Contrasting impurity-induced magnetism and dynamics in 2H-MoTe2

TL;DR

This work uses $eta$-detected NMR to study interstitial $^8$Li$^+$ in 2H-MoTe$_2$, finding no evidence for Li- or defect-induced magnetism, in contrast to muon-related signals reported previously. DFT identifies the Li stopping site at the 2a Wyckoff position in the van der Waals gap and reveals diffusion barriers along a path via the 6g site, with no induced spin polarization. Experimentally, a broad central line with faint quadrupolar satellites is observed, and a frequency-comb technique enables precise extraction of the quadrupolar frequency $ν_q$, which shows a thermally driven $T^{3/2}$-dependent reduction indicative of phonons. Spin-lattice relaxation reveals two dynamic channels and an activation energy around $0.2$ eV, consistent with local cage dynamics rather than long-range Li diffusion or magnetic ordering. Overall, the study demonstrates that impurity responses in semiconducting TMDs are highly impurity- and site-specific, and introduces a robust spectroscopic method for disentangling quadrupolar effects in complex line shapes.

Abstract

We investigate the behavior of interstitial $^8$Li$^+$ implanted near the surface of 2H-MoTe$_2$ using $β$-detected NMR. We find that, unlike the muon, $^8$Li$^+$ does not show any signature of induced magnetism. This result is consistent with density functional theory, which identifies the Li stopping site at the 2a Wyckoff position in the van der Waals gap and confirms the absence of detectable Li-induced electronic spin polarization. Both the spin-lattice relaxation and the resonance lines show evidence of strong spin dynamics above $\sim 200$ K, reminiscent of local stochastic $^8$Li$^+$ motion within a cage. The resonance line shape consists of quadrupolar satellites on top of a broad central peak. To better understand the interaction of $^8$Li$^+$ with the host material, we employ a frequency-comb measurement, by simultaneously exciting four frequencies corresponding to the first-order quadrupolar satellite transitions, $ν_0 \pm 3ν_{\mathrm{comb}}$ and $ν_0 \pmν_{\mathrm{comb}}$ around the Larmor frequency $ν_0$ as a function of $ν_{\mathrm{comb}}$. This offers an enhanced sensitivity to the quadrupolar split portion of the line. Using this method, we find a small decrease of the quadrupolar frequency with increasing temperature, showing the typical behavior associated with thermally excited phonons and the absence of any magnetic response which was observed with other defects in 2H-MoTe$_2$.

Figures (4)

• Figure 1: (a) Simulated stopping profile of $22.5keV$$^8$Li$^+$ ions in 2H-MoTe$_2$. (b) Stopping site and diffusion paths (c) Diffusion energy barriers. The high symmetry sites along the $^8$Li$^+$ diffusion path have been labeled by their Wyckoff letters of space group 194. (d) $^8$Li$^+$ spin-DOS showing absence of induced magnetization.
• Figure 2: (a) Typical $^8$Li $\beta$-NMR line shapes for 2H-MoTe$_2$ in a field of 6.55T $\parallel c$-axis. The spectra have been vertically offset for clarity. The solid lines show fits to the data and the vertical dashed line the position of the MgO reference resonance. (b)$^8$Li $\beta$-NMR spectra 2H-MoTe$_2$ at different temperatures in a field of 6.55T as a function of $\nu_{\rm{comb}}$ when using a frequency-comb to excite multiple quadrupolar transitions simultaneously, as described in the text with $\nu_0$ determined by fitting the line shapes in (a).
• Figure 3: Temperature dependence of the (a) quadrupolar frequency, (b) relative amplitude of the quadrupolar peaks, (c) resonance frequency, and (d) width of the central line. The dashed line in (a) shows a $T^{3/2}$ power law behavior, and the dashed line in (c) shows the reference frequency of MgO.
• Figure 4: (a) Typical $^8$Li SLR data measured in 2H-MoTe$_2$. The $^8$Li beam is implanted for 4s and then turned off (indicated by a gray shaded region). The solid lines show fits to the data. The temperature dependence of the slowly relaxing fraction (b), the fast SLR ($1/T_{1{\text{,f}}}$) (c) and slow SLR ($1/T_{1{\text{,s}}}$) (d). The solid line shows a fit with an activation energy of $\sim0.2eV$. For comparison, we also show in (b) also the relative amplitude of the quadrupolar peaks, $f_{\mathrm{quad}}$, from Fig. \ref{['fig:lineparams']}(b).