1 Introduction

Since the discovery of graphene in 2004, two-dimensional (2D) materials have been further explored and expanding rapidly in recent years [1, 2]. Among these materials, 2D magnets have garnered extensive attention due to their unique properties, making them promising for applications in memory and information technology, spintronics, and quantum computing [3]. In previous studies, various methods have been employed to introduce magnetic ordering into nonmagnetic 2D materials, including magnetic proximity [4, 5], doping [6–8], adsorption [9, 10], and defects [11, 12]. However, the magnetic moments in these systems are generally localized and difficult to create a long-range correlation, thus limiting their practical utility [13]. Consequently, researchers are eagerly seeking alternative 2D materials that exhibit intrinsic long-range magnetic order.

According to the Mermin–Wagner theorem, strong thermal fluctuations would suppress the magnetism in the 2D region [14]. To counteract this effect, magnetocrystalline anisotropy can be introduced to produce a spin wave gap, which stabilizes the long-range magnetic order at finite temperatures [15]. Due to the reduced crystal symmetry, many van der Waals (vdW) magnets possess magnetocrystalline anisotropy [16]. As a result, atomically thin van der Waals films like Fe3GeTe2 [17], VSe2 [13, 18], Cr2Ge2Te6 [15, 19], and CrI3 [20, 21], have been found to exhibit intrinsic long-range magnetic order. This discovery has sparked fundamental studies of 2D ferromagnetism.

Recently, as an emergent class of vdW-like materials, chromium telluride compounds (Cr(1+δ)Te2) have attracted widespread research interest [22–33]. The layered-like structure, as well as the very strong spin-orbit coupling of tellurium [34–36], should make Cr(1+δ)Te2 compounds more prone to induce strong magnetic anisotropy, thus stabilizing the long-range magnetic order in 2D systems. As expected, the high Curie temperatures and the intrinsic robust magnetic ordering have been demonstrated. The Curie temperatures of Cr(1+δ)Te2 films vary depending on the thickness and the value of δ, ranging from 100 K to 344 K [23–27, 37]. Apart from the typical ferromagnetic interaction, Cr(1+δ)Te2 thin films were also suggested to host antiferromagnetic interactions [29], tunable magnetic anisotropy [38], and noncollinear spin texture [39]. The δ represents the variation in the amount of intercalated Cr between CrTe2 transition-metal dichalcogenides (TMD) layers [37, 40]. Defining the density of transition metals in the TMD layer to 1, the δ contains a series of possible values, such as 0 (CrTe2), 0.25 (Cr5Te8), 0.33 (Cr2Te3), 0.50 (Cr3Te4), and 1 (CrTe) [41]. Rich magnetic properties and many compositions mentioned in Cr(1+δ)Te2 may bring great opportunities to explore the spintronic device applications. However, challenges remain in growing high-quality and phase-controllable samples for practical applications due to the complicated phase diagram of Cr(1+δ)Te2. Thereby, there is an urgent need to systematically study the growth mechanism of Cr(1+δ)Te2 films.

Molecular beam epitaxy (MBE) is a highly specialized technique for the fabrication of thin films with precise control over thicknesses and quality. The growth of Cr(1+δ)Te2 films by MBE has been realized on various substrates, such as metallic HOPG [25, 42], semiconductive MoS2 [37, 41], and GaAs [43], insulating Al2O3, mica [24, 44–46] and so on. Currently, most of the large-scale integrated circuitry is based on Si technologies [47]. Therefore, the combination of high-quality epitaxial magnetic thin films with semiconducting Si (111) could facilitate many applications, for example, the seamless integration of information processing and storage [48]. Additionally, its small lattice mismatch to the Cr(1+δ)Te2 compounds makes Si (111) more feasible to be the substrate to obtain high quality Cr(1+δ)Te2 films [30].

In this work, we successfully grew large-area Cr(1+δ)Te2 thin films on the Si (111) substrate by MBE. Combining the in-situ reflection high-energy electron diffraction (RHEED), scanning tunneling microscopy (STM), ex-situ X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and computation, the controllable phase transition from 1T-CrTe2 into Cr2Te3 was verified. At lower substrate temperatures (250°C–290°C), the 1T-CrTe2 was grown in the 1st and 2nd layers, which exhibited a simple hexagonal atomic lattice with 1×1 periodicity. As the thickness exceeded 2 layers (≥3 L), the Cr2Te3 phase formed, resulting in a quasi-hexagonal array and 2×1 reconstruction. For the relatively elevated growth temperature (370°C–390°C), Cr2Te3 phase was always the favorable phase regardless of the film thickness. However, a thickness-dependent growth mode was found in the grown films. Both the crystalline phase and growth mode transition are related to interfacial interactions and the phase stability of Cr(1+δ)Te2 compounds. Additionally, using ex-situ vibrating sample magnetometer (VSM), we demonstrated the ferromagnetism of the 30-nm Cr2Te3 films, with a Curie temperature of approximately 180 K.

2 Results and discussions

The Cr(1+δ)Te2 componds have NiAs-type crystals structures, where a (distorted) hexagonal close packing of Te atoms forms, with Cr atoms filled in octahedral interstices [28]. Taking Cr2Te3 as an example, as a TMD-like material, it can be viewed as two CrTe2 sheets with additional 0.33 Cr-layer interspaced in the vdW gap, as shown in Fig. 1(a) and (b). During the MBE growth, RHEED was employed to monitor the growth process and the quality of Cr(1+δ)Te2 films. From top to bottom, Fig. 1(c) shows the RHEED patterns of Si (111)–7×7 substrate and the epitaxial Cr(1+δ)Te2 films grown at 250°C, 290°C, 370°C, 390 °C, respectively. Here, the coverage of Cr(1+δ)Te2 films slightly exceeds 2 L, we denoted this thickness as 2.2 L according to the topography of the following STM studies. The RHEED patterns of Cr(1+δ)Te2 films show a sharp 1×1 streaky feature. This proves that Cr(1+δ)Te2 film grows with a high crystalline quality ranging from 250°C to 390°C, giving atomically flat surface morphologies. Figure 1(d) displays the intensity profile for RHEED patterns taken along the dashed lines shown in Fig. 1(c). It is noted that a set of reconstruction stripes arise between the 1×1 stripes. For clarity, the intensity variations of 2×1 stripes were extracted after normalization and plotted in Fig. 1(e). This unveils that the intensity of 2×1 stripes of the films grown at high temperatures is stronger than those grown at low temperatures. For 250°C and 290°C, the weak 2×1 stripes can be attributed to the formation of 0.2 L Cr2Te3 on 2 L 1T-CrTe2 and would be elucidated in detail in Fig. 2. The stronger intensity of the reconstruction stripe for the samples grown at 370°C and 390 °C indicates the formation of a 2×1 dominant structure, consistent with the typical RHEED pattern of Cr2Te3 [24, 44, 49]. This suggests that a phase transition could have occurred in the studied Cr(1+δ)Te2 film as the growth temperature increased.

Figure 1
figure 1

Structural features and RHEED evolution of Cr(1+δ)Te2. (a-b) The crystal structure of 1T-CrTe2 (a) and Cr2Te3 (b). (c) Typical in-situ RHEED patterns from the surface of Si (111)–7×7 substrates and as-grown c-oriented 2.2 L Cr(1+δ)Te2 films after growing at varying temperatures. (d) RHEED intensity profile plot of the images in (c). The black dashed box marks the 2×1 stripes. (e) The 2×1 stripes intensity of Cr(1+δ)Te2 films

Full size image
Figure 2
figure 2

Structural and composition characterization of Cr(1+δ)Te2 films grown at 250°C. (a, b, f) Large-scale STM image (a: 2 V, 0.1nA, b: 2.17 V, 0.1nA, f: 2.17 V, 0.1nA) of 2.2 L (a), 1.6 L (b), and multilayer (f) Cr(1+δ)Te2 film. The 1 L (yellow), 2 L (red), and 3 L (green) are marked by three colored circles in (b). Inset of (b) presents the line profiles corresponding to the red line inside this region and the RHEED diffraction pattern. RHEED pattern shows a strong 1×1 diffraction reflection of pure 1T-CrTe2, without a 2×1 stripe. Inset of (f) presents the RHEED pattern with strong 2×1 periodicity of Cr2Te3. (c-e) Atomic resolution STM image of the 1 L (0.43 V, 0.2nA), 2 L (0.009 V, 0.3nA), and 3 L (0.4 V, 0.2nA) film surface. Insets show the corresponding 2D FFT. (g) A set of dI/dV spectrums taken at the colored site in (b) (setting points: VTip = 0.5 V, ITip = 100 pA)

Full size image

To further investigate the phase transition, a series of scanning tunneling microscopy (STM) measurements were conducted. We focused on the low-temperature growth (e.g., 250°C) first. The large-scale topography images in Fig. 2(a) show that the 2.2 L Cr(1+δ)Te2 ultrathin film exhibits a layer-by-layer (Frank-van der Merwe) growth mode. As we mentioned above, the weak reconstruction stripes between the 1×1 stripes of 2.2 L Cr(1+δ)Te2 film (250°C–290°C in Fig. 1(c-d)) result from the 0.2 L Cr2Te3 on 2 L 1T-CrTe2. In other words, the critical thickness of phase transition from 1T-CrTe2 to Cr2Te3 is 2 L. Herein, detailed studies were carried out to support this conclusion. When slightly decreasing the coverage of Cr(1+δ)Te2 film from 2.2 L to lower than 2 L (about 1.6 L), the surface is predominantly covered with 1T-CrTe2 phase, as shown in Fig. 2(b), and the RHEED pattern (see the top right inset of Fig. 2(b)) also transfers to a 1T-CrTe2 1×1 reflection without any 2×1 superstructure reflection. The subsequent profile (see the lower right inset of Fig. 2(b)) taken along the red line inside this region reveals that the terrace heights for each adjacent layers of the film are uniquely 0.63 nm, consistent with the step height of one unit cell. Figure 2(c-d) represent the simple hexagonal structure recorded at the 1st and 2nd CrTe2 layer, exhibiting 1×1 periodicity with a lattice constant of 3.842 Å and 3.858 Å respectively, which agrees well with the previous results of 1T-CrTe2 [26, 50]. The fast Fourier transform (FFT) inset in Figs. 2(c-d) also manifests a pristine 1×1 periodicity. Notably, the 2×1 periodicity emerges from the 3rd layer with a larger lattice constant of 3.932 Å (Fig. 2(e)), implying significant changes in the top layer due to the Cr interspaced atoms. The FFT inset image can easily capture the 2×1 reconstruction except for the pristine 1×1 periodicity. The 2×1 reconstructions of multilayer are also shown in RHEED (see the inset of Fig. 2(f)). Namely, the growth of the intercalation compounds becomes more dominant as thickness increases. It is noted that the 2×1 RHEED stripe is only one of the experimental signatures for Cr2Te3. Other technologies should be combined to further confirm the Cr2Te3 phase. In our samples, such a 2×1 pattern results from an unusual Cr2Te3 lattice, which coexists with the normal Cr2Te3 lattice. Our calculated results, shown in Fig. S1(a-b), present the total energy of the unusual Cr2Te3 lattice is only 1.6 meV/atom higher than its normal counterparts. The simulated STM image of this unusual lattice shows a 2×1 superstructure (Fig. S1(c)). This indicates the 2×1 reconstruction pattern results from such an unusual structure. To explore the electronic structures of the 1T-CrTe2 and Cr2Te3, constant-height STS measurements were performed, and the spectra collected at the corresponding three layers were plotted in Fig. 2(g). The STS spectrum collected from the CrTe2 area exhibits a V shape, consistent with the previous reports [42]. While for the STS spectrum collected on the Cr2Te3 surface (3rd layer), three states can be observed, with energy levels of about −0.70, −0.13, and 0.54 eV relative to the Fermi level respectively. This displays that CrTe2 and Cr2Te3 have different features around Fermi level.

XPS was used to further analyze this phase transition. The ex-situ XPS spectrum of 1.6 L 1T-CrTe2 and multilayer Cr2Te3 films/2 L CrTe2 is presented in Figs. 3(a) and (b) respectively. Since the XPS peak of Cr 2p orbital and Te 3d orbital are very close to each other, the stronger XPS signals of Cr 2p\(_{3/2}\) and the Te 3d\(_{5/2}\) orbital were chosen to separate them out and study the chemical bond states. We fitted the spectra with three component peaks, i.e., Te 3d\(_{5/2}\), Cr 2p\(_{3/2}\), and Te from TexOy. The binding energy of each component is given in Table S1 in the Additional file 1. The XPS element analysis of multilayer Cr(1+δ)Te2 films gives the Cr/Te concentration ratio 0.37 higher than that of 1.6 L 1T-CrTe2, which is slightly larger than the ideal value of 0.33 (Cr2Te3 is 0.33 higher than CrTe2). This could be attributed to the excess Cr atoms randomly intercalating on the unoccupied sites of the Cr2Te3. To verify this hypothesis, we conducted the density-functional theory (DFT) calculation. Figure 3(c) and (d) display the intercalation of one and two excess Cr atoms (marked by pink) on the unoccupied sites of Cr2Te3. Both structures show the negative insertion energy (Ein) for single Cr atom (e.g., Ein \(=-0.490\) eV and Ein \(=-0.304\) eV), which demonstrates the Cr2Te3 with a few more Cr atoms intercalated are the more stable structures than the normal Cr2Te3 structure. Thereby, the 0.37 Cr/Te ratio can be explained by the Cr-rich Cr2Te3. These observations attest to the phase modulation of Cr(1+δ)Te2 films from 1T-CrTe2 to intercalation compounds (Cr2Te3) by precisely adjusting the thickness.

Figure 3
figure 3

XPS measurements for Cr(1+δ)Te2 films and insertion energy calculation for single Cr atom. (a-b) Cr-2p and Te-3d core-level X-ray photoelectron spectra from 1.6 L 1T-CrTe2 (a) and multilayer Cr(1+δ)Te2 (b) thin films. (c-d) The vertical view sketches of the Cr-rich Cr2Te3. There are one (c) and two (d) Cr atoms insert on unoccupied sites of Cr2Te3, which are marked by the pink ball

Full size image

As the growth temperature increases (e.g., 370°C), the STM topography of 2.2 L and thicker Cr(1+δ)Te2 film (≈8 nm) is revealed in Figs. 4(a) and (b) respectively. The inset shows the corresponding STM image with atomic resolution, where the clear 2×1 reconstruction lattice can be easily observed. Therefore, the atomic structure remains consistent in both thin and thick films, identifying the thickness-independent nature of the MBE-grown Cr2Te3 phases at elevated temperatures. This 2×1 reconstruction is still visible at 289 K, suggesting that it is not a charge density wave (Fig. S2). Figure 4(c) depicts that the height between adjacent terraces in the 2.2 L Cr(1+δ)Te2 film is about 1.33 nm, which can be attributed to the formation of an intercalated bilayer. However, the thick Cr2Te3 film shows a height profile of 0.63 nm in Fig. 4(d), which is different from their thinner counterparts. It means that the growth mode of Cr2Te3 films was modulated from bilayer to monolayer when the thickness exceeds 2 L at high temperatures. Figure 4(e) shows the measured differential conductance dI/dV curves of 2 L Cr2Te3 films, which have a wide range of energy scales from −1.8 to 2.8 eV bias. In addition, several peaks have been observed, qualitatively consistent with calculated DOS (Fig. 4(f)), as marked by the corresponding numbers (1∼6), respectively. The correspondingly calculated band structure is presented in Fig. S3. Thereby, DFT calculations further confirm the electronic structures of the 2 L Cr2Te3. A similar metallic feature is also observable on the thicker films. The crystalline phase of Cr2Te3 grown at high temperatures was further determined by XRD. As shown in Fig. 4(g), the layer peaks (002), (004), and (006) are indexed, which matches very well with the Cr2Te3 structure.

Figure 4
figure 4

Structural characterization and VSM measurements of Cr2Te3 films grown at 370°C. (a-b) Large-scale STM image of 2.2 L (a) and thicker Cr2Te3 (b) films (8 nm) (a: 1 V, 0.1nA; b: 2.14 V, 0.016nA). Inset shows the corresponding atomic STM image with 2×1 periodicity. (c-d) The line profiles correspond to the red line in panels (a) and (b) respectively, which show the bilayer growth mode for 2.2 L film and monolayer growth mode for thicker film. (e) Measured dI/dV spectra of 2 L Cr2Te3 surface (1 V, 0.2 nA). (f) Theoretical total DOS curve of 2 L Cr2Te3. (g) XRD profiles of Cr2Te3 films (30 nm). (h) Temperature-dependent magnetic susceptibility was measured on 30 nm Cr2Te3 films with an out-of-plane external magnetic field \(\mu _{0}\)H = 500 Oe. Inset shows the temperature-dependent of differential magnetic susceptibility (dM/dT) extracted from FC mode. (i) Magnetic hysteresis loops at various temperatures with the magnetic field perpendicular to the surface

Full size image

To understand the phase transition and growth mechanism from the perspective of phase stability, the DFT calculations of total energy were conducted. Table S2 shows the values of the total energy per atom for ML 1T-CrTe2, ML Cr2Te3, as well as BL Cr2Te3. It is evident that Cr2Te3 is the energetically more favored structure than ML 1T-CrTe2. Combined with the metastable properties of CrTe2 in bulk [50], which thus makes it mobile enough to undergo significant rearrangement to achieve a lower energy configuration (e.g., Cr2Te3) when the temperatures increase. Additionally, the thickness effect would lead to the switching of the crystalline phase between the 1T-CrTe2 (2 L) and Cr2Te3 (multilayer) at low temperatures. In the thin (2 L) case, 1T-CrTe2 experiences disparate ligand environments and a stronger substrate effect, and hence distinct interatomic exchange, resulting in stabilizing the 1T-CrTe2. However, as the thickness increases, the substrate effect gets weaker, and then the energetically favored phase (Cr2Te3) becomes dominated. During the growth process at high temperatures, the growth mode of Cr2Te3 converts from the bilayer to the monolayer as the thickness increases, which could also be ascribed to the thickness effect. The interfacial interactions between the thin Cr2Te3 film (2 L) and Si substrate could play a key role in the bilayer growth mode, therefore, although the total energy per atom of BL Cr2Te3 is higher than that of ML Cr2Te3 [25, 51], the BL growth mode is favored at the initial growth stage.

The magnetic characterization of the Cr2Te3 thick films was carried out by VSM experiments. Figure 4(h) shows the magnetization versus temperature curve under field-cooling (FC) and zero-field-cooling (ZFC) treatments. A 500 Oe magnetic field was applied perpendicular to the surface plane during the measurement. The magnetization in the FC curve decreases continuously with increasing temperature. While for the ZFC curve, it first increases to its maximum value and then decreases with increasing temperature. Such pronounced bifurcation between FC and ZFC curves at low temperatures suggests a spin-glass behavior, which can be explained by the randomness and spin frustration [52]. The randomness can be obtained by spin with site randomness [53]. In our sample, the disorder intercalation of rich Cr-atoms brings the site randomness. The spin frustration results from the competition between ferromagnetic and antiferromagnetic exchange interactions, which has been reported in chromium telluride system [30]. In the ZFC curve, the cusp of about 160 K corresponds to the spin-glass freezing temperature \(T_{f}\). The magnetic moments gradually freeze into random orientations below \(T_{f}\), leading to a decrease in the magnetization with the temperature. For the FC curve, however, the magnetic moments will be partially polarized by the cooling field and produce a net magnetization, causing the increasing of a magnetization. The Curie temperature (Tc) can be roughly determined by performing the first derivation of FC curves. As the inset of Fig. 4(h) shows, the valley in the dM/dT FC curves indicates the Tc is 180 K. The magnetic hysteresis loops, as illustrated in Fig. 4(i), show a typical behavior of soft ferromagnetism below Tc.

3 Conclusion

In this work, 2D ferromagnetic Cr(1+δ)Te2 thin films were successfully fabricated on semiconductive Si (111) substrates. The crystalline phase transition from CrTe2 to Cr2Te3 can be realized by increasing the thickness and growth temperature, which is supported by the in-situ RHEED, STM, ex-situ XPS, XRD measurements, and computational predictions. At higher growth temperatures, the film growth mode changes from bilayer Cr2Te3 to monolayer Cr2Te3 as the thickness increased. Such phase and growth mode transitions are derived from the interfacial effects and the energetic stability of Cr(1+δ)Te2 phases. Additionally, the electronic and magnetic properties of Cr2Te3 are also characterized. Our work provides an efficient strategy to controllably fabricate 2D ferromagnetic materials and paves the way toward the investigation of 2D magnetic systems, creating opportunities for future 2D spintronic architectures.

4 Methods

The Cr(1+δ)Te2 thin films were grown on the conductive P-doped Si (111) wafers (n-type doped, 0.01 Ω⋅cm) substrates by a typical MBE method. A standard heating and flashing procedure was carried out to obtain an atomically clean 7×7 surface reconstruction [54]. High-purity Te and Cr were evaporated from an effusion cell and an electron beam evaporator, respectively. The flux ratio between Cr and Te is 1:12. The substrate temperature was set from 250°C to 390 °C during growth. A post-annealing process with 5 minutes at growth temperature contributes to a better-quality film.

The STM measurements were performed in an ultrahigh vacuum (UHV) system (Unisoku-1300) mainly at 77 K unless described exclusively. Electrochemically etched tungsten tips were used after flashing and silver decoration in situ. The tunneling spectra were obtained by a lock-in amplifier with a modulation of 10 mV at a frequency of 991 Hz. The topographic images were processed by WSxM.

The stoichiometric ratio of Cr(1+δ)Te2 film was characterized by an ex-situ X-ray photoelectron spectroscopy (XPS: AXIS Ultra DLD). Binding energy was corrected using the C 1 s peak fixed at 284.8 eV. Each component was fitted by CasaXPS. The crystal structure of 30-nm Cr2Te3 was obtained by an ex-situ X-ray diffractometer (XRD, DX-2700BH) with Cu Kα radiation (\(\alpha = 1.542\) Å). The scanning step, voltage, and current of XRD were set to 0.02°, 40 kV, and 30 mA, respectively. The magnetization measurements of 30-nm Cr2Te3 were carried out with a VSM magnetometer in a physical property measurement system from Quantum Design. Magnetic fields were applied perpendicular to the sample surface. For the ex-situ measurements, a 10 nm Te layer was capped on the surface of Cr(1+δ)Te2 film to protect the sample from oxidation.

The ab initio calculations were performed under the framework of DFT as implemented in the Vienna ab initio simulation package (VASP) code [55]. The generalized gradient approximation (GGA) parametrized by Perdew–Burke-Ernzerhof (PBE) was used for the exchange-correlation functional [56], and the interaction of the valence electrons with the remaining ions was represented by projector-augmented-wave (PAW) method. During the calculation, the plane wave energy cutoff was set to 500 eV, and a vacuum layer above 15 Å in the Z direction was used to avoid the effect of periodicity. The energy conversion criterion is 10−5 eV. All the structures were fully relaxed until the force is less than 0.01 eV/Å, and the K-point grid is selected as 9 × 9 × 1 [57].