Open Access
Add to BibSonomy
Abstract
Material characteristics and crystallinity of germanium antimony telluride (GST), along with a design and experimental verification of a transmission filter in the mid-infrared are presented. Gradually increasing the temperature of an annealed phase change material, such as GST, controls the amount of crystallinity which allows the index of refraction to increase continuously over a significant range. GST is incorporated into a metal-dielectric sub-wavelength grating, in which the dielectric segments of the grating are spaced out sufficiently to decouple the angular-dependent grating resonance from the Fabry-Perot resonance for each individual cavity. Filling in each cavity between the metal strips with GST allows for active tuning of the transmission wavelength. Angular independence of the transmission resonance is shown up to 60 degrees for devices of an amorphous and crystalline state of GST, demonstrating that the angular independence is maintained through wavelength tuning.
1. Introduction
Applications such as multispectral imaging [1], remote chemical sensing [2], and infrared countermeasures require narrow-band spectral transmission filtering, preferably polarization-insensitive [3], that can be integrated into existing photonic structures and offer a wide angular field of view. The latter consideration requires angular insensitive transmission characteristics over a large range. Filters exhibiting sharp resonances have been demonstrated using multilayered films [4,5] or by utilizing surface plasmon polariton resonances [6–8], although these filters are inherently angle-dependent. It is the condition for the phase synchronism between the incident wave and the excitation propagating along the surface, in either plasmonic or all-dielectric multilayer structures, either flat or periodical, which makes the resonances angle-dependent. Designing a structure in which the lateral transfer of electromagnetic energy is virtually eliminated, or highly suppressed, achieves angle-independent filtering. We recently demonstrated an angular-independent transmission filter using a periodic array of high-index filled nanocavities in a metal thin film [6]. Each of these cavities acts as a Fabry-Perot resonator, whereby designing the cavities to be subwavelength decouples the resonance from the surface plasmon polariton resonance and, hence, eliminates any angular dependence of the device [6,9]. By this we achieved angular-independent transmission up to 60-degree angle of incidence. Moreover, we showed that the resonance could be configured based upon device geometry [10,11]. However, for many applications multiple filters would be required to fully analyze signatures containing many distinct spectral lines. The ability to tune the transmission resonant wavelength actively and rapidly would solve this problem.
Active tuning of optical properties has been achieved using transparent conductive oxides, such as indium tin oxide (ITO) [12–14], as well as volatile and non-volatile phase change materials (PCMs) [15–18]. ITO has been shown to have high index changes around its epsilon-near-zero wavelength, typically in the near-infrared, under intense optical excitation, but at longer wavelengths there is little change in the refractive index. PCMs undergo a reversible structural phase change between an amorphous and crystalline state brought on by a thermal impulse. The thermal stimuli is key in changing and controlling both the refractive index and complex permittivity. Ge2Sb2Te5 (GST-225, herein GST) is a well-studied PCM that can change between an amorphous state and a crystalline state by thermal annealing or more rapidly by applying an electrical or optical pulse. Indeed, these phase changes can occur on the timescale of a nanosecond, or even sub-nanosecond, when optically pulsed and will remain in that state until further biasing is applied [19].
We present here a detailed analysis of GST which shows the electrical, structural, and optical changes that occur due to thermal annealing. With how sensitive this PCM parameters are, it is crucial to measure material characteristics to account for unintentional defects. This characterization demonstrates that the films we deposit agree with those found in literature [15,16,19]. We then present a design of an angular insensitive transmission filter that achieves active tuning by incorporating GST into the nanocavity design. Applying a thermal bias to amorphous GST causes a gradual change in the crystallinity of the material, which results in a large change in refractive index. Finally, we show experimental verification of our design resulting in a transmission wavelength redshift of over 1 micrometer in the mid-IR spectral range while maintaining angular insensitivity.
2. GST characteristics
It has been shown that GST can exist in two different crystalline phases [20,21]. A standard sputter deposition process produces an amorphous structure which has a high resistivity around ${10^2}$ Ωm. An inert or atmospheric gas annealing process between 150°C and 200°C will transform this film to a face-centered cubic (FCC) crystalline state. This state’s resistivity lies between ${10^{ - 1}}$ and ${10^{ - 4}}$ Ωm, being significantly lower and broad in range due to its varying crystal density structure. Further annealing at temperatures above 250°C will produce a hexagonal close-packed (HCP) state with resistivity two orders of magnitude lower. The FCC-state is capable of reverting to an amorphous structure by exciting the film with sufficiently high energy followed by rapid thermal quenching, which is achievable through optical biasing; however, the HCP-crystalline state is not easily reversed [19].
We deposited and characterized GST thin films to facilitate accurate modeling and simulations for our filter design. GST thin films were deposited by DC-magnetron sputtering using a GST-225 target onto silicon and sapphire substrates. These films each had a nominal thickness of 600 nm. The films were divided into eight groups of varying annealed temperatures: one left as-deposited, five annealed by a hotplate for 2 minutes at 125°C, 150°C, 175°C, 200°C, and 250°C, and the last two placed in an oven (Omega Lux, LMF 3550) for 2 minutes at 300°C and 350°C. These films are not patterned and analogous in temperature treatment of the devices described later in this paper.
Sheet resistance was measured for each of the films using a 4-point probe and resistivity was calculated assuming a 600 nm thickness. The amorphous GST had a resistivity of ∼60 Ωm. The FCC-crystalline film (cGST) at 175°C measured at $1.6\textrm{x}{10^{ - 3}}$ Ωm. The HCP-GST film at 350°C had a resistivity of $8.7\textrm{x}{10^{ - 6}}$ Ωm. Figure 1 displays the resistivity for all eight GST groups and illustrates the regions where the film transitions to FCC and HCP.
Fig. 1. Resistivity of GST film as a function of annealing temperature. Classified regions of amorphous, FCC, and HCP crystalline states based on resistivity using a 4-point probe.
To determine crystallinity, the films were measured using x-ray diffraction (XRD, PANalytical Empyrean) using an asymmetric out-of-plane configuration (Fig. 2). This data matches accordingly to Matsunaga et al. [21] and Petrov et al. [22], for FCC and HCP, respectively. The amorphous film shows no clear crystalline structure for the as-deposited film as well as the film annealed at 125°C, as expected. There appears to be a transitional phase change at 150°C showing signs of mixed amorphous and polycrystalline FCC-GST. After annealing at 175°C and 200°C, FCC-crystallinity is shown with the (111) peak at 25.7°, the (002) peak at 29.5° and the (022) peak at 42.7° (PDF 01-073-7757) [21]. The dominant peak is (022), which shows texturing towards this orientation as the (002) peak at 29.5° dominates the powder diffraction reference. This peak is observed after annealing at 150°C and its intensity grows by a factor of 2.5 after the sample has been annealed at 175°C. The intensity of this peak increases another 10% after a 200°C anneal. This indicates that there is a gradual volumetric growth of polycrystalline FCC-GST from the amorphous state.
Annealing at 250°C and above produces a more rapid and complete shift to the HCP phase, as indicated by the presence of the (013), (106), (110), and (203) peaks located at 29.1°, 39.5°, 43.8°, and 52.8°, respectively, along with the elimination of any of the FCC-GST phases [22]. The intensity is greatly increased for these peaks which indicates increased crystallinity with a preferred orientation towards the (110) over the (013), which dominates the powder diffraction file (PDF 01-075-1024).
Peak width is related to crystallite size by the Scherrer equation [23], which provides a lower limit for the average crystallite size. The equation takes the form
where τ is the volume weighted grain size, Δ(2θ) is the full width at half maximum of the diffraction peak, θ is the Bragg angle, λ=1.5406 Å is the Cu K-alpha wavelength, and K ∼0.9 is the shape factor [24]. By this equation, the FCC-GST crystallite sizes are at minimum 10 nm and remain constant over the annealing temperature range 150°C- 200°C. Using the same equation on the HCP-GST film, the minimum crystallite grows to 35 nm. This range of crystallite sizes are verified to increase, from FCC to HCP, using atomic force microscopy (AFM).The complex optical permittivity was determined by spectroscopic ellipsometry from 0.5 $\mathrm{\mu }$m −8 $\mathrm{\mu }$m (JA Woollam, V-VASE and IR-VASE). For the as-deposited GST films, and films annealed up to 200°C, the data was modeled using a Cauchy term and a series of Gaussian oscillators which match absorption peaks in the imaginary part and use Kramers-Kronig equations to generate the real part. By simultaneously fitting ellipsometry data with multi-angle reflectance, both the complex permittivity and film thicknesses were fit. For the films annealed at 250°C and above, a Drude term was introduced into the model to account for the large imaginary part of the permittivity at longer wavelengths. This Drude term dominates this model, which accounts for the very different permittivity observed for the 250°C film in Fig. 3. Amorphous-GST (aGST) films have an index of refraction around n = 3.48 at 5 µm wavelength, with very low loss. The FCC-GST films show a gradual and continuous increase in the index to around 4.61 at 5 µm wavelength, while the imaginary part of the permittivity remains low. After annealing at 250°C, HCP-GST has significantly higher loss while the real part of the permittivity becomes negative (metallic) beyond 5.5 µm. This is consistent with the low resistivity measured for this film. Sarangan et al. show the index of refraction for aGST and FCC-GST thin films annealed at various temperatures between 1.25-2.45 µm [15]. At 2.5 µm, they report the aGST has an index around n = 4 + 0.01i while the cGST annealed at 175°C has n = 5.5 + 0.1i, both of which agree with the results presented here.
Fig. 3. Complex permittivity of as-deposited and annealed GST films. Real part (left) and imaginary part (right).
3. Device design and simulations
Using Lumerical software, simulations were designed for optimizing a specific resonance order and maintaining angular independence for a device that is possible to manufacture. Periodic boundary conditions are used to calculate the response of the entire device to reduce simulation time.
A thin-film structure is inherently angle-dependent, and the mechanism to achieve angle-independence lies entirely within the cavity design. The filter consists of a periodical set of metal and dielectric (GST) strips. By setting the lateral dimension of each slit cavity to be sufficiently large, the slit resonance is decoupled from the periodic grating resonance. The result is that each cavity acts as its own Fabry-Perot resonator which is inherently angular independent. This requires the dielectric sections to be narrow enough to avoid resonances associated with the lateral propagation of light within the dielectric strip. Our previous work [6] explains why and how this particular design is ideal for angular independence. When the film is broken into optically separated resonators, the phase synchronism is no longer a factor—resonant wavelengths no longer depend on the incident angle. The substrate is a low index dielectric that is highly transparent in the spectral band of interest. Therefore, the transmission wavelength of this structure is heavily dependent on the thickness, lateral cavity dimension, and index of refraction of the GST. The first two parameters can be easily designed and configured during device fabrication; the latter can be actively tuned by switching the phase of the GST material.
The permittivity of the GST is set using the measured values obtained from ellipsometry, while the permittivity of silver is defined using the Drude model with the plasma frequency of $7.25\textrm{x}{10^4}\; \textrm{c}{\textrm{m}^{ - 1}}$ and damping factor $1.45\textrm{x}{10^2}\; \textrm{c}{\textrm{m}^{ - 1}}$. The refractive index of the substrate (sapphire) is set to be 1.67 with no loss. Wavelength-dependent optical constants of GST used in the simulation were obtained by spline-interpolating ellipsometric data. Periodic boundary conditions are used.
The device simulations had dimensions of 1$\;\mathrm{\mu m}$ GST strip width and 550 nm height with equal width silver strips of height 400 nm, illustrated by Fig. 4. By changing the angles of incidence for this device, Fig. 5 shows the transmittance is independent of angle up to 60° for both TE and TM-polarized light. For TE-polarized light, transmittance for the fundamental peak shows ∼90% up to 10° at a fixed wavelength of 4.1$\;\ \mathrm{\mu }$m. In TM-polarized light, overall transmittance stays constant at ∼80% up to 10° and wavelength is fixed at 6 $\mathrm{\mu }$m for all angles. This uniformity in angular dependence arises from the size of the dielectric in the device. As the dielectric increases in width, the resonance peak broadens and becomes red-shifted.
Fig. 4. Schematic diagram of a resonant-cavity mid-infrared filter. Dielectric gratings are 1 $\mu $m in width and 550 nm in height, metal is 400 nm in height with equal width, and blue substrate is sapphire. TE and TM-polarization is illustrated as the electric field traverses in the dielectric.
Fig. 5. Simulated angle independence of resonant peak, m =1, from 0° to 60° of as-deposited GST device for TE polarized and TM polarized light. TE-polarization has a peak resonance at 4.1 µm while resonance for TM-polarization is at 6 µm.
Figure 6 presents simulated TE and TM-polarized transmittance at normal incidence for devices at varying states of crystallization. The fundamental resonance, m = 1, for TE-polarization has a peak shift from 4.1 $\mathrm{\mu }$m to 4.9$\;\ \mathrm{\mu }$m, which as shown above has a corresponding change of the refractive index. At second order, m = 2, has a shift that is 3.2 $\mathrm{\mu }$m to 3.3 $\mathrm{\mu }$m. In TM-polarization case, a significant shift from 6 $\mathrm{\mu }$m to 7.6 $\mathrm{\mu }$m as the GST is transitioned from an amorphous structure to the FCC crystalline state. A second order, m = 2, is shown with a shift from ∼3.2$\;\ \mathrm{\mu }$m to 4$\;\ \mathrm{\mu }$m. Neglecting the phase of reflection at the top and bottom interfaces and leaving aside the effect of guiding the light wave in a sub-wavelength dielectric-filled gap between the metal sections, the Fabry-Perot transmission resonances of order m are estimated to be at the wavelengths ${\lambda _m}$ such that the resonator length (film thickness $d$) would be equal to the multiple integer (factor of $m$) of the in-material wavelength ${\lambda _m}/n({{\lambda_m}} )$. This leads to an estimation of ${\lambda _m} \approx 2dn({{\lambda_m}} )/m$. For a given resonance (e.g., $m = 1$ or $m = 2$), the resonant wavelength mainly follows the index change associated with the change of the crystallinity of GST. Transition from pure amorphous phase (${n_{GST\_Am}} \approx 3.5$) to pure cubic crystalline (${n_{GST\_fcc}} \approx 4.5$) phase results in a refractive index increase as large as 28%. Correspondingly, the resonant wavelengths of the main transmission peaks in the system under discussion would experience a red shift approximately directly proportional to the index change. This expectation is not absolute because the refractive index is wavelength-dependent and the effective length of the Fabry-Perot resonator is not exactly equal to the thickness of the film. Nevertheless, change of the refractive index of GST during annealing defines direction and magnitude of the resonant wavelength change.
Fig. 6. Simulated transmission for filters in the amorphous, FCC, and HCP states at normal incidence. For TE-polarization, resonance spectrum shifts from 4.1 µm to 4.9 µm. Transmission resonance in TM-polarized light shifts from 6 µm to 7.6 µm when GST undergoes a phase transition. In the HCP regime transmission drops to near zero as expected due to complex permittivity.
Normal incidence spectral features at 3.2 $\mathrm{\mu }$m are associated with the first ($\textrm{N} = 1$) diffraction order transitioning from evanescent to propagating wave in the cover medium (air, ${\textrm{n}_\textrm{c}} = 1$) and substrate (${\textrm{n}_\textrm{s}} = 1.67$), respectively. These features near 3 $\mathrm{\mu }$m are inherently angle-dependent. At oblique incidence, each of these features is split in two, with spectral locations defined by $\mathrm{\lambda } = \mathrm{\Lambda }({\textrm{n} \pm \mathrm{sin\theta }} )/\textrm{N}$, where $\textrm{n}$ is the refractive index of the cover medium (air) ${\textrm{n}_\textrm{c}}$ or substrate ${\textrm{n}_\textrm{s}}$, $\mathrm{\theta }$ is the incident angle in air, $\mathrm{\Lambda }$ is the structure period, and $\textrm{N}$ is the order of diffraction. As incident angle increases, one of such angle-dependent spectral features located at $\mathrm{\lambda } = \mathrm{\Lambda }({{\textrm{n}_\textrm{s}} + \mathrm{sin\theta }} )$ may overlap with the transmission peak associated with the fundamental Fabry-Perot resonance. Observing peculiarities of this sort requires a well collimated incident beam.
Overall, to achieve angle-independent behavior of the filter in a wider range of incident angles, the structure should be designed properly to avoid undesirable overlapping of spectral features.
4. Fabrication
Device fabrication required careful consideration of the lithography process and substrate of choice when incorporating GST into the design. While previous work [6] involving Ge-Ag filters used BaF2 substrates which have a transmission wavelength up to 10 µm, this material is brittle and can crack under high thermal treatments, which would be required to anneal the GST. Considering this, sapphire substrates were chosen which are more robust, but forced the transmission peaks to be in the mid-IR. Optical constants for as-deposited GST were found to be consistent whether on sapphire or silicon substrates. As was previously shown, there is little change between as-deposited GST and films annealed at 125°C. Therefore, the fabrication processes were designed to keep temperatures below 125°C to avoid unintentional annealing.
To achieve the desired grating structure pattern, a layer of negative resist, AZ-5214, is applied on a $\textrm{A}{\textrm{l}_2}{0_3}$ substrate and baked at 110°C for 75 seconds. AZ-5214 was chosen due to its low baking temperature and easy lift-off after metal deposition. This resist is exposed at a wavelength of 365 nm using a projection stepper (Auto-Step 200) for 0.12 seconds, followed by a post-bake at 110°C for 60 seconds. The mask grating used has a 2$\; \mathrm{\mu m}$ period with 50% duty cycle. The sample is then flood exposed using a mask aligner (Karl Suss MA6) with a wavelength of 365 nm for an additional 60 seconds before being developed in a 1:5 ratio of AZ351:DI-water. At this stage the patterned resist has a step height of 1.3$\; \mathrm{\mu m}$ confirmed by stylus profilometry.
The GST-etching process uses a chlorine-argon mixture, ∼25% chlorine, for 190 seconds with RIE and ICP powers of 100 W and 300 W, respectively, fully etching the GST film. Silver is deposited by e-beam evaporation to a thickness of 400 nm to ensure the metal lift-off process works well with the left-over photoresist after the ICP etching. Representative cross section and top view SEM images are shown in Fig. 7 for the as-deposited device.
SEM analysis indicates that the GST is taller than the depositited Ag metal, as expected, and the gratings in the top view show uniformity in the film. The dimensions of the GST vary by anneal temperature as the crystallization process produces a volumetric change in the material. The two amorphous temperatures (unannealed and 125°C) showed the GST grating width to be 1.3$\;\ \mathrm{\mu m}$. After annealing, the FCC region averaged 1.2$\;\ \mathrm{\mu m}$ width while the HCP region averaged 1.1$\;\ \mathrm{\mu m}\;\ $width. The GST gratings are constricting up to 15% during the transition from amorphous to HCP [25,26], enough that air gaps are present in this hexagonal phase structure.
5. Results and discussion
Transmission spectra for these devices were measured using a Bruker FTIR with a reflecting microscope objective to narrow the incident beam to a small section of the patterned material. Consequently, the incident light-rays are not parallel but form a hollow cone of 12°-24° which excludes normal incidence. The sample is tilted with respect to normal incidence of this light cone, and for simplicity the sample tilt angles are reported to verify the filter angular independence. Transmittance is reported into the sapphire substrate, which has a high transmittance up to about 7 µm at which point the substrate becomes opaque. Thus, it is impossible to obtain experimental data beyond that point, and the plots are therefore cut-off at 7 µm. A bare sapphire substrate is used as the reference material. The incident light is TM-polarized with respect to the gratings.
Figure 8 presents the transmission resonance peaks at normal incidence for aGST annealed at 125°C, FCC-GST annealed at 175°C, and HCP-GST annealed at 250°C, overlayed with the simulated spectra from Fig. 6. The amorphous film shows the fundamental resonance at 6.3 µm with a transmittance of 99%. The second order (m = 2) peak is visible around 3.1 µm, with roughly 65% transmittance. Due to the light cone from the FTIR, the two simulated peaks in Fig. 6, near 3.2 µm, are averaged and combined. Although small, the third order peak (m = 3) is visible around 1.75 µm with about 35% transmittance. The FCC-GST sample shows a 1 µm wavelength shift from 3.1 µm to 4.2 µm for m’ = 2 with a 45% transmittance. The second mode matches near exactly with what is simulated in Fig. 6. Slightly blue shifted and is more noticeably shifted to ∼6.2 µm for the fundamental mode. The fundamental order is beyond the sapphire substrate cutoff and cannot be reported. As expected from ellipsometry data shown in Fig. 3, HCP has high loss in permittivity for the imaginary part and negative for the real part. Thus, the filter with HCP-GST has poor transmission in MIR resulting in permanent termination of the filter. Note that transmission in the simulated HCP-GST device, annealed at 250°C, was near 0% while experimentally it is comparably significant. This transmittance is likely a result from air gaps between GST-Ag strips as GST constricts at high annealing temperatures. Alternatively, it may be advantageous to design air gaps into the structure to be used as additional tools in controlling transmission of the spectra.
Fig. 8. Measured transmittance (solid) at normal incidence of aGST annealed at 125°C, FCC-GST annealed at 175°C, and HCP-GST annealed at 250°C accompanied by its simulated values (dashed) from Fig. 6. For aGST, the sample remains amorphous with, m = 1, resonance peak at 6.5 µm and two more order resonances at 3.1 µm and 1.7 µm. Simulated aGST shows the fundamental to be shifted ∼0.3 µm and m = 1 appears to overlap. The FCC shows that m’ = 1 resonance is beyond the cutoff frequency of the sapphire substrate. Wavelength at m’ = 2 shifted to 4.2 µm and m’ = 3 is 2.4 µm. Simulated m’ = 2 matches peak resonance at 4.2 µm. HCP clearly demonstrates the high loss in the imaginary part of permittivity and metallic like structure.
Figure 9 plots the second order transmission peaks and demonstrates angular independence of the amorphous and crystalline GST devices up to 60° incidence for samples annealed at 125°C, 150°C, and 175°C. Peak resonance shifts from 3.1 µm to 4.2 µm and transmission drops from ∼65% to 45% as the GST becomes more crystalized. This gradual annealing process shows the transitional region from amorphous to FCC from the sample annealed at 150°C in MIR.
Fig. 9. Transmittance as a function of tilt for 3 annealed GST filters at the second order (m = 2) demonstrating angular independence and a wavelength shift from 3.1 µm to 4.2 µm as the GST crystalizes. Transmitance at amorphouse state is ∼65% and at ∼45% for FCC.
These devices were annealed on a hotplate and thus had phase change from amorphous to FCC. This method was used to measure and display data on how GST can be implemented as a phase change material in a mid-IR filter design. Optical or electric tunable-switching can enable more functionality in these devices as it allows for filters to go from amorphous to FCC and back to amorphous. With silver already deposited in the device, electrical switching from amorphous to FCC should be feasible [19,27]. By reducing the GST thickness, it would be easier to change the amorphic structure to fully crystalline. As seen in Fig. 9, if the user desires to terminate the device permanently, one only needs to excite the material in the HCP region and the transmission drops below 15%.
Due to limitations using a sapphire substrate, it is impossible to see the fundamental peak in the FCC state simulated to be at 7.6 µm. The experimental devices, as expected, redshifted due to the wider GST grating structure. Even with this redshift, simulation and experimental results agree that there is a fundamental resonance at ∼6 µm for the as-deposited GST device. Simulation and experiment also support the second order amorphous peak at ∼3.2 µm and the transition to FCC at ∼4.2 µm. In the HCP regime, due to the material’s high loss in permittivity, the dielectric GST becomes more metallic and nearly all reflective.
6. Conclusions
We have demonstrated the controllability of continuous index of refraction tuning by thermal annealing of GST thin films, which will gradually transition the film from amorphous to FCC to HCP crystalline states. Between the amorphous and FCC states a total refractive index change from 3.5 to 4.6 is observed while loss remains small. Higher temperature anneals in which the crystalline phase becomes HCP dramatically increases the imaginary part of the permittivity and the material acts as a lossy metal.
These properties are what lead to the construction of a tunable angular insensitive MIR transmission filter that can be configured to a wide range of wavelengths. The transmission wavelength is shifted due to the phase change of the GST via thermal annealing. It is this phase change that results in an index shift which changes the conditions of the Fabry-Perot resonance. As each cavity in the structure is acting as its own filter, the structure is widely angular independent, up to at least 60 degrees. This structure could be integrated using standard lithography processes into an existing photonic device to enable wide field of view transmittance selectivity.
Funding
Air Force Research Laboratory (FA9550-19RYCOR048, FA9550-20RYCOR059).
Disclosures
The authors declare no conflicts of interest.
Data Availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. A. V. Kanaev, M. R. Kutteruf, M. K. Yetzbacher, M. J. Deprenger, and K. M. Novak, “Imaging with multi-spectral mosaic-array cameras,” Appl. Opt. 54(31), F149–F157 (2015). [CrossRef]
2. M. Schmidt, D. Lei, and L. Wondraczek, “Hybrid nanoparticle–microcavity-based plasmonic nanosensors with improved detection resolution and extended remote-sensing ability,” Nat. Commun. 3(1), 1108 (2012). [CrossRef]
3. S. Hu, S. Yang, Z. Liu, B. Quan, J. Li, and C. Gu, “Broadband and polarization-insensitive absorption based on a set of multisized Fabry–Perot-like resonators,” J. Phys. Chem. C 123(22), 13856–13862 (2019). [CrossRef]
4. M. Wuttig, H. Bhaskaran, and T. Taubner, “Phase-change materials for non-volatile photonic applications,” Nat. Photonics 11(8), 465–476 (2017). [CrossRef]
5. G. Perez, A. M. Bernal-Oliva, E. Marquez, J. M. Gonzalez-Leal, C. Morant, I. Genova, J. F. Trigo, and J. M. Sanz, “Optical and structural characterization of single and multilayer germanium/silicon monoxide systems,” Thin Solid Films 485(1-2), 274–283 (2005). [CrossRef]
6. I. Avrutsky, E. M. Smith, S. Vangala, R. Gibson, J. R. Hendrickson, and J. W. Cleary, “Angle- and polarization-independent mid-infrared narrowband optical filters using dense arrays of resonant cavities,” Opt. Express 27(26), 37481–37493 (2019). [CrossRef]
7. J. Zhou and L. J. Guo, “Achieving angle-insensitive spectrum filter with the slit nanoresonator array structure,” J. Nanophotonics 9(1), 093795 (2015). [CrossRef]
8. T. W. Ebbesen, H. J. Lezec, H. F. Ghaemi, T. Thio, and P. A. Wolff, “Extraordinary optical transmission through subwavelength hole arrays,” Nature 391(6668), 667–669 (1998). [CrossRef]
9. S. Hadi Badri, M. M. Gilarlue, S. G. Farkoush, and S.-B. Rhee, “Reconfigurable bandpass optical filters based on subwavelength grating waveguides with a Ge2Sb2 Te5 cavity,” J. Opt. Soc. Am. B 38(4), 1283–1289 (2021). [CrossRef]
10. S. Hadi Badri and S. G. Farkoush, “Subwavelength grating waveguide filter based on cladding modulation with a phase-change material grating,” Appl. Opt. 60(10), 2803–2810 (2021). [CrossRef]
11. N. Ali, R. R. Panepucci, Y. Xie, D. Dai, and R. Kumar, “Electrically controlled 1 × 2 tunable switch using a phase change material embedded silicon microring,” Appl. Opt. 60(13), 3559–3568 (2021). [CrossRef]
12. J. A. Burrow, M. S. Alam, E. M. Smith, and R. Yahiaoui. “Chiral phase change nanomaterials,” arXiv preprint arXiv:2111.09940 (2021).
13. F. Yi, F. Shim, A. Y. Zhu, H. Zhu, J. C. Reed, and E. Cubukcu, “Voltage tuning of plasmonic absorbers by indium tin oxide,” Appl. Phys. Lett. 102(22), 221102 (2013). [CrossRef]
14. E. M. Smith, J. R. Hendrickson, J. W. Cleary, K. Leedy, J. Guo, and S. Vangala, “Tunable indium tin oxide for metamaterial perfect absorbers and nonlinear devices,” MRS Commun. 10(4), 573–578 (2020). [CrossRef]
15. A. Sarangan, J. Duran, V. Vasilyev, N. Limberopoulos, I. Vitebskiy, and I. Anisimov, “Broadband reflective optical limiter using GST phase change material,” IEEE Photonics J. 10(2), 1–9 (2018). [CrossRef]
16. Y. Zhang, C. Folwer, and J. Liang, “Electrically reconfigurable non-volatile metasurface using low-loss optical phase-change material,” Nat. Nanotechnol. 16(1), 6–15 (2021). [CrossRef]
17. S. Abdollahramezani, O. Hemmatyar, and M. Taghinejad, “Dynamic hybrid metasurfaces,” Nano Lett. 21(3), 1238–1245 (2021). [CrossRef]
18. Y. Qu, Q. Li, and L. Cai, “Thermal camouflage based on the phase-changing material GST,” Light: Sci. Appl. 7(1), 1 (2018). [CrossRef]
19. G. A. Sevison, S. Farzinazar, J. A. Burrow, C. Perez, H. Kwon, J. Lee, M. Asheghi, K. E. Goodson, A. Sarangan, J. R. Hendrickson, and I. Agha, “Phase change dynamics and two-dimensional 4-bit memory in Ge2Sb2 Te5 via telecom-band encoding,” ACS Photonics 7(2), 480–487 (2020). [CrossRef]
20. P. Guo, A. Sarangan, and I. Agha, “A review of germanium-antimony-telluride phase change materials for non-volatile memories and optical modulators,” Appl. Sci. 9(3), 530 (2019). [CrossRef]
21. T. Matsunaga and N. Yamada, “Structure, structural phase transitions, mechanical properties, defects, etc.-Structural investigation of GeSb2Te4: A high-speed phase-change material,” Phys. Rev. B: Condens. Matter Mater. Phys. 69(10), 104111 (2004). [CrossRef]
22. I. I. Petrov and R. M. Imamov. “ZG Pinsker Electron-diffraction determination of structures of Ge2Sb2 Te5 and Ge2Sb2 Te7” Soviet Phys. Cryst. Ussr 13.3, 339 (1968).
23. A. Patterson, “The Scherrer formula for X-ray particle size determination,” Phys. Rev. 56(10), 978–982 (1939). [CrossRef]
24. J. I. Langford and A. J. C. Wilson, “Scherrer after sixty years: A survey and some new results in the determination of crystallite size,” J. Appl. Crystallogr. 11(2), 102–113 (1978). [CrossRef]
25. W. Njoroge, H. Wöltgens, and M. Wuttig, “Density changes upon crystallization of Ge2Sb2.04Te4.74 films,” J. Vac. Sci. Technol., A 20(1), 230–233 (2002). [CrossRef]
26. J. Xu, C. Qi, L. Chen, L. Zheng, and Q. Xie, “The microstructural changes of Ge2Sb2 Te5 thin film during crystallization process,” AIP Adv. 8(5), 055006 (2018). [CrossRef]
27. J. Zheng, Z. Fang, and C. Wu, “Nonvolatile electrically reconfigurable integrated photonic switch enabled by a silicon PIN diode heater,” Adv. Mater. 32, 2001218 (2020). [CrossRef]
