1. Introduction
Transition metal dichalcogenides (TMDs) feature a layered structure similar to graphene but with a direct bandgap, making them ideal semiconductor candidates for various applications [
1,
2,
3]. Structurally, TMDs can adopt three different crystalline forms: 1T, 2H, and 3R. Both the 2H and 3R phases exhibit trigonal prismatic coordination and are naturally semiconductors, but the 2H structure is more thermally stable than the 3R, making it more suitable for applications requiring stability [
4,
5,
6]. The crystal structure of a TMD monolayer belongs to the hexagonal crystal system, and its reciprocal space exhibits six-fold symmetry [
7,
8]. When the thickness of TMDs is reduced to a monolayer, the lack of inversion symmetry gives rise to the valley degeneracy in the reciprocal space to split into inequivalent K and K' points with opposite spin states [
9,
10,
11]. This valley-spintronic effect makes the electron spin and valley properties interdependent and sensitive to the lights with angular momentum [
7,
9,
12]. When specific circularly polarized light (either left- or right-handed) is incident on the TMD monolayer, it spin-dependently excites electrons in a specific valley, leading to different occupation levels in the two valleys and resulting in valley spin-polarization [
13]. This spin-dependent electron transition mechanism, by controlling valley properties, influences the optoelectronic and magnetic properties of TMDs and has given rise to a new research field known as valleytronics. Recent studies have explored various approaches to induce valley spin-polarization, such as using circularly polarized light for photoexcitation [
9,
14,
15] and the Zeeman effect in magnetic fields to break spin symmetry [
16,
17,
18]. It has been observed that valley spin-polarization can be triggered via the magnetic proximity effect (MPE) of ferromagnetic (FM) materials, which can be achieved at accessible field levels below 1 Tesla, as demonstrated by spin-sensitive spectroscopy [
19,
20,
21]. These findings provide possibilities for realizing TMD-based valley spintronics for more diverse applications. Although the MPE-induced valley spin-polarization in TMD monolayers is clear, an intriguing but unresolved question remains: Can the ferromagnetic order in antiferromagnets enable the MPE in TMD monolayers to incorporate more ingredients into future valleytronics? This issue is promising but requires a more comprehensive understanding.
This study attempts to resolve the magnetic proximity effect (MPE) induced by an antiferromagnet (AFM) NiO with a specific crystallographic orientation. NiO is a G-type AFM with a Néel temperature higher than room temperature [
22,
23]. It has been shown that the (111)-terminated surface of NiO can host the MPE with a depolarization temperature at approximately 100 K, while the (001)-terminated surface of NiO dramatically deactivates the MPE. This phenomenon revisits the importance of the FM order in AFM through uncompensated spins. This approach provides a new perspective on the study of magnetic van der Waals heterostructures using AFM materials.
2. Materials and Methods
In this study, a three-zone tube furnace was used for chemical vapor deposition (CVD) to grow monolayer MoS2 on a sapphire substrate. The molybdenum trioxide (MoO3) powders and sulfur (S) powders were used as precursors, positioned in the middle of the second zone and the front end of the first zone, respectively. The sapphire substrates were placed downstream of the MoO3 at the second zone. An argon atmosphere with a flow rate of 40 sccm was applied as the carrier gas to promote the reaction of MoO3 and S and the subsequent deposition onto sapphire substrates. During the CVD, the pressure inside the tube furnace was maintained at 4 torr. The set temperatures at the zone 1, 2, and 3, are 550 , 750 , and 800 , respectively. This creates a temperature gradient inside the tube, promoting the deposition with different distances from MoO3 at zone 2. After growing MoS2 monolayers on a sapphire substrate, a wet transfer technique was used to transfer the MoS2 samples onto the desired NiO substrates with (111)-terminated and (001)-terminated surface. Polymethyl methacrylate (PMMA) was spin-coated onto the sapphire substrate with the grown MoS2 monolayers. After a soft bake treatment at 120°C for 2 minutes, the sample was soaked in a 1M potassium hydroxide solution for 30 minutes, and then immersed in deionized water for cleaning. Due to the hydrophobic nature of PMMA, the PMMA adhered to MoS2 would lift off from the sapphire substrate and be transferred onto the surface of NiO with specific surface termination. The desired substrate is then used to pick up the PMMA/MoS2 film. After drying, the sample is soaked in acetone for 2 hours to remove the PMMA, thus completing the transfer process for subsequent characterizations. The Raman and photoluminescence (PL) spectra were collected using a Raman spectrometer with an argon laser source of 533 nm at room temperature. The laser beam was focused onto the sample’s surface and the spot size was approximately 1 m. The laser current was 0.755 A, and a 1200-line/mm grating was employed for an energy resolution of 2 meV. The integration time for collecting Raman and PL spectra was 30 and 5 seconds to yield distinguishable signals, respectively. Spin-resolved PL (SR-PL) was performed using linearly polarized light to pump and circularly polarized lights to probe the MoS2 monolayers on the NiO substrate at various temperatures.
3. Results
Figure 1a shows the morphology of the MoS
2 grown on the sapphire substrate, in which the MoS
2 domains appear to be randomly distributed and orientated on the sapphire substrate. It reveals the sapphire substrate is atomically flat, thus lacking the preferred nucleation sites on the sapphire surface during the deposition.
Figure 1b exhibits the Raman spectrum acquired from the MoS
2 domain as demonstrated in
Figure 1a. The two characteristic peaks at ~ 384 cm
-1 (
) and ~403 cm
-1 (
) correspond to the horizontal and vertical vibration modes of the MoS
2 monolayer. The monolayer nature of the grown MoS
2 is indicated by the wave number difference between
and
, approximately 19 cm
-1 [
24,
25]. An atomic force microscope (AFM) was employed to probe the step height at the edge of MoS
2 to examine the properties of the monolayer for the associated valleytronics properties.
Figure 1c shows the topography of MoS
2 grown on the sapphire substrate, and
Figure 1d shows the step profile of the investigated MoS
2 probed along the scanning trajectory, as shown by the blue line in
Figure 1c. As a result, the step height is approximately 0.74 nm, suggesting the MoS
2 in the form of monolayer fabricated by CVD.
After the fabrication of the MoS
2 monolayers and the characterization, the focus was turned to the effect of transferring MoS
2 monolayers onto the NiO substrate.
Figure 2a shows the crystal and magnetic structure of the NiO, featuring a rock-salt structure with the AFM texture in a G-type configuration [
26,
27]. Based on the magneto-structural configuration, the NiO with (111) termination yields a significant FM order on the surface as an uncompensated plane. On the contrary, the NiO with (001) termination is magnetically compensated, resulting in no considerable magnetization on the surface and serving as the natural AFM order in NiO.
Figure 2b shows the Raman spectra before and after the MoS
2 transfer onto the NiO substrate with (111) termination, denoted as NiO
(111), in which the
state didn’t vary notably but the
state appeared to be softened after the transfer. Based on the current literature, the
state is relatively sensitive to the strain/stress issue arising from the different interfacial coherency with the substrate, therefore, the transfer treatment gave rise to limited strain issues. However, the
softening suggests the electron doping effect arising from the charge transfer from the substrate [
12,
28,
29,
30]. The result suggests that interfacing MoS
2 with NiO would stabilize the n-type characteristic of MoS
2. Furthermore, the PL spectra before and after the MoS
2 transfer both reflect the exciton state ~1.83 eV as an indicator of the high monolayer quality [
31], but it shows the PL red-shifting ~10 meV upon transferring the MoS
2 monolayers onto the NiO
(111). Both Raman and PL spectra suggest the MoS
2 on NiO
(111) indeed underwent an electronic transition after the transfer. A detailed discussion regarding the electronic modification on MPE will be provided after having the results of SR-PL characterization.
Figure 3a shows the SR-PL spectra of MoS
2 on NiO
(111) with temperature dependency, probed using the right-handed (RCP) and left-handed (LCP) circularly polarized lights. SR-PL taken at 4 K exhibited two distinguishable states at 1.93 eV and 1.97 eV corresponding to the trion and exciton states, denoted as T and A, respectively. As the results, the SP-PL taken at 4 K exhibits a distinguishable magnetic circular dichroism (MCD) at both T and A states, suggesting the valley spin-polarization in the MoS
2 monolayers triggered by the MPE of NiO
(111) substrate. It could be noticed that both T and A states underwent a redshift transition upon increasing the temperature. It is because of the thermal expansion driving electrical band gap reduction, thus reducing the optical band gap [
32,
33]. In addition, it has been shown both T and A states are responsible for the valley spin-polarization while putting the MoS
2 monolayers on NiO
(111).
Figure 3b exhibits the SR-PL results of the control sample comprising MoS
2 monolayers on NiO
(001). As expected, the NiO
(001) has a fully spin-compensated surface, thus resulting in no distinguishable MCD via the probe of SR-PL. Comparing the results in
Figure 3a,b, it reveals the crystallographic control of the AFM NiO substrate appears to be a critical switch to induce the valley spin-polarization via MPE.
Figure 3c exhibits the plots of the degrees of the valley spin-polarization for the T and A states of the MoS
2 monolayers on NiO
(111) and the A state of the MoS2 monolayers on NiO(001). It shows the depolarization of the T and A states acquired by the MoS
2 monolayers on NiO
(111) is roughly at 100 K, which is relatively high in the associated studies [
19,
34]. Besides, the valley spin-polarization degree mismatch among the T (NiO
(111)), T (NiO
(111)), and the A (NiO
(001)) at 4 K may reveal two important signatures: (1) The (111)-terminated surface of NiO substrate enables activating the MPE as a crystallographic switch. (2) T state appears to be much more sensitive to the MPE, hence giving rise to the higher valley spin-polarization degree. In order to explain the latter observation, a charge-transfer-induced MPE is introduced together with the origin of the valley spin-polarization.
Let’s revisit the SR-PL results and discuss the origin of the MPE based on the spin configuration in the MoS
2/NiO heterostructure. As shown in
Figure 3a, both T and A states existed at low temperatures (<100 K) and it should be noticed that the T state is much correlated with magnetism.
Figure 4a conceptually depicts the magnetism associated with the T state, in which the spins carried by the electron and hole located in the
K valley and the spin carried by the electron in the
K’ valley are exhibited. Consequently, the spins carried by the electron and hole located in the
K valley would compensate each other in nature, thus resulting in no magnetism. However, the additional spin located in the K’ valley yields the net moment. The T states highlighted by the blue and red triangle shadows represent the local net moments with spin up and spin down, denoted as T(
) and T(
), respectively. In general, both T(
) and T(
) should populate equally and therefore leading to no magnetism. However, once the MoS
2 monolayers interface with the NiO with (111) termination, the interfacial exchange coupling between MoS
2 and NiO
(111) would favor one of the T states and induce the imbalanced T population on the net valley spin-polarization. On the counterpart, upon interfacing MoS
2 monolayers with NiO
(001), the fully compensated surface, as shown in
Figure 4b, would result in no population preference on the T state, therefore deactivating the MPE to diminish the valley spin-polarization. This observation suggests precise control over the crystallographic growth of AFM functions as an effective switch to activate the MPE in MoS
2 monolayers, which is not only scientifically interesting but also practical for valley spintronics.