Endurance of chalcogenide optical phase change materials: a review

Louis Martin-Monier; Cosmin Constantin Popescu; Luigi Ranno; Brian Mills; Sarah Geiger; Dennis Callahan; Michael Moebius; Juejun Hu

doi:10.1364/OME.456428

1. Introduction

Upon switching from the amorphous to crystalline state, the optical and electronic properties of chalcogenide PCMs, exemplified by the Ge-Sb-Te (GST) alloys, change sharply, which provides compelling active tuning capabilities. PCMs have already found widespread use in non-volatile memories. Nonetheless, if we discount their use in optical discs [1], PCMs’ immense application potential in photonics has only picked up in the past decade [2–4], motivated by applications such as optical switching [5–16], photonic memory [17–19], neuromorphic computing [20–24], active metamaterials and metasurfaces [25–37], reflective displays [38–40], and thermal camouflage [41–43]. Beyond GST, novel phase change alloy compositions specifically designed for photonic applications have also been developed, featuring low optical losses over an extended spectral band [44–47]. Extensive endurance is a key requirement for most photonic applications. While PCM endurance has been heavily vetted for electrical data storage, the distinct switching mechanisms and material compositions employed in photonics entail unresolved challenges that limit current low-loss PCMs’ endurance.

The main goal of this work is to review PCM characteristics and processing steps that impact endurance, identify the common causes for failure relevant to photonic applications and provide a roadmap for future research directions. This review is organized along two main axes. A first axis documents the various factors influencing PCM switching properties. Along a second axis, known failure mechanisms for PCMs are discussed in light of present photonic device performance.

2. On the various factors influencing optical PCM switching properties

2.1 Which phase change materials?

Among phase change chalcogenide materials, GST (often with a stoichiometry in the neighborhood of Ge₂Sb₂Te₅, i.e. GST-225) is the most widely used and studied for both its electrical (high resistance contrast) and photonic (high refractive index contrast) properties. Several alternative compositions with lower optical losses have attracted more recent focus, such as Ge₂Sb₂Se₄Te (GSST) [44,45,48–51], Sb₂Se₃ [47–54], and Sb₂S₃ [39,46,55,56]. Since their chemistry is closely related to that of GST, several conclusions on processing or oxidation can be directly translated to these alternative materials. Literature on GST is hence used as the benchmark for processing and performance of optical PCMs in the rest of this review work.

2.2 Influence of oxidation on PCM

Chalcogenides are prone to oxidation, which has a significant influence over PCM crystallization dynamics and endurance. First turning to GST, ambient oxidation commonly unfolds in two main steps: (i) an initial phase where preferential Ge and Sb oxidation [57–61] occurs, (ii) followed by a gradual formation of tellurium oxide. This evolution is apparent in Fig. 1 [57], where oxidation is visible hours following GST exposure to ambient air and becomes severe after two days, as witnessed by the significant modification in Ge 3d and Sb 4d bands in the X-ray photoelectron spectroscopy (XPS) spectra. Monitoring the surface oxygen content of freshly etched GST shows a moderate surface oxygen implantation of 2 at. % following 4 hours ambient air exposure, which gradually builds up to 53 at. % O after 30 days of aging (see Table 1) [62].

Fig. 1. XPS analysis of GST layer for different air exposure times at room temperature. Arrows indicate the energy position of various oxides based on non-relevant elements. Shadings indicate the regions associated to each element (Green↔Te, Blue↔Sb, Pink↔Ge). Ge and Sb have overlapping peaks in purple. Data adapted from [57].

Download Full Size | PDF

Table 1. Oxygen surface contamination of GST with time. Data adapted from Ref. [62].

View Table | View all tables in this article

While no direct measurement of GST native oxide thickness exists in the literature to the best of our knowledge, GeTe native oxide thickness has been previously estimated using angle resolved-XPS. Upon 90 days exposure in air, the surface of a GeTe film oxidizes into 1 nm of germanium oxide, followed by a 6 nm layer of combined tellurium and germanium oxides [57].

Ge and Sb oxidize significantly more than Te, leading to a surface stoichiometric imbalance. Annealing oxidized GST films leads to further segregation. A tendency to form a Ge-rich phase within the ∼10 nm surface layer is additionally observed by low-energy ion scattering during crystallization of a 100 nm GST-225 film at 300°C in oxygen atmosphere [63]. Sb oxide complexes are found throughout the film, while the surface layer is depleted in Te [63]. The oxidation-induced elemental segregation promotes both crystallization of the Te phase at low temperature as well as partial crystallization of residual alloy into Ge-depleted phase compositions [57,64]. Both pure Te and Ge-depleted phases (e.g. Ge₁Sb₂Te₄) exhibit reduced crystallization temperature (see Fig. 3), which lead to reduction in observed GST crystallization temperature [57–60]. Variations in phase-change properties are therefore largely a consequence of the stoichiometric evolution in the upper layer.

Oxidation of antimony sulfide (Sb₂S₃) and antimony selenide (Sb₂Se₃) shows some specificities with respect to GST. First and foremost, the oxidation process leads to the release of volatile products from both materials [65–67], which lead to a change in stoichiometry according to the following reactions:

(1)$$2S{b_2}{S_3} + 9{O_{2(g )}} \to 2S{b_2}{O_3} + 6S{O_2}(g )$$

(2)$$S{b_2}S{e_3}(s )+ 3 \cdot {O_2}({g/s} )\to 3 \cdot S{b_2}{O_3}(s )+ 3 \cdot S{e_2}(g )$$

The volatilization of oxidation by-products leads to a release of Se₂ gas in the case of Sb₂Se₃ and SO₂ in the case of Sb₂S₃. At temperatures above 600°C, bulk Sb₂S₃ becomes volatile, while Sb₂O₃ volatilizes at sufficiently high temperatures (>900°C) and relatively low oxygen content (>5%) [65–67]. As a result, and similarly to GST, Sb₂S₃ (resp. Sb₂Se₃) shows surface stoichiometric imbalance [47,68] with an enrichment in Sb and O, and depletion in S (resp. Se). Exposure of Sb₂Se₃ to ambient conditions over 5 days was shown to increase Sb₂O₃ surface content by 27%, whereas exposure to vacuum for the same amount of time instead depleted Se from the Sb₂Se₃ film [68]. Thickness of native oxide obtained in standard atmospheric conditions was estimated at 1-3 nm based on solar cell back contact interfacial measurements [68,69].

Oxidation is caused not only by exposure to oxygen, but also to other chemicals at high temperatures. Ambient oxidation of GST was shown to be most strongly influenced by water vapor in the atmosphere at room temperature, which resulted in a greater extent of oxidation than those exposed to N₂ or O₂ (64% O in 70% humidity for 30 min, vs. 52% O after 30 min oxygen exposure and 29% O after 30 min nitrogen exposure) [70]. Considering that numerous common atomic layer deposition (ALD) processes involve H₂O vapor at temperatures > 150 °C, further studies assessing whether such an environment could induce chalcogenide oxidation would help guide future choice of capping layer.

But how does oxidation negatively impact cyclability? Oxidation is well known to alter PCM chemical composition, thereby modifying both phase change properties (i.e. melting point) and optical properties (refractive index), and limiting the amount of switchable material (for films < 100 nm). Beyond inducing instantaneous changes in phase properties, oxidation may have some deep implications during cycling. Previous studies have shown that oxidation with specific GST compositions such as Ge-rich GST [60] could result in a massive redistribution of the chemical elements within the film upon crystallization, in stark contrast to the homogeneous nucleation observed in non-oxidized films. Elemental gradients GST along with presence of native oxides make such alloys prone to segregation, a major concern for photonic devices. The importance of oxidation is independent of the switching mechanisms (e.g. optical or electrical switching), given that architectures associated with both switching mechanisms may include an encapsulating layer.

2.3 Wet cleaning of the oxide layer

Wet cleaning of phase change materials is a key step to eliminate damaged layers induced by oxidative ashing or plasma etching processes, which would otherwise be detrimental for device performance. While several valid chemistries have been proposed to this end (DI water [70,71], nitric acid [72], oxalic acid [38,39], ammonium hydroxide [70,73]), the most commonly used chemistry remains diluted HF, whereby GST oxide is almost entirely removed (57% O to 2% O in XPS, see Fig. 2). Depending on the solubility of each oxide, the leftover PCM surface is heavily depleted in Ge and Sb at its surface (79% and 68% of original stoichiometry respectively after HF etching) but enriched in Te (186% of original stoichiometry after HF etching), due to preferential oxidation of Ge, and Sb [70,74,75]. Nevertheless, wet etching and development may also impact chalcogenide phase change properties, and further studies would thereby be required to assess its role on device cyclability.

Fig. 2. XPS spectra of GST following (left) HBr etching, (middle) photoresist stripping using O₂ plasma, and (right) HF wet clean. Reproduced from [75] with permission.

Download Full Size | PDF

2.4 Influence of doping (lightweight elements: N, O, C)

While oxidation decreases crystallization temperature, uniform doping of GST with lightweight elements such as N, O, and C has a complex influence on crystallization temperature [63,76–78]. At low doping levels, significant increases in crystallization temperature have been reported: from 150.6 to 300.6 °C in 8% wt. carbon doped GST-225 [77]; from 130°C to 200°C in wt. 3% nitrogen doped GST-225 [78], and from 150 °C to 200 °C with 16.7 and 21.7 wt. % oxygen incorporated into GST-225 [76]. At higher doping levels, the crystallization temperature starts to decreases again: switching from 21.7 wt. % oxygen to 30.6 wt. % oxygen incorporated into GST-225 induces a decrease in crystallization temperature from 200°C to 170°C [76].

Previous studies have demonstrated the role of microstructure in the crystallization delay. Below 10 at.%, Te, Sb and most of Ge are in metallic state and the free oxygen is located at interstitial sites [79]. More specifically, it has been shown that oxygen incorporation using ion beam sputtering deposition in GST matrix leads to the preferential formation of bulk Ge-O bonds, rather than GeO₂ phase as in the case of native oxide [63,76]. In the case of nitrogen doping, similar observations were made, with Ge preferentially binding to N to form Ge-N bonds. Dopant insertion in the GST lattice is correlated with an increase in crystallization temperature [80]. Larger oxygen concentrations (>10% [76,79,81]) leads to phase separation and the emergence of bulk Ge-deficient crystalline compositions (e.g. GeSb₂Te₄, Sb₂Te₃, c.f. Figure 3) [58,76,81], while oxygen binds to antimony to form Sb₂O₃ [76].

Fig. 3. (a) Pseudobinary GeTe-Sb2Te3 phase diagram; (b) Sb-Te binary phase diagram.

Download Full Size | PDF

Doping may also modify the occurrence of phase transitions for compounds with multiple allotropic varieties. In the case of GST, earlier works [63] showed that, in the presence of oxygen, initial transformation to the cubic atomic arrangement is hindered and consequently crystallization is delayed. While this statement applies to GST, it is important to note that not all phase change materials have allotropic varieties.

It is interesting to note that extensive oxidation (4 months in ambient atmosphere) of thin sputtered GST films (30 nm) may lead to analog microstructural changes as oxygen doping. Annealing such films at 200°C under vacuum for 1 hour was shown to lead to the accumulation of amorphous germanium oxide at grain boundaries [58,82], while the rest of the alloy is enriched in Sb and Te.

Doping GST with light elements may also prove beneficial for endurance. Laser switching of GST has been shown to increase with oxygen concentration, up to 6% at. oxygen [83], increasing by about an order of magnitude the endurance (see Fig. 4). Similar results have been observed for nitrogen doping, with concentration 2.7% at. nitrogen allows for 8 × 10⁵ cycles, up from 4 × 10⁴ cycles without nitrogen [84]. So how can doping improve cyclability? Several explanations have been suggested, among which (i) hindered elemental diffusion at the grain boundaries [84]; (ii) a reduced volume change upon phase transition [85,86], and (iii) reduced switching energy [85,87]. Nevertheless, more studies are required to shed light on the underlying causes, which may differ based on dopant type and concentration.

Fig. 4. Increase of overwrite cyclability with oxygen concentration in GST alloy. Reproduced from [83] with permission.

Download Full Size | PDF

As previously discussed, doping may endow PCMs with distinct phase change properties such as melting point and endurance. It is important to note that doping would also entail a change in PCM index, which should be carefully characterized.

2.5 Influence of encapsulation on crystallization behavior

To limit oxidation, most PCM applications involve dense capping layers to protect from oxidative environments. Comparison of the structural and chemical characteristics of films left exposed to air with those shown by encapsulated Ge-rich GST films using nitrides (SiN, TiN, TaN) [60,88] or Ta [88] as capping layer highlight how encapsulation fundamentally modifies crystallization. In air-exposed Ge-rich GST films, Ge crystallization preferentially occurs at the film surface while the Ge₂Sb₂Te₅ grains develop later, at higher temperature, and deeper in the film. This is attributed to the appearance of seed layers in or below the oxide during the early stage of annealing. When the Ge-rich GST film is encapsulated, however, the nucleation occurs homogeneously in the whole film.

The chemistry of the capping layer also influences crystallization kinetics [89,90]. By encapsulating 30 nm stoichiometric GST films between dielectric films [89], crystallization dynamics is affected. The dielectric capping layers promote nucleation within GST for Si₃N₄ and Ta₂O₅ capping layers, but inversely impede the nucleation using SiO₂ capping layer. The crystallization temperatures are modified by the capping layer within a relatively marginal range of up to 20°C, depending on the capping material. Wettability measurements infer that these variations are affected by the surface reactivity and chemical affinity of the constituent film materials. [89,91] Experimental evidence suggests that the effect of encapsulating material tends to wane with either higher temperatures or thicker encapsulated layers [91]. Previous works comparing ZnS, SiO₂ and Si₃N₄ capping layers have suggested that modulations in crystallization temperature depend on the surface reactivity and chemical affinity of the film materials, rather than physical parameters such as morphology and stress [89]. Wettability measurements indicate that the ZnS forms chemical binding state with the GST film; while SiO₂ and Si₃N₄ capping layers show limited binding to GST. This discrepancy consequently impacts the nucleation rate and ultimately the crystallization temperature. Similar effects have been reported in alternative materials [60,91,92], which suggests that this interplay is general in scope. In terms of photonic device performance, a higher crystallization temperature implies increased power requirements, which may be challenging at reduced length scales. Conversely, reduced crystallization temperatures could impact PCM durability (e.g. time to crystallization at room temperature).

Since most PCM systems are typically in sub-micron thickness ranges, confinement can have an impact on crystallization behavior. Nucleation within small volumes commonly decreases the nucleation rate [90] due to the exclusion of impurities in small volumes that assist nucleation in bulk. It is important however to decorrelate the influence of confinement and nucleation from the oxide layer. Increases of crystallization temperature upon thickness reduction below 100 nm, have been observed in GST, but strikingly the effect is much larger when the films are not oxidized, effectively suggesting that the crystallization in oxidized films is controlled by nucleation at the oxide-film interface [89,93,94]. Confinement may also impact the emergence of allotropic varieties [90]. This is particularly relevant for GST, which may directly crystallize into its stable hexagonal phase instead of transitioning through the metastable cubic phase in confined environments at sufficiently high temperatures.

From an endurance perspective, the nature of a capping layer is not expected to significantly impact performance beyond a slight variation in switching power. Since PCMs are usually characterized by a volume reduction of the order of 5%–10% at crystallization [95,96], cycling may however induce significant stress over time, which may in turn either threaten integrity of the encapsulation layer or accumulate damage within the PCM.

2.6 Influence of plasma chemistry during the etching step

To pattern chalcogenide PCMs, exposure to etching environments is often required. This can induce damage to the PCM film in several forms, including (i) physical damage (roughening), (ii) chemical damage (incorporation of etchants such as halogens, see Table 2), or (iii) stoichiometry change (due to preferential removal of one or more elements from the film). All these effects may alter the phase-change properties of GST and potentially the performance of the final device. The following section aims at providing an overview of available etching chemistries, along with their potential adverse effects on PCMs (particularly (ii) and (iii)).

Table 2. Post-etch implantation and influence on crystallization temperature. Data adapted from [70].

View Table | View all tables in this article

Halogen-based plasmas have been until now the most widely used chemistry for etching PCMs [62,70,75,97]. This type of chemistry nevertheless comes with potentially severe drawbacks such as surface halogenation, which may occur at the expense of phase change properties [98]. Empirical evidence shows that F-containing plasma result in greater halogenation of the GST than a Cl-containing plasmas [62,70]. For instance, XPS analysis showed that GST etched with CF₄ resulted in 35% F implantation post-etch, while GST etched with Cl₂ lead to 11% Cl implantation post-etch [62]. Among C/F-containing plasmas, lower F/C ratio was shown to limit fluorine implantation. Owing to the thicker C–F polymer formed on the GST interfaces during the etching, diffusion of fluorine radicals to the GST film is hindered [98,99]. The thickness of the protective C-F layer is also correlated with the etch rate: C₄F₈ plasma was reported to etch GST more slowly than CF₄ plasma, which itself etches slower than CHF₃ plasma [70,98]. The absence of a passivation layer allows for implantation of halogens, which can induce void formation upon annealing, eventually leading to stuck-set failure.

Hydrogen-based plasmas constitute another efficient etching chemistry for GST. Based on XPS measurements, comparative stoichiometric surface quantification showed that H₂ plasma induces greater stoichiometric change than SF₆/Ar plasma, less stoichiometric change than Cl₂/Ar plasma, and a higher change in crystallization temperature compared to halogen plasmas [70]. Regarding etch rate, hydrogen-based plasmas etch Sb and Te faster than Ge. This can be linked to the reactivity of individual elements to H₂ plasma rather than the etching products’ volatility (e.g. kinetically limited). Despite the loss in stoichiometry, one potential advantage of hydrogen plasmas over halogen plasmas is that the resulting etched GST material only exhibits metallic states of Ge, Sb and Te, as evidenced by XPS [70]. This might contribute to reduced elemental segregation due to oxidation. Methane-based plasmas can also etch GST, using either CH₄, CH₄/N₂, or CH₄/Ar mixtures. Using such plasma compositions results in a depletion of Sb and Te and enrichment in Ge, showing similar dynamics as for H₂ plasmas. Among the CH₄-containing plasmas, CH₄/N₂ plasma results in the smallest composition and crystallization temperature changes in GST, along with the deposition of a C-H-N passivation layer [70]. In the case of CH₄/Ar plasma, the deposition of carbon on the surface following etching is sufficient to saturate XPS signal over GST [70]. Despite its thickness, this layer was nevertheless shown not to fully protect from oxidation.

Based on Ref. [70], CH₄/N₂ and H₂ plasmas appear to be most adapted to etch GST, limiting changes in crystallization temperature and curbing surface halogenation. Nevertheless, other properties such as refractive index may be modified by etching processes. This point will require further attention from the community if these processes are to be successfully adapted to PCM manufacturing.

Given the propensity of chalcogenides to oxidize, it is also important to note that ashing can also severely alter the GST phase change properties [75,101], effectively requiring alternative processes for cleaning or resist stripping. Previous works have demonstrated that ashing yields a GST oxide layer at the surface, mainly composed of GeO₂ [75]. After ashing, the surface layer is saturated in oxygen, with no more evolution following subsequent air exposure [75]. For comparison, such oxygen saturation is observed only upon 30 days in ambient conditions [75]. Ashing can also impact other common materials such as nitrides, with potential adverse effects on their electrical properties [101]. It was for instance shown that ashing of a TiN bottom electrode introduced a TiO₂ oxide layer at the interface with GST, which adversely impacted the electrical switching properties of GST [101].

Beyond plasma chemistry, the nature of the plasma used may induce different elemental distribution within the etched material, particularly at the sidewall edges. It has for instance been shown that relying to a chlorine neutral beam maintains sidewall profile and stoichiometry, while a chlorine radiofrequency inductively coupled plasma (RF ICP) does not maintain vertical sidewalls during etching, and further leads to a depletion in Ge and Sb near the sidewall surface (see Fig. 5).

Fig. 5. TEM images and corresponding elemental dispersion spectroscopy mappings of GST etched sidewalls, using as etchant (left) a neutral chlorine beam (NBE) and (right) a chlorine RF inductively coupled plasma (ICP). Reproduced from [100] with permission.

Download Full Size | PDF

2.7 Influence of ion implantation on phase change performance during cycling

As discussed above, the use of halogen plasmas leads to ion implantation, which in turn may reduce cycling performance. Resorting to NH₃ instead of HBr plasma to etch GST for instance has been shown to result in better cycling performance [102]. HBr-etched device exhibited appreciable amounts of Br incorporated within the GST volume with penetration depth > 50 nm. Following thermal annealing below 400°C, these ions induced voids following the volatilization of etching products (GeBr₄, SbBr₂, and TeBr₂). Turning to NH₃ plasma enables lower H implantation than Br implantation using HBr plasma, which in consequence suppresses degradation of cycling performance (c.f. Figure 6). While void formation is not the primary concern for most optical applications given the small size of the voids, the release of gaseous by-products following encapsulation may threaten capping layer integrity, which may call for an intermediate annealing step.

Fig. 6. Resorting to NH₃ plasma instead of HBr improves cycling performance by over two orders of magnitude. Reproduced from [102] with permission.

Download Full Size | PDF

3. Making sense of empirical PCM performance through failure mechanisms

3.1 Switching mechanisms

Common switching mechanisms involve either (i) electrical, (ii) optical, or (iii) electrothermal stimuli. Electrical switching (where electric current passes through the PCM) has attracted the most focus for electrical non-volatile memories, allowing for outstanding endurances up to 10¹² cycles [103]. However, it is poorly adapted for optical applications. Firstly, the intrinsic filament-like crystallization induced by electrical switching induces strongly inhomogeneous material properties, which are incompatible for optical applications. Optical switching is associated with several constraints, including the influence of beam shape on crystallization profile, high required optical power, and most importantly limited scalability for on-chip device integration. Finally, electrothermal switching via heaters made of metal [104,105], ITO [9,11,106], graphene [107,108], or doped Si [109–111], has the potential to crystallize entire PCM volumes simultaneously, while discarding key failure mechanisms such as electromigration [112]. Endeavors to engineer the switching thermal profile using heater geometry are paving the way towards an homogenous control over crystallization, a stepping stone to achieve multi-level control over bulk PCM crystallization and re-amorphization [104]. While they still face hurdles, they are in theory free from several key failure mechanisms such as electromigration. The precise control over phase change properties is a key aspect to address for PCM to reach their full potential. This includes incremental control over partial PCM crystallization, as well as reproducibility of optical characteristics with cycling. Despite such advantages, reported electrothermal endurance has remained below those reported for optical and electrical switching, which calls for further investigations into its associated failure mechanisms.

It is important to note that the duration and strength of stimuli does have an influence on PCM endurance. In the case of laser switching, pulse energy, peak power and duration was observed to significantly impact endurance, although the underlying mechanisms have yet to be fully understood. First considering single laser pulses only, it has been shown that lower energy laser pulses lead to a reduced cycling durability in Sb₂Se₃ [47]. This observation was done using a single 36 nJ pulse for amorphization, decreasing maximal endurance from 4000 cycles to an unspecified cycle number. Turning to the use of multiple laser pulses, another study using Sb₂S₃ and same optical switching method shows that gradual amorphization using multiple pulses of 155 nJ (5 pulses) instead of a single pulse of 250 nJ leads to an increase in endurance by a factor of 15. It is noteworthy that the degree of crystallization may strongly influence the cycling endurance: a partial crystallization of 90%, 60%, and 20% respectively lead to maximal endurance of 30, 1000, and 7000 cycles [113]. These observations are likely related to re-arrangement of chemical bonds and potentially long range diffusion enhanced by allowing the material to fully crystallize, but further studies are required to validate this hypothesis. Electrical switching leads to distinct correlations between pulse parameters and endurance. It was demonstrated that long pulses with disproportionate energy during reset leads to lower cyclability for SbTe-based alloy [114], likely due to enhanced atomic mobility, while cycling GST with excessively weak reset pulse also generates lower endurance [115]. Pulse parameters hence should be finely optimized to obtain optimal endurance, likely a balance between segregation and self-healing effects.

3.2 Failure mechanisms

Switching failure mechanisms fall in two broad categories: stuck-set (e.g. stuck in crystalline state) or stuck-reset (e.g. stuck in amorphous state), both driven by atomic migration at higher temperatures [116]. The underlying causes behind these failure mechanisms have been mainly attributed to either (i) void formation (stuck-reset) or (ii) elemental segregation (stuck-set).The following section presents an overview of both electrical – biased based and non-electrical bias-based failure mechanisms. In both cases, the emphasis is placed on how relevant these failure mechanisms are (or not) to electrothermal switching for photonic applications.

Incremental void formation is a bias-based phenomenon, which gradually leads to delamination at the interface between electrode and PCM. Void formation & coalescence mechanisms largely account for the emergence of stuck-reset failures. GST being a relatively soft amorphous alloy (54.9 GPa based on [117] – almost as soft as tin), along with a large thermal expansion coefficient (1.7 × 10⁻⁵ K^-1) and significant volume change during phase change (7% [49,118]), it is especially vulnerable to void development [117]. Upon switching, these voids eventually coalesce and lead to delamination at the interface between electrode and GST [59,119,120]. By adding a suitable amount of an unspecified doping material into GST, void formation processes can be delayed, significantly improving cell endurance from 10⁶ to more than 10⁹ cycles, as evident in Fig. 7 [117]. Other strategies to limit void formation and coalescence include the densification of the GST layer through post-deposition anneals [121] or the incorporation of metallic liners [122,123]. Alternative strategies to mitigate the emergence of voids include the use of metallic liners, which, combined with ALD deposition, have allowed for endurance up to 2 × 10¹² cycles [103].

Fig. 7. Influence of an unspecified doping agent on GST endurance using electrical switching. The undoped sample started failing before 10⁶ cycles. The sample failure is due to the formation of an open circuit because of the pore at the interface with the electrode. Data adapted from [117].

Download Full Size | PDF

Electromigration is a common driver behind elemental segregation and stuck-SET failure mechanisms [124–128]. This bias-based phenomenon is attributed to the difference in elemental electronegativity. In the case of our reference GST alloy, this difference is substantial, with an electronegativity of 5.49 eV for Te compared to 4.6 and 4.85 eV for Ge and Sb respectively [127]. By analyzing heavily-cycled cells under bias, it has been consistently observed that Te moves toward the positive electrode (anode), while Sb segregates in the opposite direction towards the negative electrode (cathode) [124–127], leading to mechanical failure (Fig. 8). While electromigration is relatively slow in the solid phase, it is much faster in the liquid phase, e.g. upon re-amorphization, eventually leading to mechanical failure (see Fig. 6).

Fig. 8. SEM images of (left) as-prepared GST sample and the failed (right) GST upon biasing at 200°C. Reproduced from [128] with permission.

Download Full Size | PDF

Additional factors not linked to electrode bias can also account for elemental segregation, such as thermal gradients. A thermal gradient inevitably arises when the PCM is either heated up or cooled down, which may represent a significant time of the set/reset operation due to thermal inertia. These thermal gradients may either be in-plane, depending on the geometry of the heating devices [47], or out-of-plane (e.g. perpendicular to heater plane). Based on previous works, such a thermal gradient may lead to two main segregation mechanisms: (i) incongruent melting and (ii) thermodiffusion (also known as Soret effect or thermophoresis).

The origin of incongruent melting can be understood by examining, for instance, the phase diagram of the GeTe-Sb₂Te₃ pseudobinary system (Fig. 3(a)). When heated up or cooled down, GST-225 becomes a binary mixture between 632 °C (solidus temperature) and 648 °C (liquidus temperature). Within this region, GST-225 can spontaneously separate into two phases, where a GeTe-rich phase crystallizes out of the melt, while a Sb₂Te₃-rich phase remains a liquid. When this binary mixture is subject to a temperature gradient, the phase separation can lead to directional solidification and elemental segregation [129]. A viable way to circumvent the issue of incongruent melting is to heat the entire PCM volume above the liquidus temperature during reset operation with sufficient dwelling time such that any Ge-rich solid phase formed can be re-dissolved into the liquid phase for every cycle. Indeed, it has been found that melting can ‘heal’ elemental segregation in memory cells [115]. This strategy however implies longer switching times than typical switching cycles, and further depend on the system’s thermal inertia.

Thermophoresis [130], whereby distinct molecules exhibit different responses to the strength of a temperature gradient, is characterized by the Soret coefficient S_T:

(3)$${S_\textrm{T}} = {D_\textrm{T}}/D = \frac{1}{{kT}}.\frac{{\partial G}}{{\partial T}}$$

where G is the Gibbs free energy, T is the temperature, k is the Boltzmann constant, D_T is the thermodiffusion coefficient and D is the diffusion coefficient. Thermophoresis is a driver for elemental segregation that can exist in parallel with other segregation mechanisms such as electromigration. Previous studies aiming to model mass transport in electrically-switched GST have shown that thermophoresis must be taken into account for an accurate description of experimental segregation profiles [131–135]. Focusing on GST, compositional demixing was observed along the direction of the applied thermal gradient, in the case of both the polycrystalline GST solid and its corresponding melt [133]. Soret coefficients for Sb and Ge are opposite and significant, driving Ge towards the colder region and Sb towards the hotter region [131,133]. Earlier works have shown that the influence of thermal gradients in Sb and Te can be the most significant drivers for mobility in electrically-switched GST [131,135], compared to stress or electric drivers (see Fig. 9). In Fig. 10, it is shown that maximal temperature (e.g. higher thermal gradient) leads to migration of Ge towards the surface of the film (e.g. away from the heating bottom interface). Meanwhile, Sb appears only slightly depleted away from the heater for higher temperature gradients [131,136]. The Soret coefficient for Te has been shown to be less pronounced, and hence less prone to thermophoresis compared to Sb and Ge.

Fig. 9. Comparative evolution of elemental ionic velocities in biased GST at 3 µs. (a) Total ionic velocity in forward (solid line) and reverse bias (dashed line) for each element; (b) individual ionic velocities based on the three main drivers for flow (electric, stress & thermal gradients) for each element in forward bias. Data adapted from Ref. [131].

Download Full Size | PDF

Fig. 10. Atomic concentration line profile for a GST device (a) as-deposited and heated to (b) and (right) 650 K maximal temperature, after 5 × 10⁸ consecutive 100 ns pulses and (c) 1000 K maximal temperature, after 105 consecutive 100 ns pulses. The heater occupies the region z < 0. Pulses had a maximum voltage of 1.9 V, reaching a maximum current of 425 µA. Reproduced from Ref. [133] with permission.

Download Full Size | PDF

Stress-induced segregation effects have also been studied [59,131,137]. Previous simulations focusing on electrically-switched GST (see Fig. 9 for details) [131] have shown that stress-induced flux F_S in this particular system is significantly reduced in comparison with thermodiffusion F_T (for Ge and Te, F_S is about half the value of F_T). Nevertheless, previous experimental observations by transmission electron microscopy have shown evidence for Te segregation at the grain boundary in GST correlated with substantial stress release at temperatures above 200 °C, following transition from amorphous to cubic structure for GST [59]. This observation highlights the need for an accurate description of the stress distribution to predict stress-related atomic fluxes, directionality and relative strength.

Crystallization-induced segregation (CIS) is another mechanism [138] not caused by electric bias, relevant when crystallization kinetics are slow. When local elemental distribution does not allow for rapid crystallization, a modified composition with faster crystallization kinetics may nucleate and grow first, giving rise to CIS. With sufficiently high material mobility (but potentially well below melting temperature), the stoichiometry with fastest crystallization speed nucleates and grows first. CIS was evidenced by Auger spectroscopy during laser pulse-induced crystallization. Local stoichiometry highlights segregation of Te and Sb in opposite directions (see Fig. 11). CIS assumes however that the local stoichiometry shows poor crystallization kinetics, as is for instance the case with Te-rich GST [138] but may not be valid for all PCM compositions, in particular for GST, well known for its fast switching characteristics.

Fig. 11. High-resolution Auger analysis of Te-rich GST during laser induced crystallization. While Te systematically migrates into the film during optical exposure, Sb migrates towards the optically-heated region for sufficiently high pulse powers. This elemental rearrangement within the poorly crystallizing Te-rich GST alloy allows the emergence of a local stoichiometry capable of much more rapid crystallization. Reproduced from Ref. [138] with permission.

Download Full Size | PDF

As feature sizes are scaled down and large aspect ratios are considered, potential surface tension effects threaten the integrity of thin films, and must be factored in. Thin capping layers and high surface-to-volume ratios are commonly associated with higher free surface energies, which may lead to dewetting. Uncapped 40 nm thick non-encapsulated GST films [139] thick films were shown to dewet following fs-pulsed laser irradiation on a Si support, emphasizing how critical such phenomenon is during optical PCM switching. During dewetting, viscous flow (e.g. liquid dewetting) at temperatures above the melting temperature during the amorphization step, or diffusion along grain boundaries at temperatures well below the melting temperature (e.g. solid-state dewetting) during the crystallization step [140], can both lead to film break-up to minimize the overall film free surface energy. While film mobility is higher in liquid dewetting, the amorphization process also occurs on a shorter time scale than crystallization. As a consequence, it is not possible to conclude a priori which of the two dewetting types dominate in PCM dewetting. Examination of dewetting patterns that eventually arise upon failure may a posteriori point to the dominant mechanism: irregular dewetting patterns following grain boundaries are typically associated with solid-state dewetting, while regular dewetting patterns with constant curvatures are associated to liquid dewetting (see Fig. 12). To circumvent such issues, several solutions exist: (i) thick capping layers help ensure the mechanical integrity of mobile films during switching, constraining the film into its present shape; (ii) designing geometries with low surface-to-volume ratios reduces the system’s initial free surface energy, and hence limits dewetting; (iii) the choice of surrounding medium (e.g. capping layer and substrate) based on its wetting properties with the encapsulated PCM (e.g. low contact angle between PCM and surrounding medium) would further help to mitigate dewetting. This last point would also limit the emergence of additional stress during switching, thereby limiting stress-related segregation.

Fig. 12. Top scanning electron microscope view of a Sb₂Se₃ PCM patch on Si rib waveguide with a 15 nm Al₂O₃ capping layer following electrothermal cycling. The film retracts from the bottom edges and nucleates holes at the edge of the rib waveguide, indicating liquid dewetting.

Download Full Size | PDF

Several strategies can be adapted from bias-based switching to non-biased switching, and thereby contribute to improve electrothermal endurance. Duration and strength of stimuli, doping, as well as limiting the impact of oxidation are among the most critical factors that bear relevance both with and without bias. Biased-based switching studies have also unraveled several more general phenomena which are likely highly relevant for non-biased-based switching, such as stress-induced-effects, crystallization-induced segregation, and thermal migration.

The focus was placed in this section on GST, given the comparatively larger literature available on this PCM compared to other less traditional PCMs such as Sb₇Te₃, GSST, Sb₂Se₃ or Sb₂S₃. Given that these optical materials are all chalcogenides with chemically similar elements, they share several common characteristics with GST, which suggests they are prone to similar failure mechanisms. GSST and Sb₂Se₃ both exhibit significant volume change upon crystallization (4-5% for GSST [49], and 5-6% for Sb₂Se₃ based on measurements by the authors). Specifically regarding Sb₂Se₃ and Sb₂S₃, both materials show similarly large thermal expansion coefficients (3.7 × 10⁻⁵ K⁻¹ for Sb₂Se₃ [141] and up to 1.3 × 10⁻⁵ K^-1 for Sb₂S₃ [142] versus 1.7 × 10⁻⁵ K^-1 for GST [49,118]), and similar Young’s modulus (64.78 GPa for Sb₂S₃ and 73.64 GPa for Sb₂Se₃ by calculation based on [143], versus 54.9 GPa for GST based on [117]). Among the compositions cited above, Sb₇Te₃ is a quasi-eutectic composition, while Sb₂Se₃ and Sb₂S₃ undergo congruent melting (e.g. single phase system). This stands in stark contrast with GST, which undergoes incongruent melting between liquidus and solidus temperature (see Fig. 3). This observation eliminates incongruent melting as potential cause of segregation for Sb₂Se₃, Sb₇Te₃, and Sb₂S₃.

3.3 Empirical PCM endurance in photonic applications and possible limitations

To illustrate the influence of these failure mechanisms in device applications, the present section provides a bird eye’s view of reported optical PCM endurance, with a wide range of performance (see Table 3).

Table 3. Summary of reported PCM endurance and associated processing characteristics for optical applications (green shaded rows). Endurance data measured in GST electronic memory devices (yellow shaded rows) are also included for comparison. D-GST: GST doped with unspecified dopant. [26, 45, 50, 51, 52, 107, 112, 113, 116, 117, 144, 145, 146].

View Table | View all tables in this article

Unfortunately, the most widely studied materials such as GST and GSST have thus far exhibited sub-optimal endurance for optical applications. Prior works using ITO electrodes to switch highly confined 7 nm-thin sputtered GST films have also shown an endurance limited to about 150 cycles [144]. Specifically regarding electrothermal switching, thin (<40 nm) sputtered GST films fully capped by ZnS/SiO₂ have demonstrated an endurance limited to 50 cycles [26], while other works showed endurance > 1000 cycles for 10 nm-thick GST patches encapsulated in 30 nm Al₂O₃ [110]. Similarly, a wide variety of performances have been obtained for GSST. While over 1,000 switching cycles have been achieved in GSST films on small (10 µm) heaters [45], larger area GSST meta-atoms (250 nm thick GSST capped with 15 nm Al₂O₃ films on 200 µm heaters) allowed for stable cycling up to tens of cycles [104].

Sb₂Se₃ and Sb₂S₃ are two promising alloys with moderate index contrast, which have also shown different behavior under switching. Recent works cycling thermally evaporated Sb₂S₃ capped with 10 nm Al₂O₃ have shown cyclability over a minimum of 125 cycles [53]. Stable switching of sputtered Sb₂Se₃ covered by a 200 nm capping layer of ZnS:SiO₂ (20%:80%) has been reported over 4000 cycles, with a decay in reflection between 4000 and 5000 cycles, dropping to 50% of the initial switching contrast. A similar cycling test of Sb₂S₃ with an identical capping layer show a reduced endurance compared to Sb₂Se₃, with a material degradation apparent after around 1000 cycles [47]. Sb₂S₃ is particularly prone to damage at the nucleation sites during annealing, induced by the lateral migration of sulfur [47,113]. Interestingly, using these same materials in patterned geometries such as patches over ring resonator structures exhibited a cycling durability lower by an order of magnitude [52]. Cycling a 23-nm-thin Sb₂Se₃ film with an identical 200 nm ZnS:SiO₂ capping layer using a laser induces gradual damage, particularly at the pattern edges. Following initial transmission fluctuations, constant device transmission was observed over 350 cycles. Beyond 500 cycles, transmission contrast is decreased by half, driven by degradation of material properties [52]. The authors attribute the failure mechanism to significant thermal in-plane gradients and a higher maximal temperature, driven by lower local thermal conductivity of the surrounding capping layer and the abrupt interfaces.

Excellent endurance by electrothermal switching has been demonstrated for Sb₇Te₃. Electrothermal switching of Sb₇Te₃ patches deposited using sputtering directly on TiW heaters have shown endurance over 1.5 × 10⁸ cycles before showing indications of reset failures [146], setting the endurance of this PCM apart from earlier demonstrations. While no direct information is provided on the capping layer (material, thickness), patterning method (lift-off vs. etching), or dimensions in this particular reference, other works from the same group with Sb₇Te₃ suggests PCM thicknesses in the range 40-100 nm, SiN as substrate layer, and lateral dimensions < 2 µm [147]. We note that unlike GST, Sb₇Te₃ is a quasi-eutectic composition that exhibits a single crystalline phase throughout the entire temperature range according to the phase diagram in Fig. 3(b). This attribute likely contributes to the enhanced endurance through suppressing incongruent melting.

If one also considers electrical switching, prior studies have demonstrated outstanding endurance, with GST endurance over 10⁶ cycles for rewritable optical discs and up to 2 × 10¹² cycles [116] for nonvolatile memories [2,103,117,148,149].

So what level of endurance can one expect in PCM-based optical devices which largely rely on electrothermal switching? Electromigration can be ruled out as a failure mode. The associated void formation mechanisms which account for the set-stuck failure in electrical switching are unlikely to be the primary concern for optical applications, either. While these factors work in favor of electrothermal switching, photonic applications introduce challenges not anticipated by electronic memories. In particular, photonic devices typically demand PCMs with much larger volumes. As a result, temperature variations across the PCM as well as cyclic stress caused by volume change during phase transition become more severe in photonic devices. The former is linked to elemental segregation due to thermodiffusion and/or incongruent melting, whereas the latter could exacerbate elemental segregation and compromise structural integrity of capping films, possibly inducing secondary damages through dewetting, volatization, and oxidation. While the impact of these factors remains to be quantified in further studies, mitigating temperature gradients via heater geometry optimization [104,150] and adoption of capping materials with larger thickness and improved mechanical durability are plausible solutions to enhance endurance.

Our analysis also points to other research directions worth pursuing. Etching plasma chemistries may adversely impact phase change properties, noticeably changing the crystallization temperature and potentially inducing voids in the crystal structure. Further works assessing the influence of plasma etching versus the more “benign” lift-off process would help assess whether PCM endurance is critically impacted by specific etching environments beyond the reported formation of voids. Based both on this review and other works [2,148,149], it appears that the nature of the compounds immediately surrounding the PCM also can influence phase change properties. Deposition of ITO, Al₂O₃ or SiO₂, as previously implemented for PCM photonic applications [52,104,144], commonly involve the use of potentially oxidative environments, which might be detrimental for PCM endurance. Exploring alternative non-oxidative capping media (e.g. SiN or AlN) could alleviate the issue.

4. Design rules for optimal PCM endurance in optical applications

To fulfill the promises of PCMs, the stability of their phase change properties must be ensured during cycling. While no fundamental limits to optical PCM endurance have been identified in this review, several key properties and processing parameters have been singled out to further improve their durability, which are summarized through the following set of design rules (by estimated descending order of importance):

1) selecting a PCM composition that remains within single crystalline phase regions throughout the entire temperature range relevant to switching operation, and exhibits small volume shrinkage during crystallization;
2) encapsulating a PCM with an inert (non-reactive), stable material to suppress interdiffusion, surface oxidation, and dewetting;
3) optimizing the switching process and/or heater design to minimize temperature non-uniformity;
4) designing a mechanically robust device structure against cyclic stress;
5) removing PCM native oxide prior to encapsulation (using for instance diluted HF) to avoid later elemental segregation;
6) resorting to lift-off or less aggressive plasma chemistries such as N₂/CH₄ to pattern PCM, so as to avoid inducing chemical implantation and voids;
7) leveraging self-healing mechanisms in the PCM re-melting process;
8) identifying optimal pulse switch strength and duration for enhanced endurance.

5. Conclusion

In this work, GST and other chalcogenide PCMs have been reviewed in the context of extending endurance for optical applications while analyzing lessons learnt from electronic use of GST. Oxidation easily takes place in these chalcogenides and needs to be accounted for in process design. Depending on the dimensionality of PCM, it may be a significant concern and can be prevented with proper encapsulation after deposition. Fabrication steps that introduce dopants, purposefully or by accident, should be accounted for as they can affect thermal properties, in particular switching behavior, as well as optical properties. Repeated switching of the PCM may lead to failure due to be pore formation (for electrical circuits), phase segregation, electromigration in the case of high electric fields, and thermomigration in the case of large temperature gradients. Potential new research directions include analyzing the effective switching volume as a function of oxide thickness, investigating melting-induced healing the PCM via optical and electrothermal heating, and developing reactive ion etching and wet etching methods that limit PCM damage, stoichiometry change and roughening. Further development of doping in optical PCMs to extend endurance and improve material consistency is also needed.

Funding

Charles Stark Draper Laboratory; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (P500PT_203222).

Disclosures

The authors declare no conflicts of interest.

Data availability

No data were generated or analyzed in the presented research.

References

1. E. R. Meinders, A. V. Mijiritskii, L. Van Pieterson, and W. Matthias, “Optical data storage: phase-change media and recording,” Opt. Data Storage Phase-Change Media Rec., 1–173 (2006).

2. M. Wuttig, H. Bhaskaran, and T. Taubner, “Phase-change materials for non-volatile photonic applications,” Nat. Photonics 11(8), 465–476 (2017). [CrossRef]

3. T. Cao and M. Cen, “Fundamentals and applications of chalcogenide phase-change material photonics,” Adv. Theory Simul. 2(8), 1900094 (2019). [CrossRef]

4. S. Abdollahramezani, O. Hemmatyar, H. Taghinejad, A. Krasnok, Y. Kiarashinejad, M. Zandehshahvar, A. Alù, and A. Adibi, “Tunable nanophotonics enabled by chalcogenide phase-change materials,” Nanophotonics 9(5), 1189–1241 (2020). [CrossRef]

5. M. Stegmaier, C. Ríos, H. Bhaskaran, C. D. Wright, and W. H. P. Pernice, “Nonvolatile all-optical 1 × 2 switch for chipscale photonic networks,” Adv. Opt. Mater. 5(1), 1600346 (2017). [CrossRef]

6. M. Rudé, J. Pello, R. E. Simpson, J. Osmond, G. Roelkens, J. J. G. M. Van Der Tol, and V. Pruneri, “Optical switching at 1.55 µm in silicon racetrack resonators using phase change materials,” Appl. Phys. Lett. 103(14), 141119 (2013). [CrossRef]

7. C. Rios, M. Stegmaier, Z. Cheng, N. Youngblood, C. D. Wright, W. H. P. Pernice, and H. Bhaskaran, “Controlled switching of phase-change materials by evanescent-field coupling in integrated photonics [Invited],” Opt. Mater. Express 8(9), 2455 (2018). [CrossRef]

8. D. Gostimirovic, R. Soref, and W. N. Ye, “Resonant bistable 2 × 2 crossbar switches using dual nanobeams clad with phase-change material,” OSA Continuum 4(4), 1316 (2021). [CrossRef]

9. K. Kato, M. Kuwahara, H. Kawashima, T. Tsuruoka, and H. Tsuda, “Current-driven phase-change optical gate switch using indium-tin-oxide heater,” Appl. Phys. Express 10(7), 072201 (2017). [CrossRef]

10. P. Xu, J. Zheng, J. K. Doylend, and A. Majumdar, “Low-loss and broadband nonvolatile phase-change directional coupler switches,” ACS Photonics 6(2), 553–557 (2019). [CrossRef]

11. C. Wu, H. Yu, H. Li, X. Zhang, I. Takeuchi, and M. Li, “Low-loss integrated photonic switch using subwavelength patterned phase change material,” ACS Photonics 6(1), 87–92 (2019). [CrossRef]

12. Q. Zhang, Y. Zhang, J. Li, R. Soref, T. Gu, and J. Hu, “Broadband nonvolatile photonic switching based on optical phase change materials: beyond the classical figure-of-merit,” Opt. Lett. 43(1), 94–97 (2018). [CrossRef]

13. F. De Leonardis, R. Soref, V. M. N. Passaro, Y. Zhang, and J. Hu, “Broadband electro-optical crossbar switches using low-loss Ge₂Sb₂Se₄Te₁ phase change material,” J. Lightwave Technol. 37(13), 3183–3191 (2019). [CrossRef]

14. Y. Zhang, Q. Zhang, C. Ríos, M. Y. Shalaginov, J. B. Chou, C. Roberts, P. Miller, P. Robinson, V. Liberman, M. Kang, K. A. Richardson, T. Gu, S. A. Vitale, and J. Hu, “Transient tap couplers for wafer-level photonic testing based on optical phase change materials,” ACS Photonics 8(7), 1903–1908 (2021). [CrossRef]

15. J. Faneca, L. Trimby, I. Zeimpekis, M. Delaney, D. W. Hewak, F. Y. Gardes, C. D. Wright, and A. Baldycheva, “On-chip sub-wavelength Bragg grating design based on novel low loss phase-change materials,” Opt. Express 28(11), 16394 (2020). [CrossRef]

16. J. Faneca, S. Garcia-Cuevas Carrillo, E. Gemo, C. R. de Galarreta, T. Domínguez Bucio, F. Y. Gardes, H. Bhaskaran, W. H. P. Pernice, C. D. Wright, and A. Baldycheva, “Performance characteristics of phase-change integrated silicon nitride photonic devices in the O and C telecommunications bands,” Opt. Mater. Express 10(8), 1778 (2020). [CrossRef]

17. C. Rios, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer, C. D. Wright, H. Bhaskaran, and W. H. P. Pernice, “Integrated all-photonic non-volatile multi-level memory,” Nat. Photonics 9(11), 725–732 (2015). [CrossRef]

18. N. Farmakidis, N. Youngblood, X. Li, J. Tan, J. L. Swett, Z. Cheng, C. D. Wright, W. H. P. Pernice, and H. Bhaskaran, “Plasmonic nanogap enhanced phase-change devices with dual electrical-optical functionality,” Sci. Adv. 5(11), eaaw2687 (2019). [CrossRef]

19. E. Gemo, J. Faneca, S. G.-C. Carrillo, A. Baldycheva, W. H. P. Pernice, H. Bhaskaran, and C. D. Wright, “A plasmonically enhanced route to faster and more energy-efficient phase-change integrated photonic memory and computing devices,” J. Appl. Phys. 129(11), 110902 (2021). [CrossRef]

20. J. Feldmann, M. Stegmaier, N. Gruhler, C. Riós, H. Bhaskaran, C. D. Wright, and W. H. P. Pernice, “Calculating with light using a chip-scale all-optical abacus,” Nat. Commun. 8(1), 1256 (2017). [CrossRef]

21. J. Feldmann, N. Youngblood, C. D. Wright, H. Bhaskaran, and W. H. P. Pernice, “All-optical spiking neurosynaptic networks with self-learning capabilities,” Nature 569(7755), 208–214 (2019). [CrossRef]

22. M. Xu, X. Mai, J. Lin, W. Zhang, Y. Li, Y. He, H. Tong, X. Hou, P. Zhou, and X. Miao, “Recent advances on neuromorphic devices based on chalcogenide phase-change materials,” Adv. Funct. Mater. 30(50), 2003419 (2020). [CrossRef]

23. J. Feldmann, N. Youngblood, M. Karpov, H. Gehring, X. Li, M. L. Gallo, X. Fu, A. Lukashchuk, A. S. Raja, J. Liu, C. D. Wright, A. Sebastian, T. J. Kippenberg, W. H. P. P. Pernice, H. Bhaskaran, M. Stappers, M. Le Gallo, X. Fu, A. Lukashchuk, A. S. Raja, J. Liu, C. D. Wright, A. Sebastian, T. J. Kippenberg, W. H. P. P. Pernice, and H. Bhaskaran, “Parallel convolutional processing using an integrated photonic tensor core,” Nature 589(7840), 52–58 (2021). [CrossRef]

24. Z. Cheng, C. Ríos, W. H. P. Pernice, C. David Wright, and H. Bhaskaran, “On-chip photonic synapse,” Sci. Adv. 3(9), e1700160 (2017). [CrossRef]

25. Q. Wang, E. T. F. Rogers, B. Gholipour, C. M. Wang, G. Yuan, J. Teng, and N. I. Zheludev, “Optically reconfigurable metasurfaces and photonic devices based on phase change materials,” Nat. Photonics 10(1), 60–65 (2016). [CrossRef]

26. B. Gholipour, J. Zhang, K. F. MacDonald, D. W. Hewak, and N. I. Zheludev, “An all-optical, non-volatile, bidirectional, phase-change meta-switch,” Adv. Mater. 25(22), 3050–3054 (2013). [CrossRef]

27. Y. Wang, P. Landreman, D. Schoen, K. Okabe, A. Marshall, U. Celano, H.-S. P. Wong, J. Park, and M. L. Brongersma, “Electrical tuning of phase-change antennas and metasurfaces,” Nat. Nanotechnol. 16(6), 667–672 (2021). [CrossRef]

28. C. Williams, N. Hong, M. Julian, S. Borg, and H. J. Kim, “Tunable mid-wave infrared Fabry-Perot bandpass filters using phase-change GeSbTe,” Opt. Express 28(7), 10583 (2020). [CrossRef]

29. F. Yue, R. Piccoli, M. Y. Shalaginov, T. Gu, K. A. Richardson, R. Morandotti, J. Hu, and L. Razzari, “Nonlinear mid-infrared metasurface based on a phase-change material,” Laser Photonics Rev. 15(3), 2000373 (2021). [CrossRef]

30. M. Y. Shalaginov, S. D. Campbell, S. An, Y. Zhang, C. Ríos, E. B. Whiting, Y. Wu, L. Kang, B. Zheng, C. Fowler, H. Zhang, D. H. Werner, J. Hu, and T. Gu, “Design for quality: reconfigurable flat optics based on active metasurfaces,” Nanophotonics 9(11), 3505–3534 (2020). [CrossRef]

31. M. Y. Shalaginov, S. An, Y. Zhang, F. Yang, P. Su, V. Liberman, J. B. Chou, C. M. Roberts, M. Kang, C. Rios, Q. Du, C. Fowler, A. Agarwal, K. A. Richardson, C. Rivero-Baleine, H. Zhang, J. Hu, and T. Gu, “Reconfigurable all-dielectric metalens with diffraction-limited performance,” Nat. Commun. 12(1), 1225 (2021). [CrossRef]

32. M. N. Julian, C. Williams, S. Borg, S. Bartram, and H. J. Kim, “Reversible optical tuning of GeSbTe phase-change metasurface spectral filters for mid-wave infrared imaging,” Optica 7(7), 746–754 (2020). [CrossRef]

33. X. Yin, T. Steinle, L. Huang, T. Taubner, M. Wuttig, T. Zentgraf, and H. Giessen, “Beam switching and bifocal zoom lensing using active plasmonic metasurfaces,” Light: Sci. Appl. 6(7), e17016 (2017). [CrossRef]

34. C. Ruiz de Galarreta, I. Sinev, A. M. Alexeev, P. Trofimov, K. Ladutenko, S. Garcia-Cuevas Carrillo, E. Gemo, A. Baldycheva, J. Bertolotti, and C. David Wright, “Reconfigurable multilevel control of hybrid all-dielectric phase-change metasurfaces,” Optica 7(5), 476–484 (2020). [CrossRef]

35. A. Tittl, A. K. U. Michel, M. Schäferling, X. Yin, B. Gholipour, L. Cui, M. Wuttig, T. Taubner, F. Neubrech, and H. Giessen, “A switchable mid-infrared plasmonic perfect absorber with multispectral thermal imaging capability,” Adv. Mater. 27(31), 4597–4603 (2015). [CrossRef]

36. S. Abdollahramezani, O. Hemmatyar, M. Taghinejad, H. Taghinejad, A. Krasnok, A. A. Eftekhar, C. Teichrib, S. Deshmukh, M. El-Sayed, E. Pop, M. Wuttig, A. Alu, W. Cai, and A. Adibi, “Electrically driven programmable phase-change meta-switch reaching 80% efficiency,” arXiv:2104.10381 (2021).

37. J. R. Thompson, J. A. Burrow, P. J. Shah, J. Slagle, E. S. Harper, A. Van Rynbach, I. Agha, and M. S. Mills, “Artificial neural network discovery of a switchable metasurface reflector,” Opt. Express 28(17), 24629 (2020). [CrossRef]

38. Z. Ni, S. Mou, T. Zhou, and Z. Cheng, “Broader color gamut of color-modulating optical coating display based on indium tin oxide and phase change materials,” Appl. Opt. 57(13), 3385–3389 (2018). [CrossRef]

39. H. Liu, W. Dong, H. Wang, L. Lu, Q. Ruan, Y. S. Tan, R. E. Simpson, and J. K. W. Yang, “Rewritable color nanoprints in antimony trisulfide films,” Sci. Adv. 6(51), 7171–7187 (2020). [CrossRef]

40. X. Yin, M. Schäferling, A. K. U. Michel, A. Tittl, M. Wuttig, T. Taubner, and H. Giessen, “Active chiral plasmonics,” Nano Lett. 15(7), 4255–4260 (2015). [CrossRef]

41. T. Cao, X. Zhang, W. Dong, L. Lu, X. Zhou, X. Zhuang, J. Deng, X. Cheng, G. Li, and R. E. Simpson, “Tuneable thermal emission using chalcogenide metasurface,” Adv. Opt. Mater. 6(16), 1800169 (2018). [CrossRef]

42. Y. Qu, Q. Li, L. Cai, and M. Qiu, “Polarization switching of thermal emissions based on plasmonic structures incorporating phase-changing material Ge₂Sb₂Te₅,” Opt. Mater. Express 8(8), 2312 (2018). [CrossRef]

43. K. Du, L. Cai, H. Luo, Y. Lu, J. Tian, Y. Qu, P. Ghosh, Y. Lyu, Z. Cheng, M. Qiu, and Q. Li, “Wavelength-tunable mid-infrared thermal emitters with a non-volatile phase changing material,” Nanoscale 10(9), 4415–4420 (2018). [CrossRef]

44. Y. Zhang, J. Li, J. B. Chou, Z. Fang, A. Yadav, H. Lin, Q. Du, J. Michon, Z. Han, Y. Huang, H. Zheng, T. Gu, V. Liberman, K. Richardson, and J. Hu, “Broadband transparent optical phase change materials,” in 2017 Conference on Lasers and Electro-Optics, CLEO 2017 - Proceedings (OSA, 2017), Vol. 2017-Janua, pp. 1–2.

45. Y. Zhang, J. B. Chou, J. Li, H. Li, Q. Du, A. Yadav, S. Zhou, M. Y. Shalaginov, Z. Fang, H. Zhong, C. Roberts, P. Robinson, B. Bohlin, C. Ríos, H. Lin, M. Kang, T. Gu, J. Warner, V. Liberman, K. Richardson, and J. Hu, “Broadband transparent optical phase change materials for high-performance nonvolatile photonics,” Nat. Commun. 10(1), 4279 (2019). [CrossRef]

46. W. Dong, H. Liu, J. K. Behera, L. Lu, R. J. H. Ng, K. V. Sreekanth, X. Zhou, J. K. W. Yang, and R. E. Simpson, “Wide bandgap phase change material tuned visible photonics,” Adv. Funct. Mater. 29(6), 1806181 (2019). [CrossRef]

47. M. Delaney, I. Zeimpekis, D. Lawson, D. W. Hewak, and O. L. Muskens, “A new family of ultralow loss reversible phase-change materials for photonic integrated circuits: Sb₂S₃ and Sb₂Se₃,” Adv. Funct. Mater. 30(36), 2002447 (2020). [CrossRef]

48. D. Sahoo and R. Naik, “GSST phase change materials and its utilization in optoelectronic devices: A review,” Mater. Res. Bull. 148, 111679 (2022). [CrossRef]

49. K. Aryana, Y. Zhang, J. A. Tomko, M. S. Bin Hoque, E. R. Hoglund, D. H. Olson, J. Nag, J. C. Read, C. Ríos, J. Hu, and P. E. Hopkins, “Suppressed electronic contribution in thermal conductivity of Ge₂Sb₂Se₄Te,” Nat. Commun. 12(1), 7187 (2021). [CrossRef]

50. S. X. Gan, C. K. Lai, W. Y. Chong, D. Y. Choi, S. Madden, and H. Ahmad, “Optical phase transition of Ge₂Sb₂Se₄Te₁ thin film using low absorption wavelength in the 1550 nm window,” Opt. Mater. (Amsterdam, Neth.) 120, 111450 (2021). [CrossRef]

51. A. Forouzmand and H. Mosallaei, “Dynamic beam control via Mie-resonance based phase-change metasurface: a theoretical investigation,” Opt. Express 26(14), 17948 (2018). [CrossRef]

52. M. Delaney, I. Zeimpekis, H. Du, X. Yan, M. Banakar, D. J. Thomson, D. W. Hewak, and O. L. Muskens, “Nonvolatile programmable silicon photonics using an ultralow-loss Sb₂Se₃ phase change material,” Sci. Adv. 7(25), 3500–3516 (2021). [CrossRef]

53. C. Ríos, Q. Du, Y. Zhang, C.-C. Popescu, M. Y. Shalaginov, P. Miller, C. Roberts, M. Kang, K. A. Richardson, T. Gu, S. A. Vitale, and J. Hu, “Ultra-compact nonvolatile photonics based on electrically reprogrammable transparent phase change materials,” arXiv:2105.06010 (2021).

54. O. Hemmatyar, S. Abdollahramezani, I. Zeimpekis, S. Lepeshov, A. Krasnok, A. I. Khan, K. M. Neilson, C. Teichrib, T. Brown, E. Pop, D. W. Hewak, M. Wuttig, A. Alu, O. L. Muskens, and A. Adibi, “Enhanced meta-displays using advanced phase-change materials,” arXiv:2107.12159 (2021).

55. L. Lu, Z. Dong, F. Tijiptoharsono, R. J. H. Ng, H. Wang, S. D. Rezaei, Y. Wang, H. S. Leong, P. C. Lim, J. K. W. Yang, and R. E. Simpson, “Reversible tuning of Mie resonances in the visible spectrum,” ACS Nano 15(12), 19722–19732 (2021). [CrossRef]

56. Z. Fang, J. Zheng, A. Saxena, J. Whitehead, Y. Chen, and A. Majumdar, “Non-volatile reconfigurable integrated photonics enabled by broadband low-loss phase change material,” Adv. Opt. Mater. 9(9), 2002049 (2021). [CrossRef]

57. E. Gourvest, B. Pelissier, C. Vallée, A. Roule, S. Lhostis, and S. Maitrejean, “Impact of oxidation on Ge₂Sb₂Te₅ and GeTe phase-change properties,” J. Electrochem. Soc. 159(4), H373–H377 (2012). [CrossRef]

58. S. Tripathi, P. Kotula, M. K. Singh, C. Ghosh, G. Bakan, H. Silva, and C. B. Carter, “Role of oxygen on chemical segregation in uncapped Ge₂Sb₂Te₅ thin films on silicon nitride,” ECS J. Solid State Sci. Technol. 9(5), 054007 (2020). [CrossRef]

59. L. Krusin-Elbaum, C. C. Jr, K. N. Chen, M. Copel, D. W. Abraham, K. B. Reuter, S. M. Rossnagel, J. Bruley, and V. R. Deline, “Evidence for segregation of Te in Ge₂Sb₂Te₅ films: Effect on the “phase-change” stress,” Appl. Phys. Lett. 90(14), 141902 (2007). [CrossRef]

60. M. Agati, C. Gay, D. Benoit, and A. Claverie, “Effects of surface oxidation on the crystallization characteristics of Ge-rich Ge-Sb-Te alloys thin films,” Appl. Surf. Sci. 518, 146227 (2020). [CrossRef]

61. L. V. Yashina, R. Püttner, V. S. Neudachina, T. S. Zyubina, V. I. Shtanov, and M. V. Poygin, “X-ray photoelectron studies of clean and oxidized α-GeTe (111) surfaces,” J. Appl. Phys. 103(9), 094909 (2008). [CrossRef]

62. Y. Canvel, S. Lagrasta, C. Boixaderas, S. Barnola, Y. Mazel, and E. Martinez, “Study of Ge-rich GeSbTe etching process with different halogen plasmas,” J. Vac. Sci. Technol., A 37(3), 031302 (2019). [CrossRef]

63. R. Golovchak, Y. G. Choi, S. Kozyukhin, Y. Chigirinsky, A. Kovalskiy, P. Xiong-Skiba, J. Trimble, R. Pafchek, and H. Jain, “Oxygen incorporation into GST phase-change memory matrix,” Appl. Surf. Sci. 332, 533–541 (2015). [CrossRef]

64. Y. Kim, S. A. Park, J. H. Baeck, M. K. Noh, K. Jeong, M.-H. Cho, H. M. Park, M. K. Lee, E. J. Jeong, D.-H. Ko, and H. J. Shin, “Phase separation of a Ge₂Sb₂Te₅ alloy in the transition from an amorphous structure to crystalline structures,” J. Vac. Sci. Technol., A 24(4), 929–933 (2006). [CrossRef]

65. R. Padilla, A. Aracena, and M. C. Ruiza, “Kinetics of stibnite (Sb₂S₃) oxidation at roasting temperatures,” J. Min. Metall. Sect. B Metall. 50(2), 127–132 (2014). [CrossRef]

66. R. Padilla, G. Ramírez, and M. C. Ruiz, “High-temperature volatilization mechanism of stibnite in nitrogen-oxygen atmospheres,” Metall. Mater. Trans. B 41(6), 1284–1292 (2010). [CrossRef]

67. Ž. Živković, N. Štrbac, D. Živkovič, D. Grujičić, and B. Boyanov, “Kinetics and mechanism of Sb₂Se₃ oxidation process,” Thermochim. Acta 383(1-2), 137–143 (2002). [CrossRef]

68. N. Fleck, O. S. Hutter, L. J. Phillips, H. Shiel, T. D. C. Hobson, V. R. Dhanak, T. D. Veal, F. Jäckel, K. Durose, and J. D. Major, “How oxygen exposure improves the back contact and performance of antimony selenide solar cells,” ACS Appl. Mater. Interfaces 12(47), 52595–52602 (2020). [CrossRef]

69. A. Mavlonov, T. Razykov, F. Raziq, J. Gan, J. Chantana, Y. Kawano, T. Nishimura, H. Wei, A. Zakutayev, T. Minemoto, X. Zu, S. Li, and L. Qiao, “A review of Sb₂Se₃ photovoltaic absorber materials and thin-film solar cells,” Sol. Energy 201, 227–246 (2020). [CrossRef]

70. E. L. Chen, “Effects of plasma etching on GeSbTe compositional control,” University of California Los Angeles (2020).

71. Z. Zhang, J. Pan, Y. L. Foo, L. W. W. Fang, Y. C. Yeo, R. Zhao, L. Shi, and T. C. Chong, “Effective method for preparation of oxide-free Ge₂Sb₂Te₅ surface: An X-ray photoelectron spectroscopy study,” Appl. Surf. Sci. 256(24), 7696–7699 (2010). [CrossRef]

72. H. Y. Cheng, C. A. Jong, R. J. Chung, T. S. Chin, and R. T. Huang, “Wet etching of Ge₂Sb₂Te₅ films and switching properties of resultant phase change memory cells,” Semicond. Sci. Technol. 20(11), 1111–1115 (2005). [CrossRef]

73. L. Wang, B. Liu, Z. Song, S. Feng, Y. Xiang, and F. Zhang, “Basic wet-etching solutions for Ge₂Sb₂Te₅ phase change material,” J. Electrochem. Soc. 157(4), H470 (2010). [CrossRef]

74. M. Shen, T. Lill, N. Altieri, J. Hoang, S. Chiou, J. Sims, A. McKerrow, R. Dylewicz, E. Chen, H. Razavi, and J. P. Chang, “Review on recent progress in patterning phase change materials,” J. Vac. Sci. Technol., A 38(6), 060802 (2020). [CrossRef]

75. Y. Canvel, S. Lagrasta, C. Boixaderas, S. Barnola, Y. Mazel, K. Dabertrand, and E. Martinez, “Modification of Ge-rich GeSbTe surface during the patterning process of phase-change memories,” Microelectron. Eng. 221, 111183 (2020). [CrossRef]

76. M. H. Jang, S. J. Park, D. H. Lim, M. H. Cho, K. H. Do, D. H. Ko, and H. C. Sohn, “Phase change behavior in oxygen-incorporated Ge₂Sb₂Te₅ films,” Appl. Phys. Lett. 95(1), 012102 (2009). [CrossRef]

77. T. Li, J. Shen, L. Wu, Z. Song, S. Lv, D. Cai, S. Zhang, T. Guo, S. Song, and M. Zhu, “Atomic-Scale Observation of Carbon Distribution in High-Performance Carbon-Doped Ge₂Sb₂Te₅ and Its Influence on Crystallization Behavior,” J. Phys. Chem. C 123(21), 13377–13384 (2019). [CrossRef]

78. S. A. Song, W. Zhang, H. Sik Jeong, J. G. Kim, and Y. J. Kim, “In situ dynamic HR-TEM and EELS study on phase transitions of Ge₂Sb₂Te₅ chalcogenides,” Ultramicroscopy 108(11), 1408–1419 (2008). [CrossRef]

79. J. González-Hernández, P. Herrera-Fierro, B. Chao, Y. Kovalenko, E. Morales-Sánchez, and E. Prokhorov, “Structure of oxygen-doped Ge:Sb:Te films,” Thin Solid Films 503(1-2), 13–17 (2006). [CrossRef]

80. M. C. Jung, Y. M. Lee, H. D. Kim, M. G. Kim, H. J. Shin, K. H. Kim, S. A. Song, H. S. Jeong, C. H. Ko, and M. Han, “Ge nitride formation in N-doped amorphous Ge2Sb 2Te5,” Appl. Phys. Lett. 91(8), 083514 (2007). [CrossRef]

81. C. Rivera-Rodríguez, E. Prokhorov, G. Trapaga, E. Morales-Sánchez, M. Hernandez-Landaverde, Y. Kovalenko, and J. González-Hernández, “Mechanism of crystallization of oxygen-doped amorphous Ge 1Sb2Te4 thin films,” J. Appl. Phys. 96(2), 1040–1046 (2004). [CrossRef]

82. B. Kooi, W. Groot, and Jt. De Hosson, “In-situ TEM study of the crystallization of Ge₂Sb₂Te₅,” (n.d.).

83. A. Ebina, M. Hirasaka, and K. Nakatani, “Oxygen doping effect on Ge–Sb–Te phase change optical disks,” J. Vac. Sci. Technol., A 17(6), 3463–3466 (1999). [CrossRef]

84. R. Kojima, T. Kouzaki, T. Matsunaga, and N. Yamada, “Quantitative study of nitrogen doping effect on cyclability of Ge-Sb-Te phase-change optical disks,” in Optical Data Storage (SPIE, 1998), Vol. 3401, pp. 14–23.

85. R. Wang, Z. Song, W. Song, T. Xin, S. Lv, S. Song, and J. Liu, “Phase-change memory based on matched Ge-Te, Sb-Te, and In-Te octahedrons: Improved electrical performances and robust thermal stability,” InfoMat 3(9), 1008–1015 (2021). [CrossRef]

86. T. Guo, S. Song, Y. Zheng, Y. Xue, S. Yan, Y. Liu, T. Li, G. Liu, Y. Wang, Z. Song, M. Qi, and S. Feng, “Excellent thermal stability owing to Ge and C doping in Sb2Te-based high-speed phase-change memory,” Nanotechnology 29(50), 505710 (2018). [CrossRef]

87. Q. Hubert, C. Jahan, A. Toffoli, G. Navarro, S. Chandrashekar, P. Noé, V. Sousa, L. Perniola, J. F. Nodin, A. Persico, S. Maitrejean, A. Roule, E. Henaff, M. Tessaire, P. Zuliani, R. Annunziata, G. Reimbold, G. Pananakakis, and B. De Salvo, “Carbon-doped Ge₂Sb₂Te₅ phase-change memory devices featuring reduced RESET current and power consumption,” Eur. Solid-State Device Res. Conf.286–289 (2012).

88. P. Noé, C. Sabbione, N. Bernier, N. Castellani, F. Fillot, and F. Hippert, “Impact of interfaces on scenario of crystallization of phase change materials,” Acta Mater. 110, 142–148 (2016). [CrossRef]

89. N. Ohshima, “Crystallization of germanium-antimony-tellurium amorphous thin film sandwiched between various dielectric protective films,” J. Appl. Phys. 79(11), 8357–8363 (1996). [CrossRef]

90. F. C. Meldrum and C. O’Shaughnessy, “Crystallization in confinement,” Adv. Mater. 32(31), 2001068 (2020). [CrossRef]

91. R. Pandian, B. J. Kooi, J. T. M. De Hosson, and A. Pauza, “Influence of capping layers on the crystallization of doped SbxTe fast-growth phase-change films,” J. Appl. Phys. 100(12), 123511 (2006). [CrossRef]

92. W. K. Njoroge, H. Dieker, and M. Wuttig, “Influence of dielectric capping layers on the crystallization kinetics of Ag₅In₆Sb₅₉Te₃₀ films,” J. Appl. Phys. 96(5), 2624–2627 (2004). [CrossRef]

93. R. E. Simpson, M. Krbal, P. Fons, A. V. Kolobov, J. Tominaga, T. Uruga, and H. Tanida, “Toward the ultimate limit of phase change in Ge₂Sb₂Te₅,” Nano Lett. 10(2), 414–419 (2010). [CrossRef]

94. S. Raoux, J. L. Jordan-Sweet, and A. J. Kellock, “Crystallization properties of ultrathin phase change films,” J. Appl. Phys. 103(11), 114310 (2008). [CrossRef]

95. W. Zhou, L. Wu, X. Zhou, F. Rao, Z. Song, D. Yao, W. Yin, S. Song, B. Liu, B. Qian, and S. Feng, “High thermal stability and low density variation of carbon-doped Ge₂Sb₂Te₅ for phase-change memory application,” Appl. Phys. Lett. 105(24), 243113 (2014). [CrossRef]

96. W. K. Njoroge, H.-W. Wöltgens, and M. Wuttig, “Density changes upon crystallization of Ge₂Sb_2.04Te_4.74 films,” J. Vac. Sci. Technol., A 20(1), 230–233 (2002). [CrossRef]

97. G. Feng, B. Liu, Z. Song, S. Feng, and B. Chen, “Reactive ion etching of Ge₂Sb₂Te₅ in CHF₃/O₂ plasma for nonvolatile phase-change memory device,” Electrochem. Solid-State Lett. 10(5), D47–50 (2007). [CrossRef]

98. S. K. Kang, M. H. Jeon, J. Y. Park, M. S. Jhon, and G. Y. Yeom, “Etch damage of Ge₂Sb₂Te₅ for different halogen gases,” Jpn. J. Appl. Phys. 50(8), 086501 (2011). [CrossRef]

99. S. K. Kang, J. S. Oh, B. J. Park, S. W. Kim, J. T. Lim, G. Y. Yeom, C. J. Kang, and G. J. Min, “X-ray photoelectron spectroscopic study of Ge₂Sb₂Te₅ etched by fluorocarbon inductively coupled plasmas,” Appl. Phys. Lett. 93(4), 043126 (2008). [CrossRef]

100. S.-K. Kang, M. H. Jeon, J. Y. Park, G. Y. Yeom, M. S. Jhon, B. W. Koo, and Y. W. Kim, “Effect of Halogen-Based Neutral Beam on the Etching of Ge₂Sb₂Te₅,” J. Electrochem. Soc. 158(8), H768 (2011). [CrossRef]

101. D. Gao, B. Liu, Y. Li, Z. Song, W. Ren, J. Li, Z. Xu, S. Lü, N. Zhu, J. Ren, Y. Zhan, H. Wu, and S. Feng, “The effect of oxygen plasma ashing on the resistance of TiN bottom electrode for phase change memory,” J. Semicond. 36(5), 056001 (2015). [CrossRef]

102. J. H. Park, J. H. Kim, D. H. Ko, Z. Wu, D. H. Ahn, S. O. Park, and K. H. Hwang, “Use of NH3 etchant for voids suppression to enhance set cycles in CGeSbTe-based phase change memory devices,” Thin Solid Films 616, 502–506 (2016). [CrossRef]

103. W. Kim, M. Brightsky, T. Masuda, N. Sosa, S. Kim, R. Bruce, F. Carta, G. Fraczak, H. Y. Cheng, A. Ray, Y. Zhu, H. L. Lung, K. Suu, and C. Lam, “ALD-based confined PCM with a metallic liner toward unlimited endurance,” in Technical Digest - International Electron Devices Meeting, IEDM (Institute of Electrical and Electronics Engineers Inc., 2017), pp. 4.2.1–4.2.4.

104. Y. Zhang, C. Fowler, J. Liang, B. Azhar, M. Y. Shalaginov, S. Deckoff-Jones, S. An, J. B. Chou, C. M. Roberts, V. Liberman, M. Kang, C. Ríos, K. A. Richardson, C. Rivero-Baleine, T. Gu, H. Zhang, and J. Hu, “Electrically reconfigurable non-volatile metasurface using low-loss optical phase-change material,” Nat. Nanotechnol. 16(6), 661–666 (2021). [CrossRef]

105. S. Lepeshov and A. Krasnok, “Tunable phase-change metasurfaces,” Nat. Nanotechnol. 16(6), 615–616 (2021). [CrossRef]

106. H. Taghinejad, S. Abdollahramezani, A. A. Eftekhar, T. Fan, A. H. Hosseinnia, O. Hemmatyar, A. Eshaghian Dorche, A. Gallmon, and A. Adibi, “ITO-based microheaters for reversible multi-stage switching of phase-change materials: towards miniaturized beyond-binary reconfigurable integrated photonics,” Opt. Express 29(13), 20449 (2021). [CrossRef]

107. C. Ríos, Y. Zhang, M. Y. Shalaginov, S. Deckoff-Jones, H. Wang, S. An, H. Zhang, M. Kang, K. A. Richardson, C. Roberts, J. B. Chou, V. Liberman, S. A. Vitale, J. Kong, T. Gu, and J. Hu, “Multi-level electro-thermal switching of optical phase-change materials using graphene,” Adv. Photo. Res. 2(1), 2000034 (2021). [CrossRef]

108. J. Zheng, S. Zhu, P. Xu, S. Dunham, and A. Majumdar, “Modeling electrical switching of nonvolatile phase-change integrated nanophotonic structures with graphene heaters,” ACS Appl. Mater. Interfaces 12(19), 21827–21836 (2020). [CrossRef]

109. H. Zhang, L. Zhou, L. Lu, J. Xu, N. Wang, H. Hu, B. M. A. Rahman, Z. Zhou, and J. Chen, “Miniature multilevel optical memristive switch using phase change material,” ACS Photonics 6(9), 2205–2212 (2019). [CrossRef]

110. J. Zheng, Z. Fang, C. Wu, S. Zhu, P. Xu, J. K. Doylend, S. Deshmukh, E. Pop, S. Dunham, M. Li, and A. Majumdar, “Nonvolatile electrically reconfigurable integrated photonic switch enabled by a silicon PIN diode heater,” Adv. Mater. 32(31), 2001218 (2020). [CrossRef]

111. H. Zhang, L. Zhou, J. Xu, N. Wang, H. Hu, L. Lu, B. M. A. Rahman, and J. Chen, “Nonvolatile waveguide transmission tuning with electrically-driven ultra-small GST phase-change material,” Sci. Bull. 64(11), 782–789 (2019). [CrossRef]

112. Y. Zhang, C. Ríos, M. Y. Shalaginov, M. Li, A. Majumdar, T. Gu, and J. Hu, “Myths and truths about optical phase change materials: a perspective,” Appl. Phys. Lett. 118(21), 210501 (2021). [CrossRef]

113. K. Gao, K. Du, S. Tian, H. Wang, L. Zhang, Y. Guo, B. Luo, W. Zhang, and T. Mei, “Intermediate phase-change states with improved cycling durability of Sb2S3 by femtosecond multi-pulse laser irradiation,” Adv. Funct. Mater. 31(35), 2103327 (2021). [CrossRef]

114. L. Goux, D. T. Castro, G. A. M. Hurkx, J. G. Lisoni, R. Delhougne, D. J. Gravesteijn, K. Attenborough, and D. J. Wouters, “Degradation of the reset switching during endurance testing of a phase-change line cell,” IEEE Trans. Electron Devices 56(2), 354–358 (2009). [CrossRef]

115. P. Y. Du, J. Y. Wu, T. H. Hsu, M. H. Lee, T. Y. Wang, H. Y. Cheng, E. K. Lai, S. C. Lai, H. L. Lung, S. B. Kim, M. J. BrightSky, Y. Zhu, S. Mittal, R. Cheek, S. Raoux, E. A. Joseph, A. Schrott, J. Li, and C. Lam, “The impact of melting during reset operation on the reliability of phase change memory,” IEEE Int. Reliab. Phys. Symp. Proc. (2012).

116. S. B. Kim, G. W. Burr, W. Kim, and S. W. Nam, “Phase-change memory cycling endurance,” MRS Bull. 44(09), 710–714 (2019). [CrossRef]

117. C. F. Chen, A. Schrott, M. H. Lee, S. Raoux, Y. H. Shih, M. Breitwisch, F. H. Baumann, E. K. Lai, T. M. Shaw, P. Flaitz, R. Cheek, E. A. Joseph, S. H. Chen, B. Rajendran, H. L. Lung, and C. Lam, “Endurance improvement of Ge₂Sb₂Te₅-based phase change memory,” 2009 IEEE Int. Mem. Work. IMW ‘09 (2009).

118. K. Do, D. Lee, D. H. Ko, H. Sohn, and M. H. Cho, “TEM study on volume changes and void formation in Ge₂Sb₂Te₅ films, with repeated phase changes,” Electrochem. Solid-State Lett. 13(8), H284 (2010). [CrossRef]

119. S. Ho Oh, K. Baek, S. Kyu Son, K. Song, J. Won Oh, S.-J. Jeon, W. Kim, J. Hee Yoo, and K. Jeung Lee, “In situ TEM observation of void formation and migration in phase change memory devices with confined nanoscale Ge₂Sb₂Te₅,” Nanoscale Adv. 2(9), 3841–3848 (2020). [CrossRef]

120. A. Redaelli, Phase Change Memory: Device Physics, Reliability and Applications (Springer International Publishing, 2017).

121. M. B. Sky, N. Sosa, T. Masuda, W. Kim, S. Kim, A. Ray, R. Bruce, J. Gonsalves, Y. Zhu, K. Suu, and C. Lam, “Crystalline-as-deposited ALD phase change material confined PCM cell for high density storage class memory,” in Technical Digest - International Electron Devices Meeting, IEDM (Institute of Electrical and Electronics Engineers Inc., 2015), Vol. 2016-Febru, pp. 3.6.1–3.6.4.

122. W. Kim, S. Kim, R. Bruce, F. Carta, G. Fraczak, A. Ray, C. Lam, M. Brightsky, Y. Zhu, T. Masuda, K. Suu, Y. Xie, Y. Kim, and J. J. Cha, “Reliability benefits of a metallic liner in confined PCM,” in IEEE International Reliability Physics Symposium Proceedings (Institute of Electrical and Electronics Engineers Inc., 2018), Vol. 2018-March, pp. 6D.51–6D.55.

123. Y. Xie, W. Kim, Y. Kim, S. Kim, J. Gonsalves, M. BrightSky, C. Lam, Y. Zhu, J. J. Cha, Y. Xie, J. J. Cha, W. Kim, S. Kim, J. Gonsalves, M. BrightSky, C. Lam, Y. Zhu, and Y. Kim, “Self-healing of a confined phase change memory device with a metallic surfactant layer,” Adv. Mater. 30(9), 1705587 (2018). [CrossRef]

124. J.-B. Park, G.-S. Park, H.-S. Baik, J.-H. Lee, H. Jeong, and K. Kim, “Phase-change behavior of stoichiometric Ge₂Sb₂Te₅ in phase-change random access memory,” J. Electrochem. Soc. 154(3), H139 (2007). [CrossRef]

125. C. Kim, D. Kang, T. Y. Lee, K. H. P. Kim, Y. S. Kang, J. Lee, S. W. Nam, K. B. Kim, and Y. Khang, “Direct evidence of phase separation in Ge₂Sb₂Te₅ in phase change memory devices,” Appl. Phys. Lett. 94(19), 193504 (2009). [CrossRef]

126. T.-Y. Yang, I.-M. Park, H.-Y. You, S.-H. Oh, K.-W. Yi, and Y.-C. Joo, “Change of damage mechanism by the frequency of applied pulsed DC in the Ge₂Sb₂Te₅ Line,” J. Electrochem. Soc. 156(8), H617 (2009). [CrossRef]

127. T. Y. Yang, I. M. Park, B. J. Kim, and Y. C. Joo, “Atomic migration in molten and crystalline Ge₂Sb₂Te₅ under high electric field,” Appl. Phys. Lett. 95(3), 032104 (2009). [CrossRef]

128. Y. H. Huang, C. H. Hang, Y. J. Huang, and T. E. Hsieh, “Electromigration behaviors of Ge₂Sb₂Te₅ chalcogenide thin films under DC bias,” J. Alloys Compd. 580, 449–456 (2013). [CrossRef]

129. S. W. Nam, C. Kim, M. H. Kwon, H. S. Lee, J. S. Wi, D. Lee, T. Y. Lee, Y. Khang, and K. B. Kim, “Phase separation behavior of Ge₂Sb₂Te₅ line structure during electrical stress biasing,” Appl. Phys. Lett. 92(11), 111913 (2008). [CrossRef]

130. S. Duhr and D. Braun, “Why molecules move along a temperature gradient,” Proc. Natl. Acad. Sci. U. S. A. 103(52), 19678–19682 (2006). [CrossRef]

131. L. Crespi, A. L. Lacaita, M. Boniardi, E. Varesi, A. Ghetti, A. Redaelli, and G. D’Arrigo, “Modeling of atomic migration phenomena in phase change memory devices,” in 2015 IEEE 7th International Memory Workshop, IMW 2015 (Institute of Electrical and Electronics Engineers Inc., 2015).

132. Y. J. Park, T. Y. Yang, J. Y. Cho, S. Y. Lee, and Y. C. Joo, “Electrical current-induced gradual failure of crystalline Ge₂Sb₂Te₅ for phase-change memory,” Appl. Phys. Lett. 103(7), 073503 (2013). [CrossRef]

133. P. Yeoh, Y. Ma, D. A. Cullen, J. A. Bain, and M. Skowronski, “Thermal-gradient-driven elemental segregation in Ge₂Sb₂Te₅ phase change memory cells,” Appl. Phys. Lett. 114(16), 163507 (2019). [CrossRef]

134. G. Novielli, A. Ghetti, E. Varesi, A. Mauri, and R. Sacco, “Atomic migration in phase change materials,” Tech. Dig. - Int. Electron Devices Meet. IEDM (2013).

135. T. Y. Yang, J. Y. Cho, Y. J. Park, and Y. C. Joo, “Driving forces for elemental demixing of GeSbTe in phase-change memory: computational study to design a durable device,” Curr. Appl. Phys. 13(7), 1426–1432 (2013). [CrossRef]

136. A. Padilla, G. W. Burr, C. T. Rettner, T. Topuria, P. M. Rice, B. Jackson, K. Virwani, A. J. Kellock, D. Dupouy, A. Debunne, R. M. Shelby, K. Gopalakrishnan, R. S. Shenoy, and B. N. Kurdi, “Voltage polarity effects in Ge₂Sb₂Te₅-based phase change memory devices,” J. Appl. Phys. 110(5), 054501 (2011). [CrossRef]

137. I. M. Park, J. K. Jung, S. O. Ryu, K. J. Choi, B. G. Yu, Y. B. Park, S. M. Han, and Y. C. Joo, “Thermomechanical properties and mechanical stresses of Ge₂Sb₂Te₅ films in phase-change random access memory,” Thin Solid Films 517(2), 848–852 (2008). [CrossRef]

138. A. Debunne, K. Virwani, A. Padilla, G. W. Burr, A. J. Kellock, V. R. Deline, R. M. Shelby, and B. Jackson, “Evidence of crystallization–induced segregation in the phase change material Te-rich GST,” J. Electrochem. Soc. 158(10), H965 (2011). [CrossRef]

139. W. Han, K. Zhao, C. Pan, Y. Yuan, Y. Zhao, Z. Cheng, and M. Wang, “Fabrication of Ge₂Sb₂Te₅ crystal micro/nanostructures through single-shot Gaussian-shape femtosecond laser pulse irradiation,” Opt. Express 28(17), 25250 (2020). [CrossRef]

140. C. V. Thompson, “Solid-state dewetting of thin films,” Annu. Rev. Mater. Res. 42(1), 399–434 (2012). [CrossRef]

141. M. G. Herrmann, R. P. Stoffel, I. Sergueev, H. C. Wille, O. Leupold, M. Ait Haddouch, G. Sala, D. L. Abernathy, J. Voigt, R. P. Hermann, R. Dronskowski, and K. Friese, “Lattice dynamics of Sb₂Se₃ from inelastic neutron and x-ray scattering,” Phys. Status Solidi B 257(6), 2000063 (2020). [CrossRef]

142. C. K. Gan, J. R. Soh, and Y. Liu, “Large anharmonic effect and thermal expansion anisotropy of metal chalcogenides: The case of antimony sulfide,” Phys. Rev. B 92(23), 235202 (2015). [CrossRef]

143. H. Koc, A. M. Mamedov, E. Deligoz, and H. Ozisik, “First principles prediction of the elastic, electronic, and optical properties of Sb₂S₃ and Sb₂Se₃ compounds,” Solid State Sci. 14(8), 1211–1220 (2012). [CrossRef]

144. P. Hosseini, C. D. Wright, and H. Bhaskaran, “An optoelectronic framework enabled by low-dimensional phase-change films,” Nature 511, 206–211 (2014). [CrossRef]

145. M. Miscuglio, J. Meng, O. Yesiliurt, Y. Zhang, L. J. Prokopeva, A. Mehrabian, J. Hu, A. V. Kildishev, and V. J. Sorger, “Intelligent edge processing with photonic multilevel memory,” in OSA Advanced Photonics Congress (AP) 2020 (IPR, NP, NOMA, Networks, PVLED, PSC, SPPCom, SOF) (2020), Paper IM2A.4 (The Optical Society, 2020), p. IM2A.4.

146. J. Moon, H. Seo, K. K. Son, E. Yalon, K. Lee, E. Flores, G. Candia, and E. Pop, “Reconfigurable infrared spectral imaging with phase change materials,” in SPIE Proceedings (SPIE, 2019), Vol. 10982, p. 32.

147. E. Yalon, I. M. Datye, J. S. Moon, K. A. Son, K. Lee, and E. Pop, “Energy-efficient indirectly heated phase change RF switch,” IEEE Electron Device Lett. 40(3), 455–458 (2019). [CrossRef]

148. M. Wuttig and N. Yamada, “Phase-change materials for rewriteable data storage,” Nat. Mater. 6(11), 824–832 (2007). [CrossRef]

149. R. Bez, P. Cappelletti, G. Servalli, and A. Pirovano, “Phase change memories have taken the field,” 2013 5th IEEE Int. Mem. Work. IMW 201313–16 (2013).

150. Y. Zhang and J. Hu, “Reconfigurable optics-a phase change for the better,” Am. Ceram. Soc. Bull. 99, 36–37 (2020). [CrossRef]

Endurance of chalcogenide optical phase change materials: a review

Abstract

Corrections

1. Introduction

2. On the various factors influencing optical PCM switching properties

2.1 Which phase change materials?

2.2 Influence of oxidation on PCM

2.3 Wet cleaning of the oxide layer

2.4 Influence of doping (lightweight elements: N, O, C)

2.5 Influence of encapsulation on crystallization behavior

2.6 Influence of plasma chemistry during the etching step

2.7 Influence of ion implantation on phase change performance during cycling

3. Making sense of empirical PCM performance through failure mechanisms

3.1 Switching mechanisms

3.2 Failure mechanisms

3.3 Empirical PCM endurance in photonic applications and possible limitations

4. Design rules for optimal PCM endurance in optical applications

5. Conclusion

Funding

Disclosures

Data availability

References

Data availability

Cited By

Figures (12)

Tables (3)

Equations (3)

Optical Materials Express

Plasma	Cl₂/Ar	SF₆/Ar	CF₄	H₂	CH₄	CH₄/Ar	CH₄/N₂
% Halogen or carbon	10% Cl	57% F	35% F	-	40% C	78% C	33% C
ΔT_c	10	-5	N/A	15	20	50	15