Effect of sputter deposition parameters on the optical contrast and phase-change dynamics of Ge₂Sb₂Te₅ thin films for reconfigurable photonic devices

ABSTRACT

The chalcogenide phase-change material Ge₂Sb₂Te₅ (GST) is paramount for developing next-generation reconfigurable photonic devices, yet its performance is critically dependent on thin-film quality. This work presents a systematic investigation into the influence of radio-frequency (RF) magnetron sputtering parameters—specifically argon (Ar) working pressure and substrate temperature—on the structural, chemical, optical, and kinetic properties of GST thin films. Films were deposited on SiO₂/Si substrates under varying Ar pressures (0.4–3.0 Pa) and substrate temperatures (Room Temperature–160 °C). Comprehensive material characterization was performed using a suite of advanced analytical techniques. Results indicate that increasing Ar pressure leads to the formation of films with lower density and higher porosity, as confirmed by scanning electron microscopy analysis. This porous microstructure facilitates a lower crystallization temperature and faster crystallization dynamics, with switching times on the order of tens of picoseconds observed via time-resolved pump-probe reflectivity. However, this enhancement in switching speed is accompanied by a significant reduction in the optical contrast, a key figure of merit for photonic applications. Conversely, elevating the substrate temperature to an optimal 80 °C during deposition produces dense, pore-free amorphous films with a density of 6.16 g/cm³, only 1.5% lower than the crystalline phase. These films exhibit superior thermal stability and a maximized refractive index contrast (Δn) at telecommunication wavelengths. X-ray photoelectron spectroscopy confirmed that a protective capping layer is essential to prevent the rapid surface oxidation of Ge. These findings establish a clear process-structure-property-performance relationship, providing a crucial framework for tuning GST film properties to meet the divergent demands of high-speed, low-power optical memories versus high-performance, low-loss tunable photonic components.

Keywords:
Porosity Engineering; Refractive Index Modulation; Crystallization Kinetics; Surface Oxidation; Deposition Thermodynamics

1. INTRODUCTION

The field of photonics is undergoing a paradigm shift towards dynamic, tunable, and reconfigurable systems capable of manipulating light on-demand. In this context, phase-change materials (PCMs) have emerged as a uniquely compelling class of materials, enabling unprecedented control over the flow of light in integrated circuits [¹]. Among the various PCMs, the ternary chalcogenide compound Ge₂Sb₂Te₅ (GST) has been the subject of intense research and development, establishing itself as the archetypal material for this technological revolution. Its prominence stems from a remarkable combination of properties, including fast switching speeds (nanoseconds to picoseconds), a large contrast in optical and electrical properties between its phases, excellent non-volatility, and high endurance [²]. These attributes have already led to its successful commercialization in optical data storage media, such as rewritable digital versatile discs (DVDs), and position it as a leading candidate for next-generation non-volatile electronic memories (PC-RAM) and advanced photonic applications like tunable metasurfaces, all-optical switches, and neuromorphic computing hardware [³].

The functionality of GST-based devices is rooted in the rapid and reversible structural transformation between a disordered, covalently bonded amorphous phase and an ordered, resonant-bonded crystalline phase [⁴]. This transition can be triggered by precisely controlled thermal stimuli, typically delivered via nanosecond laser pulses or electrical Joule heating [⁵]. The two states are readily distinguishable by their physical properties: the amorphous phase is characterized by high electrical resistivity and low optical reflectivity, whereas the metastable face-centered cubic (fcc) crystalline phase exhibits low resistivity and high reflectivity [⁶,⁷]. This substantial contrast forms the basis for data readout. The atomistic mechanism underpinning the rapid switching is believed to involve subtle rearrangements of atoms without the need to break strong covalent bonds, as described by models like the “umbrella-flip” of Ge atoms from tetrahedral to octahedral coordination, which facilitates the ultrafast transition kinetics [⁸].

Despite its promise, the performance of GST is often limited by several intrinsic material challenges. High power consumption required for the amorphization (RESET) process, limited thermal stability of the amorphous phase which affects data retention at elevated temperatures, and the formation of voids and film delamination during repeated cycling are significant hurdles [⁹]. The latter is a direct consequence of the significant density change (~6–9%) that occurs during the phase transition, inducing mechanical stress that can lead to device failure [¹⁰]. To mitigate these issues, researchers have explored various optimization strategies, including the incorporation of dopants such as carbon, nitrogen, or tungsten to enhance thermal stability and reduce switching currents [⁹], and the engineering of film microstructure and stress through careful control of deposition conditions [¹¹].

Among various thin-film deposition techniques, magnetron sputtering is the most widely adopted method for GST in both research and industrial settings. It is favored for its ability to produce large-area, highly uniform, and smooth films with excellent stoichiometry control, which are prerequisites for reliable device fabrication [¹²]. Crucially, the parameters of the sputtering process—such as the working gas pressure, substrate temperature, and RF power—are not merely operational variables but powerful levers for tuning the fundamental properties of the deposited film. It is well-established that these parameters directly influence the kinetic energy of sputtered atoms and their surface mobility, thereby dictating the resulting film’s density, microstructure, internal stress, and texture [¹⁰]. These physical characteristics, in turn, govern the material’s ultimate optical performance and phase-change dynamics. Therefore, a comprehensive understanding of the relationship between deposition conditions and material properties is essential for the rational design of high-performance GST-based photonic devices. This work aims to provide such an understanding through a systematic and exhaustive investigation into the effects of two critical sputter deposition parameters: argon working pressure and substrate temperature. By employing a wide array of advanced characterization techniques, we establish clear, quantitative process-structure-property-performance relationships. We connect the deposition conditions to the film’s morphology, crystallography, chemical state, optical contrast, and ultrafast phase-change dynamics, thereby offering a detailed guide for optimizing GST thin films for the specific demands of next-generation reconfigurable photonics.

2. MATERIALS AND METHODS

A series of Ge₂Sb₂Te₅ (GST) thin films were deposited in an AJA International ATC-2200 radio-frequency (RF) magnetron sputtering system equipped with a load-lock chamber and confocal target geometry. The films were grown on 4-inch Si (100) wafers coated with a 300 nm thick layer of thermally grown SiO₂, which serves as an electrical insulator and provides a smooth, well-defined substrate for optical characterization. A 3-inch stoichiometric Ge₂Sb₂Te₅ compound target with 99.99% purity was used as the sputtering source [¹³]. Before each deposition, the chamber was evacuated to a base pressure below 2×10⁻⁴ Pa to minimize contamination from residual gases [¹⁴].

Two sets of experiments were conducted to systematically investigate the influence of key sputtering parameters, as detailed in Table 1. In the first set, the argon (Ar) working pressure was varied from 0.4 Pa to 3.0 Pa, while the substrate was held at room temperature (RT, ~20 °C) and the RF power was fixed at 50 W. This pressure series was used to identify a regime in which the films remain fully amorphous at RT but are as dense and smooth as possible. As summarized in Table 2, the film grown at 0.6 Pa exhibits a density of 5.98 g/cm³ and an RMS roughness of 1.2 nm, very close to the 0.4 Pa film (6.05 g/cm³, 0.8 nm) and substantially denser than the 3.0 Pa film (5.75 g/cm³, 2.1 nm), indicating that 0.6 Pa lies in a dense, low-porosity regime. In the second set, the substrate temperature was therefore varied from RT to 160 °C, while the Ar pressure and RF power were held constant at this intermediate value of 0.6 Pa and 50 W, respectively. Using a single, well-characterized pressure condition allows the influence of substrate temperature on structure, optical contrast and phase-change dynamics to be separated from the strong pressure-induced changes in density and porosity reported in both our data and earlier work by Dieker and Wuttig on sputtered GST [⁹]. For depositions at elevated temperatures, substrates were heated for one hour prior to sputtering to ensure thermal stability and uniformity. A target film thickness of approximately 100 nm was maintained for all samples, controlled in situ by a quartz crystal microbalance and subsequently verified ex situ using stylus profilometry. For a subset of samples used in oxidation studies, a 10 nm thick SiO₂ capping layer was deposited in situ immediately following GST deposition without breaking vacuum to prevent atmospheric exposure [¹⁵]. The annealing process has been carried out at 500 °C. To quantify the structural and physical parameters, a suite of complementary characterization techniques was employed. Grazing-incidence X-ray diffraction (GIXRD, Cu Kα radiation) was used to identify the phase of the as-deposited films and to determine the crystallite size after annealing [¹⁶]. The average crystallite size was extracted from the full width at half maximum of the fcc (111) diffraction peak using the Scherrer equation with a shape factor of 0.9, after correction for instrumental broadening, following established procedures for Ge₂Sb₂Te₅ thin films [¹⁷]. The density of the amorphous films was obtained from X-ray reflectivity (XRR) measurements by fitting the specular reflectivity curves with a multilayer optical model based on Parratt’s formalism. The resulting densities are consistent with values reported for dense, pore-free amorphous Ge₂Sb₂Te₅ prepared by sputtering, where the amorphous density is only about 1.5% lower than that of the crystalline NaCl-type phase [¹⁰]. Film thicknesses used in the XRR fits were cross-checked against stylus profilometry and the quartz-crystal microbalance calibration. Surface roughness was characterized by atomic force microscopy operated in tapping mode. For each sample, height images were acquired over 1 µm × 1 µm scan areas at several locations, and the root-mean-square (RMS) roughness was calculated from the height distribution and averaged to obtain the values reported in Table 2. This AFM-based evaluation of RMS roughness follows common practice in the characterization of GST thin films [¹⁸]. The crystallization temperature Tc of the amorphous films was determined by differential scanning calorimetry (DSC). Approximately 5–10 mg of GST-coated SiO₂/Si pieces were heated from room temperature to 350 °C at a constant rate of 10 K/min under nitrogen. The onset temperature of the exothermic crystallization peak, obtained by the intersection of tangents to the baseline and the leading edge of the peak, was taken as Tc. This method for extracting Tc is consistent with that used in previous studies of Ge₂Sb₂Te₅ thin films [²].

Table 1
Sputter deposition parameters for Ge₂Sb₂Te₅ thin films.

PARAMETER	SERIES 1 (PRESSURE VARIATION)	SERIES 2 (TEMPERATURE VARIATION)
Target	Ge2Sb2Te5 (99.99%)	Ge2Sb2Te5 (99.99%)
Substrate	Si (100) with 300 nm SiO2	Si (100) with 300 nm SiO2
RF Power	50 W	50 W
Ar Pressure	0.4, 0.6, 1.0, 3.0 Pa	0.6 Pa
Substrate Temperature	Room Temperature (~20 °C)	RT, 80 °C, 120 °C, 160 °C
Target Thickness	100 nm	100 nm
Base Pressure	<2 × 10−4 Pa	<2 × 10−4 Pa

Table 2
Summary of structural and physical properties of GST films deposited under various conditions.

DEPOSITION CONDITION	PHASE (AS-DEPOSITED)	CRYSTALLITE SIZE (nm) (ANNEALED)	FILM DENSITY (g/cm3) (AMORPHOUS)	SURFACE ROUGHNESS (nm, RMS)	Tc (°C)
0.4 Pa, RT	Amorphous	18.5	6.05	0.8	158
1.0 Pa, RT	Amorphous	15.2	5.91	1.4	151
3.0 Pa, RT	Amorphous	12.8	5.75	2.1	142
0.6 Pa, RT	Amorphous	16.1	5.98	1.2	155
0.6 Pa, 80 °C	fcc	16.5	6.16	0.7	171
0.6 Pa, 120 °C	fcc	22.4	6.25	1.8	–
0.6 Pa, 160 °C	fcc	22.6	6.27	1.7	–

Time-resolved pump–probe reflectivity measurements were performed in reflection geometry using a Ti:sapphire femtosecond laser system (central wavelength 800 nm, pulse duration ~150 fs, repetition rate 1 kHz). The laser output was split into pump and probe arms. The pump beam (800 nm) was intensity-modulated by a mechanical chopper and focused onto the film, while the probe beam was frequency-doubled to 400 nm (β-BaB₂O₄ crystal) to minimize pump–probe spectral crosstalk. The probe spot was centered within the pump spot to ensure that the probed region experienced a quasi-uniform excitation. Pump fluence was calibrated from the pulse energy and the measured beam spot size at the sample plane, and was varied around the SET/RESET thresholds reported in Section 3.4; the probe fluence was kept at least one order of magnitude lower to avoid inadvertently inducing a phase transition. The reflected probe was detected with a fast photodiode and demodulated using a lock-in amplifier referenced to the chopper frequency. The normalized transient reflectivity was calculated as ΔR/R₀ = (R_pumped − R_unpumped)/R_unpumped. The pump–probe delay was controlled with a motorized optical delay line. The instrument response function, estimated from the pump–probe cross-correlation at the sample plane, was ~200 fs, well below the characteristic tens-of-picoseconds crystallization dynamics observed in this work. Delay steps of 0.2 ps were used in the first 50 ps to capture the rise dynamics, followed by 2–5 ps steps out to 2 ns for the slower relaxation. Each data point represents an average over multiple laser shots, and each transient was reproduced in independent runs; the uncertainties in extracted time constants were estimated from the spread across repeated measurements, following established GST pump–probe protocols.

3. RESULTS AND DISCUSSION

3.1. Influence of sputtering parameters on film structure and morphology

The structural evolution and morphology of the GST films are fundamentally dictated by the kinetic energy and flux of particles arriving at the substrate, which are directly controlled by the Ar working pressure and substrate temperature.

The GIXRD patterns of films deposited at room temperature under various Ar pressures are presented in Figure 1(A). All as-deposited films are predominantly X-ray amorphous, exhibiting a broad diffraction halo centered at approximately 2θ ≈ 28°, which indicates the absence of well-developed long-range crystalline order [¹⁰]. For the samples deposited at 0.6 Pa and 1.0 Pa, however, very weak features can be discerned at 2θ ≈ 30° and 42°. These low-intensity shoulders are close to the background level but can be indexed to the (200) and (220) reflections of the metastable fcc GST phase, in line with literature reports where the fcc (200) and (220) peaks appear near 30° and 42° after partial crystallization [¹³]. We therefore interpret the 0.6 Pa and 1.0 Pa films as largely amorphous films containing a small volume fraction of nanocrystalline fcc nuclei, while the 0.4 Pa and 3.0 Pa films remain fully amorphous within the detection limit [¹⁹]. This behavior is consistent with previous studies on sputtered GST, which show that subtle variations in deposition conditions can promote incipient crystallization or intermediate-range order even at nominally ‘amorphous’ growth temperatures [¹⁹].

Figure 1
(A) GIXRD patterns of GST films deposited at room temperature under various Ar pressures. (B) GIXRD patterns of GST films deposited under various temperature.

The effect of substrate temperature on the as-deposited film structure is shown in Figure 1(B). Films deposited at RT are largely amorphous and show only the broad halo around 2θ ≈ 28°, consistent with the GIXRD data in Figure 1(A). At a deposition temperature of 80 and 120 °C, distinct diffraction peaks emerge, signaling in-situ crystallization during growth. The peaks at 2θ values of approximately 28.5°, 29.5°, and 42.2° are indexed to the (111), (200), and (220) planes of the metastable face-centered cubic (fcc) phase of GST, respectively, in good agreement with previous XRD studies of crystalline GST films [²⁰]. As the temperature is further increased to 160 °C, additional peaks corresponding to the stable hexagonal close-packed (hcp) phase begin to appear, indicating the onset of the second structural transformation [²¹]. A summary of key structural properties derived from XRD and other techniques is provided in Table 2. It can be seen that at RT the 0.6 Pa film has a density and roughness very similar to the 0.4 Pa film, while the 3.0 Pa film is noticeably less dense and rougher, consistent with the established decrease of GST film density with increasing sputtering pressure [¹²]. This observation motivated the use of 0.6 Pa as the reference pressure for the temperature-variation series in Section 3.1 and 3.4, ensuring that the temperature dependence of crystallization behavior and optical contrast is evaluated under a deposition condition representative of dense, low-porosity GST.

The influence of deposition parameters on film morphology is directly visualized by electron microscopy. Figure 2 shows cross-sectional TEM images comparing films deposited at low (0.4 Pa) and high (3.0 Pa) Ar pressure. At low pressure, atoms arrive at the substrate with higher kinetic energy, leading to a dense, featureless film. Conversely, at high pressure, sputtered atoms undergo more collisions in the plasma, lose energy, and arrive with lower mobility, resulting in a pronounced columnar microstructure with significant inter-columnar porosity [²²]. This porous structure is a direct consequence of atomic self-shadowing effects.

Figure 2
Cross-sectional TEM images comparing the morphology of films deposited at low (0.4 Pa) and high (3.0 Pa) Ar pressure.

Figure 3 demonstrates the profound effect of substrate temperature. The film deposited at RT exhibits the aforementioned columnar structure, whereas the film grown at 80 °C is fully dense and homogeneous, with no evidence of pores or voids [²³]. This result is critically important. By providing moderate thermal energy during growth, adatom surface mobility is enhanced sufficiently to overcome shadowing effects and fill in potential voids, leading to a dense amorphous structure. The density of this film, measured to be 6.16 g/cm³, is only 1.5% lower than that of the crystalline fcc phase (6.25 g/cm³). This is a substantial improvement over typical porous amorphous films, which can have a density mismatch of up to 9%. Reducing this density change is paramount for mitigating mechanical stress during phase cycling, thereby enhancing the endurance and reliability of photonic devices [²⁴]. Figure 4 compares the plan-view morphology of films deposited at 0.4 Pa (RT), 3.0 Pa (RT), 0.6 Pa (80 °C), and 0.6 Pa (120 °C). The 0.6 Pa, 120 °C specimen, which crystallizes directly into the metastable fcc phase, exhibits a dense nanocrystalline microstructure with well-connected grains and negligible intergranular porosity. This morphology is intermediate between the highly porous, columnar structure of the 3.0 Pa sample and the dense amorphous film grown at 80 °C, and it confirms that the in situ crystallization inferred from GIXRD is accompanied by the formation of a continuous fcc grain network. This trend is consistent with previous studies on sputtered GST films, where increasing substrate temperature and optimised Ar pressure promote dense fcc grains and suppress void formation.

Figure 3
TEM images comparing the morphology of films deposited at RT and 80 °C.

Figure 4
Plan-view SEM images of Ge₂Sb₂Te₅ films illustrating the effect of sputtering pressure and substrate temperature on the crystalline morphology. (A) Film deposited at 0.4 Pa and room temperature (RT) before and after annealing. (B) Film deposited at 3.0 Pa and RT before and after annealing. (C) Film deposited at 0.6 Pa and 80 °C before and after annealing. (D) Film deposited at 0.6 Pa and 120 °C, showing the dense nanocrystalline morphology of the fcc phase after annealing. All images are acquired at the same magnification to allow direct comparison of grain size and porosity.

To probe the structure at the atomic level for the same representative film used in Figure 4, HRTEM and SAED were employed on the GST layer deposited at 0.6 Pa Ar pressure and 80 °C substrate temperature. Figure 5 presents a composite of images and diffraction patterns characterizing the different structural states of this film. The as-deposited sample (Figure 5(A)) shows a featureless HRTEM image and a diffuse halo in the SAED pattern, confirming its amorphous nature. After annealing at 500 °C, well-defined crystalline grains are observed. The SAED patterns in Figure 5(B), taken from individual grains, can be indexed to the fcc and hcp crystal structures, respectively, providing unambiguous nanoscale phase identification that complements the bulk XRD data and is in line with earlier TEM studies of GST crystallization [²⁵].

Figure 5
High-resolution TEM images and selected-area electron diffraction (SAED) patterns of a Ge₂Sb₂Te₅ thin film deposited at 0.6 Pa Ar pressure and 80 °C substrate temperature. (A) As-deposited amorphous film, showing a featureless HRTEM contrast and a diffuse halo in SAED. (B) The same film after annealing at 500 °C, displaying crystalline grains whose SAED patterns can be indexed to the fcc and hcp GST phases, respectively.

3.2. Compositional and chemical integrity

Energy-dispersive X-ray spectroscopy (EDX) was employed to verify the compositional fidelity of the sputtered GST films, because even slight deviations from the Ge₂Sb₂Te₅ stoichiometry can measurably alter optical constants and phase-change kinetics. The quantitative EDX analysis yields an atomic composition of Ge: 22.1 at.%, Sb: 22.7 at.%, and Te: 55.2 at.%, which closely matches the nominal Ge₂Sb₂Te₅ target ratio. In addition to the average composition, spatial uniformity is critical for reconfigurable photonic devices, since micron-scale segregation can introduce local refractive-index fluctuations and non-uniform switching behavior. As shown in the EDX elemental maps (Figure 6), Ge, Sb, and Te signals are evenly distributed across the mapped region, with no observable Te-rich/Sb-rich clustering or phase-separated domains at the micron scale. This homogeneous elemental distribution indicates that the sputtering conditions used here provide robust stoichiometry transfer and good compositional uniformity, supporting the reliability and reproducibility of the optical contrast and phase-change dynamics discussed in the subsequent sections.

Figure 6
EDX compositional analysis and elemental mapping of a representative Ge₂Sb₂Te₅ thin film. (A) SEM image indicating the mapped area. (B–D) Corresponding EDX elemental maps of Ge, Sb, and Te, respectively. (E) Overlay map showing the co-localized distribution of all three elements.

While the bulk composition is correct, the chemical state of the surface is critical for device performance. XPS was used to analyze the chemical bonding environment. Figure 7 displays the high-resolution core-level spectra of Ge 2p, Sb 3d, and Te 3d for a pristine film protected by an in-situ deposited SiO₂ cap. The spectra can be deconvoluted into primary components corresponding to the desired chemical bonds within the chalcogenide matrix. The Ge 2p peak is centered at a binding energy of ~1252 eV and 1220 eV, the Sb 3d₅/₂ peak at ~531.0 eV, and the Te 3d₅/₂ peak at ~572.5 eV, which are characteristic of Ge-Te, Sb-Te, and Te-(Ge,Sb) bonds, respectively [²⁶].

Figure 7
High-resolution XPS core-level spectra of (A) Ge 2p, (B) Sb 3d, and (C) Te 3d for a pristine film protected by a SiO₂ cap.

The high reactivity of GST, particularly with atmospheric oxygen, is a significant concern for practical applications. Figure 8 directly compares the Te 3d spectrum of a capped film with that of an uncapped film exposed to ambient air for 24 hours. The uncapped sample exhibits a prominent additional peak at a higher binding energy of ~587.1 eV. This peak is attributed to the formation of TeO₂, indicating the preferential oxidation of Te atoms at the surface [²⁷]. To confirm that this oxidation is a surface-limited phenomenon, XPS depth profiling was performed on an aged, uncapped sample. As shown in Figure 9, the intensity of the O 1s signal and the oxide-related components in the Ge 3d and Sb 3d spectra diminish rapidly after just a few nanometers of Ar⁺ ion sputtering, revealing the pristine GST composition underneath [²⁸]. This analysis underscores a critical point for device engineering: the GST surface is highly susceptible to oxidation, forming a stable, insulating oxide layer that can degrade optical properties and inhibit electrical contact. Therefore, the use of a suitable capping layer, such as SiO₂ or Si₃N₄, is not merely beneficial but essential for fabricating stable and reliable reconfigurable photonic devices [²⁹].

Figure 8
A direct comparison of the Te 3d spectrum for a capped film versus an uncapped film after 24 hours of air exposure.

Figure 9
XPS depth profiling of an aged, uncapped sample, showing that the oxide layer is a surface-limited phenomenon.

The local atomic structure and bonding arrangement within the amorphous network were investigated using FTIR spectroscopy. Figure 10 compares the far-infrared transmission spectra of an amorphous film and a crystalline (fcc) film. The amorphous spectrum is characterized by broad absorption bands, which is typical for disordered materials where a wide distribution of bond lengths and angles exists. In contrast, the crystalline spectrum shows sharper, more defined vibrational modes due to the long-range periodic order of the lattice [³⁰]. The broad feature in the amorphous phase centered around 160 cm⁻¹ can be assigned to vibrational modes of structural units like GeTe₄ tetrahedra and SbTe₃ pyramids, which are considered the primary building blocks of the amorphous network [³¹]. Upon crystallization, these features evolve into sharper peaks corresponding to the phonon modes of the fcc lattice.

Figure 10
FTIR spectra of amorphous GST film and a crystalline GST film.

3.3. Optical contrast and refractive index tuning

The performance of a reconfigurable photonic device is fundamentally determined by the change in its optical properties upon phase transition. Figure 11 shows the Vis-NIR transmittance and reflectance spectra for a representative GST film in its as-deposited amorphous and annealed crystalline states. A significant change is evident across the entire spectrum: upon crystallization, the transmittance decreases while the reflectance increases, signifying a transition to a more optically absorbing and reflective state [³²].

Figure 11
(A) Vis-NIR transmittance and (B) reflectance spectra for a GST film in its as-deposited amorphous and annealed crystalline states.

The absorption coefficient derived from these spectra was used to generate T_auc plots, as shown in Figure 12(A), to estimate the optical bandgap (E_g). The amorphous phase exhibits a wider bandgap of approximately 0.72 eV, which decreases to about 0.66 eV in the crystalline phase. This bandgap narrowing is consistent with the formation of resonant bonds in the crystalline state, which leads to delocalization of electronic states and a reduction in the energy required for electronic excitation. For a more precise quantification of the optical properties, spectroscopic ellipsometry was employed to determine the complex refractive index, n + ik. Figure 12(B) presents the dispersion of the refractive index (n) and extinction coefficient (k) for films deposited under low (0.4 Pa) and high (3.0 Pa) Ar pressure, measured in both amorphous and crystalline states. In all cases, both n and k increase significantly upon crystallization, which is the origin of the large optical contrast.

Figure 12
(A) T_auc plots derived from absorption spectra, used to estimate the optical bandgap for the amorphous and crystalline phases. (B) The dispersion of the refractive index (n) and extinction coefficient (k) for films deposited under low (0.4 Pa) and high (3.0 Pa) Ar pressure, in both amorphous and crystalline states.

The key figure of merit for most phase-shifter-based photonic devices is the refractive index contrast, Δn = ncryst − namorph. Figure 13 plots this contrast at the important telecommunication wavelength of 1550 nm as a function of the deposition parameters. The trend is unequivocal: films that are physically denser exhibit a higher optical contrast. The highest Δn of 2.65 is achieved for the film deposited at 80 °C, which was shown to be dense and pore-free. In contrast, the porous film deposited at 3.0 Pa shows a significantly lower Δn of 1.9.

Figure 13
The refractive index contrast (Δn) at 1550 nm as a function of deposition parameters, showing that denser films exhibit higher optical contrast.

This result establishes a critical trade-off in GST film design. The lower density and porosity of films grown at high Ar pressure, which arise from the inclusion of voids (with a refractive index of n = 1), effectively reduce the refractive index of both the amorphous and crystalline states according to effective medium theory. This inherently diminishes the achievable contrast between the two states [¹⁹]. Therefore, while high-pressure deposition may offer advantages in switching kinetics, it comes at the cost of reduced optical modulation depth. For high-performance, low-loss reconfigurable components, maximizing the intrinsic material density is paramount.

3.4. Ultrafast phase-change dynamics

The ultimate performance metric for many GST applications is the speed at which the phase transition can be induced. Time-resolved pump–probe reflectivity measurements were used to quantify the crystallization (SET) and amorphization (RESET) dynamics on ultrafast timescales. To improve reproducibility, the laser parameters, temporal resolution, and delay scanning strategy are specified in Section 2, and the time-constant extraction procedure is detailed below.

Figure 14(A) displays the transient reflectivity changes (ΔR/R₀) corresponding to the laser-induced SET process. The traces compare a low-density film (deposited at 3.0 Pa) with a high-density film (deposited at 0.4 Pa). After excitation with a pump pulse above the crystallization threshold, the reflectivity increases as the material transforms from the low-reflectivity amorphous state to the high-reflectivity crystalline state. The characteristic crystallization time was extracted by fitting the rising portion of each transient to a single-exponential growth model, ΔR/R₀(t) = A[1 − exp(−(t − t₀)/τ_SET)] + C for t > t₀, where t₀ is time-zero and τ_SET is the time constant. Fitting was performed after baseline correction using the pre-pump region (t < 0), using nonlinear least-squares minimization. Using this procedure, the low-density porous film exhibits faster crystallization kinetics (τ_SET ≈ 30 ps) than the dense film (τ_SET ≈ 82 ps). Consistent with the faster kinetics, the porous film crystallizes at a lower threshold fluence (5.86 mJ/cm²) compared to the dense film (7.04 mJ/cm²), highlighting the role of reduced thermal mass and enhanced heat-flow pathways in porous microstructures.

Figure 14
(A) Transient reflectivity changes showing the laser-induced crystallization dynamics for a low-density (3.0 Pa) and a high-density (0.4 Pa) film. (B) A typical reflectivity trace illustrating the sub-nanosecond RESET process induced by a high-fluence pump pulse.

Figure 14(B) shows a representative transient for the RESET process. A higher-fluence pump pulse rapidly heats the crystalline material above its melting point; the subsequent rapid quenching freezes the disordered liquid-like state, producing a sharp drop in reflectivity to a stable value characteristic of the amorphous phase [⁴]. The RESET timescale was quantified by the 10–90% fall time of the reflectivity drop within our delay window, consistent with prior ultrafast optical studies that resolve melt–quench-driven amorphization and relaxation in GST-based resonantly bonded materials.

The key optical and switching performance metrics for films prepared under different conditions are summarized in Table 3. This table encapsulates the central trade-off uncovered in this study. Deposition at high Ar pressure yields films that are faster and more energy-efficient to switch, which is desirable for applications like high-speed memory. However, this comes at the expense of a lower optical contrast (Δn) and higher optical loss (k), which are detrimental for high-performance photonic components. Conversely, deposition at low pressure or an optimized elevated temperature (80 °C) produces dense films with superior optical contrast and lower loss, but which require more energy and time to crystallize. This comprehensive analysis provides a clear design space for GST films. The choice of sputtering parameters is not arbitrary but a deliberate engineering decision to navigate the trade-off between switching kinetics and optical performance, allowing for the tailoring of GST films to the specific requirements of the target photonic application.

Table 3
Comparison of optical and switching performance metrics for GST films.

DEPOSITION CONDITION	Δn @ 1550 nm	Δk @ 1550 nm	CRYSTALLIZATION TIME (ps)	THRESHOLD FLUENCE (mJ/cm2)
0.4 Pa, RT	2.48	1.85	82	7.04
3.0 Pa, RT	1.90	1.42	30	5.86
0.6 Pa, 80 °C	2.65	2.01	95	7.51

4. CONCLUSION

This study has conducted a systematic and comprehensive investigation into the effects of RF magnetron sputtering parameters—argon working pressure and substrate temperature—on the multifaceted properties of Ge₂Sb₂Te₅ thin films. A clear and direct process-structure-property-performance relationship has been established, providing crucial insights for the design of GST-based reconfigurable photonic devices. The key findings can be summarized as follows. First, the Ar working pressure during sputtering profoundly influences the film’s microstructure and density. Increasing Ar pressure from 0.4 Pa to 3.0 Pa results in progressively more porous, lower-density films with a distinct columnar morphology. Deposition at an optimized, elevated substrate temperature of 80 °C was found to be highly effective in eliminating porosity, yielding fully dense amorphous films with a smooth surface and a density only 1.5% lower than the crystalline phase. These dense films exhibit enhanced thermal stability and a maximized optical contrast, with a refractive index change (Δn) of 2.65 at 1550 nm. Third, XPS analysis highlighted the high surface reactivity of GST, particularly the preferential oxidation of Ge, underscoring the necessity of a protective capping layer for any practical device implementation. These results translate into direct, actionable recommendations for device engineering. For applications where switching speed and low power consumption are the primary concerns, such as in high-density optical memories, films deposited at higher Ar pressures offer a distinct advantage. Conversely, for high-performance, low-loss reconfigurable photonic components such as tunable filters, phase shifters, and metasurfaces, where maximizing the optical modulation depth and minimizing scattering loss is critical, films should be deposited under low Ar pressure and at an optimized temperature of ~80 °C to achieve maximum density and optical contrast.

5. ACKNOWLEDGMENTS

This work has been supported by National Natural Science Foundation of China (62205263).

6. BIBLIOGRAPHY

• ^[1] DU, J., ZHOU, J., ZHANG, L., et al, “Investigation of the crystallization characteristics of intermediate states in Ge2Sb2Te5 thin films induced by nanosecond multi-pulsed laser irradiation”, Nanomaterials, v. 12, n. 3, pp. 536, Feb. 2022. doi: http://doi.org/10.3390/nano12030536. PubMed PMID: 35159881.
  » https://doi.org/10.3390/nano12030536
• ^[2] YAKUBOV, A., SHERCHENKOV, A., LAZARENKO, P., et al, “Contact resistance measurements for the Ge2Sb2Te5 thin films”, Chalcogenide Letters, v. 17, n. 1, pp. 1–8, 2020. doi: http://doi.org/10.15251/CL.2020.171.1.
  » https://doi.org/10.15251/CL.2020.171.1
• ^[3] SHERCHENKOV, A., KOZYUKHIN, S., BABICH, A., et al, “Crystallization mechanism and kinetic parameters in Ge2Sb2Te5 thin films for the phase change memory application”, Chalcogenide Letters, v. 15, n. 1, pp. 45–54, 2018.
• ^[4] SUN, X., EHRHARDT, M., LOTNYK, A., et al, “Crystallization of Ge2Sb2Te5 thin films by nano- and femtosecond single laser pulse irradiation”, Scientific Reports, v. 6, n. 1, pp. 28246, Jun. 2016. doi: http://doi.org/10.1038/srep28246. PubMed PMID: 27292819.
  » https://doi.org/10.1038/srep28246
• ^[5] FARIA, D.L., LOPES, T.A., MENDES, L.M., et al, “Valorization of wood shavings waste for the production of wood particulate composites”, Matéria (Rio de Janeiro), v. 25, n. 3, pp. e1, 2020. doi: http://doi.org/10.1590/s1517-707620200003.1080.
  » https://doi.org/10.1590/s1517-707620200003.1080
• ^[6] ALI, G., “Extraction of acetaldehyde from mineral water using 2,4-dinitrophenylhydrazine as derivatizing agent”, Matéria (Rio de Janeiro), v. 25, n. 3, pp. e2, 2020. doi: http://doi.org/10.1590/s1517-707620200003.1083.
  » https://doi.org/10.1590/s1517-707620200003.1083
• ^[7] XU, Y., WAN, L., ZARE, N., et al, “Cr2C MXene modification of an electrochemical platform allows for highly selective and sensitive detection of PSMA, a prostate cancer biomarker”, Journal of Nanostructure in Chemistry, v. 15, n. 5, pp. 152517, Oct. 2025.
• ^[8] ARYANA, K., KIM, H.J., ISLAM, M.R., et al, “Optical and thermal properties of Ge2Sb2Te5, Sb2Se3, and Sb2S3 for reconfigurable photonic devices”, Optical Materials Express, v. 13, n. 11, pp. 3277–3286, 2023. doi: http://doi.org/10.1364/OME.503178.
  » https://doi.org/10.1364/OME.503178
• ^[9] BALA, N., KHAN, B., SINGH, K., et al, “Recent advances in doped Ge2Sb2Te5 thin film based phase change memories”, Materials Advances, v. 4, n. 3, pp. 747–768, Feb. 2023. doi: http://doi.org/10.1039/D2MA01047J.
  » https://doi.org/10.1039/D2MA01047J
• ^[10] ZHANG, Q., LIAN, C., XU, Q., et al, “Density of amorphous sputtered Ge2Sb2Te5 thin films”, AIP Advances, v. 13, n. 1, pp. 015014, Jan. 2023. doi: http://doi.org/10.1063/5.0133477.
  » https://doi.org/10.1063/5.0133477
• ^[11] SIMPSON, R.E., KRBAL, M., FONS, P., et al, “Toward the ultimate limit of phase change in Ge2Sb2Te5”, Nano Letters, v. 10, n. 2, pp. 414–419, Feb. 2010. doi: http://doi.org/10.1021/nl902777z. PubMed PMID: 20041706.
  » https://doi.org/10.1021/nl902777z
• ^[12] DIEKER, H., WUTTIG, M., “Influence of deposition parameters on the properties of sputtered Ge2Sb2Te5 films”, Thin Solid Films, v. 478, n. 1, pp. 248–251, May. 2005. doi: http://doi.org/10.1016/j.tsf.2004.08.019.
  » https://doi.org/10.1016/j.tsf.2004.08.019
• ^[13] BERTELLI, M., DÍAZ FATTORINI, A., DE SIMONE, S., et al, “Structural and electrical properties of annealed Ge2Sb2Te5 films grown on flexible polyimide”, Nanomaterials, v. 12, n. 12, pp. 2001, Jun. 2022. doi: http://doi.org/10.3390/nano12122001. PubMed PMID: 35745340.
  » https://doi.org/10.3390/nano12122001
• ^[14] XU, J., QI, C., CHEN, L., et al, “The microstructural changes of Ge2Sb2Te5 thin film during crystallization process”, AIP Advances, v. 8, n. 5, pp. 055006, May. 2018. doi: http://doi.org/10.1063/1.5025204.
  » https://doi.org/10.1063/1.5025204
• ^[15] SANTALA, M.K., REED, B.W., TOPURIA, T., et al, “Nanosecond in situ transmission electron microscope studies of the reversible Ge2Sb2Te5 crystalline ⇔ amorphous phase transformation”, Journal of Applied Physics, v. 111, n. 2, pp. 024309, Jan. 2012. doi: http://doi.org/10.1063/1.3678447.
  » https://doi.org/10.1063/1.3678447
• ^[16] BANO, T., SAJJAD, S., LEGHARI, S.A.K., et al, “Dimensionality-driven CQD/TiO2-g-C3N4 ternary composites for boosting photocatalytic and PEC performance”, Journal of Nanostructure in Chemistry, v. 15, n. 5, pp. 152521, Oct. 2025.
• ^[17] SATOH, H., SUGAWARA, K., TANAKA, K., “Nanoscale phase changes in crystalline Ge2Sb2Te5 films using scanning probe microscopes”, Journal of Applied Physics, v. 99, n. 2, pp. 024306, Jan. 2006. doi: http://doi.org/10.1063/1.2163010.
  » https://doi.org/10.1063/1.2163010
• ^[18] KOZYUKHIN, S., VOROBYOV, Y., SHERCHENKOV, A., et al, “Isothermal crystallization of Ge2Sb2Te5 amorphous thin films and estimation of information reliability of PCM cells”, Physica Status Solidi. A, Applications and Materials Science, v. 213, n. 7, pp. 1831–1838, Jul. 2016. doi: http://doi.org/10.1002/pssa.201532930.
  » https://doi.org/10.1002/pssa.201532930
• ^[19] SIMANDAN, I.-D., SAVA, F., BURUIANA, A.-T., et al, “Influence of deposition method on the structural and optical properties of Ge2Sb2Te5”, Materials, v. 14, n. 13, pp. 3663, Jan. 2021. doi: http://doi.org/10.3390/ma14133663. PubMed PMID: 34209141.
  » https://doi.org/10.3390/ma14133663
• ^[20] WU, H., HAN, W., ZHANG, X., “Ultrafast dynamics of different phase states Ge2Sb2Te5 film induced by a femtosecond laser pulse irradiation”, Materials, v. 15, n. 19, pp. 6760, Jan. 2022. doi: http://doi.org/10.3390/ma15196760. PubMed PMID: 36234103.
  » https://doi.org/10.3390/ma15196760
• ^[21] ADNANE, L., DIRISAGLIK, F., CYWAR, A., et al, “High temperature electrical resistivity and Seebeck coefficient of Ge2Sb2Te5 thin films”, Journal of Applied Physics, v. 122, n. 12, pp. 125104, Sep. 2017. doi: http://doi.org/10.1063/1.4996218.
  » https://doi.org/10.1063/1.4996218
• ^[22] SONG, S.A., ZHANG, W., SIK JEONG, H., et al, “In situ dynamic HR-TEM and EELS study on phase transitions of Ge2Sb2Te5 chalcogenides”, Ultramicroscopy, v. 108, n. 11, pp. 1408–1419, Oct. 2008. doi: http://doi.org/10.1016/j.ultramic.2008.05.012. PubMed PMID: 18657907.
  » https://doi.org/10.1016/j.ultramic.2008.05.012
• ^[23] D’ARRIGO, G., MIO, A., FAVARO, G., et al, “Mechanical properties of amorphous Ge2Sb2Te5 thin layers”, Surface and Coatings Technology, v. 355, pp. 227–233, Dec. 2018. doi: http://doi.org/10.1016/j.surfcoat.2018.02.050.
  » https://doi.org/10.1016/j.surfcoat.2018.02.050
• ^[24] KUMAR, M., VORA-UD, A., SEETAWAN, T., et al, “Study of pulsed-DC sputtering induced Ge2Sb2Te5 thin films using facile thermoelectric measurement”, Materials & Design, v. 98, pp. 254–261, May. 2016. doi: http://doi.org/10.1016/j.matdes.2016.03.046.
  » https://doi.org/10.1016/j.matdes.2016.03.046
• ^[25] PARK, Y.J., LEE, J.Y., KIM, Y.T., “In situ transmission electron microscopy study of the nucleation and grain growth of Ge2Sb2Te5 thin films”, Applied Surface Science, v. 252, n. 23, pp. 8102–8106, Sep. 2006. doi: http://doi.org/10.1016/j.apsusc.2005.10.026.
  » https://doi.org/10.1016/j.apsusc.2005.10.026
• ^[26] WANG, G., WANG, H., LOTNYK, A., et al, “Unique interface-driven crystallization mechanism and element-resolved structure imaging of ZnO-Ge2Sb2Te5 nanocomposites”, Ceramics International, v. 44, n. 18, pp. 22497–22503, Dec. 2018. doi: http://doi.org/10.1016/j.ceramint.2018.09.020.
  » https://doi.org/10.1016/j.ceramint.2018.09.020
• ^[27] MIAO, L., CHEN, L., “Evidence for Ge2Sb2Te5 phase formation upon crystallization of Ge-doped Sb2Te3 phase change films”, Vacuum, v. 227, pp. 113411, Sep. 2024. doi: http://doi.org/10.1016/j.vacuum.2024.113411.
  » https://doi.org/10.1016/j.vacuum.2024.113411
• ^[28] GOURVEST, E., PELISSIER, B., VALLÉE, C., et al, “Impact of Oxidation on Ge2Sb2Te5 and GeTe Phase-Change Properties”, Journal of the Electrochemical Society, v. 159, n. 4, pp. H373–H377, Jan. 2012. doi: http://doi.org/10.1149/2.027204jes.
  » https://doi.org/10.1149/2.027204jes
• ^[29] YIN, Q., CHEN, L., “Microstructure and crystallization kinetics of Ge2Sb2Te5–Sn phase change materials”, Journal of Materials Science Materials in Electronics, v. 29, n. 19, pp. 16523–16533, Oct. 2018. doi: http://doi.org/10.1007/s10854-018-9746-0.
  » https://doi.org/10.1007/s10854-018-9746-0
• ^[30] DORY, J.-B., CASTRO-CHAVARRIA, C., VERDY, A., et al, “Ge–Sb–S–Se–Te amorphous chalcogenide thin films towards on-chip nonlinear photonic devices”, Scientific Reports, v. 10, n. 1, pp. 11894, Jul. 2020. doi: http://doi.org/10.1038/s41598-020-67377-9. PubMed PMID: 32681142.
  » https://doi.org/10.1038/s41598-020-67377-9
• ^[31] NĚMEC, P., NAZABAL, V., MOREAC, A., et al, “Amorphous and crystallized Ge–Sb–Te thin films deposited by pulsed laser: Local structure using Raman scattering spectroscopy”, Materials Chemistry and Physics, v. 136, n. 2, pp. 935–941, Oct. 2012. doi: http://doi.org/10.1016/j.matchemphys.2012.08.024.
  » https://doi.org/10.1016/j.matchemphys.2012.08.024
• ^[32] XU, Z., CHEN, C., WANG, Z., et al, “Optical constants acquisition and phase change properties of Ge2Sb2Te5 thin films based on spectroscopy”, RSC Advances, v. 8, n. 37, pp. 21040–21046, Jun. 2018. doi: http://doi.org/10.1039/C8RA01382A. PubMed PMID: 35542376.
  » https://doi.org/10.1039/C8RA01382A

Publication Dates

• Publication in this collection
  10 Apr 2026
• Date of issue
  2026

History

• Received
  19 July 2025
• Accepted
  11 Feb 2026