Open Access Article
Nguyen
Thi Han
a and
Vo Khuong
Dien
*b
aDepartment of Basic Science, Hung Yen University of Technology and Education, Hung Yen, Vietnam
bFPT University, Can Tho campus, Can Tho city, Vietnam. E-mail: vkdien@fpt.edu.vn
First published on 30th October 2025
Using first-principles calculations, we systematically studied the spin–orbit coupling (SOC)-induced valley spin splitting in a pristine HfN2 monolayer (HfN2-ML) and in Cr- and Ni-doped systems. The pristine HfN2-ML is revealed to host a direct band gap at the K and K′ points of the Brillouin zone. The valley spin splitting reaches 350 meV for the conduction bands and 83.5 meV for the valence bands. Furthermore, the exciton binding energy of the HfN2-ML is estimated to be approximately 0.90 eV. The exciton ground states belong to the Wannier–Mott type, which are governed by the electron and hole band edge states. More importantly, the valley spin at the K and K′ points could substantially change the effects when Cr or Ni is doped into the HfN2-ML. Consequently, the Cr-doped HfN2 monolayer exhibits a pronounced Zeeman splitting of approximately 300 meV. These findings highlight the promise of the HfN2-ML and related two-dimensional (2D) materials for prospective applications in valleytronic and spintronic devices.
Beyond TMDs, other 2D materials have also emerged as promising candidates for valleytronic and spintronic applications, owing to their rich electronic structures, tunable interlayer couplings, and the possibility of engineering symmetry breaking through external fields or selective doping.9 Like TMDs, transition metal nitride semiconductors have garnered considerable interest10–12 owing to their exceptional properties.13–16 Moreover, extensive efforts have been devoted to doping in 2D systems as an effective strategy to tailor their electronic and magnetic properties, thereby opening avenues for prospective applications in spintronics and valleytronics devices.17
Hafnium dinitride (HfN2) represents one of the materials attracting substantial interest. Bulk HfN2 has been reported to exhibit remarkable ductility, and thin films of HfNx have been successfully fabricated using the through-silicon-via technology.18 Although experimental efforts toward the synthesis of 2D HfN2 are still ongoing, the stability of the HfN2-ML has been established theoretically.19 It's generating significant interest due to its unique properties, such as optical, electronic, and structural characteristics.19–21 In particular, the HfN2-ML is emerging as a candidate for exploring new possibilities in valleytronics.22,23 Using first-principles calculations, substrate-induced excellent electronic properties in HfN2-ML have been observed in the literature, such as CrS2/HfN224 and MoTe2/HfN2.25 However, there has been a scarcity of comprehensive investigations into HfN2-ML up to now, encompassing both theoretical calculations and experimental measurements related to the properties of valley electronics and excitonic properties. Additionally, the effects of doping transition metal elements on the HfN2-ML-induced enriched valley properties have not yet been explored.
In this work, we systematically investigate the intrinsic electronic and excitonic properties of pristine HfN2-ML, along with the remarkable electronic modifications induced by chromium (Cr) and nickel (Ni) doping, using first-principles calculations. The HfN2-ML exhibits several key characteristics: (i) a moderate direct band gap of 4.20 eV at the K and K′ valleys, accompanied by distinct conduction- and valence-band splittings; (ii) strong excitonic effects arising from its unique band dispersion and non-uniform dielectric screening; and (iii) pronounced valley–spin polarization at the K and K′ points upon Cr and Ni doping, leading to a significant Zeeman-type splitting. These results highlight HfN2-ML as a promising 2D material for next-generation spintronic, valleytronic, and optoelectronic applications. Moreover, Cr dopants introduce net magnetic moments into the otherwise nonmagnetic HfN2-ML, offering an effective route to realize spin-polarized functionalities for practical spintronic device applications.
On the other hand, the Berry curvature38,39 is defined as
, where un,k and k represent the Bloch function and wave vector, respectively. For an applied in-plane electric field, this Berry curvature gives rise to an anomalous velocity
, which drives the charge carriers at the K and K′ points in opposite directions. The Berry curvature is calculated from the wave functions via WANNIER90;40
The electronic band structures, calculated at the GW level including spin–orbit coupling (SOC), reveal a direct band gap of 4.20 eV located at the K and K′ points, as shown in Fig. 2a. The blue and red curves represent the spin-up and spin-down states, respectively. The band dispersion of the HfN2-ML exhibits a pronounced valley spin splitting (VSS) of approximately 350 meV in the conduction band and a smaller VSS of about 83.5 meV in the valence band. Interestingly, this trend is opposite to that observed in conventional transition-metal dichalcogenides (TMDs), where the valence-band VSS dominates. For comparison, the VSS in monolayer MoS2 and WSe2 is about 150 meV and 430 meV, respectively, while the corresponding conduction-band splitting is nearly negligible.42 The distinct SOC-induced splittings in the valence and conduction bands originate from the different orbital characters of the underlying Bloch states. Specifically, the valence band is primarily derived from the N-(2px, 2py) orbitals, whereas the conduction band is dominated by the Hf-(5dxy, 5dx2–y2) orbitals. These orbital contributions are clearly reflected in the band-decomposed charge densities and the orbital-resolved projected density of states (PDOS), as shown in Fig. 1b and c.
The spin ordering at the K and K′ valleys is opposite; consequently, carriers in these valleys can be selectively excited by circularly polarized light following opposite optical selection rules, as indicated by the black arrows in Fig. 2a. It is worth noting that the spin-split states at the K and K′ valleys remain energetically degenerate rather than significantly lifted. This subtle degeneracy stems from the underlying time-reversal symmetry that intrinsically connects the two valleys. Nevertheless, the absence of inversion symmetry in HfN2-ML leads to opposite Berry curvatures for charge carriers at the K and K′ points, as clearly shown in Fig. 2b and c. This valley-contrasting Berry curvature not only reveals the nontrivial topological nature of the electronic bands, but also offers a promising route for manipulating valley-dependent charge transport and optical selection in this system.
Fig. 3 shows the optical absorption spectra of HfN2-ML obtained from GW and GW–BSE calculations including SOC. In Fig. 3a, the imaginary part of the dielectric function, ε2(ω), computed within the GW-RPA framework (black), is compared with that obtained from the GW-BSE approach (red), which explicitly incorporates electron–hole interactions. Two pronounced absorption peaks, labeled A and B and located near 5.8 and 6.0 eV, respectively, correspond to bright resonant excitons that dominate the optical response above the quasiparticle band gap. The inclusion of electron–hole interactions markedly enhances these features, highlighting the strong excitonic effects in this material. Fig. 3b focuses on the low-energy region, where four bound excitonic states (E1–E4) are identified below the GW quasiparticle band gap with a large binding energy of approximately 0.90 eV, primarily arising from transitions at the K and K′ valleys. This value exceeds those reported for typical transition-metal dichalcogenides (TMDs), such as MoS2 (0.54 eV)43 and WS2 (0.71 eV),44 indicating enhanced Coulomb interactions and reduced dielectric screening in HfN2-ML. The excitons E1–E4 exhibit relatively small but finite oscillator strengths, as shown in Fig. 3c, suggesting that they are weakly bright (quasi-dark) due to spin–orbit-induced mixing and the lack of inversion symmetry. Fig. 3d–k illustrates the real-space exciton wavefunctions of these four lowest bound states from both side and top views. The excitonic wavefunctions are spatially well confined within a few unit cells, confirming their Wannier–Mott character, consistent with the exciton localization observed in monolayer TMDs.45,46
To investigate the influence of doping on the valleytronic properties of HfN2-ML, we substitute transition-metal atoms into the pristine system. Specifically, Cr and Ni atoms are selected as dopants, with a single Cr or Ni atom replacing one Hf atom in a 4 × 4 × 1 supercell of HfN2-ML to model the doping effect. The choice of Cr and Ni is motivated by their partially filled 3d orbitals, which can introduce localized magnetic moments and significantly enhance SOC. These characteristics are crucial for valleytronics, as they enable the breaking of time-reversal and inversion symmetries, leading to valley splitting and spin–valley polarization. Upon doping, the Cr-doped HfN2-ML becomes spin-polarized, whereas the Ni-doped system remains non-spin polarized. The calculated total magnetic moments are −1.937μB for Cr-doped HfN2-1L and 0.000μB for Ni-doped HfN2-ML, respectively. This contrasting behavior originates from their distinct electronic configurations: the Cr–3d orbitals lie near the Fermi level and strongly hybridize with Hf–d and N–p states, resulting in exchange splitting and stabilization of a ferromagnetic ground state. In contrast, the Ni–3d orbitals are positioned deeper in the valence region, exhibiting weak hybridization near the band edges, which suppresses spin polarization and leads to a nonmagnetic ground state.
The charge density difference (Δρ) is plotted in Fig. 4a, and defined as Δρ = ρCr-/Ni–HfN2 − ρCr/Ni – ρHfN2, where ρCr-/Ni–HfN2, ρHfN2, and ρCr/Ni represent the charge densities of the Cr- or Ni-doped HfN2-ML, the pristine HfN2-ML, and the isolated Cr or Ni atoms, respectively, in the same spatial configuration. The resulting maps reveal pronounced charge accumulation and depletion regions primarily localized at the doping interface, confirming strong orbital hybridization between the dopant and the host lattice. Bader charge analysis further shows that approximately 2.50 (2.40) electrons are transferred from Cr (Ni) to the HfN2-ML, indicating significant charge redistribution and covalent bonding characteristics. To gain deeper insight into the magnetic behavior, the spin density distribution of the Cr-doped HfN2-ML is presented in Fig. 4b. The spin density is mainly concentrated around the Cr dopant site, while contributions from the surrounding Hf and N atoms are relatively minor, confirming that the magnetic moment predominantly originates from the localized Cr-3d states.
![]() | ||
| Fig. 4 (a) Side view and top view of the charge density differences of Cr-/Ni-doped HfN2-1L, and (b) spin distributions of Cr-doped HfN2, respectively. Note: the pink and cyan iso-surfaces are 0.002. | ||
Fig. 5a presents the density of states (DOS) of the pristine HfN22-ML, calculated at the PBE level without SOC. For comparison, Fig. 5b and c display the DOS of the Cr- and Ni-doped HfN2-ML, respectively, obtained using the same theoretical level of theory. Upon Cr doping (Fig. 5b), a ferromagnetic ground state emerges, as indicated by the distinct spin-resolved orbital contributions of Cr, Hf, and N atoms in both the valence and conduction regions. This behavior aligns with the spin-density distributions in Fig. 4b, where a clear asymmetry between spin-up and spin-down channels is evident. The induced magnetism originates from spin-dependent splitting of the valence states and the localization of Cr-3d orbitals near the Fermi level (EF), which enhances exchange interactions and mediates spin polarization among neighboring Hf and N atoms. Such orbital hybridization and energy-level rearrangement reveal that Cr doping not only stabilizes long-range ferromagnetic order but also substantially modifies the electronic structure near EF, thereby offering a mechanism to tailor the spintronic functionality of HfN2-ML. Specifically, the adsorption of Cr introduces an n-type character, as Cr-3d orbitals donate additional electrons close to the conduction-band edge, shifting the Fermi level upward and producing a metallic response in the spin-polarized DOS. Ni-doped HfN2-ML in Fig. 5c, on the other hand, retains nonmagnetic and semiconducting characteristics. The nearly symmetric van Hove singularities in the spin-up and spin-down DOS closely mirror those of the pristine HfN2-ML, confirming the absence of exchange splitting and net magnetic moment. The electronic structure thus remains dominated by the host Hf–d and N–p states. Consequently, while Cr incorporation enables the formation of spin-polarized states and magnetic ordering, Ni doping primarily alters the band dispersion, serving instead as an effective route to tune the electronic transport properties without breaking time-reversal symmetry.
![]() | ||
| Fig. 5 (a) The van Hove singularities in the density of states (DOS) without (w/o) SOC of pristine HfN2, corresponding to (b) and (c) doping Cr, and Ni atoms in HfN2-ML, respectively. | ||
Fig. 6a–c present the PBE + SOC band structures of the pristine, Cr-doped, and Ni-doped HfN2-ML. The corresponding magnified views focusing on the valence band dispersions near the Fermi level are shown in Fig. 6d–f. As illustrated in Fig. 6a, the pristine HfN2-ML exhibits distinct band-edge features around the K and K′ valley points. The calculated parameters, including the valley Zeeman splitting (Ez), valley splitting (Δv), and spin splitting
at the K and K′ valleys for the pristine, Cr-doped, and Ni-doped HfN2-ML are summarized in Table 1. The valley Zeeman splitting is given by Ez = Δ+opt − Δ−opt, where Δ+opt and Δ−opt correspond to the optical transition energies under left- and right-circularly polarized light (σ+ and σ−), respectively. Notably, in the Cr-doped HfN2-ML, the emergence of magnetic ordering breaks time-reversal symmetry and lifts the valley degeneracy, resulting in a pronounced Zeeman splitting of Ez = 300 meV. In contrast, the Ni-doped system, being nonmagnetic, preserves the valley degeneracy, consistent with its absence of spin polarization.
![]() | ||
| Fig. 6 (a) Energy band structure of the pristine HfN2-ML obtained with PBE + SOC; (b) and (c) unfolded electronic band structures of the Cr- and Ni-doped HfN2-ML, respectively, obtained using the vaspkit code.47 Panels (d)–(f) show the corresponding enlarged band structures in the valence bands. | ||
, and the valley Zeeman splitting (Ez). The symbols “+” and “−” represent quantities at the K and K′ valleys, respectively, while “c” and “v” denote the conduction and valence bands. All energies are expressed in meV
| System | Δ +,vspin | Δ −,vspin | Δ +,cspin | Δ −,cspin | Δ +opt | Δ −opt | E z |
|---|---|---|---|---|---|---|---|
| Pristine HfN2 | 40.0 | 40.0 | 400 | 400 | 1450 | 1450 | 0 |
| Cr-doped HfN2 | 35.0 | 30.0 | 300 | 800 | 1150 | 850 | 300 |
| Ni-doped HfN2 | 25.0 | 25.0 | 350 | 350 | 1350 | 1350 | 0 |
| This journal is © The Royal Society of Chemistry 2025 |