Epitaxial calcium fluoride (CaF2) layers with a nominal thickness up to 10 nm on the (111)-oriented Silicon (Si) are obtained. Surface topographies of the fluoride films are recorded and the current-voltage characteristics of the Au/CaF2/Si structures are studied. On a qualitative level, these structures exhibited all the features usual for metal-insulator-semiconductor systems. The current-voltage curves of the samples were reproduced by modeling considering a finite (0.1-1 nm) value of the standard thickness deviation of the dielectric CaF2 film. Keywords: calcium fluoride, thin films, MIS structure, leakage current.

Graphene is a promising material for applications as a channel in graphene field-effect transistors (GFETs) which may be used as a building block for optoelectronics, high-frequency devices and sensors. However, these devices require gate insulators which ideally should form atomically flat interfaces with graphene and at the same time contain small densities of traps to maintain high device stability. Previously used amorphous oxides, such as SiO 2 and Al 2 O 3 , however, typically suffer from oxide dangling bonds at the interface, high surface roughness and numerous border oxide traps. In order to address these challenges, here we use 2 nm thick epitaxial CaF 2 as a gate insulator in GFETs. By analyzing device-to-device variability for about 200 devices fabricated in two batches, we find that tens of them show similar gate transfer characteristics. Our statistical analysis of the hysteresis up to 175 o C has revealed that while an ambient-sensitive counterclockwise hysteresis can be present in some devices, the dominant mechanism is thermally activated charge trapping by border defects in CaF 2 which results in the conventional clockwise hysteresis. We demonstrate that both the hysteresis and bias-temperature instabilities in our GFETs with CaF 2 are comparable to similar devices with SiO 2 and Al 2 O 3 . In particular, we achieve a small hysteresis below 0.01 V for equivalent oxide thickness (EOT) of about 1 nm at the electric fields up to 15 MV cm −1 and sweep times in the kilosecond range. Thus, our results demonstrate that crystalline CaF 2 is a promising insulator for highly-stable GFETs.

Hybrid ferromagnet-semiconductor systems possess new outstanding properties, which emerge when bringing magnetic and semiconductor materials into contact. In such structures, the long-range magnetic proximity effect couples the spin systems of the ferromagnet and semiconductor on distances exceeding the carrier wave function overlap. The effect is due to the effective p-d exchange interaction of acceptor-bound holes in the quantum well with d-electrons of the ferromagnet. This indirect interaction is established via the phononic Stark effect mediated by the chiral phonons. Here, we demonstrate that the long-range magnetic proximity effect is universal and observed in hybrid structures with diverse magnetic components and potential barriers of various thicknesses and compositions. We study hybrid structures consisting of a semimetal (magnetite Fe3O4) or dielectric (spinel NiFe2O4) ferromagnet and a CdTe quantum well separated by a nonmagnetic (Cd,Mg)Te barrier. The proximity effect is manifested in the circular polarization of the photoluminescence corresponding to the recombination of photoexcited electrons with holes bound to shallow acceptors in the quantum well induced by magnetite or spinel itself, in contrast to interface ferromagnet in case of metal-based hybrid systems. A nontrivial dynamics of the proximity effect is observed in the studied structures due to recombination-induced dynamic polarization of electrons in the quantum well. It enables the determination of the exchange constant Δexch ≈ 70 μeV in a magnetite-based structure. The universal origin of the long-range exchange interaction along with the possibility of its electrical control offers prospects for the development of low-voltage spintronic devices compatible with existing solid-state electronics.

Thin films of calcium fluoride with a nominal thickness from 2 up to 20 nm were grown on silicon-(111) surface by means of molecular beam epitaxy. The interest to fluorite insulating films is stipulated by the potential applications in field-effect transistors with a 2D channel. The studied Au/CaF2/Si heterostructures exhibited behavior which is typical for a metal-insulator-semiconductor system. In order to attain a quantitative agreement between the measured current-voltage curves and the theoretical model, spatial fluctuations of the fluoride layer thickness had to be accounted for. The standard deviation of the CaF2 layer thickness was found to be between 0.2 and 1.2 nm tending to increase with the nominal thickness. These values are in agreement with root mean square surface roughness measured by atomic force microscopy. A satisfactory reproducibility of the electric characteristics of the studied films makes CaF2 a promising insulating material for electronic devices.

Epitaxial calcium fluoride (CaF2) layers with a nominal thickness up to 10 nm on the (111)-oriented Silicon (Si) are obtained. Surface topographies of the fluoride films are recorded and the current-voltage characteristics of the Au/CaF2/Si structures are studied. On a qualitative level, these structures exhibited all the features usual for metal-insulator-semiconductor systems. The current-voltage curves of the samples were reproduced by modeling considering a finite (0.1–1 nm) value of the standard thickness deviation of the dielectric CaF2 film.

Two-dimensional (2D) electronics can enable FETs down to a few nanometers. However, these devices require scalable insulators which should form high-quality interfaces with 2D channels and maintain low gate leakage currents for sub-1nm equivalent oxide thickness (EOT). Previously used amorphous oxides result in poor interfaces with 2D materials, while hBN has mediocre dielectric properties ( ε < 5, E g = 6eV) [1] . As a promising alternative, we suggest the use of the crystalline ionic insulator CaF 2 ( ε = 8.43, E g = 12.1eV) which forms van der Waals interfaces with 2D semiconductors [2] . At the moment, CaF 2 can be grown by molecular-beam epitaxy (MBE) down to a few nanometers thickness [3] and appears promising for chemical vapour deposition (CVD) [4] and atomic-layer deposition (ALD) [5] . Here we discuss our recent progress [3] , [6] , [7] on ultra-thin CaF 2 which presents a universal platform for 2D devices. In particular, we demonstrate nanoscale MoS 2 FETs with L =50-60nm and a record-thin ~ 2nm CaF 2 insulator (EOT~ 0.9nm) which exhibits near-ideal subthreshold swing (SS).

It is shown for the first time that Co subnanometer coatings deposited by molecular-beam epitaxy on the (0001) surface of the topological insulator BiSbTeSe2 at a temperature of 330°C open an energy gap in the spectrum of topological surface states in the region of the Dirac point with a shift of the Dirac-point position caused by the preliminary deposition of an adsorbate at room temperature. The gap width is 21 ± 6 meV. Temperature-dependent measurements in the range of 15–150 K show no changes in the energy-gap width.

It was shown for the first time that Co subnanometer coaverage, being deposited by molecular beam epitaxy method onto the (0001) surface of the BiSbTeSe2 topological insulator at 330 °C, opens an energy band gap in the spectrum of topological surface states in the region of the Dirac point, with a shift in the position of the Dirac point caused by preliminary deposition of the adsorbate at room temperature. The gap band width is 21 +/- 6 meV. Temperature-dependent measurements in the 15-150 K range did not show any width changes.

Fabrication of ultrascaled 2D field effect transistors (FETs) requires 2D semiconductors with sizable bandgaps and high carrier mobilities as well as competitive insulators to separate the channel from the gate. However, so far the focus has been put on channel materials, while insulators suitable for ultrascaled 2D FETs have not been identified. The most common choice is oxides known from Si technologies, e.g. SiO2, Al2O3, and HfO2. However, despite the respectable performance of some devices, the interfaces of these oxides with 2D channels are of poor quality and contain numerous defects which degrade their performance and reliability. On the other hand, while 2D insulators such as hBN form well-defined van der Waals interfaces with 2D channels, the poor dielectric properties of hBN make it unsuitable for scaling. As an alternative, we have recently suggested the use of calcium fluoride (CaF2) as an insulator for 2D FETs [1]. Few-nanometers thin CaF2 layers can be grown on Si(111) by molecular beam epitaxy (MBE) which forms an F-terminated inert surface with no dangling bonds [2]. This results in a quasi van der Waals interface with 2D materials (Fig.1a), similar to those known from hBN. Furthermore, due to its good dielectric properties, the tunnel currents through CaF2 are lower than for most high-k oxides with equal EOT, not to mention SiO2 and hBN [1]. The gate currents of our CVD-grown MoS2 FETs with epitaxial CaF2 insulators of record-small thickness of only about 2 nm (EOT less than 1nm) are small compared to the drain current (Fig.1c). Thus, already in the first bare channel prototypes we achieve competitive on/off current ratios of up to 10 and SS down to 90 mV/dec (Fig.2a). At the same time, the hysteresis in our devices is even smaller than in Al2O3 encapsulated SiO2(25nm)/MoS2 FETs [3] (Fig.2b). In Fig.2c we compare the normalized hysteresis widths for various technologies where only CaF2 gives comparable results to Si/high-k FETs. In addition, we analyzed the reliability of our devices and found it to be significantly improved compared to available 2D FETs [4]. In summary, we demonstrated the feasibility of FETs with CaF2 insulators and found that the virtually defect-free nature of CaF2 leads to good performance, small hysteresis, and excellent reliability.

The currents flowing in metal–CaF 2 – n -Si and metal–SiO 2 –CaF 2 – n -Si structures with the same (about 1.5 nm) fluoride thickness are compared in the reverse-bias mode. It is revealed that the current in the case of a two-layer dielectric can be notably higher within a certain voltage range. Such unexpected behavior is associated with the coexistence of both electron and hole components of the current as well as with the configuration of the SiO 2 –CaF 2 barrier through which tunneling occurs. The results of measurements and explanatory simulation data are presented.

Two-dimensional (2D) materials can potentially provide a route to overcome the limitations of Si technologies by enabling nanoscale more than Moore FETs. Fabrication of these devices requires i) 2D semiconductors with sizable bandgaps and high carrier mobilities and ii) competitive insulators to separate the channel from the gate. However, up to now attention has been mostly paid to the channel materials, while insulators fully suitable for 2D FETs have not been identified. For instance, native oxides of 2D semiconductors, which would go along with them as well as SiO2 goes with Si, either do not exist or cannot be easily synthesized. As a result, 2D FETs currently face the same problems as many other emerging technologies (e.g. Ge, IIIV or GaN FETs), and commercial devices do not yet exist. The absence of native insulators for 2D FETs requires the use of other materials. The most obvious of them are oxides known from Si technologies (e.g. SiO2, Al2O3, HfO2), which have been used in most 2D FETs. However, despite a respectable performance of some devices, the typical thickness of the oxides used ranges from tens to even hundreds of nanometers. When being scaled down to the equivalent oxide thicknesses (EOT) below 1nm, as required for end-of-the-roadmap devices, these oxides become amorphous. Thus, their interfaces with 2D channels are of poor quality and contain numerous defects which degrade the performance and reliability of 2D FETs. Another solution is the use of 2D insulators which form well-defined van der Waals interfaces with 2D channels. The best known of them is hBN which indeed resulted in the improvement of 2D FETs. However, when scaled down to several nanometers thickness, hBN exhibits significant tunnel leakage currents. This is due to the moderate dielectric properties (ε < 5, EG = 6 eV) of hBN, which appears to be unsuitable for scaling. As for other 2D insulators, such as mica or oxide nanosheets, their usability in devices has not been demonstrated. As an alternative, we have recently suggested to use calcium fluoride (CaF2) as an insulator for 2D devices [1]. Few-nanometers thin CaF2 layers can be grown on Si(111) by molecular beam epitaxy (MBE) which forms an Fterminated inert surface with no dangling bonds [2]. This results in a quasi van der Waals interface with 2D materials (Fig.1a), similar to those formed by hBN. Furthermore, due to its good dielectric properties (ε = 8.43, EG = 12.1eV), the tunnel currents through CaF2 are lower than for most high-k oxides with equal EOT, not to mention SiO2 and hBN. We fabricated hundreds of CVD-grown MoS2 FETs with epitaxial CaF2 insulators of record-small thickness of only about 2 nm (EOT less than 1nm), see Fig.1b. The gate currents in our devices are small compared to the drain current (Fig.1c). Thus, already in the first bare channel prototypes we achieve competitive on/off current ratios of up to 10 and SS down to 90 mV/dec (Fig.2a). At the same time, the hysteresis in our devices is even smaller than in Al2O3 encapsulated SiO2(25nm)/MoS2 FETs [3] (Fig.2b). In Fig.2c we compare the hysteresis widths normalized by the insulator field factor ΔVG/dins, where ΔVG is the width of the gate voltage sweep range and dins the insulator thickness. For CaF2 the hysteresis is comparable to that in Si/high-k FETs. In Fig.3 we show the results for bias-temperature instabilities (BTI) in our CaF2(2nm)/MoS2 FETs. The negative BTI (NBTI, Fig.3a) and positive BTI (PBTI, Fig.3b) at insulator fields which are typically used for 2D FETs (2.5 to 5MV/cm) are relatively weak. This is likely due to the lack of insulator defects in the crystalline CaF2 insulator. However, PBTI stress at the insulator field of 7.5MV/cm leads to a negative drift of the threshold voltage (Fig.3c). This kind of degradation has never been observed for 2D FETs with thick insulators, which routinely operate at lower insulator fields. Recently we found that the negative Vth shift after PBTI stress observed in our CaF2/MoS2 FETs is very similar to that in MoS2 FETs with 4nm thick hBN insulators [4]. Thus, we suggest that this is due to the activation of breakdown mechanisms in ultrathin insulators. However, we note that a reasonable on current in our CaF2/MoS2 FETs can be achieved already at a gate voltage of 1V (Fins of 5MV/cm), allowing safe device operation. We also analyzed the thermal stability of our devices and found that it strongly depends on the quality of the MoS2 channel. If the CVD-grown MoS2 film is formed by few nanometers sized grains (Process 1, Fig.4a), baking of the devices at 100C introduces a strong hysteresis (Fig.4b). However, this hysteresis is observed only when using small sweep times tsw and disappears for slow sweeps (Fig.4c). We suggest that the origin of this behavior is thermally enhanced creation of S vacancies in MoS2. These S vacancies can cause a hysteresis when interacting with adsorbates [5], which can penetrate to the CaF2/MoS2 interface through the numerous grain boundaries in the small grain MoS2. In contrast to charge trapping by slow insulator defects, which seems to be missing in CaF2, interaction between S vacancies and adsorbates is a fast process. This explains the fast sweep hysteresis which we observe. In contrast, in the CaF2/MoS2 FETs with larger grains of the MoS2 film (Process 2, Fig.5a) the hysteresis remains small even after baking at 165C (Fig.5b), though a negative shift of Vth attributed to the creation of S vacancies is present. However, these S vacancies remain passive, since the number of grain boundaries in large grain MoS2 is small and thus the penetration of adsorbates to the interface is less efficient. At the same time, the number of active defects in crystalline CaF2 is small, independently of the MoS2 quality. Thus, the hysteresis in our Process 2 devices remains small within the whole range of sweep times, while being smaller than in Al2O3 encapsulated SiO2(25nm)/MoS2 FETs (Fig.5c). In summary, we examined the reliability and thermal stability of the MoS2 FETs with CaF2 insulators of record small 2nm thickness. We found that the virtually defect-free nature of CaF2 insulators leads to a small hysteresis and BTI, while the thermal stability of our devices strongly depends on the quality of the MoS2 channel.

Sub-angstrom Co coverage, being deposited on BiSbTeSe2(0001) surface at 200-330 C, opens a band gap at the Dirac point, with the shift of the Dirac point position caused by RT adsorbate pre-deposition. Temperature dependent measurements in 15-150 K range have shown no band gap width change. This fact indicates the nonmagnetic nature of the gap which may be attributed to the chemical hybridization of surface states upon the introduction of Co adatoms, which decrease crystallographic symmetry and eliminate topological protection of the surface states.

AbstractThe currents flowing in metal–CaF_2– n -Si and metal–SiO_2–CaF_2– n -Si structures with the same (about 1.5 nm) fluoride thickness are compared in the reverse-bias mode. It is revealed that the current in the case of a two-layer dielectric can be notably higher within a certain voltage range. Such unexpected behavior is associated with the coexistence of both electron and hole components of the current as well as with the configuration of the SiO_2–CaF_2 barrier through which tunneling occurs. The results of measurements and explanatory simulation data are presented.

The effect of increasing the tunnel current in a metal–calcium fluoride–silicon structure with addition of a silicon dioxide layer between fluoride and metal (which seems paradoxical at first glance) has been considered. This effect of nonmonotonic change in the tunnel conductivity with an increase in the insulator thickness may occur at a relatively high bias at the structure and is related to the tunnel-barrier deformation, at which electrons are tunneling through its part formed by the oxide. At low biases, the occurrence/ thickening of an additional layer leads to a natural decrease in the current. Similar behavior is possible in principle for some other combinations of materials.

AbstractThe effect of increasing the tunnel current in a metal–calcium fluoride–silicon structure with addition of a silicon dioxide layer between fluoride and metal (which seems paradoxical at first glance) has been considered. This effect of nonmonotonic change in the tunnel conductivity with an increase in the insulator thickness may occur at a relatively high bias at the structure and is related to the tunnel-barrier deformation, at which electrons are tunneling through its part formed by the oxide. At low biases, the occurrence/ thickening of an additional layer leads to a natural decrease in the current. Similar behavior is possible in principle for some other combinations of materials.

The growth and crystal structure of NiF2 layers on CaF2/Si(111) heteroepitaxial substrates have been investigated. It has been shown that molecular beam epitaxy at temperatures of 350–450°C provides a stable epitaxial growth of the metastable orthorhombic NiF2 phase (structural type CaCl2) with a nickel fluoride layer thickness up to 1 μm in the metastable phase. According to X-ray diffraction, the unit cell parameters in layers of orthorhombic nickel fluoride are a = 4.5680(1) Å, b = 4.7566(3) Å, and c = 3.0505(2) Å, which are very close to the known values for this phase. It has been established that the condition \((100)_{NiF_2 } \left\| {(111)_{CaNiF_2 } } \right.\) holds over a wide range of growth parameters, which agrees with the results of the qualitative crystallographic analysis of the elements of similarity of the structures under consideration. The formation of a domain texture, the character of which depends on the growth temperature and nickel fluoride layer thickness, has been observed in the heterojunction plane.

LaF3 films in the 5-40 nm thickness range were grown on Si(111) by molecular beam epitaxy. The substrates were kept at 450 degrees C during deposition. The films were investigated by high-energy X-ray photoemission flanked by conventional X-ray photoemission, reflection high-energy electron diffraction, and atomic force microscopy. The film growth was layer-by-layer. The surface of the films presented flat terraces, similar to 100 nm wide, separated by monatomic steps, reproducing the morphology of the substrate. La 3d, F Is, P Is, and Si 2p core levels and valence band were measured by high-energy photoemission to investigate the reactivity of the system and the surface and bulk composition of the films, following varying sample treatments (X-ray irradiation, sputtering, heating). The fresh prepared films resulted of high purity, with no traces of reaction or intermixing at the buried interface between the substrate and the trifluoride. The X-ray beam was seen to induce F depletion at the surface and promote oxide formation. F depletion enhancement was obtained through Ar ion sputtering. An irreversible variation of the film composition was finally observed for samples heated above 300 degrees C, with the development of La oxides and oxofluorides. These effects were related to the high mobility of F ions in the LaF3 lattice and to the high tendency of defects formation involving F sites.

Cobalt nano-structured ultrathin films were grown on orthorhombic MnF(2) by molecular beam epitaxy on CaF(2) epitaxial layers deposited on Si(111) substrates. The Co film was grown at room temperature. It was found to be polycrystalline, forming nano-islands with height≈diameter≤10 nm. X-ray absorption evidences the chemical stability of the Co/MnF(2) interface. Remarkably, x-ray magnetic circular dichroism (XMCD) demonstrates that the Co induces a net magnetization on the Mn ions close to the interface. The magnetic moments of these Mn ions couple antiparallel to the Co and rotate upon field reversal following the magnetization of the Co both below and high above the Néel temperature of MnF(2) (T(N) = 67 K). The density of coupled Mn moments is found to be temperature dependent, with an equivalent thickness of ~1.5 MnF(2) monolayers at 20 K, decreasing to about ~0.5 ML as the temperature is raised to 300 K. Interestingly, the intensity of the Mn XMCD signal appears to be related to the coercivity of the Co layer. This behavior is interpreted in terms of the competition between thermal fluctuations, exchange coupling between Co and Mn at the interface and, at low temperature, the antiferromagnetic order in MnF(2).

The longitudinal conductivity of La 1 − x Sr x F 3 − x solid solution films ( x = 0–0.24) with thicknesses of 40–260 nm grown on glass ceramics at temperatures from room temperature to 300°C and frequencies of 10 −1 –10 6 Hz was studied by impedance spectroscopy. The concentration dependence of film conductivity on the SrF 2 content had a maximum near x = 0.05. An equivalent circuit was constructed on the basis of the impedance plots to describe migration processes. The DC conductivity was evaluated for all samples under study. The activation energies were estimated from the temperature dependences of the DC conductivities of the films. The resulting dependences of electrophysical parameters were compared with those for bulk materials in terms of the relaxation conductivity model.