Air‐Stable Atomically Encapsulated Crystalline‐Crystalline Phase Transitions in In2Se3

The layered semiconductor In2Se3 has a low temperature crystalline–crystalline (α → β) phase transformation with distinct electrical properties that make it a promising candidate for phase change memory. Here, using scanning tunneling microscopy, correlative in situ micro‐Raman, and electrical measurements, it is shown that the β phase can persist in bulk crystals at room temperature in non‐oxidative environments. Of particular note, the stability of β phase crystals in ambient conditions under encapsulation of graphene and similar passivation layers, is reported for the first time. The strategy of encapsulation to ensure the persistence of β phase overlaps with efforts to passivate switching materials. It is further demonstrated that degradation from the elevated temperatures required for the phase change is slowed through examination of Raman signatures. These results demonstrate an alternative method of phase manipulation with a new stabilization of β‐In2Se3 in ambient conditions potentially extendable to other polymorphic materials, and the importance of passivation in In2Se3 memory devices.


Introduction
The pursuit of neuromorphic computing technology has garnered significant interest by researchers in the last decade to explore non-volatile memory devices that have faster speeds, [1][2][3] lower power consumption, [4][5][6] and better scalability [7,8] to replace existing von-Neumann architecture.[14] Materials such as In 2 Se 3, [15,16] MoTe 2 [17,18] , and VO 2 , [19,20] exhibit structural transitions between crystalline phases potentially affording an insulator-metal transition at temperatures closer to room temperature.23] Phase engineering of materials for better control of phase transitions and phase stability in bulk and two-dimensional (2D) materials is a critical task for materials science.Structural phase transitions between the various crystalline phases have been extensively studied in bulk In 2 Se 3 , [16,24,25] and previous reports have found a thickness scaling behavior of the phase transformations with increased transformation temperatures and diffe ring stabilities of phases at the thin film limit. [26,27]Furthermore, In 2 Se 3 phase-change materials utilizing the - [28] and - [29] phase change have been demonstrated, however multilevel switching has not been extensively explored within this single material system.Bettering our understanding of the crystalline-crystalline phase transitions allows for identification of conditions required for multiple stable phases which are important for multi-level PCM applications.
Here we report studies of the - transformation in layered In 2 Se 3 fabricated using mechanical exfoliation and stability of the  phase in multi-layers at room temperature under encapsulation.The - transition was triggered by thermal excitations between 200-350 °C and the corresponding changes between the two phases were studied using scanning tunnelling microscopy, in situ Raman spectroscopy, and electrical measurements.By triggering the transition in different environments, we show that  phase, which was previously found to not persist at room temperature, stable at room temperature in non-oxidative environments.Observing the stability of the  phase under vacuum in the scanning tunneling microscope, the reverse transformation occurred once the sample was returned to atmospheric conditions.We then show that  can persist in atmospheric conditions if the flake is encapsulated by another material, in this case we demonstrate this with graphene and hexagonal boron nitride (hBN).We also show the benefits of encapsulation with regards to environmental stability.The accessibility of the  phase at room temperature, with distinct electrical properties than the  phase, provides the basis for multi-level phase-change materials in a single material system.

Results & Discussion
In 2 Se 3 flakes were exfoliated from a bulk crystal in  phase onto stainless steel foil for structural studies by scanning tunnelling microscopy (STM).Figure 1a shows the STM image of the  phase which agrees with simulated STM image of the layer plane of  phase obtained from density functional theory (DFT) calculations Figure S2 (Supporting Information).The lattice constant obtained by lattice indexing the fast Fourier transform (FFT) inset in Figure 1a agrees with lattice constants of bulk polycrystalline  phase In 2 Se 3 . [26]Scanning tunnelling spectroscopy (STS) measurements on the defect-free regions show a band gap of ≈1.27 eV (Figure 1c) agreeing with literature [30] and DFT calculations (Figure S1, Supporting Information).These measurements were conducted after the sample was annealed at 120 °C for 2 h to remove residual molecules and water once inside the STM chamber.This thermal treatment was conducted below the - transition temperature of 200 °C [16] which is confirmed by the previous characterization.The sample was then removed from the STM chamber to another ultra-high vacuum (UHV) chamber and annealed at 200 °C for 2 h to trigger phase change.However, STS measurements indicate the flake remained  phase (Figure 1c).It was reported that the In 2 Se 3 - transition temperature is thickness dependent and increases as thickness decreases. [26]It is estimated that this effect begins ≈300 nm so it would not apply as the sample used was on the order of 300 microns.However, the non-transition is explained as hysteresis in crystalline-crystalline phase changes are expected. [31]he phase change from  to  was achieved after annealing the flake at 300 °C for 2 h. Figure 1b shows an STM image of the  phase surface with the FFT inset.FFT shows a superstructure characteristic in  phase In 2 Se 3 and with the lattice constant extracted agreeing with literature confirming that the sample underwent the phase change to  In 2 Se 3 after annealing at 300 C. [26] Additionally, STS shown in Figure 1c provides electronic characterization of the  phase showing a band gap of ≈0.52 eV. [30]hese samples were conducted after cooling to room temperature, counter to what was previously reported where  phase reverts to  phase in bulk single-crystal In 2 Se 3 similar to our sample. [16]The  phase was stable at room temperature in UHV, as verified by STS measurements, ≈3 months after annealing.However, Raman and photoluminescence spectra of the sample shortly after it was removed from the UHV STM chamber is shown in Figure 1d and e and can be indexed to  phase. [24,32]his matches the Raman and photoluminescence spectra of the bulk -phase In 2 Se 3 crystal from which it was exfoliated seen in Figure 1d and e, contrasting with the Raman and photoluminescence spectra of -phase In 2 Se 3 also included in Figure 1d and e.The present experimental results support previous reports that bulk  phase single-crystal In 2 Se 3 reverts to  phase upon cooling to room temperature, but with the new condition that  phase will be stable under UHV.In-situ studies of the - transformation using micro-Raman measurements under different annealing conditions were conducted to investigate the stability of the  phase.In 2 Se 3 flakes were exfoliated from a bulk crystal onto SiO 2 /Si substrates, and the thickness of the flakes was by different contrasts on the SiO 2 /Si substrate in the optical image with calibration by atomic force microscopy (AFM).Previous studies have shown a thickness dependence to the - transition temperature and stability. [26]The increased transition temperature is believed to be tied to the reduced material dimension, but also the interface of the film and the surrounding environment.To further investigate the effect of the surrounding environment on the stability of the  phase outside of vacuum as seen in the STM studies, samples were loaded into a closed heating stage under ambient, high-purity nitrogen, and high-purity argon (see Methods).The Raman spectrum collected as the same sample was heated and cooled under ambient and then Ar is shown in Figure 2a,b and Figure S3 (Supporting Information).Starting at room temperature, the spectrum of the sample was indexed to  phase and the - transformation can be observed in both conditions between 220 and 240 C identified by the A 1 (LO) phonon mode of -In 2 Se 3 at 204 cm −1 [24] .During the cooling cycle, the reverse transformation is present under ambient conditions once the sample cools to room temperature supported by the sharp  phase A 1 (LO+TO) phonon mode at 104 cm −1 shown in Figure 2a. [24,32]This contrasts with the heating and cooling cycle done under an inert environment in Figure 2b where the -phase phonon modes persist upon cooling to room temperature.This was repeated with samples of similar thicknesses separately under the different environments to the same results.Additionally, this was supported with electrical resistance of a thick flake measured before and after annealing under high-purity nitrogen.The electrical resistance of the sampleafter annealing under ambient conditions is ≈2 orders of magnitude lower than that of the -phase confirming -In 2 Se 3 following the inert annealing.
Figure 2b shows that the stability of the  phase under an inert environment is like the behavior observed in UHV.Likewise, the stability of the sample follows as the  phase will revert to  phase once returned to an ambient environment.Polycrystalline In 2 Se 3 in powder form, has been observed to maintain a mixture of  and  In 2 Se 3 upon cooling to room temperature potentially attributed to the small crystal size.To investigate the phase purity of the samples during the phase transition, Raman mapping and spectra at various points on the sample were collected to confirm the single  phase (Figure S4, Supporting Information).In addition to the in-situ studies under different environments, Raman spectra were collected before and after triggering the - phase transformation in ambient environment, vacuum, in Ar and with a forming gas (4% H 2 in N 2 ) to study the stability of  phase.After annealing the samples in a furnace with the different environments, all samples aside from one annealed in ambient air remained  after annealing but was only stable in ambient air for up to a week.Altogether this suggests that an oxidative environment is responsible for enabling the phase transition as summarized in Figure 4. Thus, passivating the flake with an encapsulating layer was attempted to control the phase of In 2 Se 3 flakes.
Heterostructures with a van der Waal material such as graphene or hexagonal boron nitride encapsulating the In 2 Se 3 flake were fabricated by exfoliating from bulk crystals and transferring onto SiO 2 /Si using a PDMS stamp.Figure 2c shows the Raman spectra of the encapsulated sample.The - transformation is observed at 250 °C and after heating to 325 °C and cooling to room temperature the resulting Raman spectra is indexed to  phase.To ensure the sample was sufficiently thick such that the stability of the  phase was not due to thickness effects, a sample of similar thickness (within 20 nm) was also heated and cooled to show the relaxation of  phase to .The reverse transformation prevented by passivation is further supported by these results as the persistence of the  phase coincided with the encapsulation of the surface of the flake.Additionally, another sample was prepared stacking hBN onto In 2 Se 3 and similar results were obtained showing material is not important but rather coverage of the surface.
Previous studies on the - transformation in In 2 Se 3 show the mechanism for the structural transition is assisted by a shear phonon mode leading to intra-quintuple layer (QL) rearrangement. [28,33]For the reverse - transformation in bulk In 2 Se 3, there is not a direct transformation when cooling to room temperature, but instead a distorted ' phase is formed. [34]This intermediate ' phase then transforms to the original  phase, but this is not the case in thin layers. [26]In 2D layers the reverse transformation is blocked by a large entropy barrier caused by a random distribution of the middle layer Se atoms due to the shear glide & compression. [33]Thus,  phase has been demon-strated to be stable after cooling down to room temperature in thin films (>150 nm down to monolayer), and ' was first shown to appear in thin films after cooling  phase to 180K. [26,35]Later studies show that by introducing Se vacancies in  phase during CVD growth by adding to InSe to the precursor, ' can be grown, and subsequent DFT calculations agree that under high temperature and at least 3% Se vacancy -In 2 Se 3 can spontaneously transform into '-In 2 Se 3 . [36]Additional studies examining the relationship between '-In 2 Se 3 and -In 2 Se 3 found that '-In 2 Se 3 is obtainable from the  phase through electrical stimulus. [37]hey also demonstrated reversible phase transition between  and '-In 2 Se 3 via strain relaxation and tension on a 2D In 2 Se 3 flake which was similarly observed in a thick multi-layer flake (Figure S4, Supporting Information).
We discuss the - transformation in bulk In 2 Se 3 and prevention of the reverse -'- transformation that was observed in Figures 1 and 2. As previously stated, bulk -In 2 Se 3 requires the ' phase to transform back to  phase, but this transformation to ' requires a sufficient Se vacancy.With Se vacancies introduced and at high temperatures '-In 2 Se 3 can be formed.However, an external strain input is required to order the ' phase and achieve the full reverse transformation to -In 2 Se 3. This, in combination with the results from the above STM and Raman experiments suggests that the stability of the  phase can be achieved in bulk and thick layers by preventing the formation of Se vacancies.] This was achieved by heating  phase flakes under non-oxidative conditions such as vacuum or Ar.From larger scale STM images in Figure 3, two -In 2 Se 3 samples, one annealed in vacuum and the other in ambient air, defect densities were calculated.We find the sample annealed in vacuum to have a defect density of 10 10 cm -2 , and the sample annealed in ambient air to have a larger defect density on the order of 10 11 cm -2 supporting the notion that a non-oxidizing environment slows defect formation.and encapsulated bulk.
Figure 4 shows the updated phase transition paths regarding -In 2 Se 3, -In 2 Se 3 , and '-In 2 Se 3 depending on thickness and encapsulation.More studies probing the effect of the oxide on the structure of both  and  In 2 Se 3 could be conducted with TEM.Additionally, substrate effects have been hypothesized and could be investigated such double encapsulating the flake with a bottom layer although studies have been conducted on the effect of substrates on the growth of In 2 Se 3 showing little significance between SiO 2 and Si.
Encapsulation via van der Waals layer passivation has been demonstrated elsewhere to prevent the oxidation of thin film semiconductors for device applications. [40,41]This work demonstrates that the fabrication of In 2 Se 3 /graphene-based devices allows for phase control of  phase In 2 Se 3 at room temperature in thick films.However, improving environmental stability is also a benefit of the encapsulation as sample degradation is measured with spectroscopy after annealing at high temperatures in different environments.Optical images of the encapsulated samples are shown in Figure 5b-f after each annealing.After stacking the graphene flake onto ≈200 nm In 2 Se 3, the In 2 Se 3 flake was transformed into  phase by annealing at 250 °C and cooling to room temperature.In attempt to trigger the reverse - transformation, the heterostructure was heated to higher temperatures in which degradation due to oxidation and heat would occur for bare In 2 Se 3 .In bulk In 2 Se 3 there is a higher temperature phase change from  to  between 350 °C -650 °C. [16]It has also been reported that despite the  phase persisting after cooling to room temperature in single-crystal thin films, the reverse transformation can be triggered with an optical excitation which was estimated to result in a phase change after reaching a temperature of 640 °C. [28]Thus, the encapsulated In 2 Se 3 was heated to temperatures of 450, 600, and 650 °C with the latter being done under Ar flow while the other done in atmospheric conditions.After heating to 450 °C, no phase change occurred as the sample remained in  phase, and no noticeable degradation was observed optically and in the Raman spectra compared to typical damage when heating bare In 2 Se 3 above 300 °C. [25]When heating to 600 °C, degradation was noticeable, but the phase transition was not present.Finally, after heating to 650 °C in Ar, the sample was completely degraded with no Raman signal obtained.While the investigations into a reverse - transition was inconclusive perhaps due to the transformation requiring more than a global thermal excitation, [33] the encapsulation layer was shown to improve the environmental stability of In 2 Se 3 up to high temperatures of 450 °C (Figure 5).

Conclusion
In summary, we demonstrated that the persistence of the  phase in bulk and multi-layer flakes can be achieved by preventing oxidation of the In 2 Se 3 surface even after cooling to room temperature.This was found by annealing the flakes in inert environments including samples under encapsulating layers passivating the surface and observing the persistence of the  phase.The transformation and stability of the  phase was confirmed by STS and Raman spectroscopy.This observation demonstrates encapsulation by graphene as a new viable strategy to ensure a stable -In 2 Se 3 in ambient air and elevated temperatures.Increasing the reliability in which room temperature stable crystalline phases with differing electrical and optical properties in In 2 Se 3 are present can lead to the development of multi-level PCM utilizing In 2 Se 3 .Furthermore, this method of phase control can potentially be extendable to other layered materials with crystallinecrystalline phase transformations.

Experimental Section
Scanning Tunnelling Microscopy: The experiments in this study were carried out in two ultrahigh vacuum (UHV) chambers connected by an in-vacuum sample transfer, with base pressures of 1×10 −10 Torr.A thick In 2 Se 3 flake (Sigma-Aldrich) was mechanically exfoliated with the "Scotchtape" method onto a stainless-steel foil after being mounted in the UHV preparation chamber.The sample was then annealed at 120 °C for 2 h to remove residual water and ensure a clean surface.For STM characterization, the sample was transferred to the STM measurement chamber equipped with an Omicron LT STM.All measurements were performed under UHV conditions and at room temperature.STM images were recorded with a tungsten (W) STM tip with tunnelling parameters: U = 1.0 V and I = 200 pA unless otherwise specified with the tip condition verified on an Au (111) surface.For the STS measurements, a modulation signal of 200 mV at 1 KHz was used.Following the initial characterization after annealing at 120 °C, the procedure was repeated after annealing the sample at 200, 300, and 350 °C.For comparison of defect densities, a thin In 2 Se 3 flake was exfoliated and annealed at 300 o C in ambient conditions before being transferred into the STM.
Crystalline-Crystalline Phase Transition Characterization: Raman spectroscopy was performed in a Renishaw InVia Raman spectrometer using a 532 nm laser excitation.In situ Raman spectra were collected with samples mounted on a Linkam heating stage and heated from room temperature to 350 °C.When operating the Linkam stage under a closed inert atmosphere, the stage was purged with ultra-high purity Argon or Nitrogen at a flow rate of 1.5L mi −1 n and then sealed.Samples were heated and cooled at a rate of 10 °C min −1 and Raman spectrum were collected after 3 min at the target temperature to allow for thermal equilibrium.Electrical resistance was measured with devices that had exfoliated  and  layers without contacts to avoid resistance changes due to annealing using a Cascade probe station with an Agilent 4156 semiconductor parameter analyzer under ambient conditions.

Figure 1 .
Figure 1.Atomically resolved STM images of bulk In 2 Se 3 a) before, -phase and b) after,  phase annealing with FFT inset.Scale bar: 1 nm; c) dI/dV spectrum taken after annealing at various temperatures and cooling to RT (Vsample = 0.5 V, I = 100pA).STS calculated bandgap.Black dashed lines are guides for the eye to mark the bandgap; d,e) Raman and PL spectra of bulk In 2 Se 3 that was stored in air (black), bulk In 2 Se 3 immediately after removing from vacuum in the STM (red), and -In 2 Se 3 (blue).

Figure 2 .
Figure 2. In situ micro-Raman spectra of In 2 Se 3 multi-layers at various temperatures and various environments with a) measured in atmosphere, b) 99.9% purity argon, and c) also measured in atmosphere but with the In 2 Se 3 encapsulated by multi-layer graphene.d) Room temperature I-V characteristics of an -phase film and a -phase film with calculated resistance inset.

Figure 3 .
Figure 3. Large scale STM image (I = 50pA, V = 1 V, scale bar 10 nm) of -In 2 Se 3 annealed in vacuum a) and ambient conditions b).The bright and dark features correspond to single point defects in -In 2 3 .Scale bar, 10 nm.

Figure 5 .
Figure 5. a) Schematic of the In 2 Se 3 film with an encapsulating graphene layer assembled with a viscoelastic poly-dimethyl siloxane (PDMS) stamp via micromanipulator.Comparison of degradation of encapsulated sample after heat treatment for 1 h in the conditions listed via Raman spectra and optical images b-f).