller cracks are not present and that the majority of cracking occurs during drying (inset scale marker 200 µm)

The optical micrographs shown in Fig. 5 not only demonstrate the increase in domain size as temperature is increased but also suggest a change in the domain shape or direction of cracking. However, this is somewhat misleading and in fact there is no change in the direction in which cracking occurs within these films. This is established clearly by examination of the SEM images for thin films grown at the three temperatures (Fig. 6). It is apparent that at all temperatures investigated the cracking always occurs along the close packed <110> directions. At 25 °C the cracking is isotropic with crack widths of 5 µm. At 45 °C, the SEM image (Fig. 6) shows that the cracking is still isotropic, but the optical micrographs (Fig. 5) suggest that the crack width is less uniform. In the sample grown at 65 °C, it is clear that while the cracks are still aligned along the close packed <110> directions, the widths of the cracks are no longer uniform. Cracks that extend up to lengths of 1 mm are observed in the growth direction, i.e. perpendicular to the meniscus. These cracks are still typically 5 µm wide. Analysis by SEM shows that the cracks, which in Fig. 5 appear to be parallel to the growth front, are in fact narrow cracks (<1 µm) that alternate between two of the close packed directions (Fig. 6). These narrow cracks join together the long cracks and form pseudo-rectangular domains. The reduced width of these cracks may indicate that the stresses associated with drying are reduced in this direction. At 65 °C domains of typical dimension of 100 × 300 µm can be grown with all of the sphere sizes investigated.

	Fig. 6 FEG-SEM images of typical cracks observed at the temperatures indicated. The images reveal that there is no change in the direction of the cracking with change in growth temperature.

The fast Fourier transforms (FFTs) calculated from the SEM images have been used to measure the long-range order within the thin films. The FFT of an image is essentially a two-dimensional spatial frequency diagram, which when indexed can be used to determine the crystallographic orientation of the colloidal crystals. Fig. 7 shows two typical patterns calculated from SEM images obtained from samples grown at 25 and 65 °C, respectively. FFTs of the images from samples grown at low temperature generally display weak spots, and occasionally ring patterns, indicating that only short-range order is present. In those calculated from samples grown at greater temperatures higher-order spots are observed and ring patterns are seldom seen. This indicates there is an increase in long-range order as temperature is increased. The rotation of the <220> spots relative to the growth direction has been measured (a) at each of the five analysis positions (Fig. 1) to estimate local angular variations, and (b) at the center of the three analysis spaced 10 mm apart on the substrate to estimate the long-range correlations. It was established that the rotation between adjacent domains is typically <1°, and over the entire 40 mm length of a sample the rotation is typically ±5°. The smallest angular variation is observed at the highest growth temperatures. This confirms that the crystallographic orientation of the material is preserved regardless of the presence of drying cracks, which would indicate that the cracks occur after self-assembly and during the drying process

	Fig. 7 2-D Spatial frequency diagrams obtained at 25 and 65 °C with schematic illustration of the measurement technique used to determine the angle of rotation.

It has been reported that on increasing the growth temperature the evaporation rate increases and there is a greater influx of colloidal spheres into the meniscus region with no net change in sample thickness.⁹ This is not entirely consistent with our observations. In fact we have detected changes in thickness when the growth temperature is raised from 25 to 45 °C, with little thickness difference on further increase. Using a 1% v.f. dispersion the thickness difference between samples grown at 25 and 45 °C is about 66%, as shown in Fig. 8. Given that there is no further thickness increase when the temperature is raised to 65 °C suggests that sedimentation of the spheres is competing with self-assembly at low temperatures. It has been demonstrated that by introducing convection currents sedimentation of larger colloidal spheres during capillary growth can be eliminated, and although the temperature across the evaporation vessel is constant, a similar effect may be occurring.²¹

	Fig. 8 SEM micrographs of thin film samples grown from 300 nm spheres. The sample grown at 25 °C is around 30 layers thick and that grown at 45 °C is approximately 50 layers (the 45 °C sample has been lifted from the substrate to facilitate imaging.

There is also a noticeable change in the adhesion of the films to the substrate as the temperature is changed. At higher temperatures the films appear to be bonded more strongly to the substrate and the film does not appear to flake or delaminate on manipulation of the samples. We have made no attempt to quantify the adhesion or the interfacial characteristics.

Shrinkage of the polystyrene colloidal spheres upon drying is considered to be one of the factors that might contribute to the cracking observed in these materials. It was hypothesised that by increasing the amount of moisture present in the drying chamber the shrinkage process might be slowed, the drying/shrinkage stresses reduced and hence, the amount of cracks decreased.

In this project all of the specimens were grown at low RH (10–20%) unless otherwise stated. For comparative purposes two samples were grown at 45 and 65 °C, respectively, under high RH (40–50%) conditions. The resultant films exhibited extremely poor adhesion to the substrate, very small areas of opalescence and an uneven coverage of the substrate. The equivalent experiment was not carried out at 25 °C given that the predicted time for the growth cycle was in excess of 60 days and possible fluctuations in both temperature and RH may have introduced potential errors to the analysis. Based on these experiments it is concluded that the RH must be low at higher temperatures to favour the formation of good quality colloidal crystals.

The results of the investigation of two different angles, 90 and 75°, between the substrate and the meniscus are reported. Visual characterisation of the optical microscope images does not reveal any changes in the domain size of samples grown at these two orientations. However, when the domain size measurements are analysed statistically, it can be shown that the domain size at 75° is larger than that grown at 90°. This is discussed in detail in the quantitative analysis section. However, the use of a 75° angle compromises the length of the film that can be produced because as the growth proceeds the distance between the vessel wall and the substrate decreases. When the solvent level becomes sufficiently low there is interaction of the meniscus with the sides of the vial resulting in the deposition of the spheres in a V shaped pattern at the bottom of the slide. This effect is not observed when the substrate is placed at an angle of 90°.

The thickness of the thin films increased in proportion to the concentration of the colloid. However, as the v.f. of polymer was increased above 1% to 5%, the adhesion of the films to the substrate was significantly degraded. The films tended to delaminate or flake on handling. A volume fraction of 1% was identified as optimum for the fabrication of mechanically stable thin films. At 0.5% and below, the films became very thin and optically transparent, which caused difficulty in reflectivity measurements. Although it has been reported that if thicker films are required multiple growth cycles can be repeated on the same substrate,⁹ initial attempts to repeat this in a polystyrene colloidal crystal have proved unsuccessful. Re-immersion of the thin films into a polymer colloid damages the films and the result is a sample with an extremely uneven surface topography and a high percentage of defects.

The sphere diameters investigated (Table 1) have similar growth characteristics and their thin films have almost identical properties. Based on the analysis carried out sphere diameter is not considered to be a significant factor, when working with such small colloids. The sphere diameters correlate directly with the reflectance peak position as given by eqn. (3) (Table 1).

A systematic trend in the measured n_eff was observed for different sphere sizes (Table 1) The highest value determined, 1.60, is close to that of bulk polystyrene (1.58–1.59).²² One can speculate on the origins of this observation: higher filling factor, i.e. >74% volume occupied; wavelength dependant variation of n_eff; n of spheres is greater than that of bulk polystyrene. The latter explanation is appealing if the possibility that n is also dependant on diameter is considered. If the spheres exhibited a core–shell structure this may explain the relationship between n_eff and sphere diameter.

Design of Experiments (DoE) methodology has been successfully employed to optimize processes where many physical factors may be potentially active.²⁰ The methodology relies on a series of experimental conditions where a high and low value of each experimental parameter is varied in an orthogonal manner (i.e. each individual measurement has a different combination of the high and low values of the experimental parameters). The response, or yield, of each experiment is noted and, because the experiments are orthogonal, the response of each parameter and combination of parameters may be extracted from the full data set, known as the full factorial design. With N factors 2^N experimental runs are required for a full factorial design. A commercial program was used to prepare the analysis.²³

In this study the response measured was domain size independently measured. Five parameters were initially chosen namely growth temperature, sphere diameter, position on the substrate, substrate orientation and RH. Since only lower values of RH produced measurable results this parameter was not carried forward in the analysis. Although the DoE method only requires a high and low value for each parameter, in general three values of temperature and spheres size were measured with the results analysed in pairs.

The effect of a particular parameter is obtained by measuring the average values when the parameter is high, minus the average values when the parameter is low. To determine if a parameter has an effect, as opposed to just statistical variation the whole experimental space can be replicated. A less time consuming alternative is to use the analysis method of Lenth.²⁴ The method relies of effect sparcity i.e. there are only a small number of effects active in the process. Lenth gives a method to compute two levels; the Pseudo Standard Error (PSE) and the Margin of Error (ME). Factors that fall below the first level show little evidence of being active. Those factors between the levels are possibly active, and factors showing contrasts outside the two levels are probably active. The main advantage of this method is that real units rather than statistically derived units are used in the plots.

Fig. 9(a) shows a Lenth plot where the change in domain size in the growth direction (y) is plotted for the different experimental parameters. The dashed lines closest to the origin are the ME and the dotted lines are the PSE. The most significant effect is temperature. The large positive value indicates that high temperature produces larger domain sizes in the growth (y) direction. A change in the average domain length of 212 ± 80 µm over the temperature range 25–65 °C. A further analysis of data for temperature range 45–65 °C indicates the majority of the temperature effect is in this range.

	Fig. 9 (a) and (b): Lenth plot of y and x domain size change for different experimental parameters. The dashed lines indicate the margin of error, the dotted lines the pseudo-random error.

The Lenth plot in Fig. 9(a) shows sphere diameter to be the second most significant factor affecting domain size. The plot covers a change in sphere diameter from 230 to 376 nm and is outside the PSE range and likely to be an active factor effecting the domain size. However, while the measurements were associated with a large change in y domain size, 88 µm the measurements possessed an even larger standard deviation ±170 µm and therefore the change cannot be conclusively associated with sphere diameter.

Fig. 9(b) shows the Lenth plot for domain size change perpendicular to the growth (x) direction. Variation of the substrate angle between 90 and 75° produces an unexpected effect on the size of the domain perpendicular to the growth (x) direction. Fig. 9(b) shows the Lenth plot with a negative ordinate for the substrate orientation indicating that an angle of 75° produces larger domain size of 52 ± 22 µm compared to 90° (vertical) and is statistically significant. It is possible that the change in contact angle between the substrate and the meniscus may change either the growth kinetics of the capillary growth process or alternatively modify the drying rate and hence the stresses that cause the formation of cracks. Further study to understand the influence of the substrate angle is in progress.

DoE can also identify combinations of parameters that may act to give a greater effect than their linear combination i.e. synergistic effects. These second-order effects also shown in the Lenth plots in Fig. 9. Since none of the second-order effects exceed the ME it is concluded that the synergistic effects are not statistically significant.

We define quality as the sum of several principal factors, enabling comparison between different samples to be made. The definition of a perfect sample would be large domain size, low crack density, low angular rotation between domains, good mechanical strength and well-resolved reflectance peaks in the visible spectral region. Temperature is clearly the most significant factor: the ability to control and maintain a desired temperature is as important as the chosen temperature. Fluctuations in temperature can lead to samples of exceptionally low quality being produced. We have improved the quality of the thin films by increasing the growth temperature to an optimum value of 65 °C and growing at low values of relative humidity. The growth characteristics and the quality of the thin films produced from colloidal spheres of different sizes are similar, however a noticeable increase in n_eff. is observed with an increase in sphere size and to a lesser degree temperature. The angle of the substrate leads to an unexpected increase in the domain size parallel to the meniscus.

Consolidation of all of the results obtained has allowed identification of an optimum set of growth conditions for the fabrication of large-area defect free thin film photonic crystals. By carefully increasing the growth temperature we have been able to increase the overall quality of the photonic crystals, lowering defects, increasing mechanical strength and at the same time reduce the growth period to less than 5 days.

Footnote

Now at, Department of Materials, Imperial College London, Exhibition Road, London SW7 2AZ (d.mccomb@imperial.ac.uk)