Charge-transfer-enhanced d–d emission in antiferromagnetic NiPS₃

Van der Waals (vdW) magnetic materials have gained tremendous attention since the discovery of intrinsic ferromagnetic order in atomically thin chromium triiodide (CrI₃) and chromium germanium telluride (Cr₂Ge₂Te₆).^1–6 Particularly, antiferromagnetic metal phosphorus trichalcogenides (MPX₃, M = Mn, Fe, Ni; X = S, Se) are becoming attractive for their promising potential for next-generation spintronics.^7–10 Among this family of vdW magnetic materials, nickel phosphorus trisulfide (NiPS₃) has been extensively studied in recent years with many intriguing phenomena reported, such as phonon–magnon coupling,¹¹ charge-spin correlation,¹² and spin-correlated photon emission.^13–15 However, although the unpaired d electrons in transition metal atoms of magnetic materials play crucial roles in determining their properties, studies on the transition between d electronic states of NiPS₃, especially the d–d emission, were very limited even though they possess the promising potential to couple to the magnetic property.^16,17 For example, previous studies have observed the photoluminescence (PL) due to the d–d transition in vdW magnetic materials CrX₃ (X = I, Br), which is spontaneous circularly polarized light determined by the magnetization direction.^18,19 However, researchers have revealed the weak absorption of the d–d transitions that are forbidden by selections rules even in bulk NiPS₃,^20–22 making it arduous to investigate the d–d emission.

To enhance PL intensity, researchers have invented several approaches, such as constructing dielectric or plasmonic nanostructures,^23,24 using optical cavity,^25,26 and applying external strain.^27,28 However, these methods usually require multiple steps of lithography and the PL enhancement that usually happen in localized positions. Compared to the aforementioned methods, vdW heterostructures constructed by stacking vdW materials are used as excellent platforms to tune the properties of the materials due to their convenient fabrication and unique interlayer interactions, such as charge transfer,^29–34 proximity effect,^35–37 and moiré potential.^38–41 Among all types of heterostructures, type-I heterostructure has a straddling band alignment formed by different materials.⁴² Thus, charge carriers generated in the broader-bandgap layer will transfer to the narrower-bandgap layer when the two layers are in close contact, leading to the increase in the carrier concentration and consequently PL enhancement in the narrower-bandgap layer. Utilizing this feature, the PL enhancement in the narrower-bandgap layer has been reported in many vdW stacked heterostructures, such as MoTe₂/WSe₂,³¹ MoSe₂/FePS₃,³² and InSe/MoS₂.⁴³ 

Here, we report the observation of d–d emission in NiPS₃ and its enhancement in type-I vdW heterostructure between NiPS₃ and monolayer tungsten disulfide (WS₂). A broad emission from NiPS₃ with energy of around 1.7 eV is observed. This emission is assigned to the d–d transition of NiPS₃ from the ³T_1g to the ³A_2g state in Ni²⁺ ions under an octahedral crystal field. Compared to the bare NiPS₃ itself, the d–d emission in the WS₂/NiPS₃ heterostructure is enhanced by ∼10 − 50 times under different energies of laser excitation. The excitation wavelength and power dependence of the emission are investigated, showing that the charge transfer between NiPS₃ and WS₂ is the origin of the enhancement. We further investigate this d–d emission to reveal the fine energy level structure of the d electrons by performing temperature dependent measurements. A splitting of the d–d emission peak into two peaks is observed below the Néel temperature (i.e., 152 K), showing an energy difference of ∼105 meV between the split ³A_2g and ³E_g states due to the trigonal crystal field. Furthermore, polarization measurements on the d–d emission suggest a non-polarized property distinct from the spin-correlated emission reported previously.^13–15 This work deepens our understanding about the physical properties of the emerging vdW antiferromagnetic NiPS₃ and could serve as a model for the study of other antiferromagnets in the family.

The NiPS₃ flakes were prepared by mechanical exfoliation of its bulk single crystals that were synthesized using the chemical vapor transport (CVT) method.⁴⁴ Detailed characterizations of the bulk crystals are shown in Fig. S1. We used an all-dry transfer method to stack monolayer WS₂ onto the top of the NiPS₃ flake that was mechanically exfoliated on an SiO₂/Si substrate.⁴⁵ Figure 1(a) shows the schematic of a WS₂/NiPS₃ van der Waals heterostructure. Figure 1(b) shows a typical optical image of a monolayer WS₂ stacked on an NiPS₃ flake, showing the heterostructure region and the individual NiPS₃ and WS₂ regions for a comparison study. Raman spectra and atomic force microscopy (AFM) image of the heterostructure (Fig. S2) show that the WS₂ flake is a monolayer, and the NiPS₃ flake is about 122 nm thick. Figure 1(c) shows PL spectra under the excitation of 2.33 eV collected from three different regions of the sample at 78 K. The spectrum from monolayer WS₂ shows a pronounced excitonic emission at 2.02 eV. The spectrum from the bare NiPS₃ region features signals in the range of 1.3–1.5 eV, including the spin-correlated emission (peak X) reported recently.^13–15 Note that a very weak emission at 1.7 eV can be observed in the bare NiPS₃ down to six layers (Fig. S3), which has not been reported before. The energy of this emission at 1.7 eV matches well with one of the d–d transitions in NiPS₃ reported in the literature.¹⁶ When stacking WS₂ and NiPS₃ together, this emission is ∼13 times enhanced under excitation of 2.33 eV in the heterostructure region compared to that in the bare NiPS₃. The PL intensity color map in Fig. 1(d) shows that this enhancement occurs in the whole WS₂/NiPS₃ region, excluding the possibility of a localized phenomenon. The excitonic emission of WS₂ is strongly quenched with an energy blue shift in the heterostructure (Fig. S4), indicating a significant interlayer interaction between WS₂ and NiPS₃.⁴⁶ These observations clearly show that the interaction between WS₂ and NiPS₃ plays a key role in enhancing the d–d emission of NiPS₃. In addition, we also notice that the emission in the range of 1.3–1.5 eV of NiPS₃ does not show changes in the heterostructure, which needs further investigation in future studies.^13–15,47

NiPS₃ crystal has the layered van der Waals structure belonging to the monoclinic space group C_2/m, with the honeycomb structure of Ni²⁺ ions distributed in the (P₂S₆)⁴⁻ dimers (Fig. S1).^48,49 The (P₂S₆)⁴⁻ dimers create a trigonally distorted octahedral crystal field to the Ni²⁺ ions, leading to the splitting of the d-orbitals on Ni. Specifically, the degenerated d orbitals of the Ni²⁺ ion (which forms the ³F state) splits into the ground state ³A_2g and the excited states ³T_2g and ³T_1g under the O_h crystal field. The ³T_1g state will further split into ³E_g and ³A_2g states because of the trigonal crystal field (D_3h) formed by the distortion of the O_h crystal field [Fig. S5(a)].^21,50 The electronic transitions between these splitting states, also known as the d–d transitions, are governed by the spin and Laporte selection rules with light excitation.⁵¹ Specifically, only the transitions without multiplicity change (ΔS = 0) and with parity change (Δl = ±1) are allowed to occur. The d–d transition between the ³T_1g and ³A_2g states in NiPS₃ with the transition energy of ∼1.7 eV is spin allowed yet Laporte forbidden, resulting in a weak absorption from the room temperature (300 K) to the cryogenic temperature (4.2 K),²¹ which also accounts for the hardly observable d–d emission [Fig. S5(b)]. However, in the type-I WS₂/NiPS₃ heterostructure, the ³T_1g and ³A_2g states of NiPS₃ locate inside between the conduction band minimum (CBM) and the valence band maximum (VBM) of WS₂ [Fig. 2(a)],⁵² making the photo-excited electrons and holes transfer readily from the WS₂ layer to the NiPS₃ layer. This interlayer transfer increases the carrier population on the d states in NiPS₃, leading to the enhancement of the d–d emission in NiPS₃.

This hypothesis is supported by the excitation wavelength-dependent measurements at 78 K. As shown in Fig. 2(b), we compare the PL spectra from the bare NiPS₃ and WS₂/NiPS₃ heterostructure under the excitations of 1.91, 2.18, 2.33, and 2.71 eV. When the excitation energy (e.g., 2.18, 2.33, and 2.71 eV) is larger than the WS₂ bandgap energy (1.95 eV),⁵² a strong enhancement of the d–d emission is observed in the WS₂/NiPS₃ heterostructure; when the excitation energy (e.g., 1.91 eV) is lower than the WS₂ bandgap energy, no enhancement is observed, where the d–d emission is as weak as that from the bare NiPS₃. The reason for this phenomenon is that, only when the excitation energy is higher than the bandgap energy of WS₂, the photo-excited electrons and holes can be generated and transferred to NiPS₃ for the d–d emission enhancement. As discussed before, the charge transfer from WS₂ to NiPS₃ also leads to remarkable quenching of the WS₂ excitonic emission in the heterostructure region. Figure 2(c) plots the intensity ratio (R) of WS₂ emission from the heterostructure region and the bare WS₂ region under excitation from 2.18 to 2.71 eV, and the corresponding d–d emission enhancement factor (EF) defined as the intensity ratio of the d–d emission from the heterostructure and the bare NiPS₃ region. Note that smaller R corresponds to stronger quenching of the WS₂ excitonic emission and larger degree of charge transfer at the WS₂/NiPS₃ interface. Figure 2(c) clearly shows the correlation between R and EF: the R value reaches the minimum at 2.5 eV where EF reaches the maximum, suggesting that larger degree of charge transfer between the layers would lead to stronger d–d emission of NiPS₃.

We further investigate the excitation power dependence of the d–d emission to gain more insight into the photo-physics of the heterostructure. As shown in Fig. 3(a), we measure PL spectra of the WS₂/NiPS₃ region using a 2.33 eV excitation with power from 10 μW to 2 mW. The PL features of the heterostructure, including the NiPS₃ emissions in the energy range of 1.3–1.5 eV, the WS₂ excitonic emission at 2.02 eV, and the d–d emission at around 1.7 eV, are observed in the power range we measured. Their intensities all increase with the increase in excitation power. As a comparison, the bare NiPS₃ exhibits a weaker d–d emission in the same power range shown in Fig. 3(b). Figure 3(c) plots the integrated PL intensity as a function of the excitation power, which is fitted by the power law $IPL∝Pexck$⁠, where $IPL$ is the integrated PL intensity, and $Pexc$ is the power of the excitation laser. For near-band edge emissions of semiconductors, the dimensionless exponent k is known to be 1 < k < 2 for free-exciton or bound-exciton luminescence, and k < 1 for free-to-bound exciton or donor–acceptor pair recombination luminescence.^53,54 Our results show that the exponent k for WS₂ exciton in the heterostructure is 1.25, consistent with the previous reports.^55,56 The k value for the d–d emission from an uncovered NiPS₃ is obtained as 1.13. However, the k value for the d–d emission from the WS₂/NiPS₃ heterostructure region is obtained as 0.56, suggesting a different mechanism from the d–d emission in the bare NiPS₃. Note that such a small k (i.e., k < 1) is a characteristic property of donor–acceptor pair recombination,^57–59 which is in consistence with the charge-transfer mechanism we proposed in the type-I NiPS₃/WS₂ heterostructure in Fig. 2(a): since the ³A_2g and ³T_1g states of NiPS₃ are the narrower components in the type-I band alignment, these states act as the extrinsic acceptor and donor states of the WS₂ layer, respectively. As a contrast, the k values of peak X are 0.92 for both the bare NiPS₃ and the WS₂/NiPS₃ heterostructure, matching well with the reported value of 0.916.¹⁴ In addition, no noticeable enhancement of peak X emission is observed from the heterostructure compared to the bare NiPS₃. The unchanged k value and emission intensity of peak X in the heterostructure imply that the d–d emission may not be coupled to the peak X emission.

Leveraging the enhanced d–d emission in NiPS₃, we further investigate the fine d-orbital splitting in NiPS₃ by performing temperature-dependent measurements. Figure 4(a) shows the PL spectra of the WS₂/NiPS₃ heterostructure from 300 K down to 78 K under the excitation of 2.33 eV, where the fitted d–d emission peaks are shown in shadows. When the temperature is below 150 K, the d–d emission splits into two peaks, D₁ and D₂, with the splitting energy of ∼105 meV [Fig. 4(b)]. These two peaks show blue shift at low temperature, in accordance with the empirical Varshni equation $E(T)=E(0)−αT2/(T+β)$⁠, where $E(0)$ is the energy at 0 K, and $α$ and $β$ are fitting constants (Table S1).⁶⁰ As we propose that the d–d emission originates from the transition from the ³T_1g to ³A_2g states, the observation of these two peaks indicates a further splitting of the d electronic states under the trigonal crystal field as shown in Fig. 4(c). The degenerate ³T_1g state will split into two states of ³E_g and ³A_2g, and the energy difference between these two states has been reported to be around 110 meV by absorption spectra,^20,21,61 reflection spectra,²² and calculations.^50,61 Here, the D₁ and D₂ peak positions, as well as their energy difference match well with the transition of ³A_2g → ³A_2g and ³E_g → ³A_2g, respectively, further indicating the d–d transition origin of the emission. Figure 4(d) plots the intensity changes of the D₁ and D₂ peaks, as well as their intensity ratio (D₁/D₂), as a function of temperature. The intensities of D₁ and D₂ are well fitted by the Boltzmann distribution (Table S2).⁶² As discussed above, these two d–d emissions are supposed to be Laporte forbidden by the selection rules. However, there are two mechanisms that can relax this selection rule. First, the d electronic states can couple with the vibrations (vibronic coupling) in NiPS₃, creating an admixture between the states with even and odd parity, thereby partially lifting the parity selection rule.^18,63,64 Second, the local inversion symmetry breaking of NiPS₃ due to the spin structure emergence also activates the d–d transitions below the Néel temperature (T_N),⁶⁵ which accounts for the abrupt intensity change of the d–d emission at T_N for both the bare NiPS₃ and WS₂/NiPS₃ heterostructure as shown in Fig. S6. Note that the d-orbital ³T_2g splitting of NiPS₃ under the trigonal crystal field has also been reported by electronic Raman spectroscopy, showing that the d-orbital splitting exists at room temperature and does not originate from the magnetic phase transition of NiPS₃.⁶⁴ Thus, we attribute the invisible D₂ peak above the T_N to the weak signal rather than the magnetic phase transition.

Since the d electron properties play a significant role in the magnetic properties, there remains a question whether the d–d emissions in NiPS₃ are correlated with the magnetic configuration. Previous reports have shown that the sharp PL emission at 1.476 eV is linearly polarized due to coupling with the ordered spin in antiferromagnetic NiPS₃.^13–15 Here, we investigate the polarization of the d–d emissions by detecting the signals from different collective polarizations at temperature of 78 K which is below the Néel temperature (i.e., 152 K). As shown in Fig. S7, we fix the polarization of the excitation laser and selectively allow the pass of certain polarized emission signals arrive to the detector. Figure 5(a) shows the top view of the zigzag configuration of the spin order. Figure 5(b) shows the contour color map of the polarization-resolved PL spectra of the WS₂/NiPS₃ heterostructure, where the peak X at ∼1.476 eV and d–d emission peaks D₁ and D₂ exhibit distinct dependences on the collection angles. By fitting their integrated intensities using the function: $I(θ)=I0+I1 sin2θ$ (⁠$θ$ is the rotational angle), peak X shows a clear sinusoidal dependence on the polarization angle [Fig. 5(c)], which was attributed to the correlation with the zigzag spin order as reported in our previous work.¹⁴ However, both D₁ and D₂ peaks do not show polarization characteristics as shown in Fig. 5(c), indicating that they are not correlated with the zigzag spin order. This is consistent with a recent study on the spin dynamics of the same d–d transitions by time-resolved pump-probe measurements.¹⁶ 

In summary, we report the observation of d–d emission in NiPS₃ and its strong enhancement through the charge transfer in a type-I WS₂/NiPS₃ heterostructure. We attribute this d–d emission to the transition between the ³T_1g and ³A_2g states, which are d states split by the octahedral crystal field. The charge-transfer enhancement mechanism is supported by the excitation wavelength and power dependent measurements. This observation enables in-depth investigation of the d-state properties of NiPS₃. Specifically, the low-temperature measurements show that the ³T_1g further splits into the ³A_2g and ³E_g states with a splitting energy of ∼105 meV due to the trigonal crystal field, evidenced by the splitting of the d–d emission peak into D₁ and D₂ peaks. The polarization investigation of the d–d emission shows that they are not correlated with the spin configuration, which is distinct from the spin-correlated sharp peak X reported before. Our results reveal rich electronic and optical properties of vdW antiferromagnetic NiPS₃, providing a new model system for the study of other antiferromagnetic metal phosphorus trisulfide like manganese phosphorus trisulfide and iron phosphorus trisulfide. The readily formed heterostructure by stacking vdW materials together also shows the convenience and flexibility for property tuning, which offers a powerful platform for both the fundamental study and the application development of antiferromagnetic materials.

Single crystals of NiPS₃ were grown via the chemical vapor transport (CVT) technique.⁴⁴ A stoichiometric quantity of elements (molar ratio Ni:P:S = 1:1:3, 1 g in total) and pure iodine (around 20–30 mg) were sealed into a quartz ampule under pressure pumped down to 10⁻⁴ Torr. Then, the ampule was placed in a two-zone furnace and kept at 650–600 °C for a week. The layered NiPS₃ crystals were obtained in a low-temperature zone and characterized using multiple tools (Fig. S1). Bulk NiPS₃ crystals were exfoliated onto SiO₂(90 nm)/Si substrates using the scotch-tape method. Single crystals of WS₂ were purchased from HQ Graphene. Monolayer WS₂ flakes were exfoliated using a simplified gold-assisted method.^45,66 WS₂ crystals were first exfoliated by a tape, and then, gold was evaporated and deposited directly onto the tape. Another tape was used to assist the further exfoliation by gold. The structures of tape-gold-crystal were put onto SiO₂/Si substrates and then were heated at 120 °C. After 5 min of heating, tapes were peeled off at 120 °C, and then, gold were removed by etching, leaving a large area of monolayer WS₂ flakes on substrates (Fig. S8). Heterostructures were prepared by dry-transferring⁴⁵ monolayer WS₂ onto the top of the NiPS₃ flakes.

Optical measurements were conducted on a micro-Raman spectrometer (Horiba-JY T64000) and collected by a spectrometer with a liquid nitrogen cooled CCD camera. Samples were put in a cryostat (Cryo Industries of America, Inc.) with temperature ranging from 78 to 300 K. A 532 nm laser line was generated by MSL-FN-532 (CNI), and all other lasers were generated by a Kr⁺/Ar⁺ ion laser (Coherent Innova 70C Spectrum). All Raman signals were collected with a 1800 g/mm grating, and all PL signals were detected with a 150 g/mm grating.

The polarization angle resolved PL measurements were performed by adding a linear polarizer and a half-wave plate in the collective path. A 532 nm laser was used as incident excitation. Samples were kept at 78 K, and the signals were collected by a 50× long-working-distance objective and a 150 g/mm grating.

See the supplementary material for additional information and figures.

This material is based upon work supported by the National Science Foundation (NSF) under Grant No. 1945364. X.L. acknowledges the membership of the Photonics Center at Boston University. T.L. and J.C. acknowledge the support from the Department of Energy (DOE) under Award No. DE-SC0021064.

The authors have no conflicts to disclose.

Q.T. and X.L. conceived the idea of this work. Q.T. and W.L. prepared and characterized monolayer WS2. Q.T., T.L., J.C., H.K. and X.W. prepared and characterized NiPS₃ samples. Q.T. and T.L. fabricated the heterostructure by dry transfer method. Q.T. and H.K. conducted the optical measurements. Q.T. and X.L. analyzed and interpretated the data. Q.T. and X.L. wrote the manuscript with contributions from all authors.

Qishuo Tan: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Weijun Luo: Data curation (equal); Formal analysis (supporting); Writing – review & editing (equal). Tianshu Li: Data curation (supporting); Writing – review & editing (supporting). Jun Cao: Data curation (supporting); Writing – review & editing (supporting). Hikari Kitadai: Data curation (supporting); Formal analysis (supporting); Writing – review & editing (supporting). Xingzhi Wang: Data curation (supporting); Formal analysis (equal); Methodology (equal); Supervision (equal); Writing – review & editing (equal). Xi Ling: Conceptualization (lead); Data curation (supporting); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Project administration (lead); Resources (equal); Supervision (lead); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support this study are available from the corresponding author upon reasonable request.