Origin of exotic ferromagnetic behavior in exfoliated layered transition metal dichalcogenides MoS2 and WS2†

The room temperature ferromagnetism is closely related to the structural disorder on the sheet edges and the presence of a conductive metallic phase. The origin of room temperature ferromagnetism in transition metal dichalcogenides (TMDCH: MoS2 and WS2) was investigated in detail. The ferromagnetism is strongly dependent on the concentration of defects predominantly located on the TMDCH sheets edges as well as on the formation of the metallic 1T phase. Signifi cant diff erences between the ferromagnetism of TMDCH exfoliated using butyllithium and sodium naphtalenide originate from diff erences in the exfoliation degree. Nanoscale


Introduction
Layered transition metal dichalcogenides (TMDs) have received revived interest in the past few years being dubbed as the "next graphene" due to their possible applications in energy generation and their highly anisotropic material properties. [1][2][3][4] In contrast to graphene, where the layer is single atom thick and truly 2-dimensional, MoS 2 (or WS 2 ) sheets consist of transition metals sandwiched between two layers of chalcogen atoms. These triple layers are bonded to each other by van der Waals interactions. 5,6 Such a "triple-layer" can exhibit three different crystallographic arrangements, the most common semiconducting 2H (trigonal prism), metallic 1T (octahedral) and semiconducting 3R (trigonal prism). 7,8 The broad selection of metal elements and chalcogens in TMD as well as the large variety of different crystal structures of the materials yield a large number of combinations for tailoring the TMD properties. One of them is the magnetic behaviour. Similarly to graphite, MoS 2 or WS 2 in their bulk forms do not exhibit ferromagnetic or paramagnetic behavior. The magnetic properties of TMDs have been studied both theoretically and experimentally. Theoretically it has been predicted that the introduction of sulphur vacancies onto the S-edges of MoS 2 nanoribbons would lead to a non-zero net magnetic moment. 9 Furthermore zigzag and armchair terminated edges of MoS 2 nanoribbons were studied by density functional theory (DFT) calculations. It was predicted that the zigzag terminated nanoribbons are metallic and show ferromagnetic properties while armchair terminated nanoribbons are semiconducting and nonmagnetic. 10 A magnetic ordering has also been theoretically described for WS 2 nanosheets. The results of DFT calculations have shown that the adsorption of non-metallic elements such as H, B, O or F onto the edges of TMDs induces magnetic properties. 11 More recently experimental studies have been emerging. While the synthesis of magnetic TMD nanoribbons has not been reported, probably because of the difficulties in their synthesis, 12,13 only a few reports on TMD magnetic nanosheets exist. The first report on magnetic MoS 2 nanosheets was based on thermal evaporation synthesis onto silicon or tantalum foil substrates. 14 However the role of the substrate and the interface between MoS 2 and the substrate remains unclear and may have influenced the magnetic properties of the prepared samples. 15,16 Further studies have shown the preparation of MoS 2 by a two-step hydrothermal synthesis, which introduces sulphur vacancies responsible for magnetic behaviour. 17 Magnetic WS 2 nanosheets have also been prepared by exfoliation in DMF. 18 However, the frequently used organometallic intercalation followed by exfoliation has not been employed for the synthesis of the above mentioned TMD nanomaterials. These methods are easily scalable and relatively undemanding. Here we wish to show that exfoliated MoS 2 and WS 2 exhibit exotic ferromagnetic properties, which we consider as an unexpected behavior in view of the fact that bulk TMDs exhibit diamagnetic properties.

Materials
Molybdenum disulphide (99.8%) and tungsten disulphide (99.8%) were obtained from Alfa Aesar. n-Butyllithium (1.6 M in hexane), benzophenone and sodium were obtained from Sigma-Aldrich, Czech Republic. Tetrahydrofuran (THF), hexane and naphthalene were delivered by Lach-Ner, Czech Republic. Hexane was dried with sodium before use and THF was dried by distillation from a sodium/benzophenone mixture under an argon atmosphere before use.

Synthesis procedure
The exfoliation of MoS 2 and WS 2 with butyllithium was performed at room temperature. 3 g of MoS 2 was stirred with butyllithium solution for 72 hours. Subsequently the intercalated MoS 2 and WS 2 , respectively, were filtered out and washed with hexane. All these operations were performed under an argon atmosphere in a glovebox. Intercalated MoS 2 and WS 2 were dispersed in water and repeatedly centrifuged and redispersed in water. Finally the exfoliated material was dried in a vacuum oven for 48 hours at 50°C before further use.
Sodium naphthalenide used as an intercalation compound was prepared by stirring sodium metal (0.35 g) with naphthalene (1.8 g) for 2 h in 50 ml of THF under an argon atmosphere. After the addition of transition metal dichalcogenide (TMD) powders (3 g), intercalation was carried out by stirring the solutions at room temperature for 24 h under an argon atmosphere. The prepared intercalates were then suction filtered and washed several times with THF. Separated samples were dispersed in water and centrifuged several times. Finally the exfoliated material was dried in a vacuum oven for 48 hours at 50°C before further use.

Analytical techniques
Morphology studies were performed by using a scanning electron microscope (SEM) equipped with a FEG source of electrons (Tescan Lyra Dual Beam Microscope). The sample composition was determined by means of an energy dispersive spectroscopy (EDS) analyzer (X-MaxN) with a 20 mm 2 SDD detector (Oxford Instruments). Data were evaluated using AZtecEnergy software. Before the measurement, samples were placed onto a conductive carbon tape to avoid charging effects. All the measurements were carried out with 15 kV acceleration voltage.
High-resolution transmission electron microscopy (HRTEM) was carried out on a transmission electron microscope JEOL JEM 2200FS operated at 200 kV (autoemission Schottky gun, point resolution 0.19 nm) with an in-column energy Ω-filter for EELS/EFTEM analyses, a STEM unit, and an Energy Dispersive X-ray (EDX) SDD detector (Oxford Instruments X-Max attached). Images were recorded on a Gatan CCD camera with a resolution of 2048 × 2048 pixels using the Digital Micrograph software package. EDX analyses were performed and treated with the INCA software package.
High resolution X-ray photoelectron spectroscopy (XPS) was performed with an ESCAProbeP (Omicron Nanotechnology Ltd, Germany) spectrometer using a monochromatic aluminium X-ray radiation source (1486.7 eV). A wide scan survey of all elements was performed with consequent high-resolution scans of the Mo, W and S core levels. Relative sensitivity factors were used to determine the chalcogen-to-metal ratio and the quantitative distribution of 2H and 1T phases. The samples were placed onto a carbon conductive tape prior to the measurement.
Characterization by Atomic Force Microscopy (AFM) was performed on the NT-MDT Ntegra Spectra from NT-MDT in tapping mode. The samples were dispersed in isopropyl alcohol (0.5 mg mL −1 ), sonicated for 10 minutes and then dropped onto a freshly cleaved mica substrate before the measurement.
An InVia Raman microscope (Renishaw, England) was used for Raman spectroscopy measurements. The spectrometer operates in backscattering geometry with a CCD detector. A Nd-YAG laser (532 nm, 50 mW) was used with 50× magnification objective. The instrument was calibrated on a silicon reference which gives a peak position at 520 cm −1 . The laser power did not exceed 5% of the maximum 50 mW power in order to avoid damage to the sample.
The magnetic measurements were performed using a SQUID magnetometer MPMS-XL (Quantum Design). The hysteresis loops were recorded between −10 and 10 kOe at T = 300 K. The magnetization measurements at T = 2 K were conducted between 0 and 70 kOe.
The surface areas were measured using adsorption of methylene blue. About 0.05 g of each TMD material (weighed exactly to within ±0.01 mg) was dispersed in 70 mL of deionized water by ultrasonication (400 W, 30 min). A defined amount of 1 mM of methylene blue aqueous solution was added to the suspension and then filled to 100 mL. The concentration of unabsorbed methylene blue was determined by UV-Vis spectroscopy at λ max = 665 nm (Cary 50, Agilent). For the calculation of surface areas, the value of 1.05 nm 2 per methylene blue molecule was used. 19

Results and discussion
Here we have investigated the effect of the exfoliation process on the magnetic and structural properties of molybdenum and tungsten disulfide. We have used two agents for chemical exfo-liation of bulk MoS 2 and WS 2 : butyllithium (BuLi) and sodium naphthalenide (NAPH). While butyllithium is commonly used for the chemical exfoliation of transition metal dichalcogenides (TMDs), 20,21 only little has been reported on TMD exfoliation using sodium naphthalenide. 22 Further on, we will show that sodium naphthalenide is not only more efficient in TMD exfoliation than butyllithium for both MoS 2 and WS 2 but also induces magnetic ordering. Before the discussion of exotic magnetic properties of exfoliated TMDs we thoroughly examined the exfoliated TMDs by scanning electron microscopy (SEM), atomic force microscopy (AFM), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, high-resolution transmission electron microscopy (HR-TEM), selective area electron diffraction (SAED) and surface area measurement by methylene blue adsorption. We will show that the magnetic properties measured by using a superconducting quantum interference device (SQUID) strongly depend on the character and concentration of the induced defects. Simultaneously, the phase transition from the 2H to 1T form, which sets in as a result of intercalation (likely due to charge carrier doping), has to be correlated with the peculiar magnetic behavior.
The morphology of the exfoliated and bulk TMDs was examined by SEM as shown in the ESI (Fig. SI1 †). One can observe that for BuLi exfoliated WS 2 , most of the sheets remained stacked with sharp edges, indicating an incomplete intercalation and the subsequent exfoliation has been incomplete. It is known that intercalation of Li to WS 2 is incomplete at room temperature and becomes more pronounced at higher temperatures. 23 NAPH exfoliation of WS 2 , on the other hand, leads to sheets with smaller lateral sizes and slight edge wrinkling, which points to a significantly higher degree of exfoliation than that in the case of WS 2 ·BuLi. A similar trend can be observed for MoS 2 , however, with more substantial edge wrinkling, especially for NAPH exfoliated MoS 2 .
XRD measurements were further conducted (Fig. SI2 †) to confirm the presence of pure TMDs. We have not detected any diffraction lines originating from molybdenum or tungsten oxides. This indicates a low degree of oxidation of MS 2 to MO 2 / MO 3 . Even though some increase in MO 3 was observed by XPS measurements (see XPS discussion), it was only on the surface and was only detectable by XPS. It is also obvious from the data that NAPH exfoliated TMDs show less intense and broadened reflections. This is especially apparent for NAPH·MoS 2 which only shows weak and very broad reflections indicating a high degree of exfoliation. As for NAPH·WS 2 we have detected much higher tendency for restacking (even though AFM and Raman spectroscopy indicate a similar level of exfoliation as for MoS 2 and all XRD measurements were performed within a couple of weeks after preparation) which is obvious from its quite sharp XRD pattern. Also the BuLi exfoliated TMDs show much lower degree of exfoliation which is in good agreement with the data discussed later on.
We used AFM to determine the sheet thickness and to further investigate the morphology of the exfoliated MoS 2 and WS 2 . Fig. 1 shows the AFM images documenting that the use of BuLi for exfoliation of MoS 2 and WS 2 leads to multilayered materials with sheet thicknesses of ∼20-25 nm. While WS 2 retains its sharp edged structure observed by SEM, MoS 2 consists of sub-micron sized flakes. The application of NAPH led to a high degree of exfoliation of both MoS 2 and WS 2 manifested by a typical sheet thickness (∼0.7 nm) corresponding to a single layer. This result is in good agreement with the previously reported values of the monolayer thickness of MoS 2 and WS 2 , respectively. 5,24 To obtain insight into the structural properties of exfoliated TMDs, we performed Raman spectroscopy. It is well known that Raman spectroscopy is a very useful tool for identifying structural properties as well as the number of layers as has been shown several times for various TMDs including MoS 2 and WS 2 . 24-26 WS 2 shows two first-order modes denoted as E 1 2g and A 1g at 356 cm −1 and 418 cm −1 , respectively and a disorder activated mode denoted as LA at 352 cm −1 , which dominates the WS 2 Raman spectrum. These modes can be effectively used to determine the number of layers. Raman spectra show no apparent difference between BuLi exfoliated WS 2 and the spectrum of bulk materials (Fig. 2). This would suggest that the material was delaminated only to a small extent, which is in good agreement with the results previously observed by SEM and AFM. The spectrum of NAPH exfoliated WS 2 shows major differences. We can note the decrease in the I(A 1g )/I(E 1 2g ) ratio from ∼2 in the bulk material to ∼0.35 in  NAPH exfoliated WS 2 . This has been previously reported as a characteristic feature of a single atom layered WS 2 . 24 It should be noticed that the band centered at around 354 cm −1 is broadened, which is associated with an activation of LA mode due to disorder introduced during the exfoliation (e.g. edge wrinkling observed by SEM and HR-TEM). In addition we have located low intensity peaks in the low frequency region (see Fig. SI3 †). These peaks correspond to the 1T phase and are denoted as J 1 , J 2 and J 3 . 27 Even though these modes confirm the presence of the 1T form since they are not allowed in the 2H phase structure, it is more accurate to use other methods, such as XPS, to calculate the 1T concentrations. 27 It is noticeable that these modes are also visible even in the spectrum of BuLi·WS 2 where AFM and XRD showed a very low, if any, degree of exfoliation. This suggests that even though the intercalation of BuLi into WS 2 and its subsequent exfoliation are highly inefficient, some charge carrier doping takes place, leading to a slight increase in the 1T phase concentration (see XPS for a detailed discussion). The modes denoted as E 1 2g at 383.5 cm −1 and A 1g at 408.6 cm −1 in the spectra of MoS 2 are of analytical value. The peak separation Δ is used in the case of MoS 2 for the determination of the number of layers. We have obtained Δ = 26.1 cm −1 for bulk MoS 2 and Δ = 26 cm −1 for BuLi exfoliated MoS 2 . These values are nearly identical and they signify that the thickness of BuLi exfoliated MoS 2 is more than four layers. 28 By contrast, we have shown by AFM measurements that the average sheet thickness is lower than that of the bulk material. More importantly, NAPH exfoliated MoS 2 shows the peak separation of Δ = 21.4 cm −1 . This value is close to that of monolayer MoS 2 documenting that the high efficiency of NAPH was highly efficient in MoS 2 exfoliation. 28,29 In addition to E 1 2g and A 1g modes, several additional features can be seen in the low wavenumber region of MoS 2 and WS 2 spectra (see Fig. SI3 †). Firstly, modes denoted as J 1 , J 2 and J 3 ( pink regions) are clearly visible in the spectra of MoS 2 ·BuLi and MoS 2 ·NAPH. These modes are not allowed in 2H modification but become allowed in 1T MoS 2 polymorph. 30,31 This indicates a successful 2H to 1T phase transition in BuLi·MoS 2 and NAPH·MoS 2 samples. In the spectrum of bulk MoS 2 none of these modes (J 1 , J 2 , J 3 ) can be detected. Even though the presence of J 1 , J 2 and J 3 phonon modes prove the presence of the 1T phase, other methods such as XPS are more suitable for determining the relative 1T phase concentrations. 30 The peak at ∼280 cm −1 is related to the presence of MoO 3 . 32 The presence of MoO 3 on the surface has been proven by XPS measurements (see XPS discussion). The increase in the intensity of the peak located at ∼280 cm −1 , compared to the bulk material, for MoS 2 ·BuLi and MoS 2 ·NAPH is likely caused by slightly higher amounts of MoO 3 on the surface of these samples, also observed by XPS.
So far we have shown by means of SEM, AFM and Raman measurements that sodium naphthalenide is more efficient in exfoliation of bulk MoS 2 and WS 2 than the usually used butyllithium. In order to obtain better information on the chemical changes taking place within the prepared materials, we con-ducted XPS measurements. Detailed high-resolution spectra of core-level Mo 3d and W 4f after Shirley background subtraction are displayed in Fig. 3.
It is well known that bulk materials consist mainly of the semiconducting 2H phase. During the exfoliation varying amounts of metallic 1T phase emerge as a consequence of electron doping introduced during intercalation. 33 The relative changes of the transition from the 2H to 1T phase are summarized in Table 1. For WS 2 , the spectra consist of W 4+ 4f 5/2 and W 4+ 4f 7/2 peaks at approximately 32 eV and 35.1 eV, respectively, corresponding to the 2H phase. A contribution of the 1T phase can be also found in the spectra at around 31.1 eV and 34.2 eV as well as the pair corresponding to W 6+ at ∼36.8 eV and ∼38.5 eV. The deconvolution of those components is essential when trying to identify the chemical changes occurring during the exfoliation.
For BuLi exfoliated WS 2 it is clearly visible that there is only a small contribution from the metallic 1T phase of about 10 at%. This reflects the low degree of exfoliation (and the low degree of electron doping) also supported by electron microscopy and Raman measurements. Then for NAPH exfoliated WS 2 a much more pronounced increase in the 1T phase to ∼37% was observed. The presence of W 6+ is also notable from the spectra for both samples.
Let us note that even in bulk materials there is a certain amount of tungsten atoms (as well as Mo in MoS 2 ) in the  It is also possible to track chemical changes in molybdenum and tungsten disulphides by deconvolution of the S 2p core-level, but it is rather difficult to determine the changes with satisfying accuracy because of the low proximity and overlap of individual components. The S 2p high-resolution core-level spectra are shown in Fig. SI4. † In addition the survey spectra are shown in the ESI (Fig. SI5 †) together with the individual element concentrations (Table SI2 †). The HR-TEM measurement was performed in order to further investigate the structure of exfoliated MoS 2 and WS 2 at the atomic level. The SAED in combination with EDS was applied to confirm the structure and elemental composition of exfoliated TMDs. Both materials, exfoliated with butyllithium and sodium naphthalenide, are composed of both single and few layer sheets. All exfoliated TMDs with the exception of the WS 2 ·BuLi sample exhibit wrinkling close to the rim of the sheets (Fig. SI6 †). The EDS analysis yielded the S/Mo ratio close to the theoretical value of 2 (2.09 for MoS 2 ·BuLi and 1.93 for MoS 2 ·NAFT). The EDS analysis of WS 2 exfoliated with butyllithium and sodium naphthalenide gave the S/W ratio of 0.41 and 1.23, respectively. Such a large difference in the S/W ratio can be explained by the large difference in tungsten and sulfur molar mass. The selective area electron diffraction (SAED) was performed on selected sheets of TMDs which were further investigated by HR-TEM microscopy. The diffraction shows the presence of both single and few layer TMDs with the hexagonal symmetry of the structure. The results of SAED are shown in the ESI (Fig. SI7 †). The HR-TEM can be used to distinguish between 2H and 1T phases. In order to identify different phases, detailed images were obtained. In comparison with BuLi exfoliated TMDs we can observe a high degree of disorder on the rims of TMD sheets exfoliated with sodium naphthalenide. This can be seen in detail in Fig. 4 for exfoliated MoS 2 and WS 2 . The disorder in the MoS 2 and WS 2 structures was theoretically predicted to be responsible for room temperature ferromagnetism 34 and also experimentally reported for MoS 2 layers with a high concentration of edges. 35 Since the samples exfoliated with sodium naphthalenide exhibit a significantly larger degree of exfoliation with a higher concentration of edges and a high degree of disorder, the relationship with the observed room temperature ferromagnetism appears to be quite straightforward.
In order to recognize the presence of the 2H and 1T phases from HR-TEM images, we used the method reported in the literature. 30 The MoS 2 ·NAPH showed the dominant presence of 1T modifications while for the sample MoS 2 ·BuLi, we can see both the 1T and 2H structures. This is in qualitative agreement with XPS results, where the higher concentration of the 1T phase was observed for samples exfoliated with sodium naphthalenide. The HR-TEM images of exfoliated MoS 2 and WS 2 are shown in Fig. 5.  In order to better quantify the degree of MoS 2 and WS 2 exfoliation we performed the surface area measurement using methylene blue adsorption. This method was reported to be useful for surface area measurement of TMDs and other layered compounds such as clay silicates. 36,37 The obtained surface area of MoS 2 and WS 2 exfoliated with butyllithium was 62.1 m 2 g −1 and 36.1 m 2 g −1 , respectively. A significantly larger area was obtained for MoS 2 and WS 2 exfoliated using sodium naphthalenide, 88.3 m 2 g −1 for MoS 2 ·NAPH and 190.2 m 2 g −1 for WS 2 ·NAPH. The observed differences in the surface area of MoS 2 and WS 2 exfoliated with sodium naphthalenide and butyllithium are in good agreement with the results obtained by microscopic techniques, where a substantially higher degree of exfoliation was observed for NAPH samples.
The magnetic measurements of butyllithium and sodium naphthalenide exfoliated TMDs were carried out using a SQUID magnetometer. At room temperature (T = 300 K) the field dependence of the magnetization was found to be given by a superposition of a diamagnetic moment proportional to the magnetic field (the corresponding susceptibilities X D are listed in Table 2) and a small ferromagnetic (FM) contribution m FM , for which the hysteresis loops are plotted in Fig. 6a.
The hysteresis curve is characterized by a very small coercivity of the order of 100 Oe and by the saturation magnetization m S . The FM contribution is roughly independent of temperature. For samples WS 2 ·BuLi and MoS 2 ·BuLi the FM magnetization, if any, lies under a limit of 2.5 × 10 −5 emu g −1 given by the accuracy of the measurement. At T = 2 K all samples exhibit a paramagnetic (PM) behaviour (Fig. 6b) which can be approximately described by the sum of the Brillouin contribution for spin S with the saturated value m PM0 and a small linear contribution with a susceptibility χ 1.
The value m PM0 , the estimated spin S as well as the susceptibility χ 1 are given in Table 2. In addition we performed a careful magnetic characterization of the starting bulk MoS 2 and WS 2 . For both bulk materials the room temperature hysteresis curves corrected for a diamagnetic contribution with the susceptibility χ D are shown in the ESI (Fig. SI8 †). The ferromagnetic (FM) magnetization is observed only for the sample MoS 2 , for which the value of the saturation magnetization m S (see Table 2) reliably exceeds the error level 2.5 × 10 −5 emu g −1 .
In this case the hysteresis curve exhibits a finite remanent magnetization of 8.1 × 10 −5 emu g −1 and a coercivity of 240 Oe. However this value is one order of magnitude lower compared to the exfoliated MoS 2 . For the WS 2 sample we can also formally speak about the FM contribution but its value is lower than the error level. At the lowest temperature T = 2 K we measured the magnetization curves providing information about the paramagnetic (PM) contribution to the magnetic moment (Fig. 6b). The m(H) curves can be well approximated by the sum of the PM Brillouin contribution for spin S = 2, g = 2 and a linear susceptibility χ 1 ( Table 2). First we shall discuss the values of diamagnetic susceptibility contributions. The theoretical diamagnetic susceptibility for MoS 2 and WS 2 can be calculated using the data for Mo 4+ , W 4+ and S 2− ions. 38 This yields χ D (MoS 2 ) = −3.436.10 −7 emu (g Oe) −1 and χ D (WS 2 ) = −2.46.10 −7 emu (g Oe) −1 . Whereas for the bulk materials MoS 2 and WS 2 we could speak about a rough agreement with the measured values of χ D , a clear disagreement occurs for the sample MoS 2 ·NAPH (having the largest FM magnetization) where the measured value of χ D is of the order of 10 −8 emu (g Oe) −1 . The observed FM magnetization in the bulk is significantly lower compared to the material exfoliated with sodium naphthalenide. As far as the PM magnetization is concerned it is interesting that for the MoS 2 bulk sample the Brillouin approximation yields S = 2 in contrast to S = 1.5 for MoS 2 ·NAPH. This difference originates from the high degree of exfoliation and the 2H-1T phase transition observed by XPS. In the case of so small FM contributions, it is necessary to take into account the possible influence of impurities Fig. 6 (a) T = 300 K hysteresis loops for MoS 2 and WS 2 , (b) T = 2 K, magnetization curves for MoS 2 and WS 2 (symbolsmeasured data, lines calculation for S = 1 and the corresponding values m PM0 , χ 1 given in Table 2).  39 For bulk MoS 2 and WS 2 no FM ordering has been observed so far to the best of our knowledge. However, detailed magnetic measurement of starting MoS 2 and WS 2 showed no or extremely weak magnetization (MoS 2 ) at room temperature and also the ferromagnetism of samples MoS 2 ·BuLi and WS 2 ·BuLi was below the detection limit of the SQUID device. This leads us to the conclusion that the room temperature ferromagnetism observed in sodium naphthalenide exfoliated MoS 2 and WS 2 is associated with several effects which take place during the TMD exfoliation. First we have to take into account the 2H-1T phase transition which is significantly larger compared to the samples exfoliated with butyllithium. The transformation of MoS 2 and WS 2 from 2H to 1T modification was observed by HR-TEM and well documented by XPS spectroscopy from deconvolution of Mo 3d and W 4f HR-XPS spectra, respectively. Moreover, a larger surface area imposes a higher concentration of edges, where a large amount of defects can be observed in HR-TEM images. Since NAPH samples possess a high degree of exfoliation, this will be one of the main factors necessary for the room temperature ferromagnetic ordering caused by the high concentration of edges related to a high degree of exfoliation and possibly also connected to the defects present on the edges. These defects may also lead to a PM contribution which superimposes on the intrinsic diamagnetic component. This manifests itself in the reduction of the measured susceptibility χ D as was observed, e.g., in the case of the sample MoS 2 ·NAPH. Let us mention that the FM state induced by defects was predicted for various types of defects in MoS 2 and also experimentally observed on nanostructured MoS 2 layers with a high concentration of defects. 35,40 We also have to note one possible origin of ferromagnetism, which has been recently reported on another transition metal dichalcogenide -ReSe 2 . The room temperature ferromagnetism was observed on the wrinkled area of ReSe 2 . 41 Such a wrinkling of the layers was also observed in the TEM images of MoS 2 and WS 2 sheets exfoliated with sodium naphthalenide. Although this feature was also observed in the MoS 2 exfoliated with butyllithium, a significantly lower degree of exfoliation led only to a minor influence on the magnetic behavior. Another interesting finding is the presence of a small PM contribution corresponding to the concentration of 10 −3 -10 −4 of PM centers with a spin S. The spin S = 1 could be connected with the occurrence of a spin-pairing of d 2 centers mentioned by other authors. 2 Our results suggest larger values of spins for the exfoliated WS 2 samples.

Conclusions
The detailed investigation of MoS 2 and WS 2 exfoliated using various alkali metal intercalating compounds like butyllithium and sodium naphthalenide yields materials with significant differences in the degree of exfoliation. The exfoliation using butyllithium predominantly resulted in a few layer materials, while the exfoliation using sodium naphthalenide led to almost complete exfoliation and formation of single layer TMDs. Such an intensive process of intercalation and subsequent exfoliation is associated with the 2H to 1T phase transition and formation of defects on the edges of TMD sheets. The presence of a metallic phase (1T) and formation of defects in the TMD structure can be responsible for the exotic ferromagnetic behavior which was observed up to room temperature. Even though the room temperature ferromagnetism is of the order of 1 × 10 −3 emu g −1 , this effect was not observed in bulk MoS 2 and WS 2 or even in the butyllithium exfoliated material. Butyllithium exfoliated samples exhibit a significantly lower degree of exfoliation and subsequently a lower concentration of defects on the edges of individual sheets. This led us to draw a conclusion that the room temperature ferromagnetism is closely related to the structural disorder on the sheet edges and the presence of a conductive metallic phase. Our finding can have a strong impact on the future applications of transition metal dichalcogenides in the construction of novel spintronic devices.