Engineering Band Alignment in MoS₂/HfO₂ Heterojunctions via Controlled Nitridation

Background

Layered transition‑metal dichalcogenides (TMDCs) have attracted intense research interest due to their unique electronic and optical properties that are highly tunable with layer thickness [1, 2]. Molybdenum disulfide (MoS₂) is especially promising as a channel material for post‑7‑nm scaling, offering a direct bandgap of 1.8 eV in the monolayer and an indirect bandgap of 1.2 eV in bulk form [7]. Unlike graphene’s zero bandgap, MoS₂’s thickness‑dependent bandgap enables both optical and electrical device applications [3, 8]. The density of states in few‑layer MoS₂ is roughly three times that of the monolayer, which translates to higher ballistic drive currents [8]. Consequently, few‑layer MoS₂ can provide significant advantages for transistor design compared with its monolayer counterpart [3].

Meanwhile, silicon‑oxide‑based dielectrics are approaching their physical limits due to their low dielectric constant. To achieve thin equivalent‑oxide thicknesses (EOT) and high‑performance devices, high‑k dielectrics such as Al₂O₃, ZrO₂, HfO₂, and h‑BN have been integrated with MoS₂ [10–14]. For example, DiStefano et al. reported CBO and VBO values of 3.3 ± 0.2 eV and 1.4 ± 0.2 eV for MoS₂ on amorphous BN [13], while Tao et al. measured CBOs of 3.56 eV (Al₂O₃) and 1.22 eV (ZrO₂) for monolayer MoS₂ [15]. McDonnell et al. found a CBO of 2.09 ± 0.35 eV and a VBO of 2.67 ± 0.11 eV for the MoS₂/HfO₂ interface [12]. HfO₂ is particularly attractive because of its high dielectric constant (~20) and compatibility with poly‑SiGe, TaN, and poly‑crystalline silicon gates [16]. However, HfO₂ suffers from poor thermal stability, high leakage, and significant trap densities [17], motivating passivation strategies such as interface nitridation or fluorination [18, 19].

In this work, we investigated the band alignment of few‑layer MoS₂ on HfO₂ dielectrics with and without nitrogen plasma treatment, characterizing the interfacial chemistry by XPS.

Methods

SiO₂/Si wafers (280 nm SiO₂) were cleaned sequentially in acetone and isopropanol (10 min each) by ultrasonic agitation, followed by DI water rinse and N₂ dry. Few‑layer MoS₂ films were grown on SiO₂/Si by chemical vapor deposition (CVD) using MoO₃ (0.08 mg, 99 % Alfa Aesar) and S powder (1 g, 99 %) [20, 21]. After growth, the MoS₂ films were transferred onto HfO₂/Si substrates using a PMMA‑assisted wet transfer method (Figure 1a). PMMA was spin‑coated onto MoS₂/SiO₂/Si, the SiO₂ layer was etched in KOH, and the floating MoS₂/PMMA stack was scooped onto HfO₂/Si before dissolving PMMA in acetone.

HfO₂ layers were deposited by atomic layer deposition (ALD) at 200 °C using tetrakis(ethylmethylamido)hafnium (TEMAH) and H₂O as precursors [23, 24]. To optimize the nitrogen incorporation, secondary ion mass spectrometry (SIMS) indicated that 70 s of N₂ plasma introduced excessive nitrogen into the oxide, degrading its quality, whereas 30 s produced no detectable surface nitrogen. For the nitridated samples, a 50 s N₂ plasma treatment (3 Pa) was applied to HfO₂/Si prior to MoS₂ transfer. This condition delivered a nitrogen dose of ~8.4 × 10¹⁴ atoms cm⁻², corresponding to ~1.5 % nitrogen concentration as confirmed by XPS. Four samples were prepared for XPS analysis: (1) MoS₂ on SiO₂/Si, (2) bulk HfO₂ on Si, (3) MoS₂ on as‑grown HfO₂, and (4) MoS₂ on N₂‑plasma‑treated HfO₂.

a Process flow of PMMA‑assisted wet transfer for the MoS₂/ALD‑HfO₂ heterojunction. b Raman spectra of as‑grown and transferred MoS₂. Inset: cross‑section TEM of MoS₂ on SiO₂/Si.

Results and Discussions

Raman spectroscopy (RENISHAW InVia) confirmed that the transfer process preserved the crystal quality of few‑layer MoS₂. Two prominent peaks at 382.86 cm⁻¹ (E₂g¹) and 406.43 cm⁻¹ (A₁g) showed no significant shift after transfer, indicating minimal strain or defect introduction. The frequency separation (Δk ≈ 23.57 cm⁻¹) corresponds to four‑to‑five layers, corroborated by HRTEM measurements of ~2.8 nm thickness (Figure 1b).

SIMS depth profiling of the nitrided sample (Figure 2a) revealed a nitrogen concentration gradient, with nitrogen diffusing into the HfO₂ layer and accumulating near the HfO₂/Si interface. The N 1s XPS spectra (Figure 2b) showed a low‑intensity peak at ~395.8 eV in the nitrided heterojunction, indicative of Mo‑N bond formation [31].

a SIMS depth profiles of Mo, N, Hf, and Si for MoS₂ on nitrided HfO₂/Si. b N 1s XPS spectra for MoS₂/HfO₂ heterojunctions with and without nitridation.

To extract the VBO, XPS spectra of Mo 3d and Hf 4f core levels were recorded with 0.05 eV step size using a VG ESCALAB 220i‑XL system (Al Kα, 1486.6 eV). The binding‑energy differences (BEDs) between core levels and the valence‑band maximum (VBM) were determined as 228.49 ± 0.1 eV for Mo 3d₅/₂ in MoS₂ (sample 1) and 14.10 ± 0.1 eV for Hf 4f₇/₂ in bulk HfO₂ (sample 2). For the heterojunctions, the Mo 3d₅/₂ peak shifted from 229.45 ± 0.05 eV (sample 3) to 229.90 ± 0.05 eV (sample 4) upon nitridation, reflecting charge transfer across the Mo‑N interface. Likewise, the Hf 4f₇/₂ peak moved from 17.40 ± 0.05 eV to 17.60 ± 0.05 eV, and the O 1s peak shifted by 0.20 eV to higher binding energy (Figure 3d). These shifts signify downward band bending in HfO₂, consistent with nitrogen‑induced donor‑like defects [34].

a XPS spectra of Mo 3d and valence band for few‑layer MoS₂. b XPS spectra of Hf 4f and valence band for bulk HfO₂. c XPS spectra of Mo 3d, Hf 4f, and d O 1s for MoS₂/HfO₂ heterojunctions with and without nitridation.

Using the Kraut method, the VBO (ΔE_V) was calculated as 2.09 ± 0.1 eV for the nitrided heterojunction and 2.34 ± 0.1 eV for the unnitrided one. The CBO (ΔE_C) was derived from the bandgap difference (HfO₂: 5.9 ± 0.1 eV; MoS₂: 1.4 ± 0.1 eV) and the VBO: 2.41 ± 0.1 eV (nitrided) versus 2.16 ± 0.1 eV (unnitrided) (Eq. 2). Band diagrams (Figure 4) illustrate that both heterojunctions provide strong confinement for electrons and holes, essential for MoS₂‑based FETs. The nitrided interface, with its higher CBO, is particularly favorable for n‑channel devices.

Band diagrams of MoS₂/HfO₂ heterojunctions: a without nitridation and b with nitridation.

Conclusions

Our XPS study demonstrates that a brief N₂ plasma treatment of HfO₂ before MoS₂ deposition can significantly modify the interfacial band alignment. The unnitrided heterojunction exhibits a VBO of 2.34 ± 0.1 eV and a CBO of 2.16 ± 0.1 eV, whereas nitridation reduces the VBO to 2.09 ± 0.1 eV and raises the CBO to 2.41 ± 0.1 eV. The observed Mo‑N bonding and nitrogen‑induced donor states account for the band bending and the improved electron confinement. These findings provide a clear route for band‑engineering of 2D/ high‑k interfaces, paving the way for more reliable and high‑performance MoS₂‑based electronics.

Availability of Data and Materials

The datasets supporting the conclusions of this manuscript are included within the manuscript.

Abbreviations

ALD:

Atomic layer deposition

BE:

Binding energy

BED:

Binding energy difference

CBO:

Conduction band offset

CL:

Core level

CVD:

Chemical vapor deposition

FET:

Field‑effect transistor

HfO₂:

Hafnium oxide

HRTEM:

High‑resolution transmission electron microscope

MoS₂:

Molybdenum disulfide

PMMA:

Poly (methyl methacrylate)

SIMS:

Secondary ion mass spectrometry

SL:

Single‑layer

TEMAH:

Tetrakis (ethylmethylamido) hafnium

TMDC:

Transition metal dichalcogenide

VBM:

Valence band maximum

VBO:

Valence band offset

XPS:

X‑ray photoelectron spectroscopy