1. Introduction

Sculptured-thin-film (STF) technology is rapidly maturing as an optical nanotechnology since its inception in the early 1990s [1, 2, 3]. Although theoretical research outpaced experimental research during the first decade, many optical STF-based devices have been designed, fabricated and tested [3, 4, 5, 6]. One type of STFs have become ever more valuable as a platform for optical devices. These are chiral STFs, which are best described as assemblies of upright, parallel, helical columns that may be isolated from each other or be intertwined. Their helicoidal morphology, being periodic, engenders the circular Bragg phenomenon: when a circularly polarized (CP) plane wave is normally incident on a chiral STF, it is highly reflected — provided (i) the handedness of the incident plane wave is the same as the structural handedness of the chiral STF, (ii) the wavelength lies within a regime called the Bragg regime, and (iii) the chiral STF is sufficiently thick [3, 4]. In contrast, a CP plane wave of the other handedness is reflected very little. The CP-discriminatory characteristic of reflection by a chiral STF is displayed over a wide range of incidence angles [3, 7], and is called the circular Bragg phenomenon (CBP). For use as a CP filter, a chiral STF must be sufficiently thick [3, 4 ] for the CP-discrimination to be significant.

One way to reduce the thickness needed to observe well-developed CBP is to increase the local linear birefringence of the chiral STF [8, 9]. A well-developed CBP should display high selective reflection of CP light and be confined to a defined spectral regime, typically 50-100 nm in bandwidth. The original technique for fabricating STFs is to obliquely direct vapor from a fixed source towards a rotating substrate, in some low-pressure environment [2, 11]. A few years ago, the serial bideposition (SBD) technique was adapted for chiral STFs [10], with consequent broadening of the Bragg regime and enhancement of the CP-discriminatory reflection by chiral STFs only a few periods thick. In the SBD technique, subdeposits are made from two oppositely oriented sources (either both real or one real and the other simulated by a rapid 180° rotation of the substrate). Once the pair of subdeposits have been made, the substrate is incrementally stepped a small angle Δδ, and the subdeposit-pairs repeated, thus creating structural chirality. Slight variations in the deposition rate are averaged through the thousands of sub-deposits needed to fabricate a chiral STF, and any deleterious effects on the desired optical response properties can be eliminated through feedback systems. Consistently producing high-quality STFs, the SBD technique is capable of commercial application [12].

In this communication, we focus on the effect that post-deposition annealing has on the optical response properties of SBD chiral STFs. Amorphous, and we suspect oxygen-deficient, thin films are not stable over time in atmosphere, and post-deposition annealing is necessary to make these films stable. Annealing modifies material composition and crystallinity [13], thereby improving material stability [14], and therefore can be expected to alter optical response properties.

Whereas annealed STFs may be expected to mostly retain those optical response properties that originate from their helical pitch and intercolumnar separation, other optical response properties must change substantially. Preliminary evidence on the effect of annealing on the Bragg regime has come from experiments on chiral STFs fabricated using the original deposition technique [15], but not the more optically relevant SBD technique. Clearly, in order to exploit the beneficial effects of annealing, the consequent changes in the optical response properties must be considered a priori in the design process. The signature of a well-developed CBP allows experimental analysis of the post-deposition annealing effects on the optical response.

The plan of this communication is as follows: Section 2 provides a description of the STF fabrication and optical characterization methods. Section 3 contains the experimental results obtained and a discussion thereof to show that annealing leads to a blue-shift of the CBP.

2. Experimental

Five TiO₂ SBD chiral STFs (samples 1–5) were initially fabricated for this study. TiO₂ was chosen for its high (bulk) index of refraction and to compare with published results [10,15,16,17]. All SBD chiral STFs are structurally left-handed. The thickness of each equals eight or more helical pitches, and the center-wavelengths ${\lambda }_{\mathrm{o}}^{\text{Br}}$ lie in the range 450–670 nm before annealing. Optical characterization experiments revealed that all five samples display well-developed CBP. Nine other SBD chiral STFs were fabricated in a second batch and optically characterized as well; as the conclusions drawn therefrom were the same as for the first batch, experimental results for the second batch are not provided here. A third batch of eight SBD chiral STFs (samples 6–13) were characterized also, results for which are provided as appropriate here.

2.1. STF fabrication

All samples were deposited using a Semicore evaporation chamber, at a base pressure less than 5×10^-6 Torr, and the Telemark electron beam evaporation equipment. Although oxygen can be flowed into the deposition chamber to successfully deposit stoichiometric films [18], due to equipment constraints oxygen was not utilized in these depositions. The source material was TiO₂ (99.99 %) pellets from International Advanced Materials. Two unheated Corning 7059 glass substrates, a distance d₁ apart, were centered a distance d₂ above a 7-cc crucible on vacuum-rated motors that were controlled external to the vacuum chamber with a Windows-based computer interface. Substrates were loaded onto a 5-cm diameter holder and their centers placed in a 4 cm diameter circle centered on the axis of rotation. Four measurements made from four different substrates located radially at 0 mm, 7 mm, 14 mm, and 21 mm from the center show no significant change in the CBP, thereby establishing the uniformity of the SBD chiral STFs. The operating pressure was set below 5×10^-5 Torr. All substrate rotation was in the counter-clockwise sense looking at the substrate, thus endowing the growing SBD chiral STFs with structural left-handedness. The electron beam was maintained at 8 kV and 0.1 A manually for a constant deposition rate of 0.25 nm s^-1, as measured by a quartz crystal monitor at normal incidence located near the substrate. A nearly constant deposition rate allows for uniform rotational periods and hence a constant pitch. While fabricating samples 1–5, the crucible was refilled with the evaporant material prior to the deposition of each sample. On the other hand, samples 6–13 were fabricated during two depositions each of which started with completely new evaporant material. Deposition parameters can be seen in Table 1.

2.2. Annealing

Annealing at 500°C for one hour is sufficient to change amorphous TiO₂ to an anatase crystal structure [16]. For chiral STFs fabricated using the original deposition technique (i.e, non-SBD), the CBP appears to stabilize after roughly 5 hr of annealing [15]. Therefore, the SBD chiral STFs were annealed in an oxygen environment at 425°C for 8 hours. This period of annealing time was found to be sufficient to change amorphous TiO₂ into anatase TiO₂, based on grazing-incidence X-ray diffraction (XRD) as seen in Fig. 1.

Table 1. Summary of deposition parameters for 13 different SBD chiral STF samples.

View Table | View all tables in this article

2.3. Optical characterization

All SBD chiral STFs were characterized to determine optical responses. The transmission of CP light by the samples was measured over the visible regime. Light from a 100-W Leica halogen lamp was successively passed through focusing optics, a Glan-Thompson linear polarizer, a quarter-wavelength Fresnel rhomb, the chiral STF on Corning 7059 glass, another quarter-wavelength Fresnel rhomb, another Glan-Thompson linear analyzer, and a fiber optic leading to one of three channels of an Ocean Optics spectrometer. The environment was not controlled during measurement in the laboratory. Samples 1–5 were characterized and annealed more than two months after deposition and were stored in a cabinet. Samples 6–13 were characterized and annealed two months after deposition. Optical measurements of samples 6–13 were taken following submersion in de-ionized water for 2 min followed by 1 min on a warm hot plate. This process was used for measurements before and after annealing. Total measurement time for each sample was less than 10 min. This procedure was used to partially control the role of environmental moisture because the optical properties of porous thin films have been shown to change due to moisture uptake [17] Measurements were made and samples were annealed within 72 hr.

The incident light could be either left or right-CP (LCP or RCP), and the intensities of LCP and RCP components of the transmitted light were measured. Thus, the four transmittances T_LL, T_RL, T_RR and T_LR were measured as the (free-space) wavelength λ _o was varied from 400 to 700 nm; here, for instance, T_RL is the relative intensity (expressed as a percentage) of the RCP transmitted light when the incident light is LCP. Each transmittance was found using the spectrometer software and the definition

where S is the measured intensity when the SBD chiral STF is present, R is the reference intensity transmitted only through the uncoated Corning 7059 glass, and D denotes the dark intensity. All four transmittances were measured before and after annealing.

 

Fig. 1. X-ray diffraction spectrum of an annealed SBD chiral STF with background removed. The Miller indices corresponding to the x-ray diffraction peaks are shown on the lower graph. These peaks indicate the anatase phase of TiO₂, and would be absent in the x-ray diffraction spectrum of amorphous TiO₂.

Download Full Size | PDF

From the measurement of T_LL of structurally left-handed STFs as a function of the wavelength, the needed constitutive properties can be determined. In addition to structural handedness, a chiral STF possesses five more quantities that describe its constitutive properties [3, Chap. 9]: the angle χ, which is the rise angle of the helical columns in non-SBD films but is a columnar rise angle for the subdeposits in SBD films (the columns in SBD films with symmetric subdeposits are virtually upright); the pitch P of the helical columns; and three relative permittivity scalars ε_a, ε_b, and ε_c, which delineate local orthorhombicity of the chiral STF. (Note that ε_a, ε_b, and ε_c characterize the entire STF, and not simply the column material.) For normal incidence of light, it is appropriate to define ε_d = ε_a ε_b/(ε_a cos²χ + ε_b sin²χ). With the assumption of negligible dispersion and dissipation, the center-wavelength of the Bragg regime for normally incident light is estimated as

and the full-width-at-half-maximum (FWHM) bandwidth as

The quantity |√ε_c - √ε_d| is the local linear birefringence. After estimating ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$ and (Δλ_o)^Br from the measured transmittance spectrums, Eqs. (2) and (3) could be used for film analysis. However, measuring the FWHM bandwidth graphically can become difficult if irregularly shaped troughs are present in the transmittance spectrum. However, the following protocol was used consistently throughout which provides validity for comparison: First, the wavelength of minimum T_LL was determined and designated as ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{min}}$. Next, T_RR (${\mathrm{\lambda }}_{\mathrm{o}}^{\text{min}}$) was determined. A horizontal line was drawn midway between T_LL (${\mathrm{\lambda }}_{\mathrm{o}}^{\text{min}}$) and T_RR (${\mathrm{\lambda }}_{\mathrm{o}}^{\text{min}}$). The intercept of this straight line with the trough in the spectrum of T_LL was taken to be (Δλ_o)^Br, while the central wavelength of the intercept was designated as . Note that ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$ and ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{min}}$ may not coincide but are in close proximity for all 13 samples. The preferred and more accurate method would be a combination of ellipsometry [19] and mathematical modeling [20].

2.4. Crystallinity and thickness

Pre- and post-annealing characterization of the crystal structure of all SBD chiral STFs was done using XRD grazing angle measurements on Scintag Inc. equipment. Measurements were made with a step size of 0.02° from 2° to 64° over one hour. The incidence angle was set to 2°. A field emission scanning electron microscope (FESEM) was used to characterize the morphology of SBD chiral STFs deposited on silicon, so as to determine the approximate morphology changes upon annealing. The thickness of each film was measured using the Tencor P-10 profilometer along an edge of the material defined with a kapton tape mask. Three or four measurements were made and the average value recorded. We ascertained that the residue from the kapton tape can create an uncertainty of 12 nm in values of P pre-annealing and 6 nm after annealing. The thickness of sample 5 alone was estimated from FESEM images.

Table 2. Summary of changes in measured data due to annealing. Top value is pre-annealing, middle value is post-annealing, and bottom value is percent change.

View Table | View all tables in this article

Table 3. Summary of changes in inferred/calculated data due to annealing. Top value is pre-annealing, middle value is post-annealing, and bottom value is percent change.

View Table | View all tables in this article

3. Results and discussion

As mentioned previously, 13 chiral STFs of TiO₂ were fabricated using the SBD technique. XRD analysis revealed that all samples were amorphous before annealing, and that the samples changed to anatase crystal structure after annealing. Figure 2 contains high-resolution cross-sectional FESEM images of a SBD chiral STF before and after annealing. All samples appeared slightly bluish, on transmission to the naked eye before annealing, which suggests an oxygen deficiency in the films [20]. Although the SBD chiral STFs are very porous and were annealed in oxygen, this coloration persisted after annealing. We note that a reviewer stated that oxygen-deficient titania films observed by him/her to be slighty brown; this distinction could be due to widely varying stochiometries of titania films obtained from different sources.

Transmittance spectrums from four SBD chiral STFs are shown in Fig. 3. Cross-polarized transmittances T_LR and T_RL were either unaffected by annealing or decreased slightly. Measured data is presented in Table 2. While conclusions drawn from the transmittance measurements are summarized in Table 3. The center-wavelengths ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$ blue-shifted approximately 3.4–8.5 % of the pre-annealing value and (Δλ_o)^Br increased 3–20 %. Samples 6–9 that were deposited on the same substrate holder showed a blue-shift from 652 ± 2 nm to 630 ± 2 nm. Samples 10–13 showed a blue-shift from 624 ± 3nm to 601 ± 1 nm. The difference |T_R-T_L|, where T_R= T_RR + T_LR and T_L = T_LL + T_RL, increased 1–8 %. However, sample 1 had a decrease in |T_R-T_L| of 1 % that occurred due to blue-shift upon annealing into a highly dispersive short wavelength regime that reduced T_RR substantially. The structural period P decreased 1–3.8 %. Except sample 8 where P increased 0.9 %, this increase is attributed to a high amount of residue from kapton tape that was removed during annealing at 425°C.

Using the values of ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$ and (Δλ_o)^Br estimated from the transmittance spectrums, as well as from the measured values of P, we found from Eq. (2) and Eq. (3) that (√ε_c + √ε_d)/2 decreased 0–5 % while the local linear birefringence |√ε_c - √ε_d| increased 7–29 %. Samples 6-9 had pre-annealing values of 1.44 ± 0.04 and 0.147 ± 0.003 and post- annealing values of 1.42 ± 0.02 and 0.164 ± 0.003 of (√ε_c + √ε_d)/2 and |√ε_c - √ε_d| respectively. Samples 10–13 had pre-annealing values of 1.41 ± 0.01 and 0.132 ± 0.001 and post-annealing values of 1.39 ± 0.02 and 0.143 ± 0.002 of (√ε_c + √ε_d)/2 and |√ε_c - √ε_d| respectively. The local linear birefringence values of samples 1–5 are higher than those previously published [10] which may be attributed to (i) different process parameters, such as the incremental SBD angle, Δξ, and (ii) stochiometric differences between the evaporant materials coming from different suppliers. Also, as thickness are underestimated from tilted cross-sectional SEM images, quite likely P of sample 5 was underestimated, which led to an over-estimation of (√ε_c + √ε_d)/2 and |√ε_c - √ε_d|.

The major conclusion from our experiments is the blue-shifting of the CBP after annealing, as is clear from Fig. 3. Now, the void regions of a thin film should not be thought of as being vacuous; instead, they are filled with low-density material in comparison to the columns [21, 22]. Infiltration of void regions with a material whose bulk refractive index is even lower [8], columnar thinning [23], and a decrease in the helical pitch [3] are three factors that will blue-shift ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$. Let us examine these factors in turn.

The first factor is the reduction of the helical pitch P. Suzuki et al. [16] observed that TiO₂ columnar STFs, grown with vapor directed at an angle χv = 20° with respect to the substrate plane, contain closely spaced columns. These columnar STFs, which are not structurally chiral, can decrease as much as 10 % in thickness upon 1 hr annealing at 500 °C. In contrast, non-SBD chiral STFs fabricated with χv = 8° contain isolated helical columns and were reported not to decrease in thickness upon annealing [16]. To a large extent, the vapor incidence angle χv dictates the separation of the columns. The SBD chiral STFs, grown at χv = 11° and 12.5°, for this study contain closely spaced columns, as seen in Figs. 2 and 4. Therefore the slight decrease in P (1–3.8 %) cannot suffice to explain the reduction of ${\lambda }_{\mathrm{o}}^{\text{Br}}$ by 3.4-8.5 %. In addition, the FWHM bandwidths increased by 3–20 %, which also indicates that other factors are at work.

Figure 2 shows high-resolution FESEM images of a SBD chiral STF fabricated on silicon during deposition of samples 6–9; Fig. 2(a) is pre-annealing and Fig. 2(b) is post-annealing. The average column diameter is ~38 nm in Fig. 2(a) and is ~33 nm in Fig. 2(b). These values were estimated by measuring 10 specific column diameters and calculating the average. Note that due to the helical shape of the columns, the diameter in these images is not uniform through the thickness hence, measurements were performed in those vertical sections wherein neighboring columns can be easily distinguished from each other. This diameter reduction results in a ~13% columnar thinning. From Fig. 4 it is clear that large voids appear as a result of annealing. From Figs. 4(c) and 4(d), a conservative estimate of the void size is 20 nm, whereas the columns before annealing were 70–80 nm in diameter but 60–70 nm in diameter after annealing. Thus, a ~15% decrease in column diameter indicates columnar thinning due to annealing. Columnar thinning has been theoretically shown to blue-shift ${\lambda }_{\mathrm{o}}^{\text{Br}}$[23]: roughly, 10% columnar thinning leads to a 9% blue shift. But ${\lambda }_{\mathrm{o}}^{\text{Br}}$ reduced by only 3–8%, which indicates that some other factors are at work to partially mitigate the effect of columnar thinning.

 

Fig 2. High-resolution cross-sectional FESEM images of the column base similar to those of samples 6–9 (a) pre-annealing image and (b) post-annealing image. The faint ~10 nm periodic characteristic is due to microscope vibration.

Download Full Size | PDF

 

Fig. 3. Measured spectrums of the transmittances T_LL, T_RL, T_RR, and T_LR of four different SBD chiral STFs pre-and post-annealing, (a) sample 2, (b) sample 3, (c) sample 10, and (d) sample 6.

Download Full Size | PDF

 

Fig. 4. FESEM images of sample 5(a) before annealing, and (b, c) after annealing; (d) is a magnified version of (c) to show columnar thinning and void expansion due to annealing. The half pitches of the helical columns are highlighted with lines in (c) and some of the voids are highlighted in (d).

Download Full Size | PDF

The clear implication is that the optical permittivity of the column material must have increased due to annealing (which would change ε_a, ε_b, and ε_c) , any significant change in χ being ruled out by the very small reduction in P. Indeed, in a preliminary study on annealing of non-SBD chiral STFs [15], the authors attributed a 1.7 % red-shift of ${\mathrm{\lambda }}_{\mathrm{o}}^{\text{Br}}$ upon annealing for 8 hr at 450 °C to an increase in the bulk refractive index of TiO₂ upon transformation from amorphous to anatase crystal structure.

The increase in FWHM bandwidth and the increase in selective transmission can be attributed to increased local linear birefringence, which can be caused by columnar thinning and an increase in the bulk refractive index of TiO₂.

The red-shift of the CBP reported for non-SBD chiral STFs [15] in contrast to the blue-shift reported here for SBD chiral STFs deserves comment. First, the 1.7 % red-shift is considerably smaller than the 3.4–8.5 % blue-shift. Next, the number of helical pitches in the non-SBD chiral STFs annealed was 3 in contrast to 8 for SBD chiral STFs, and it is known that non-SBD chiral STFs have considerably smaller local linear birefringences than comparable SBD chiral STFs [10]. Therefore, in contrast to the SBD chiral STFs presented here, those non-SBD chiral STFs may not be expected to exhibit a well-developed CBP [2]. This is in fact borne out by data on |(T_L - T_R)|, whose value at oBr was reported in Ref. 15 to be ~0.22 for an annealed non-SBD chiral STF but is 0.81–0.97 for the annealed SBD chiral STFs in this study. Quite likely, the relative importances of the three factors identified by us are different of the sole sample presented in Ref. 15. In any case, the considerable difference between Ref. 15 and our results on 13 samples strongly suggests that conclusions about the effects of annealing on optical performance should only be drawn with data on chiral STFs that exhibit well-developed CBP, if the objective is to use those chiral STFs as CP filters [24].

To conclude, we have shown that, on annealing, (i) a small reduction in helical pitch and (ii) columnar thinning together lead to an overall blue-shift of well-developed CBP exhibited by SBD chiral STFs, in spite of a concurrent tendency towards red-shifting by an increase in the relative permittivity of the column material due to a change from amorphous to crystalline nature. As post-deposition annealing significantly improves the environmental stability of thin films, the spectral changes caused by annealing must be taken into account when deciding upon the process parameters for deposition. Further research should focus on other factors that may play subtle roles in annealing.