Co‐ and Nd‐Codoping‐Induced High Magnetization in Layered MoS2 Crystals

Magnetic 2D‐layered materials are promising for the applications of spintronic devices and compact magnetic devices. There are only a few reported intrinsic 2D‐based magnetic materials. Therefore, introducing magnetic element into 2D‐layered materials is an effective strategy to synthesize 2D‐layered magnetic materials. Recently, ferromagnetism has been realized by doping single transition‐metal or rare‐earth element, whereas, there is no report by codoping of both transition‐metal and rare‐earth elements. Herein, Co and Nd are codoped into MoS2‐layered crystals by ion implantation. An extremely high magnetization of 6916.3 emu cm−3 at 10 K and 80.3 emu cm−3 at 300 K is achieved. The high magnetization is attributed to the contribution from both transition‐metal and rare‐earth elements as well as defects, such as vacancy of cation ions, anions, or interstitials. Hence, herein, a useful strategy may be opened to develop high‐performance magnetic materials based on 2D‐layered materials.

to introduce ferromagnetism include creating native defects or edges, doping of magnetic elements, and utilizing proximity effect of external ferromagnet. [5,[8][9][10] Doping with magnetic elements has been considered as an effective way to achieve ferromagnetism in semiconductors for spintronic devices. [11] Therefore, it has also been employed for the research of ferromagnetism in 2D materials. Initially, The ferromagnetic behavior and origin of the ferromagnetism in magnetic element-doped 2D materials were investigated by theoretical calculations, [5] while experimentally it is difficult to achieve effective doping in 2D materials. Therefore, doping magnetic into layered materials, such as MoS 2 , WSe, has been extensively researched, [10,12] which is similar to the research of diluted magnetic semiconductors. [13] Among these reports, many of the materials show weak ferromagnetism with very low coercivity and saturation and researchers challenge whether it is a true ferromagnetism. [12] More recently, large coercivities and high saturation magnetization have been reported in certain magnetic element-doped 2D-layered materials and in other complex layered materials. [4,14] It is well established that a high-room-temperature magnetization and a large coercivity are crucial ingredients for technological applications as either soft or hard magnetic materials. Atomic scale geometry of 2D-layered magnetic materials makes them favorable for fabrication of compact advanced nano-devices. [15] Therefore, the control of coercivity and high magnetization are essential for future compact spintronic devices.
Previous research for magnetic element-doped 2D-layered materials only involved single magnetic element or one magnetic element together with one nonmagnetic element. In this work, transition-metal element Co and rare-earth element Nd were doped together into layered 2H-MoS 2 single crystal by employing ion-implantation technique. Rare-earth element (Nd) is chosen as a dopant owing to its strong spin-orbit coupling and ability to effectively tailor the magnetic properties of semiconductors. [16] In addition, doping of transition metal (Co) also has significant influence on the magnetic properties of MoS 2 . [17] In this research work, 1 at% Nd-and 4 at% Co-codoped MoS 2 demonstrates a colossal saturation magnetization of 6916.3 emu cm À3 , which is four times higher than metallic Co (%1500 emu cm À3 ). The very high saturation magnetization is ascribed to the dopant, native/implantation-induced defects and associated complexes. Moreover, the increase of Nd concentration induces a relatively large coercivity of 700 Oe suggesting spin-orbit coupling-induced anisotropy by rare-earth element doping. The work may offer a potential pathway for engineering 2D-layered-based magnet.   www.advancedsciencenews.com www.pss-rapid.com respectively. The choosing of 4 at% Co is based on previous work. [10] All the observed sharp peaks can be linked to the MoS 2 , which shows that the hexagonal crystal structure of MoS 2 is not influenced by Co and Nd doping. However, the intensity becomes lower, suggesting formation of defection inside the single crystals. Moreover, no secondary phase related to Co/Nd or impurities is detected as evident from the absence of characteristic peaks related to Co/Nd species in all samples. This further demonstrates that all Co and Nd atoms are undergone substitutional doping inside the MoS 2 lattice to form Co x Nd y Mo 1ÀxÀy S 2 . It should be noted that XRD is not sensitive enough, which cannot detect very small number of impurities. Hence, a little fraction of agglomeration without substitution may exist. It is observed that the [002] peak intensity decreases after the Co and Nd doping, which may be attributed to the implantation-induced lattice distortion or disorder. [18] Figure 1b displays the peak shift of enlarged [002] peak. Peaks of 4 at% Co-MoS 2 and 1 at% Nd, 4 at% Co-MoS 2 samples exhibit red shift. This may be associated with the smaller ionic radius of Mo þ5 (75 pm) and Mo þ6 (73 pm), as compared to Mo þ4 (79 pm), produced during ion-implantation process as evidenced by the XPS analysis, which will be discussed later. However, blue shift of peak for 2 at% Nd, 4 at% Co-MoS 2 sample may be linked to higher concentration of Co þ2 (88.5 pm) and Nd þ3 (112.3 pm) with larger ionic radius. Moreover, further increase in the Nd concentration in 5 at% Nd, 4 at% Co-MoS 2 sample shifts the peak to higher diffraction angle due to the formation of more concentration of Mo þ5 (75 pm) and Mo þ6 (73 pm).

Results and Discussion
To detect the structural defects and to identify phases in pristine and doped MoS 2 samples, Raman spectroscopy was performed as depicted in Figure 1c. In un-doped MoS 2 , prominent peaks at 382.2 and 408.0 cm À1 may be attributed to the E 1 2g (in-plane Mo-S phonon mode) and A 1g (out-of-plane Mo-S mode), respectively. These peaks also demonstrate a good crystallinity of MoS 2 [19,20] and support XRD results. Terrace and edge-terminated structures are responsible for E 1 2g and A 1g vibration modes. [21] The 4 at% Co doping into MoS 2 induces the peak shift of E 1 2g and A 1g to 378.8 and 404.6 cm À1 , whereas, it remains unaffected by further Nd codoping. This peak shift in Raman spectra of MoS 2 is due to transition-metal doping, which has also been reported earlier. [21,22] Reduction in peak intensity, peak broadening, and noticeable peak shift may be ascribed to the doping-induced defects, lattice disorder, and size distribution. [18,23,24] Similar to XRD results, no other peak is observed in the Raman spectra, which indicates the absence of impurity phases and uniform distribution of Co and Nd atoms at the doping sites of MoS 2 . Furthermore, Figure 1d displays the sharp decrease in intensity ratios of A 1g /E 1 2g with the 4 at% Co doping, which further drops with the incorporation of Nd (1 at% and 2 at%) atoms into 4 at% Co-MoS 2 . Further increase in concentration of Nd (5 at%) tends to increase the A 1g /E 1 2g ratio. Figure 1d provides the rich texture information of the un-doped and doped MoS 2 samples and demonstrates that transition-metal  www.advancedsciencenews.com www.pss-rapid.com (Co) and rare-earth-metal (Nd) codoping in MoS 2 promotes E 1 2g vibration mode and hence terrace-terminated structure. [20,21] Oxidation state and chemical bonding of Co-and Nd-codoped samples were examined by employing XPS. Figure 2a,d exhibits XPS spectra of Mo(3d), S(2p), Co(2p), and Nd(4d) core levels. Figure 2a shows a prominent Mo(3d) doublet against pure MoS 2 with the corresponding binding energy (E b ) of 229.6 eV of Mo (3d 5/2 ), which may be associated with the Mo 4þ ions. The 4 at% Co doping into MoS 2 introduces three new doublets in addition to Mo þ4 with the corresponding binding energies of 228.89, 230.62, and 233.0 eV, which may demonstrate the presence of Mo 3þ , Mo 5þ , and Mo 6þ ions, respectively. [24,25] Figure 2b, the pristine MoS 2 sample displays a single S(2p) doublet at 162.6 eV (2p 3/2 ), which may correspond to a S 2À . Co alone and Co/Nd codoping into MoS 2 introduce two new feeble doublets which may be attributed to the S valence state other than S 2À . The S (2p) core level peaks of S(2p 3/2 ), for 1 at% Nd, 4 at% Co-MoS 2 , 2 at% Nd, 4 at% Co-MoS 2 , and 5 at% Nd, 4 at% Co-MoS 2 samples at 162. 25 Similarly, Figure 2c depicts the XPS spectra of Co(2p) core level. Co(2p 3/2 ) peaks at binding energies of 779.05, 779.02, 778.75, and 778.5 eV may be ascribed to CoMoS; whereas peaks at 781.52, 781.33, 781.35, and 781.36 eV for 4 at% Co-MoS 2 , 1 at% Nd, 4 at% Co-MoS 2 , 2 at% Nd, 4 at% Co-MoS 2 , and 5 at% Nd, 4 at% Co-MoS 2 , respectively, correspond to Co 2þ . [23,24,26] XPS spectra of Nd(4d) core level are presented in Figure 2d. The peaks of Nd(4d) at binding energies of 122.60, 122.13, and 122.18 eV, in 1 at% Nd, 4 at% Co-MoS 2 , 2 at% Nd, 4 at% Co-MoS 2 , and 5 at% Nd, 4 at% Co-MoS 2 samples respectively indicate the presence of Nd 3þ . [27] It demonstrates that high doping concentration of Nd leads to more Nb 3þ and agrees with the peak shift in XRD spectra. Figure 3 illustrates the S/Mo ratio variations in all five samples based on XPS results. Un-doped MoS 2 exhibits the highest S/Mo ratio of 2.27 suggesting the presence of Mo (cation) vacancies, which tends to sharply decrease with the 4 at% Co doping and reach to 1.34, indicating that Co doping induces S vacancies in MoS 2 . The 1 at% Nd codoping into 4 at%Co-MoS 2 shows relative improvement in S/Mo ratio (1.60), which is an evidence of removal of anion vacancies. However, a decrease in S/Mo ratio is observed with the increment in Nd doping (2 at% and 5 at%) and it reaches its minimum value of 1.23, hence exhibiting the existence of maximum S vacancies. These results demonstrate that increasing doping concentration stimulates S vacancies. The formation of chalcogen (S) vacancies in TMDCs may be a consequence of exposure to the radiation during the ion-implantation process. [28] Co and Nd depth profile and distribution in pristine and doped MoS 2 samples were investigated by utilizing SIMS, as shown in Pure MoS 2 is generally known as a diamagnetic material, which may exhibit ferromagnetism attributing to zigzag edges at grain boundaries and/or sulfur vacancy. [29] Doping of transition element is also an effective way to induce magnetism in TMDCs. In this study, Co was initially doped into MoS 2 and subsequently Nd was codoped with Co into MoS 2 to explore its magnetic properties. Figure 5a-d illustrates the hysteresis loops of Co doped alone and Co/Nd-codoped MoS 2 at different temperatures. At 5 K, 4 at% Co-doped MoS 2 displays prominent ferromagnetic signal with a saturation magnetization of 601.9 emu cm À3 (at 50 kOe applied magnetic field), as shown in Figure 5a. The ferromagnetic signal is reduced to 30.0 emu cm À3 at room temperature. The diamagnetic signal from the bulk MoS 2 substrate was removed from the results of all samples for better visualization of the ferromagnetic signal. After codoping of 1 at% Nd into 4 at% Co-MoS 2 , the M-H loops, at 10 K, demonstrate a significantly high saturation magnetization of 6916.3 emu cm À3 at 50 kOe, which has never been reported in the research of diluted magnetic semiconductors, as depicted in Figure 5b. The saturation magnetization does not change if taken at the temperature of 5 K. As a comparison, the highest saturation magnetization reported for a doped magnetic semiconductor (Co-doped MoTe 2 ) is around 2231 emu cm À3 . [30] Moreover, observed magnetic signal in current work is %4 times higher than metallic Co (%1500 emu cm À3 ). However, at 300 K, magnetic signal sharply declines to 2.8 emu cm À3 at 50 kOe, indicating strong temperature dependence. Assuming the magnetic moment is from Co,  www.advancedsciencenews.com www.pss-rapid.com  www.advancedsciencenews.com www.pss-rapid.com the magnetic moment of Co will be around 3,700 μ B /Co atom. This is huge compared to theoretical value of 1.7 μ B /Co. Hence, the magnetic moment should be associated with other sources. From our previous work and theoretical calculations, [10,31] this colossal saturation magnetization at 10 K may be associated with the Co and Nd impurities, S vacancies as illustrated by XPS results in Figure 3, doping induce defects, such as cation and anion vacancies, interstitials, implantationinduced holes, terrace terminated structures as depicted by low A 1g /E 1 2g in Raman results (Figure 1d), and exchange coupling of dopants (Co and Nd). In addition, the giant magnetization may be linked to Nd-doping-induced polarization of Mo ions, as that Mn-doping-induced spin polarization of Ga ions in GaN [32] and uncompensated 4d and 5f electrons of Mo and Nd, respectively. Figure 5c,d exhibits that further increase in Nd atoms concentration (2 and 5 at%) leads to significant decrease in ferromagnetic signals to 1778 and 349.9 emu cm À3 at 5 K. However, at room temperature, in comparison to 1 at% Nd,4 at%Co-MoS 2 , the saturation magnetization increases to 80.3 and 72.2 emu cm À3 for 2 at% Nd,4 at%Co-MoS 2 and 5 at% Nd, 4 at% Co-MoS 2 , respectively. In addition, an evident coercivity of 700 Oe was also observed in 5 at% Nd, 4 at% Co-MoS 2 sample at 50 K as shown in inset of Figure 5d, which tends to decrease with the further increase in temperature to a value of 399.6 Oe at room temperature. Enlarged view of the hysteresis loop is displayed in the insets, indicating relatively low coercivity in all other samples. Moreover, in 5 at% Nd, 4 at% Co-MoS 2 sample, two magnetic phases could be observed, which may be associated with the high-doping-concentration-induced impurity phase.
To understand the ferromagnetism mechanism, annealing of as-prepared samples was performed at 500°C for 10 min in Ar environment to remove the ion-implantation-induced defects. Ferromagnetic signals are detected in all samples as illustrated in Figure 6a-d. In comparison to unannealed 4 at% Co-MoS 2 sample, hysteresis loops of annealed 4 at% Co-MoS 2 demonstrates an improvement in saturation magnetization with values of 1146.1 emu cm À3 at 5 K and 45.5 emu cm À3 at room temperature at the applied field of 50 kOe, as depicted in Figure 6a. A sharp decline in magnetic signal is observed in annealed 1 at% Nd, 4 at%Co-MoS 2 sample. As shown in Figure 6b, with the applied field of 50 kOe, M-H loops exhibit a saturation magnetization of 346.7 and 71.4 emu cm À3 at 5 and 300 K, respectively. However, Figure 6c shows that at 5 and 300 K, the magnetic signal of annealed 2 at% Nd, 4 at%Co-MoS 2 sample was significantly enhanced to 1432.9 and 71.5 emu cm À3 (at 50 kOe), respectively. The saturation magnetization value is comparable to metallic Co (%1500 emu cm À3 ). Further increase in Nd concentration (5 at% Nd) results in relatively decrease in saturation magnetization with a value of 999.2 and 40.5 emu cm À3 (at 50 kOe) at 5 K and room temperature, respectively, as displayed in Figure 6d, which may be attributed to the antiferromagnetic coupling of dopants at higher doping concentration. [33] All annealed samples show low coercivity as illustrated in inset of Figure 6a-d. Annealing is an effective way to remove dopinginduced lattice disorders and strain/stress. The ferromagnetism www.advancedsciencenews.com www.pss-rapid.com in annealed samples may be attributed to the annealing-induced S vacancies, impurities, and terrace-terminated structure. Magnetization from defects after annealing is now a known phenomenon and cannot be ignored. In our recent reports, WS 2 and MoS 2 nanoparticles have shown ferromagnetism when annealed at different temperatures. [9,34] It should be noted that in this situation, the major of the magnetization may also come from the MoS 2 crystals due to the formation of defects. For the 1 at% Nd-and 4%Co-codoped sample, the saturation magnetization was strongly reduced, which supports that the high magnetization is associated with defects, such as cation vacancies and holes. [10] But the enhanced magnetization after annealing may be complicated. As discussed previously, annealing-induced S vacancies or other possible edge defects may be associated with the magnetization. With codoping of Co and Nd, the annealing behavior may be different from that of pure 2D single crystals. [9,34] Figure 7a,b is the summary of the temperature dependence of saturation magnetization of as-prepared and annealed Co-doped and Nd/Co-codoped MoS 2 samples. Both of the samples show higher magnetization at low temperature, but drastic decline at higher temperature, suggesting existence of a large number of paramagnetic-like phases. It is a general phenomenon for diluted magnetic semiconductors.
Compared to as-fabricated sample, all the annealed samples show the similar trend in magnetization temperature dependence curves. The annealing may remove vacancies (i.e., cation vacancies), stress, and defects or induce edge state and vacancies (i.e., S vacancies), which alter the magnetization behaviors, as discussed in the previous paragraphs. However, we observed that the as-prepared 1% Nd, 4% Co-MoS 2 sample exhibits the highest magnetization at 5 K, suggesting that there are more defects in 4 at% Co-2 at% Nd-MoS 2 sample, which contribute to the high magnetization, but is removed after annealing.
The work has demonstrated that codoping of Nd and Co can induce a giant magnetization in TMDC-based layered materials. This phenomenon of introducing ferromagnetism in 2D-layered materials has been widely reported. However, such high magnetization is rarely reported. The recent reported magnetization and coercivity in 2D-layered materials doped with transition and/or rare-earth metal are summarized in Table 1. It is evident from the Table 1 that the current work has demonstrated the highest saturation magnetization. This work further indicates that codoping two different elements (transition and rare earth) may be an effective strategy to achieve high magnetization in the research of ferromagnetic 2D-layered materials, which provides a novel approach to develop 2D-layered magnet with desired properties.

Conclusion
We have codoped Co and Nd elements into MoS 2 single crystal by employing the ion-implantation technique. It was found that for Co-alone-doped MoS 2 , a prominent magnetic signal is observed at both room and low temperature. However, Nd and Co codoping can introduce a very high saturation magnetization (6916.3 emu cm À3 ) at 10 K. The high magnetization is ascribed to the Co and Nd impurities, cation and anion vacancies, doping-induced defects, interstitials, terrace-terminated structures, and exchange coupling of dopants (Co and Nd). In addition, the giant magnetization may be linked to Nd-dopinginduced polarization of Mo ions and uncompensated 4d/5f

Experimental Section
Single crystal molybdenum (VI) disulfide (MoS2) was used to explore the doping effect. The high purity (>99%) single crystal MoS2 substrates were purchased from Kejing Corp., Hefei, China. All samples of MoS 2 crystal were of 20-30 mm 2 size. Physical ion-implantation technique was utilized to obtain Co-and Nd-codoped MoS 2 . The process was performed by employing low-energy ion implanter facility at GNS Science, New Zealand. [41] Cobalt and neodymium were then doped into MoS 2 single crystal by ion implantation with 20 keV Co þ and 30 keV Nd þ , respectively. Monte Carlo DYNAMIC-TRIM [42] calculations of 20 keV Co þ implantation into MoS 2 for 3 Â 10 15 Co cm À2 yields peak concentration of %4 at% at an average depth of 13.2 nm. The 30 keV Nd fluences were varied between 1 Â 10 15 Nd cm À2 and 4.8 Â 10 15 Nd cm À2 to obtain peak Nd concentration of 1 and 5 at% at an average depth of 12.6 nm.
X-ray diffractometry (XRD, PANalytical Xpert Multipurpose X-ray Diffraction System) with Cu Kα radiation was used to identify the crystalline phases of pristine and doped MoS 2 samples. Bonding characteristics and phase crystallization were examined by Raman spectrometry using Renishaw inVia Raman microscope, fitted with a diffraction grating of 1800 lines mm À1 , excited with radiation of 514 nm argon ion laser and calibrated with Si single crystal. X-ray photoelectron spectroscopy (XPS) with Thermo Scientific ESCALAB 250i XPS (calibrated by C1s = 284.8 eV) was utilized to characterize the valance state and element composition. Dopant distribution and its depth profile in MoS 2 substrate were investigated by secondary-ion mass spectroscopy (SIMS) using Cameca IMS 5fE7 SIMS instrument operated with O 2 þ ion gun (8 nA ion current), an impact energy of 7.5 keV and rastered with 180 Â 180 um 2 region of the surface. Magnetic properties were determined by a superconducting quantum interference device (SQUID, Quantum design-XL-5).