Need for complementary techniques for reliable characterization of MoS2-like layers

Abstract

The observation of characteristic A1g and E2g1 peaks, at around 408 and 382 cm−1, respectively, in Raman spectroscopy is considered the evidence of 2H-structured MoS2, probably the most extensively studied transition-metal dichalcogenide. Here, using a combination of x-ray diffraction, x-ray photoelectron spectroscopy, and resonant Raman spectroscopy, we show that the detection of A1g and E2g1 modes in Raman spectra alone may not necessarily imply the presence of MoS2. A series of Mo–S films, ≈ 20-nm-thick, are grown on single-crystalline Al2O3(0001) substrates at 1073 K as a function of H2S partial pressure, pH2S (= 0, 0.01%, 0.1%, and 1% of total pressure) via ultra-high vacuum dc magnetron sputtering of a Mo target in 20 m Torr (2.67 Pa) Ar/H2S gas mixtures. In pure Ar discharges and with pH2S up to 0.1%, i.e., pH2S ≤ 2.67 × 10−3 Pa, we obtain body centered cubic (bcc), 110-textured films with lattice parameter a increasing from 0.3148 nm (in pure Ar) to 0.3151 nm (at pH2S = 2.67 × 10−4 Pa), and 0.3170 nm (at pH2S = 2.67 × 10−3 Pa), which we attribute to increased incorporation of S in the Mo lattice. With 1% H2S, i.e., pH2S = 2.67 × 10−2 Pa, we obtain 000l oriented 2H-structured MoS2.0±0.1 layers. Raman spectra of the thin films grown using 0.1% (and 1%) H2S show peaks at around 412 (408) and 380 cm−1 (382 cm−1), which could be interpreted as A1g and E2g1 Raman modes for 2H-MoS2. By comparing the Raman spectra of MoS2.0±0.1 and Mo:S thin films, we identify differences in A1g and E2g1 peak positions and intensities of defect-sensitive peaks relative to the A1g peaks that can help distinguish pure MoS2 from non-stoichiometric MoS2−x and multiphase Mo:S materials.