III. RESULTS AND DISCUSSION

To verify the crystal structure of MoS₂ NCs, XRD was used. Figure 1(a) shows the powder XRD pattern of MoS₂ NCs, which can be indexed to hexagonal MoS₂ (PDF 73-1508). Four main peaks are observed at 14, 33, 40, and 59° corresponding to the (002), (100), (103), and (110) planes for MoS₂ NCs, respectively.^{Reference Gao, Si, Li, Zhang, Zhang, Yang and Xue15,Reference Zhang, Li, Qi and Gao16} The broad peak appeared at 25° is found to arise from the glass holder used for placing MoS₂ NCs for XRD measurements. Ojha et al. reported^{Reference Ojha, Saha, Banerjee and Ganguli40} a similar XRD pattern for MoS₂ NCs synthesized by the hydrothermal process. The peak broadening clearly shows the NC nature of MoS₂. The Scherrer formula, D _hkl = kλ/β_hkl cos θ, was used to estimate the crystallite size of MoS₂. Here, k is a coefficient (k = 0.9), β_hkl is the full-width-at-half maximum of the (002) peak, and λ is the wave length (λ = 1.5405 Å). The calculated crystallite size from XRD line broadening is of 10 nm.

FIG. 1.

XRD pattern of MoS₂ nanocrystallites.

Morphological studies were conducted using TEM. Figures 2(a) and 2(b) show the low magnification bright field and the high resolution lattice images, respectively. The indexed selected area diffraction pattern (SAED) is shown as the inset to the bright field image. As can be seen in Fig. 2(b), the MoS₂ NCs (size of 25–30 nm) show the hexagonal structures made up of a few layers of MoS₂ basal planes. The apparent difference in the sizes estimated from XRD and TEM might be because XRD measures crystallite size whereas the TEM shows the grain size. The SAED pattern shown as the inset to Fig. 2(a) contains rings with uniform intensity—confirmed the NC nature of the material. The interlayer spacing measured from the lattice images is 6.45 Å, consistent with the previous reports^{Reference Kibsgaard, Chen, Reinecke and Jaramillo1–Reference Liu, Robertson, Li, Kuo, Darby, Muhieddine, Lin, Suenaga, Stamatakis, Warner and Tsang6} (6.3 Å) of monolayer MoS₂ grown by the CVD methods, and it corresponds to the (002) planes of MoS₂.

FIG. 2.

(a) Bright field TEM image of MoS₂ nanocrystallites. The inset shows the SAED patterns confirming the NC nature of the material. (b) High resolution transmission electron microscopy image of MoS₂ nanocrystallites. The spacing between the layers along the (002) plane is about 0.645 nm.

XPS survey spectra of MoS₂ NCs indicate the presence of peaks with binding energies (BEs) consistent with molybdenum and sulfur elements besides a lower intensity oxygen and carbon. Absence of other elements demonstrates the purity of the sample. To further probe the chemical composition and surface chemistry, high-resolution core-level XPS spectra were collected on the selected BE intervals. The BE positions were referenced and adjusted to adventitious carbon because of the charging effects on the sample. Shirley method background signal was subtracted and BE peaks were curve-fitted (deconvolution) using Gaussian/Lorentzian functions. Multiple BEs associated with C1s spectra are observed at 284.4, 286.3, and 287.9 eV which are assigned to adventitious carbon and C–O impurities from the hydrothermal process. The O1s spectral region peak appearing at 532.6 eV is assigned to C–O bonding while the lower BE peak at 530.8 eV is ascribed to Mo–O bonding. The deconvoluted Mo3d region had multiple peaks observed at 228.9 (Mo3d _5/2), 229.9 (Mo3d _5/2), and 232.48 (Mo3d _5/2) and their corresponding spin–orbit splitting doublet peaks were noticed at 232.08 (Mo3d _3/2), 233.1 (Mo3d _3/2), and 235.6 (Mo3d _3/2) respectively. In addition, the peak observed at 226.3 eV was assigned to S2s. Spin–orbit split peak separation is found to be around 3.16 eV, which is consistent with the existing literature.^{Reference Spevack and Mclntyre41–Reference Spevack and Mclntyre43} The peaks corresponding to the S 2p _3/2 and S 2p _1/2 orbital of divalent sulfide ions (S²⁻) are observed at 161.8 and 162.98 eV associated with another set of deconvoluted peaks appearing at 163.6 and 164.8 eV with an estimated spin–orbit splitting of 1.18 eV. Multiple BE peaks observed in this sample are apparently indicating different phases. The high intensity peak at 228.9 eV (Mo3d _5/2) resulting from the Mo⁴⁺ oxidation state is ascribed to the MoS₂ phase^{Reference Spevack and Mclntyre41} with the corresponding S2p _3/2 161.8 eV. In addition, the BE peak at 229.9 (Mo3d _5/2) is associated with higher oxidation state molybdenum, along with a secondary peak that corresponds to S2p _3/2 (161.8 eV) ascribed to the MoS₃ phase.^{Reference Zhang, Zhou, Zheng, Hong, Pan and Xu42} Furthermore, the BE peak at 232.48 eV (Mo3d _5/2) relating to the higher oxidation state of Mo⁶⁺ in conjunction with O1s (530.8 eV) is attributed to a mostly stable MoO₃ phase.^{Reference Spevack and Mclntyre43} XPS technique could be used to quantify the ratio of different phases within a given sample. Based on the deconvoluted Mo3d peaks, it is estimated that there is about 65% MoS₂ phase and the remaining is a combination of MoS₃ and MoO₃.

Raman spectrum was used to confirm the few-layered MoS₂ NCs. Raman spectra measured on MoS₂ NCs showed two distinct characteristic^{Reference Chiappe, Asselberghs, Sutar, Iacovo, Afanas’ev, Stesmans, Balaji, Peters, Heyne, Mannarino, Vandervorst, Sayan, Huyghebaert, Caymax, Heyns, De Gendt, Radu and Thean44} peaks at 384 and 408 cm⁻¹, which are associated with the in-plane vibrational (E¹_2g) and the out-of-plane vibrational (A_1g) modes of Mo–S bonds in 2H–MoS₂, respectively. The Raman peaks are in agreement with that of MoS₂ bulk crystals.

Figure 3 presents an overview of the representative first derivative X-band ESR spectrum measured at 7 K. The spectrum exhibits a wealth of information pertaining to intrinsic paramagnetic spin centers. To authors’ knowledge, this is the first single richest spectrum reported to date on any MoS₂ NCs and amorphous/less-crystalline compounds. ESR spectral parameters provide characteristic signatures of paramagnetic defects by analyzing the signal position, width, and shape. From this spectrum, at least seven well-resolved ESR lines are identified, marked from 1 to 7. Only those peaks that are clearly observable are indicated. Since the signals show considerable overlap, it is difficult to determine the ESR parameters of each signal accurately. Nevertheless, within experimental accuracy, the g-values of all the well-resolved peaks and their peak-to-peak line widths that could be detected separately are tabulated in Tables I and II, respectively. The g-values were determined using the resonance condition hν = gβH, where h = Planck’s constant, ν = microwave frequency, g = g-value, β = Bohr magneton, and H = resonance field (measured at the turning points). Collected at 7 K, on the as-prepared MoS₂ NCs, at least seven ESR signals are detected, associated with g ₁ = 2.1144, g ₂ = 2.0300, g ₃ = 1.9863, g ₄ = 1.9417, g ₅ = 1.8984, g ₆ = 1.8541, and g ₇ = 1.8146. No half-field lines were detected that might correspond to electron spin >1/2. This infers that, most likely, the signals that are observed in the present work arise from spin 1/2 paramagnetic spin centers. Although it has been intensely searched for, no hyperfine interaction originating from ³³S (nuclear spin I = 3/2; 0.75% abundance) and ⁹⁵Mo and ⁹⁷Mo (both I = 5/2; 25.5% added abundance) could be traced. No external impurities^{Reference Spackman45–Reference Stesmans, Iacovo and Afanas’ev47} from MoS₂ NCs are observed. In addition, as verified from the controlled measurements, no similarity between the ESR spectra measured on MoS₂ NCs and ESR spectra recorded from sulfur and Mo precursors could be found, which clarifies that the unreacted compound could not account for the observed ESR spectra. It has been observed that the ESR spectra do not resemble to the spectra collected from the Teflon tape, quartz tube, and empty cavity. In the absence of relevant ESR literature studies, some of the low intense overlapping ESR peaks appeared on the low magnetic field side could not be identified with certainity. Further high frequency ESR spectroscopic measurements in collaboration with National High Magnetic Field Laboratory (NHMFL) Electron Magnetic Resonance (EMR) facility at Florida will be performed in resolving the overlapping ESR signals.

FIG. 3.

First-derivative representative X-band (9.40 GHz) ESR spectra measured on MoS₂ nanocrystals at 7 K. One can see that there are at least seven well-resolved ESR lines as marked from 1 to 7. The spectrum is plotted on both the magnetic field and g-value axes.

TABLE I.

Temperature dependences of g-values.

TABLE II.

Temperature dependences linewidths.

The attention will be paid on the well-resolved ESR signals. To better understand the behavior of observed paramagnetic centers, temperature-dependent ESR measurements (7–40 K) were performed on the as-prepared MoS₂ NCs. The data are presented in Fig. 4. As presented in Tables I and II, the g-values and the linewidths of all the ESR signals did not vary significantly as a function of temperature (7–40 K). This observation is consistent with the previous reports^{Reference Konings, van Doormen, Koningsberger, De Beer, Farragher and Schuit48,Reference Khulbe, Mann and Ternan49} in support of Mo⁵⁺ species (discussed later), whose ESR spectra were shown to be temperature-independent. This leads us to believe that these are isolated intrinsic paramagnetic defects and do not interact with each other.

FIG. 4.

Temperature dependence (7–40 K) of ESR spectra measured on MoS₂ nanocrystals.

Annealing MoS₂ NCs under Ar ambience can improve the crystallinity and phase formation, but, at the expense of grain growth. The number of ESR signals decreased due to the decrease in the surface area of MoS₂ NCs as a result of Ar annealing (4 h, 300 °C). Now, the modification of ESR spectra upon post Ar annealing will be considered. Because of Ar annealing, very distinct paramagnetic centers are found. Interestingly, the rich ESR spectrum observed from MoS₂ NCs disappeared. Instead, as presented in Fig. 5, a broad signal (g-value = 2.0895; peak-to-peak line width = 821 G) and a narrow signal (g-value = 1.9289; peak-to-peak line width = 69 G) are noticed. Upon closer inspection, there are few more weak unresolved narrow signals overriding the broader signal were also detected. This implies that the paramagnetic defect sites observed from pristine MoS₂ NCs still exist, however, with much reduced defect density.

FIG. 5.

ESR spectra collected on Ar-annealed (4 h, 300 °C) MoS₂ nanocrystals measured at 7 K.

Now, the possible origin of these paramagnetic defect centers involved in the observed ESR spectra will be discussed. The signals in the low field region (<3000 G) are very weak in intensity, also not well resolved compared to the rest of the spectra. Hence, that part of the spectra is excluded from the present discussion. Therefore, the efforts will be focused in analyzing the ESR signals numbered from 1 to 7 in the spectra. Although Mo⁶⁺ is detected in XPS spectra, it is diamagnetic; not expected^{Reference Khulbe, Mann and Ternan49} to give ESR signal, and hence this possibility can be ruled out. Even it is paramagnetic and abundant as can be seen from XPS, Mo⁴⁺ would give a quite broad signal due to large zero field splitting. This possibility can also be ruled out as all the signals that are observed have relatively narrower line widths ranging from 40 to 95 G. Mo³⁺ is relatively unstable and cannot be captured using X-band ESR. Hence, Mo⁵⁺ is the only ion, which is paramagnetic, ESR-detectable, and highly stable. In this work, no attempt was made to derive the absolute spin concentrations of all these paramagnetic defect sites, as it is not straightforward to estimate, particularly, given the strong overlapping of the signals. This might require simulation of entire powder spectra that needs usage of much computational work and modeling,^{Reference Bensimon, Belougne, Giuntini and Zanchetta33,Reference Deroide, Bensimon, Belougne and Zanchetta34} which is beyond the scope of the present work.

Previous ESR studies reported^{Reference Bensimon, Belougne, Giuntini and Zanchetta33,Reference Deroide, Bensimon, Belougne and Zanchetta34} on MoS₂ compounds will help in identifying the ESR signals observed in the present work. These authors collected ESR spectra when (amorphous/less crystalline) MoS₂ was exposed to water, oxygen, and by varying the S to Mo ratio. Part of the spectra observed in the present work resembles to the previous work reported by Zanchetta's group.^{Reference Bensimon, Belougne, Giuntini and Zanchetta33,Reference Deroide, Bensimon, Belougne and Zanchetta34} Their work identified three classes of paramagnetic spin centers with variable intensity, whose ESR signals are of Lorentzian in character. In addition, the previous ESR studies^{Reference Khulber, Mann and Ternan31,Reference Bensimon, Belougne, Giuntini and Zanchetta33,Reference Louis and Che35} documented on the bulk MoS₂ supported (Al₂O₃ and SiO₂) catalysts will also assist here in correctly interpreting the spectra. From the analysis presented here together with the help of literature reports,^{Reference Khulber, Mann and Ternan31–Reference Louis and Che35,Reference Sobczynski and Zmierczak50} the signal appeared at g = 2.1144 was assigned to oxygen adsorbed species (as measurements were performed in open air), though it was not observed in most of the previous studies.^{Reference Khulber, Mann and Ternan31–Reference Louis and Che35} The signal appeared at the g-value of 2.0300 was assigned to sulfur chains in MoS₂ NCs. This is consistent with the previous observations reported^{Reference Bensimon, Belougne, Giuntini and Zanchetta33} on irradiated amorphous sulfur, where the electron hole is localized on the sulfur atom. It can be noted this signal is much narrower than that (120–150 G) of the similar sulfur-related signals reported in the literature.^{Reference Silbernagel, Pecoraro and Chianelli32–Reference Louis and Che35} Perhaps, this might be due to the strong exchange narrowing effect. The ESR signals appeared at g = 1.9863, g = 1.9417 could come from Mo⁵⁺ (4d ¹ configuration) paramagnetic centers surrounded by sulfur-rich environment (MoS₃). The ESR observations appeared at g = 1.8984, 1.8541, and 1.8146 could be ascribed to Mo⁵⁺ paramagnetic centers surrounded by oxygen-rich environment (MoO₃).

Now, ESR spectra of Ar-annealed MoS₂ NCs will be analyzed. The ESR signals are shown in Fig. 5. To trace the origin of the broad signal, the work done^{Reference Arcon, Zorko, Cevc, Mrzel, Remškar, Dominko, Gaberšček and Mihailovic30} by Arcon and co-authors will help us. The authors conducted X-band ESR study on Li-doped MoS₂ nanotubes. They observed two quite different components. A broad (800–2600 G) signal associated with the g-value of 2.15–2.52 and a narrow (∼4 G) signal appeared at the g-value of 2.0029 were detected. The g-value of the broad signal is farther from the free electron g-value (2.0023) which suggests the effect of strong spin–orbit coupling from Mo. The authors attributed the broad signal to (d-band in Mo) conduction electrons coupled to defects through exchange interaction. As the g-value and line width of the broad signal observed in the present work are in close agreement with their observations, the broad signal observed in the present work could be assigned to the conduction electron spins. It is natural to expect a similar situation that might have occurred in the current MoS₂ NCs during the Ar annealing process. This results in significant surface reorganization toward bulk-like structure, and, hence, MoS₂ becomes more conducting at the expense of reduction in defect density. On the other hand, the narrower and much weaker signal could be assigned to oxo-Mo⁵⁺ centers, which might be present in the form of small spin clusters. It is evident that this particular oxo-Mo⁵⁺ site still survived even after the Ar annealing treatment. This finding implies that the observed spin centers are of surface-related, and probably they are located on the edges of the layers. The spectra remain unaffected even at further higher measurement temperatures (25 and 40 K).

Some of the paramagnetic centers such as sulfur vacancies and Mo-related defects are known to be catalytically active. One can harness them to improve the materials’ catalytic performance. As discussed previously, the XRD pattern did not show any evidence of MoS₃/MoO₃ residual phases. However, from the current experimental findings derived from ESR spectroscopy, it is believed that MoS₂ NCs do contain unavoidable MoS₃ and MoO₃ fractions. This observation is consistent with the XPS data as well, where the presence of MoS₃ and MoO₃ fractions was evident. These defects can be exploited for their magnetic and catalytic properties. In fact, the initial magnetic measurements performed on MoS₂ NCs have revealed the presence of ferromagnetism at room temperature. Future work will look into the simulation of the observed X-band ESR spectra and perform very high frequency ESR measurements MoS₂ NCs to gain further insights.