Imaging Spin Dynamics in Monolayer WS2 by Time-Resolved Kerr Rotation Microscopy Elizabeth J. Bushong,1 Yunqiu (Kelly) Luo,1 Kathleen M. McCreary,2 Michael J. Newburger,1 Simranjeet Singh,1 Berend T. Jonker,2 Roland K. Kawakami1* 1 2
Department of Physics, The Ohio State University, Columbus OH 43210, USA Naval Research Laboratory, Washington DC 20375, USA
*e-mail:
[email protected] Transition metal dichalcogenides (TMDs) are of great interest due to their unique band structure with large valley-dependent spin-orbit splittings, Berry curvature, and spin/valley-selective optical selection rules.1-10 While these properties make TMDs extremely attractive for spintronics, utilization of these properties in nanoscale devices demands a fundamental understanding as obtained, for example, through imaging of spin dynamics with high spatial resolution. Here, we report the first spatially-resolved images of spin dynamics in monolayer WS2 using time resolved Kerr rotation11 (TRKR) microscopy with ~1 micron resolution. We discover a complex spatial dependence of spin density varying on the micron length scale, with spin lifetimes exceeding 5 ns. Comparing micro-photoluminescence and TRKR microscopy reveals an unexpected anticorrelation between strong A exciton luminescence and high spin density, which provides new insights on the transfer of spin angular momentum from short-lived excitons to long-lived spin states of resident conduction electrons.12-14 We also find that the spin lifetime in WS2 is robust against external magnetic fields, in contrast to MoS2,12 which confirms predictions that larger spin-orbit coupling will stabilize spins against relaxation, contrary to conventional materials.4,7,12,15-17 These results demonstrate high resolution imaging of spin dynamics as a powerful technique for investigating spindependent physics in 2D materials. The band structure of monolayer WS2 is characterized by a direct gap and large spin splittings with opposite polarities in the +K and –K valleys (Figure 1a).6,7,18-20 This produces spin-
1
valley coupling that is predicted to suppress spin-flip scattering and spin dephasing, resulting in long spin and valley lifetimes.4,7,12,15-17 Because the helicity of light couples to the spin and valley polarization via optical selection rules,4,5 optical probes have been utilized to investigate the spin
and
valley
dynamics
in
TMDs.
Early
studies
utilizing
polarization-resolved
photoluminescence spectroscopy found high retention of circular polarization in MoS2 monolayers (up to 99%), suggesting valley lifetimes exceeding 1 ns.6-9 While initial timeresolved optical techniques observed short lifetimes for valley polarization of excitons (tens of ps or less),21-27 more recent TRKR measurements observed nanosecond spin lifetimes of resident conduction electrons in n-type monolayer MoS2 and WS2, and resident holes in p-type monolayer WSe2.12-14 To investigate the origin and character of the long lived spins, we utilize TRKR microscopy to directly image the spin dynamics in a monolayer TMD with unprecedented spatial resolution (~1 micron) and ~150 fs temporal resolution (see Methods). Experiments are performed on high quality monolayer WS2 grown by chemical vapor deposition (CVD) on SiO2/Si substrates.28 As shown in Figure 1b, a given WS2 sample has small isolated triangles, typically 10-40 microns in size, which are believed to be single-crystalline. We verify the quality and monolayer nature of the WS2 using photoluminescence (PL) spectroscopy. The PL spectrum measured at 6 K (Figure 1c) shows a strong peak at 630 nm corresponding to the A exciton, while the absence of indirect gap PL emission at longer wavelengths, which is present in bilayer and bulk samples, confirms the films are monolayer.2,3 Transport measurements indicate that the as-grown material is n-type. Spin dynamics in monolayer WS2 are investigated using TRKR microscopy, depicted in Figure 1d. The sample is held at 6 K in a low vibration, closed-cycle optical cryostat. Ultrafast pulses from an optical parametric oscillator (~150 fs, 625 nm, 76 MHz) are split into pump and probe pulses, each of which is focused onto the sample with ~1 μm spot size. The circularlypolarized pump pulse creates valley-polarized excitons, each consisting of a spin polarized electron and hole, as shown in the band diagram (Figure 1a). The time-delayed, linearly2
polarized probe pulse measures the combined spin and valley polarization through the Kerr rotation of its linear polarization axis. Figure 1e shows the Kerr rotation as a function of time delay between the pump and probe pulses. As shown in the inset of Figure 1e, there is an initial rapid exponential decay of 3 ps (curve fit in green), which is due to the loss of valley polarization of excitons, consistent with previous TRKR studies.22,25,26 However, a substantial Kerr rotation remains beyond the initial decay and persists beyond several nanoseconds. Excitons could not produce this signal because they recombine within the first few hundred picoseconds,21,24,29 so the long lifetime must therefore originate from the spin and/or valley polarization of resident conduction electrons of the n-type material. For the conduction electrons, the Kerr rotation is due to spin and valley polarization and is given, to lowest order, by ↑,
−
↓,
+
↑,
−
↓,
is the spin imbalance and
=
↑,
+
↓,
~ −
+ ↑,
, where −
↓,
=
is the
valley imbalance of the electron density. We find that the Kerr rotation is well described by a biexponential decay with time constants of τshort = 320 ps and τlong= 5.4 ns (the curve fit is the solid red line in Figure 1e). Such bi-exponential behavior agrees, for example, with a spin relaxation model based on fast intervalley scattering in monolayer TMDs.12 In this model, which we discuss again later, fast intervalley scattering between the K and –K valleys produces Dyakonov-Perel-like spin relaxation (i.e. spin dephasing by a fluctuating spin-orbit field). By scanning the overlapped pump and probe spots relative to the sample at a fixed time delay, a spatial image of the spin density is obtained. By repeating this mapping for different time delays, the microscopic evolution of spin dynamics can be directly revealed. Figure 2 shows a series of Kerr rotation images taken on a triangular island of monolayer WS2 at different time delays. The sequence of images illustrates a previously undiscovered complex spatial dependence of the spin density on the WS2 island. Areas with a large spin density (colored in yellow/orange/red) are separated by only a few microns from regions with almost no spin density (colored blue/black). The spatial distribution is striking, with a central core of low spin density surrounded by regions of higher spin density. It is worthwhile to note that even at 3
11 ns, there is still a measurable spin density. Interestingly, the images at longer time delays appear to show the spin density evolving to a more symmetric distribution within the triangle. To gain further insight into the spatial dependence of the Kerr rotation, we investigate its relationship with the photoluminescence (PL) spectra. Figures 3b and 3c show a series of TRKR delay scans and PL spectra taken along a linecut within the triangular island, as indicated in Figure 3a. The most notable trend is that in the region with strongest PL, the TRKR has a fast decay (
