WorksPhysics and Astronomy
Role of seed layer in growing atomically flat TiTe2/Sb2Te3 heterostructure thin films at the wafer scale
Abstract
Chalcogenide phase-change materials (PCMs) are a leading candidate for advanced memory and computing applications. Epitaxial-like growth of chalcogenide thin films at the wafer scale is important to guarantee the homogeneity of the thin film but is challenging with magnetron sputtering, particularly for the growth of phase-change heterostructure (PCH), such as TiTe2/Sb2Te3. In this work, we report how to obtain highly textured TiTe2/Sb2Te3 heterostructure thin films with atomically sharp interfaces on standard silicon substrates. By combining atomic-scale characterization and ab initio simulations, we reveal the critical role of the Sb2Te3 seed layer in forming a continuous Si-Sb-Te mixed transition layer, which provides a wafer-scale flat surface for the subsequent epitaxial-like growth of TiTe2/Sb2Te3 thin film. By gradually reducing the thickness of the seed layer, we determine its critical limit to be ~2 nm. Non-negligible in-plane tensile strain was observed in the TiTe2 slabs due to the lattice mismatch with the adjacent Sb2Te3 ones, suggesting that the chemical interaction across the structural gaps in the heterostructure is stronger than a pure van der Waals interaction. Finally, we outline the potential choices of chalcogenides for atomically flat seed layers on standard silicon substrates, which can be used for wafer-scale synthesis of other high-quality PCM or PCH thin films.
Introduction
Chalcogenide phase-change materials (PCMs) are one of the leading candidates for non-volatile memory and neuromorphic in-memory computing applications [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. The most important family of PCM consists of the Ge-Sb-Te alloys along the GeTe-SbTe pseudo-binary line [13, 14, 15, 16, 17, 18, 19, 20, 21], e.g. GeSbTe (GST), which has been extensively used in electronic, photonic as well as hybrid optoelectronic devices [22, 23, 24, 25, 26, 27, 28, 29, 30]. The basic working principle is the large contrast in electrical resistance or optical transmission between the crystalline and amorphous phases of PCM for memory encoding [1]. Given its metastable nature, the amorphous phase undergoes spontaneous structural relaxation, causing a steady increase in electrical resistance with time, known as the resistance drift [31, 32, 33, 34]. For binary storage, this aging effect is not a serious issue, as the contrast window is enlarged instead of reduced with time. But for the multilevel programming scheme adopted in neuromorphic devices, the drift issue is a major hindrance for high-precision encoding of multiple resistance states [35, 36, 37].
The structural origin of the resistance drift in amorphous PCMs stems from the disappearance of structural defects that are generated during the rapid cooling process and the reinforcement of Peierls-like distortions between pairs of octahedral bonds with time [38, 39, 40, 41]. Scaling down the film thickness to exploit volume effects [42, 43, 44, 45] or alloying PCM with impurities can help reduce the drift coefficient effectively [46, 47, 48, 49]. Another promising approach to suppress drift is based on heterostructures composed of alternating thin layers of two different PCMs or of a PCM and a confinement material (CM). Such heterostructures are typically grown by heterogeneous sputtering or molecular beam epitaxy. Besides the low drift, they also offer low switching energy and fast switching speed. Prominent examples include the GeTe(GST)/SbTe [50, 51, 52, 53, 54, 55, 56, 57, 58, 59] superlattice heterostructures, which are made of two standard PCMs with similar melting point, and the TiTe/SbTe phase-change heterostructures (PCH) [60, 61, 62, 63, 64, 65]. In the latter, TiTe serves as CM due to its much higher melting point than SbTe, effectively reducing the drift coefficient of the PCM layers. More heterostructure combinations were predicted [66] and developed [67, 68, 69, 70] for memory applications with CM = TiTe, MoTe, NiTe or HfTe, and PCM = GeTe, GST, GeSbTe or TaSbTe.
For practical use, it is important to form sharp interfaces between the CM layers and PCM layers during growth. Otherwise, accumulation of structural defects at the interfaces upon extensive cycling could turn the PCM into conventional transition metal doped alloys, causing device degradation [60, 61]. In this work, we report how to deposit TiTe/SbTe heterostructure thin films with atomically sharp interfaces at the wafer scale. The key to obtaining high-quality heterostructure thin films relies on the quality of the seed layer deposition and the in situ heating during subsequent co-sputtering of the heterostructure. We determine the critical (minimum) thickness of the seed layer required to ensure the formation of sharp interfaces on a standard silicon substrate, and provide in-depth understanding of this interface region via atomic-scale structural characterization experiments and ab initio calculations. We also find that the TiTe crystal lattice is under non-negligible tensile strain due to the lattice mismatch with crystalline SbTe. At last, we provide guidelines on the choice of seed layer that can guarantee highly textured thin film deposition.
Experimental and simulation details
The TiTe/SbTe heterostructure thin films were deposited on silicon wafers by magnetron sputtering (AJA, Orion-8). The silicon wafers were cleaned by sonication in acetone, ethanol, and deionized water for 10 min, respectively, and then baked at 80°C for 30 minutes. The base pressure in the deposition chamber was maintained at Torr. The layers were deposited using a (99.99%) target at a pressure of 3 mTorr, and the layers were deposited by co-sputtering with Ti (99.995%) and Te (99.99%) targets at a pressure of 4.7 mTorr. The cross-section specimens of the thin films were prepared using a FEI Helios NanoLab 600i focus ion beam (FIB) system with a Ga ion beam at 30 kV beam energy, and were thinned and polished at 5 kV / 20 pA and 2 kV / 10 pA, respectively. The spherical aberration corrected scanning transmission electron microscopy (STEM) high-angle annular dark-field (HAADF) and energy dispersive X-ray (EDX) mapping experiments were performed on a JEM-ARM300F2 STEM with a probe aberration corrector, operated at 300 kV. The high-resolution transmission electron microscopy (HRTEM) and EDX experiments were performed on a Talos-F200X operated at 200 kV. The atomic force microscopy (AFM) experiments were measured using a SPM-9700HT. The X-ray diffraction (XRD) experiments were performed by a Bruker D8 ADVANCE with CuK radiation () in the range 5°~60° with the scanning step size of 0.02°. The Raman spectra were collected by using a Renishaw inVia Qontor Raman microscope with a solid-state 532 nm laser for the excitation. The laser power was set to 0.25 mW, and the exposure time was 1 s with 100 cycles. Density functional theory (DFT) calculations were carried out using the Vienna Ab-initio Simulation Package (VASP) code [71]. The Perdew-Burke-Ernzerhof (PBE) functional [72], the projector augmented plane-wave (PAW) pseudopotentials [73], and the Grimme’s D3 dispersion correction [74] were applied in the calculations. An energy cutoff of 500 eV was used. We constructed and supercells that both contained 60 atoms, and used k-point mesh of points used to sample the Brillouin zone. The Löwdin charges were calculated using the LOBSTER code [75, 76, 77].
Results and discussion
We deposited several PCH thin films on 7 cm 7 cm silicon wafers following the procedure shown in Figure 1a. Before deposition, the silicon substrate was cleaned using Ar-plasma etching with radio frequency at 35 Watt over 30 minutes in the high vacuum sputtering chamber. This cleaning process is needed to remove the native oxides on top of the substrate, and is frequently termed as the reverse sputtering process [78, 79, 80]. As soon as the etching was completed, a 5 nm seed layer was deposited at room temperature. Next, the substrate was heated at 280 °C for 60 min to fully crystallize the seed layer, forming a trigonal phase with ordered quintuple-layer (QL) blocks (each QL has a thickness of 1 nm) [81, 82, 83]. Subsequently, a PCH unit made of a 5 nm slab, corresponding to 7–8 trilayer (TL) blocks, followed by a 5 nm slab, were deposited via alternate sputtering of the , Ti and Te targets under the in situ heating condition. In total, four PCH units were deposited. As the last step, the PCH thin film with a thickness of 45 nm (including the seed layer) was annealed at 280 °C for another 30 min, and was naturally cooled to room temperature in the high vacuum chamber. The whole thin film was covered with a 10$ nm-thick ZnS-SiO capping layer to prevent oxidation. The cross-sectional structural properties and elemental distributions of this thin film were characterized by combining the HRTEM with EDX experiments. As shown in Figure 1b, the HRTEM image provides an overview of the PCH thin film, which has well-aligned lattice fringes parallel to the substrate. In the corresponding EDX mapping, the Ti (red) and Sb (green) signals are well separated, while the Te (blue) signal is uniformly distributed throughout the PCH thin film. Only the Sb and Te signals were detected within the bottom $\sim$5 nm region in contact with the substrate.

Figure 1. The deposition procedure and characterization of the thin film. (a) The schematic diagrams of the deposition process in four steps. (b) The HRTEM and EDX mapping images of the PCH thin film.
The epitaxial-like growth of the PCH thin film depends on the quality and out-of-plane texture of the seed layer. In Figure 2, we characterized the surface morphology of a seed layer and a full PCH thin film. The optical images of the two thin films on the silicon wafers are presented in Figure 2a and 2c. We selected 16 uniformly distributed regions across the two thin films and measured the surface roughness of each region by using AFM. For each AFM scan, the size of the measured region was set to . Figure 2b presents the AFM images of six regions of the seed-layer sample (the remaining 10 AFM images are included in Figure S1 of the supplementary material). The respective root-mean-square roughness values provided in the figure range from to , indicating a highly smooth surface. The 16 AFM images of the PCH thin films are shown in Figure 2d and Figure S1. The values range from to , which is slightly broader than that of the seed layer but is still smaller than the thickness of a single QL block. Therefore, we conclude that we have obtained a high-quality PCH thin film with smooth surface.

Figure 2. Surface morphology characterization of the seed layer and the PCH thin film. (a) The optical image of the seed layer with a thickness of nm on a silicon wafer. (b) The AFM images taken at 6 different locations on the surface of the seed layer. (c) The optical image of the thin film on a silicon wafer. (d) The AFM images taken at 6 different locations on the surface of the PCH thin film. is the measured root-mean-square roughness.

Figure S1. Statistical histogram of the root-mean-square roughness for 16 different regions, and the AFM images for ten of them. (a) 5 nm thick SbTe seed layer and (b) 45 nm thick SbTe/TiTe film. The of SbTe seed layer and SbTe/TiTe film are observed to be below 5.2 Å and 7.6 Å, respectively.
We carried out XRD and Raman spectroscopy experiments to investigate the structural properties of the obtained PCH thin films. We also prepared pristine and thin films of thickness for comparison, for which a seed layer was used consistently. Figure 3a and Figure S2 show the measured XRD patterns at 16 locations of the PCH thin film. No obvious difference can be observed in these XRD curves. In comparison with the two pristine thin films, the PCH thin film exhibited a series of (00)-oriented diffraction peaks, namely, the (003), (006), (009), (0015) and (0018) peaks of the slabs, and the (001) and (002) peaks of the slabs. Note that the (009) and (002) peaks overlap due to their close peak locations at and , leading to a smaller shoulder in the primary peak of the PCH thin film. These well-defined (00)-oriented diffraction patterns confirm the epitaxial-like growth with out-of-plane texture across the entire wafer.

Figure 3. XRD and Raman characterizations. The (a) XRD and (b) Raman characterizations carried out on various locations of the PCH thin film. The XRD and Raman curves of the pristine thin film and pristine thin film are shown for comparison.

Figure S2. (a) XRD patterns and (b) Raman spectra for ten different sampling spots in SbTe/TiTe thin films.
Figure 3b and Figure S2 present the Raman spectra measurements of the three thin films within a range of 50–180 using a 532 nm solid-state excitation laser. The measured local Raman spectra at 16 locations of the PCH thin film show three major peaks. By comparing with pristine [84, 85, 86], the three vibrational modes are identified as an out-of-plane mode (1) at , an in-plane mode (2) at and a second out-of-plane mode (2) at for the PCH thin film. For the pristine thin film, three major vibrational modes [87] were found at for , for and for , respectively. However, these -related vibrational modes were not visible in the PCH thin film due to the much weaker Raman response. The highly consistently XRD patterns and Raman spectra across the entire thin film are indicative of wafer-scale uniformity.
Next, we conducted STEM-HAADF imaging experiments to gain a better understanding of the atomic interfaces between and in the heterostructure thin film. Figure 4 presents the HAADF images of the pristine , pristine and thin films in the direction. The brightness of each spot is roughly proportional to , where represents the average atomic number of the atomic column along the incident direction of the electron beam [88]. The atomic layers in the blocks exhibit uniform image intensity, owing to the comparable atomic number of Sb () and Te (). In contrast, the central atomic layer of the blocks displays markedly reduced intensity due to the much smaller atomic number of Ti (). The size of the block in the pristine thin film and in the heterostructure thin film is comparable, but for the blocks, the in-plane atomic spacing is enlarged from Å in the pristine thin film to Å in the PCH thin film. This in-plane expansion indicates that the lattice mismatch between and induces a tensile strain in the slabs of the PCH thin film. Similar behavior was observed in thin films, where the tensile strain could reach in the GeTe slab as compared to bulk GeTe [89].

Figure 4. Atomic-scale characterization of the lattice parameters along the direction. The HAADF images of (a) the pristine thin film, (b) the pristine thin film and (c) the thin film.
The observation of lattice expansion in the slab suggests that the alternately stacked and slabs are connected not only by pure van der Waals forces but also additional chemical interaction across the structural gaps. Therefore, the structural gaps in the PCH should also be regarded as “vdW-like” gaps rather than pure vdW gaps, similar to those in heterostructures, and those in trigonal bulk and GST [89, 90, 91]. In a recent work, we revealed the critical role of weak chemical interactions across the vdW-like gaps in stabilizing metavalent bonding (MVB) [92, 93, 94, 95] in trigonal and GST, and explained how to tailor the degree of this in-gap interaction for altering the optical properties via uniaxial strain along the out-of-plane direction [96]. Although pristine is regarded as a vdW material like other transition metal dichalcogenides [97], the inter-slab coupling can be significantly increased by co-sputtering with MVB-type layered solids, which induces in-plane strain affecting the electronic structure [98, 99, 100].
The synthesis of the PCH thin film with wafer-scale uniformity is rendered possible by high-quality deposition of the seed layer. The HRTEM image in Figure 5a shows an overall view of the multiple interfaces between different layers. We focus on the two interfaces indicated by the red and blue arrows, the corresponding atomic-scale HAADF images are shown in Figure 5b and Figure 5c. We used a standard silicon substrate with the crystal orientation that was naturally covered with a – nm layer. The initial Ar-plasma etching only removed native oxygen at the topmost of the substrate, not the full layer. The HAADF image (Figure 5c) and the corresponding EDX maps (Figure 5d–h) revealed a mixed Si-Sb-Te region of nm with no clear atomic columns except for the top three Sb-Te atomic layers. Above this transition region, we observed ordered QL blocks of the seed layer. Clearly, the ordered QL blocks are not directly contacted with the Si-100 surface. This transition region was formed because we conducted deposition of the seed layer immediately after Ar-plasma etching of the substrate. This procedure resulted in an amorphous Si layer of – nm [78, 79, 80] with gradually reduced density from the substrate towards the seed layer (Figure 5g), and the Si atoms intermixed with the subsequent sputtered Sb and Te atoms. It is important to note that the Ar-plasma etching process cannot be subjected to in situ heating, which markedly increases the probability of re-forming . In such a case it is no longer feasible to obtain a flat seed layer for high-quality growth of PCH thin films (Figure S3).

Figure 5. Structural and elemental characterizations of the thin film. (a) The large-scale HRTEM image of the multiple interfaces between the PCH thin film and the silicon substrate. (b) and (c) The atomic-resolution HAADF images of the interfaces marked by the red and yellow arrows in (a). (d)-(h) The corresponding EDX mapping of the transition region shown in (c).

Figure S3. The TEM characterizations of the TiTe/SbTe thin film, for which the growth of the SbTe seed layer was also subjected to in situ heating condition.
The atomic-scale characterization of the seed layer suggests that its thickness () can be further optimized. We repeated the sputtering process shown in Figure 1a but reduced () to nm, nm and finally 0 nm. In the latter case, the first nm layer served as the seed layer. The HRTEM measurements showed that the PCH thin film with the nm seed layer exhibited well-aligned lattice fringes parallel to the substrate, as indicated by the white arrows in Figure 6a. In the other two cases, the lattice fringes of the PCH thin films were no longer flat and smooth, as shown in Figure 6b and Figure 6c. The atomic-scale HAADF images of the seed layer region are shown in Figure 6d-f for the three PCH thin films. The first PCH thin film showed a similar Si-Sb-Te transition region with top 2–3 ordered Sb or Te atomic layers and a full SbTe QL block, consistent with that in the PCH thin film with (SbTe) 5 nm. However, the transition region in the other two PCH thin films failed to promote the formation of well-ordered atomic layers at the top edge. Therefore, the critical minimum thickness of the seed layer is 2 nm for SbTe to generate a flat surface, i.e., a full intact SbTe QL block, for the subsequent growth of high-textured thin films on top. However, it is less suitable to use TiTe as the seed layer, as sketched in Figure 6 g-i.

Figure 6. The critical thickness of the SbTe seed layer. Large-scale HRTEM images of the PCH thin films with SbTe seed layers with a thickness of (a) 2 nm, (b) 1 nm and (c) 0 nm. (d)-(f) Atomic-scale HAADF images of the transition region at the interface with the substrate of the three PCH thin films. (g-i) Sketches of the growth scenarios of the three PCH thin films.
We carried out DFT calculations to gain a better understanding of this phenomenon. As discussed before, the Ar-plasma etching of the silicon substrate leads to an ultrathin amorphous silicon slab with dangling bonds and marginal oxygen at the top edge. We constructed hexagonal supercell models of crystalline TiTe and SbTe and computed the defect formation energy for Si impurities. As shown in Figure 7, each model contains 60 atoms with 1 Si atom being the substitutional defect, and the cell volume and atomic positions of the models were fully relaxed. We calculated the defect formation energy as

Figure 7. Defect energy calculations for Si substitution. The Si, Ti, Sb, and Te atoms are rendered as yellow, red, green, and blue spheres, respectively. For Te substitution, two non-equivalent sites were considered, denoted as Te and Te.
where , , and represent the energy of the supercell with defect, the pristine phase of the supercell without defect, one Si atom in its bulk ground state, and the one atom being substituted (namely, Ti, Sb or Te atom in the their bulk ground state), respectively. For the substitution in the cation sublattice, the value is 2.48 eV to replace one Ti atom with one Si atom in TiTe but 0.79 eV to replace one Sb atom with one Si atom in SbTe. To replace one Te atom in the anion sublattice, the value is 2.08 eV for TiTe, and 1.64 eV (edge Te atom) and 1.99 eV (center Te atom) for SbTe, respectively. Clearly, the defect energies to accommodate Si impurities are smaller in SbTe than in TiTe. We also considered models with Si atoms as interstitial defects, which however yielded much too high energy costs as compared to substitutional ones (Figure S4) and are ruled out for further discussions. Given the higher substitutional defect energy and the thinner atomic slabs, it is more difficult to form a mixed transition region with TiTe than with SbTe to generate a smooth surface for the subsequent growth of PCH thin film. These calculations are consistent with our experimental observations shown in Figure 5 and Figure 6.

Figure S4. Defect formation energy () of Si interstitial sites. The value is calculated as , where , and represent the energy of the supercell with one interstitial Si atom, the pristine phase of the supercell, and the energy of one Si atom in its bulk ground state. The Si, Ti, Sb, and Te atoms are rendered as yellow, red, green, and blue spheres, respectively. The Si interstitial atom adds much higher energy costs, and would induce large atomic displacement of the surrounding atoms upon relaxation.
We also carried out charge transfer analyses based on the DFT-calculated wavefunctions [75], and the obtained Löwdin charge distributions of the five supercell models are displayed in Figure 8, where the charge values of the Si atom are indicated by the black arrows. In the substituted TiTe model, the average charge of Ti atoms is close to 0.6 e, and that of Te atoms is approximately e. When the Si atom is incorporated in the Ti sublattice, it tends to gain negative charge rather than to lose it, in contrast to the Ti atoms. This trend becomes more evident for the Te substitution, as the charge of the Si atom reaches e. The charge transfer between cation-like Sb and anion-like Te atoms is smaller in SbTe, and the Si atom acquires negative charge for all the three substitutional sites. In short, the Si impurities act as anion-like atoms rather than cation-like atoms in both SbTe and TiTe, owing to the relatively large electronegativity of Si.

Figure 8. Löwdin charge analysis. The charge values of Si, Ti, Sb, and Te atoms are depicted with green, red, orange, and blue dots, respectively. The charge value of the Si impurity is marked by a black arrow for the five models.
Although it is more frequent to grow as the seed layer for the high-quality growth of superlattice thin films [101, 102, 103], it could still be possible to generate a flat interface with the silicon substrate when using as the seed layer. In a recent work, an IBM team has reported high-quality growth in a mushroom device with being the first layer [104], but no details on how to obtain the highly textured seed layer were given. We note that it should be feasible to cover the higher energy costs for mixing with silicon atoms given a higher annealing temperature or a longer annealing time.
After Ar-plasma etching, Noé et al. deposited directly on heated silicon substrates, which resulted in Te deficiency in the seed layer [105, 106, 107]. The Te-poor seed layer could lead to a disorientation by a few degrees of the atomic planes of the superlattice with respect to the substrate [105]. Hence, they proposed to increase the concentration of Te for the seed layer growth, and Te-rich layers were clearly captured by HAADF-EDX measurements [107]. This additional compositional tuning can be avoided by heating the substrate after the deposition of the seed layer is fully completed. Various research teams also reported a complete polishing of the silicon substrates to remove the oxide layers, and epitaxial growth of , GST and thin films was achieved on top of passivated Si(111) surface with a few Sb atomic layers [108, 109, 110, 111] or a single Sb/Te atomic layer [112, 113, 114]. Nevertheless, full polishing of the native oxide layer adds fabrication costs, and is not necessary if the procedure sketched in Figure 1a is adopted.
For our approach, the seed layer needs to be chemically unreactive and intact. Therefore, it should be feasible to use , and other GST compositions along the () - () pseudo-binary line for the seed layer, as long as a well-ordered trigonal phase can be formed, e.g., GeSbTe, GeSbTe, GeSbTe, GeSbTe, BiTe, GeBiTe, SnSbTe, SnBiTe, etc. [14, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126]. However, the compositions between GeSbTe and GeTe may not be good choices as they only form cubic rocksalt-like structure, which could provide reactive surfaces to interact with TiTe layers.
Conclusion
In conclusion, we discussed in detail how to achieve epitaxial-like growth of TiTe/SbTe thin film at the wafer-scale, with a particular focus on the essential role of the seed layer. By combining atomic-scale structural characterization and ab initio modelling, we revealed the underlying growth mechanism yielding an atomically flat SbTe seed layer via a transition slab characterized by a Si-Sb-Te mixture with gradual compositional change. We determined the critical minimum thickness of the seed layer to be nm with at least one intact ordered SbTe QL block at the top edge. Importantly, in situ heating should be avoided during the initial Ar-plasma etching and seed layer growth process, since it promotes the re-formation of SiO on top of the silicon substrate. However, sufficient post annealing of the seed layer is necessary to form well-ordered SbTe QL blocks, prior to the alternate sputtering of the PCH thin film under the in situ heating condition. We also observed non-negligible in-plane tensile strain of % in the TiTe blocks induced by the upper and lower SbTe slabs with larger lattice parameters, which indicates that weak covalent coupling is present across the vdW-like gaps in the PCH thin film. The ultrathin SbTe seed layer with critical thickness of nm could serve as a general starting configuration for highly-textured growth of other chalcogenide thin films at the wafer-scale on a standard silicon substrate with relatively rough SiO. We also anticipate further exploration of PCH combinations with alternate stacking of transition metal dichalcogenides as confinement layers and main group chalcogenide semiconductors as memory layers, as long as the latter can form chemically unreactive and intrinsically stable atomic blocks.
Competing interests
The authors declare no competing interests.
Data Availability Statement
Data supporting this work will be available upon requests to the corresponding author.
Acknowledgments
The authors thank the support of XJTU for their work at CAID. W.Z. thanks the support of National Natural Science Foundation of China (62374131). J.-J.W. thanks the support of National Natural Science Foundation of China (62204201). The authors acknowledge Chaobin Zeng from Hitachi High-Tech Scientific Solutions (Beijing) Co., Ltd. and Yuanbin Qin for technical support on STEM experiments. The authors also thank Jia Liu at Instrument Analysis Center of XJTU for the assistance regarding Raman spectroscopy experiments. We acknowledge the HPC platform of XJTU and the Computing Center in Xi'an for providing computational resources. The International Joint Laboratory for Micro/Nano Manufacturing and Measurement Technologies of XJTU is acknowledged. R.M. gratefully acknowledges funding from the PRIN 2020 project “Neuromorphic devices based on chalcogenide heterostructures” funded by the Italian Ministry for University and Research (MUR).
Supplementary materials
Supplementary material associated with this article can be found in the online version at …
Supporting Information
References
- [1]M. Wuttig, N. Yamada, Nat. Mater. 2007, 6, 824.
- [2]W. Zhang, R. Mazzarello, M. Wuttig, E. Ma, Nat. Rev. Mater. 2019, 4, 150.
- [3]A. Sebastian, M. Le Gallo, R. Khaddam-Aljameh, E. Eleftheriou, Nat. Nanotechnol. 2020, 15, 529.
- [4]F. Aguirre, A. Sebastian, M. Le Gallo, W. Song, T. Wang, J. J. Yang, W. Lu, M. F. Chang, D. Ielmini, Y. Yang, A. Mehonic, A. Kenyon, M. A. Villena, J. B. Roldan, Y. Wu, H. H. Hsu, N. Raghavan, J. Sune, E. Miranda, A. Eltawil, G. Setti, K. Smagulova, K. N. Salama, O. Krestinskaya, X. Yan, K. W. Ang, S. Jain, S. Li, O. Alharbi, S. Pazos, M. Lanza, Nat. Commun. 2024, 15, 1974.
- [5]N. Youngblood, C. A. Ríos Ocampo, W. H. P. Pernice, H. Bhaskaran, Nat. Photon. 2023, 17, 561.
- [6]W. Zhou, X. Shen, X. Yang, J. Wang, W. Zhang, Int. J. Extrem. Manuf. 2024, 6, 022001.
- [7]M. Xu, X. Mai, J. Lin, W. Zhang, Y. Li, Y. He, H. Tong, X. Hou, P. Zhou, X. Miao, Adv. Funct. Mater. 2020, 30, 2003419.
- [8]X. D. Li, N. K. Chen, B. Q. Wang, M. Niu, M. Xu, X. Miao, X.-B. Li, Adv. Mater. 2024, 36, 2307951.
- [9]B. Liu, K. Li, J. Zhou, Z. Sun, Adv. Funct. Mater. 2024, 34, 2407239.
- [10]X. Wang, R. Wang, S. Sun, D. Xu, C. Nie, Z. Zhou, C. Wen, J. Zhang, R. Chu, X. Shen, W. Zhou, Z. Song, J. Wang, E. Ma, W. Zhang, arXiv: 2503.21446 2025.arxiv.org/abs/2503.21446
- [11]X. Shen, R. Chu, Y. Jiang, W. Zhang, Acta Metall. Sin. 2024, 60, 1362.
- [12]X. Shen, S. Zhang, Y. Jiang, T. Huang, S. Sun, W. Zhou, J. Wang, R. Mazzarello, W. Zhang, MGE Adv. 2024, 2, e62.
- [13]N. Yamada, E. Ohno, N. Akahira, K. Nishiuchi, K. Nagata, M. Takao, Jpn. J. Appl. Phys. Part 1 1987, 26, 61.
- [14]N. Yamada, E. Ohno, K. Nishiuchi, N. Akahira, M. Takao, J. Appl. Phys. 1991, 69, 2849.
- [15]I. Friedrich, V. Weidenhof, W. Njoroge, P. Franz, M. Wuttig, J. Appl. Phys. 2000, 87, 4130.
- [16]B. J. Kooi, J. T. M. De Hosson, J. Appl. Phys. 2002, 92, 3584.
- [17]B. Zhang, W. Zhang, Z.-J. Shen, Y.-J. Chen, J.-X. Li, S.-B. Zhang, Z. Zhang, M. Wuttig, R. Mazzarello, E. Ma, X.-D. Han, Appl. Phys. Lett. 2016, 108, 191902.
- [18]T. Siegrist, P. Jost, H. Volker, M. Woda, P. Merkelbach, C. Schlockermann, M. Wuttig, Nat. Mater. 2011, 10, 202.
- [19]W. Zhang, M. Wuttig, R. Mazzarello, Sci. Rep. 2015, 5, 13496.
- [20]T.-T. Jiang, X.-D. Wang, J.-J. Wang, Y.-X. Zhou, D.-L. Zhang, L. Lu, C.-L. Jia, M. Wuttig, R. Mazzarello, W. Zhang, Acta Mater. 2020, 187, 103.
- [21]A. Lotnyk, M. Behrens, B. Rauschenbach, Nanoscale Adv. 2019, 1, 3836.
- [22]H.-S. P. Wong, S. Raoux, S. B. Kim, J. Liang, J. P. Reifenberg, B. Rajendran, M. Asheghi, K. E. Goodson, Proc. IEEE 2010, 98, 2201.
- [23]X. Li, H. Chen, C. Xie, D. Cai, S. Song, Y. Chen, Y. Lei, M. Zhu, Z. Song, Phys. Status Solidi RRL 2019, 13, 1800558.
- [24]H. Y. Cheng, F. Carta, W. C. Chien, H. L. Lung, M. BrightSky, J. Phys. D: Appl. Phys. 2019, 52, 473002.
- [25]P. Cappelletti, R. Annunziata, F. Arnaud, F. Disegni, A. Maurelli, P. Zuliani, J. Phys. D: Appl. Phys. 2020, 53, 193002.
- [26]Q. Wang, Y. Wang, Y. Wang, L. Jiang, J. Zhao, Z. Song, J. Bi, L. Zhao, Z. Jiang, J. Schwarzkopf, S. Wu, B. Zhang, W. Ren, S. Song, G. Niu, InfoMat 2024, 6, e12543.
- [27]C. Ríos, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer, C. D. Wright, H. Bhaskaran, W. H. P. Pernice, Nat. Photon. 2015, 9, 725.
- [28]P. Hosseini, C. D. Wright, H. Bhaskaran, Nature 2014, 511, 206.
- [29]W. Zhou, B. Dong, N. Farmakidis, X. Li, N. Youngblood, K. Huang, Y. He, C. David Wright, W. H. P. Pernice, H. Bhaskaran, Nat. Commun. 2023, 14, 2887.
- [30]W. Zhang, R. Mazzarello, E. Ma, MRS Bull. 2019, 44, 686.
- [31]D. Ielmini, A. L. Lacaita, D. Mantegazza, IEEE Trans. Electron. Dev. 2007, 54, 308.
- [32]S. Braga, A. Cabrini, G. Torelli, Applied Physics Letters 2009, 94, 092112.
- [33]M. Boniardi, D. Ielmini, S. Lavizzari, A. L. Lacaita, A. Redaelli, A. Pirovano, IEEE Trans. Electron. Dev. 2010, 57, 2690.
- [34]M. Boniardi, D. Ielmini, Appl. Phys. Lett. 2011, 98, 243506.
- [35]A. Sebastian, M. Le Gallo, G. W. Burr, S. Kim, M. BrightSky, E. Eleftheriou, J. Appl. Phys. 2018, 124, 111101.
- [36]M. Le Gallo, A. Sebastian, R. Mathis, M. Manica, H. Giefers, T. Tuma, C. Bekas, A. Curioni, E. Eleftheriou, Nat. Electron. 2018, 1, 246.
- [37]S. Ambrogio, P. Narayanan, H. Tsai, R. M. Shelby, I. Boybat, G. W. Burr, Nature 2018, 558, 60.
- [38]J.-Y. Raty, W. Zhang, J. Luckas, C. Chen, C. Bichara, R. Mazzarello, M. Wuttig, Nat. Commun. 2015, 6, 7467.
- [39]S. Gabardi, S. Caravati, G. C. Sosso, J. Behler, M. Bernasconi, Phys. Rev. B 2015, 92, 054201.
- [40]K. Konstantinou, F. C. Mocanu, T. H. Lee, S. R. Elliott, Nat. Commun. 2019, 10, 3065.
- [41]W. Zhang, E. Ma, Mater. Today 2020, 41, 156.
- [42]B. Chen, X. P. Wang, F. Jiao, L. Ning, J. Huang, J. Xie, S. Zhang, X. B. Li, F. Rao, Adv. Sci. 2023, 10, 2301043.
- [43]F. Jiao, B. Chen, K. Ding, K. Li, L. Wang, X. Zeng, F. Rao, Appl. Mater. Today 2020, 20, 100641.
- [44]P. Ma, H. Tong, M. Xu, X. Cheng, X. Miao, Appl. Phys. Lett. 2020, 117, 022109
- [45]X. Shen, Y. Zhou, H. Zhang, V. L. Deringer, R. Mazzarello, W. Zhang, Nanoscale 2023, 15, 15259.
- [46]J. Huang, B. Chen, G. Sha, H. Gong, T. Song, K. Ding, F. Rao, Nano Lett. 2023, 23, 2362.
- [47]B. Liu, K. Li, W. Liu, J. Zhou, L. Wu, Z. Song, S. R. Elliott, Z. Sun, Sci. Bull. 2021, 66, 2217.
- [48]Y. Jiao, G. Wang, A. Lotnyk, T. Wu, J. Zhu, A. He, J. Alloys Compd. 2023, 953, 169970.
- [49]A. He, J. Zhu, G. Wang, A. Lotnyk, S. Cremer, Y. Chen, X. Shen, Appl. Phys. Lett. 2024, 124, 221602.
- [50]T. C. Chong, L. P. Shi, R. Zhao, P. K. Tan, J. M. Li, H. K. Lee, X. S. Miao, A. Y. Du, C. H. Tung, Appl. Phys. Lett. 2006, 88, 122114.
- [51]R. E. Simpson, P. Fons, A. V. Kolobov, T. Fukaya, M. Krbal, T. Yagi, J. Tominaga, Nat. Nanotechnol. 2011, 6, 501.
- [52]R. Wang, V. Bragaglia, J. E. Boschker, R. Calarco, Cryst. Growth Des. 2016, 16, 3596.
- [53]A. V. Kolobov, P. Fons, Y. Saito, J. Tominaga, ACS Omega 2017, 2, 6223.
- [54]A. Lotnyk, I. Ilhmi, U. Ross, B. Rauschenbach, Nano Res. 2017, 11, 1676.
- [55]L. Zhou, Z. Yang, X. Wang, H. Qian, M. Xu, X. Cheng, H. Tong, X. Miao, Adv. Electron. Mater. 2019, 5, 1900781.
- [56]C. Yoo, J. W. Jeon, B. Park, W. Choi, G. Jeon, S. Jeon, S. Kim, C. S. Hwang, ACS Appl. Electron. Mater. 2023, 5, 5794.
- [57]X.-B. Li, N.-K. Chen, X.-P. Wang, H.-B. Sun, Adv. Funct. Mater. 2018, 28, 1803380.
- [58]B. Sa, J. Zhou, Z. Sun, J. Tominaga, R. Ahuja, Phys. Rev. Lett. 2012, 109, 096802.
- [59]O. Cojocaru-Mirédin, J.-C. Bürger, N. Polin, A. Meledin, J. Mayer, M. Wuttig, A. Daus, ACS nano 2025, , DOI: 10.1021/acsnano.4c13450.
- [60]K. Ding, J. Wang, Y. Zhou, H. Tian, L. Lu, R. Mazzarello, C. Jia, W. Zhang, F. Rao, E. Ma, Science 2019, 366, 210.
- [61]J. Shen, S. Lv, X. Chen, T. Li, S. Zhang, Z. Song, M. Zhu, ACS Appl. Mater. Interfaces 2019, 11, 5336.
- [62]X. Wang, K. Ding, M. Shi, J. Li, B. Chen, M. Xia, J. Liu, Y. Wang, J. Li, E. Ma, Z. Zhang, H. Tian, F. Rao, Mater. Today 2022, 54, 52.
- [63]X. Wang, Y. Wu, Y. Zhou, V. L. Deringer, W. Zhang, Mater. Sci. Semicond. Process. 2021, 135, 106080.
- [64]G. Han, F. Liu, Y. Zhang, J. Li, W. Li, Q. Chen, Y. Li, X. Xie, J. Alloys Compd. 2021, 877, 160270.
- [65]J. Park, H. Sung, S. Son, S. Ju, H. Lee, Phys. Status Solidi RRL 2023, 17, 2200451.
- [66]R. Piombo, S. Ritarossi, R. Mazzarello, Adv. Sci. 2024, 11, 2402375.
- [67]T. H. Kim, S. W. Park, H. J. Lee, D. H. Kim, J. Y. Choi, T. G. Kim, Small 2023, 19, 2303659.
- [68]T. H. Kim, K. J. Yoo, T. H. Kim, H. J. Lee, A. C. Khot, K. A. Nirmal, S. H. Hong, T. G. Kim, Appl. Surf. Sci. 2023, 626, 157291.
- [69]S. W. Park, H. J. Lee, K. A. Nirmal, T. H. Kim, D. H. Kim, J. Y. Choi, J. S. Oh, J. M. Joo, T. G. Kim, J. Mater. Sci. Technol. 2025, 204, 104.
- [70]X. Wu, A. I. Khan, H. Lee, C.-F. Hsu, H. Zhang, H. Yu, N. Roy, A. V. Davydov, I. Takeuchi, X. Bao, H. S. P. Wong, E. Pop, Nat. Commun. 2024, 15, 13.
- [71]G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558.
- [72]J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.
- [73]G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758.
- [74]S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132, 154104
- [75]R. Nelson, C. Ertural, J. George, V. L. Deringer, G. Hautier, R. Dronkowski, J. Comput. Chem. 2020, 41, 1931.
- [76]S. Maintz, V. L. Deringer, A. L. Tchougreeff, R. Dronskowski, J. Comput. Chem. 2016, 37, 1030.
- [77]V. L. Deringer, A. L. Tchougreeff, R. Dronskowski, J. Phys. Chem. A. 2011, 115, 5461.
- [78]Y. Saito, P. Fons, A. V. Kolobov, J. Tominaga, Phys. Status Solidi B 2015, 252, 2151.
- [79]Y. Saito, P. Fons, L. Bolotov, N. Miyata, A. V. Kolobov, J. Tominaga, AIP Adv. 2016, 6, 045220.
- [80]Y. Saito, P. Fons, A. V. Kolobov, K. V. Mitrofanov, K. Makino, J. Tominaga, S. Hatayama, Y. Sutou, M. Hase, J. Robertson, J. Phys. D: Appl. Phys. 2020, 53, 284002.
- [81]T. L. Anderson, H. B. Krause, Acta Crystallogr. B 1974 30 1307.
- [82]Y. Zheng, M. Xia, Y. Cheng, F. Rao, K. Ding, W. Liu, Y. Jia, Z. Song, S. Feng, Nano Res. 2016, 9, 3453.
- [83]Y. Xu, X. Wang, W. Zhang, L. Schäfer, J. Reindl, F. vom Bruch, Y. Zhou, V. Evang, J. J. Wang, V. L. Deringer, E. Ma, M. Wuttig, R. Mazzarello, Adv. Mater. 2021, 33, 2006221.
- [84]W. Richter, C. R. Becker, Phys. Status Solidi B 1977, 84, 619.
- [85]K. M. F. Shahil, M. Z. Hossain, V. Goyal, A. A. Balandin, J. Appl. Phys. 2012, 111, 054305.
- [86]G. C. Sosso, S. Caravati, M. Bernasconi, J. Phys. Condens. Matter. 2009, 21, 095410.
- [87]X. Feng, Z. Li, G. Chen, H. Yue, Y. Gao, X. Zhang, Z. Guo, W. Yuan, CrystEngComm 2023, 25, 5399.
- [88]S. J. Pennycook, P. D. Nellist, Scanning Transmission Electron Microscopy Imaging and Analysis, Springer, 2011.DOI
- [89]R. Wang, F. R. L. Lange, S. Cecchi, M. Hanke, M. Wuttig, R. Calarco, Adv. Funct. Mater. 2018, 28, 1705901.
- [90]J. Wang, I. Ronneberger, L. Zhou, L. Lu, V. L. Deringer, B. Zhang, L. Tian, H. Du, C. Jia, X. Qian, M. Wuttig, R. Mazzarello, W. Zhang, Nanoscale 2018, 10, 7363.
- [91]Y. Cheng, O. Cojocaru-Mirédin, J. Keutgen, Y. Yu, M. Küpers, M. Schumacher, P. Golub, J.-Y. Raty, R. Dronskowski, M. Wuttig, Adv. Mater. 2019, 31, 1904316.
- [92]J. Y. Raty, M. Schumacher, P. Golub, V. L. Deringer, C. Gatti, M. Wuttig, Adv. Mater. 2019, 31, 1806280.
- [93]B. J. Kooi, M. Wuttig, Adv. Mater. 2020, 32, 1908302.
- [94]M. Wuttig, C. F. Schon, J. Lotfering, P. Golub, C. Gatti, J. Y. Raty, Adv. Mater. 2023, 35, 2208485.
- [95]M. Wuttig, C. F. Schon, D. Kim, P. Golub, C. Gatti, J. Y. Raty, B. J. Kooi, A. M. Pendas, R. Arora, U. Waghmare, Adv. Sci. 2023, 11, 2308578.
- [96]W. Zhang, H. Zhang, S. Sun, X. Wang, Z. Lu, X. Wang, J.-J. Wang, C. Jia, C. F. Schon, R. Mazzarello, E. Ma, M. Wuttig, Adv. Sci. 2023, 10, 2300901.
- [97]M. Chhowalla, H. S. Shin, G. Eda, L. J. Li, K. P. Loh, H. Zhang, Nat. Chem. 2013, 5, 263.
- [98]J. Qi, X. Qian, L. Qi, J. Feng, D. Shi, J. Li, Nano Lett. 2012, 12, 1224.
- [99]J. Li, Z. Wan, E. Ma, MRS Bull. 2014, 39, 108.
- [100]W. Li, X. Qian, J. Li, Nat. Rev. Mater. 2021, 6, 829.
- [101]J. Momand, F. R. L. Lange, R. Wang, J. E. Boschker, M. A. Verheijen, R. Calarco, M. Wuttig, B. J. Kooi, J. Mater. Res. 2016, 31, 3115.
- [102]A. I. Khan, A. Daus, R. Islam, K. M. Neilson, H. R. Lee, H.-S. P. Wong, E. Pop, Science 2021, 373, 1243.
- [103]S. Prili, V. Bragaglia, V. P. Jonnalagadda, J. Luchtenveld, B. J. Kooi, F. Arciprete, A. Sebastian, G. S. Syed, arXiv: 2411.19343 2024.arxiv.org/abs/2411.19343
- [104]G. M. Cohen, A. Majumdar, C. W. Cheng, A. Ray, D. Piatel, L. Gignac, C. Lavoie, A. Grun, H. Y. Cheng, Z. L. Liu, H. L. Lung, H. Miya Zoe, R. L. Bruce, M. BrightSky, Phys. Status Solidi RRL 2024, 18, 2300426.
- [105]P. Kowalczyk, F. Hippert, N. Bernier, C. Mocuta, C. Sabbione, W. Batista - Pessoa, P. Noé, Small 2018, 14, 1704514.
- [106]F. Hippert, P. Kowalczyk, N. Bernier, C. Sabbione, X. Zucchi, D. Térébenec, C. Mocuta, P. Noé, J. Phys. D: Appl. Phys. 2020, 53, 154003.
- [107]V. Sever, N. Bernier, D. Térébénec, C. Sabbione, J. Paterson, F. Castioni, P. Quéméré, A. Jannaud, J. L. Rouvière, H. Roussel, J. Y. Raty, F. Hippert, P. Noé, Phys. Status Solidi RRL 2024, 18, 2300402.
- [108]J. Momand, R. Wang, J. E. Boschker, M. A. Verheijen, R. Calarco, B. J. Kooi, Nanoscale 2015, 7, 19136.
- [109]R. Wang, R. Calarco, F. Arciprete, V. Bragaglia, Mater. Sci. Semicond. Process. 2022, 137, 106244.
- [110]V. Bragaglia, F. Arciprete, W. Zhang, A. M. Mio, E. Zallo, K. Perumal, A. Giussani, S. Cecchi, J. E. Boschker, H. Riechert, S. Privitera, E. Rimini, R. Mazzarello, R. Calarco, Sci. Rep. 2016, 6, 23843.
- [111]V. Bragaglia, F. Arciprete, A. M. Mio, R. Calarco, J. Appl. Phys. 2018, 123.
- [112]I. Hilmi, A. Lotnyk, J. W. Gerlach, P. Schumacher, B. Rauschenbach, APL Materials 2017, 5, 050701.
- [113]M. Behrens, A. Lotnyk, U. Roß, J. Griebel, P. Schumacher, J. W. Gerlach, B. Rauschenbach, CrystEngComm 2018, 20, 3688.
- [114]H. Bryja, J. W. Gerlach, A. Prager, M. Ehrhardt, B. Rauschenbach, A. Lotnyk, 2D Materials 2021, 8, 045027.
- [115]J.-W. Park, S. H. Eom, H. Lee, J. L. F. Da Silva, Y.-S. Kang, T.-Y. Lee, Y. H. Khang, Phys. Rev. B 2009, 80, 115209.
- [116]J. Da Silva, A. Walsh, H. Lee, Phys. Rev. B 2008, 78, 224111.
- [117]Z. Sun, Y. Pan, J. Zhou, B. Sa, R. Ahuja, Phys. Rev. B 2011, 83, 113201.
- [118]Y. Gan, J. Zhou, Z. Sun, J. Mater. Inf. 2021, 1, 2.
- [119]Y. Gan, Y. Huang, N. Miao, J. Zhou, Z. Sun, J. Mater. Chem. C 2021, 9, 4189.
- [120]J. J. Wang, H. M. Zhang, X. D. Wang, L. Lu, C. Jia, W. Zhang, R. Mazzarello, Adv. Mater. Technol. 2022, 7, 22000214.
- [121]J.-J. Wang, J. Wang, H. Du, L. Lu, P. C. Schmitz, Johannes Reindl, A. M. Mio, C.-L. Jia, E. Ma, R. Mazzarello, M. Wuttig, W. Zhang, Chem. Mater. 2018, 30, 4770.
- [122]H. Zhang, D. T. Yimam, S. de Graaf, J. Momand, P. A. Vermeulen, Y. Wei, B. Noheda, B. J. Kooi, ACS nano 2021, 15, 2869.
- [123]H. Zhang, J. Momand, J. Levinsky, Q. Guo, X. Zhu, G. H. ten Brink, G. R. Blake, G. Palasantzas, B. J. Kooi, Nano Res. 2021, 15, 2382.
- [124]D. Hsieh, Y. Xia, D. Qian, L. Wray, F. Meier, J. Dil, J. Osterwalder, L. Patthey, A. Fedorov, H. Lin, A. Bansil, D. Grauer, Y. Hor, R. Cava, M. Hasan, Phys. Rev. Lett. 2009, 103, 146401.
- [125]S. Howard, A. Raghavan, D. Iaia, C. Xu, D. Flötotto, M.-H. Wong, S.-K. Mo, B. Singh, R. Sankar, H. Lin, T.-C. Chiang, V. Madhavan, Phys. Rev. Mater. 2022, 6, 044201
- [126]T. Schäfer, P. M. Konze, J. D. Huyeng, V. L. Deringer, T. Lesieur, P. Müller, M. Morgenstern, R. Dronkowski, M. Wuttig, Chem. Mater. 2017, 29, 6749.