Open Access
Add to BibSonomy
Abstract
Time-resolved Kerr rotation microscopy is used to generate and measure spin valley polarization in MOCVD-grown monolayer tungsten diselenide (WSe2). The Kerr signal reveals bi-exponential decay with time constants of 100 ps and 3 ns. Measurements are performed on several triangular flakes from the same growth cycle and reveal larger spin valley polarization near the edges of the flakes. This spatial dependence is observed across multiple WSe2 flakes in the Kerr rotation measurements but not in the spatially resolved reflectivity or microphotoluminescence data. Time-resolved pump-probe overlap measurements further reveal that the Kerr signal’s spatial dependence is not due to spin diffusion on the nanosecond timescale.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Transition metal dichalcogenide (TMD) monolayers have been the focus of intense study over the past decade. While bulk TMDs have an indirect bandgap making them less suitable for optical measurements, when these materials are thinned down to a single van der Waals layer, the band structure transitions to a direct bandgap at the K point [1,2]. Another interesting property of TMD monolayers is that spatial inversion symmetry breaking combined with strong spin-orbit coupling leads to valley-dependent optical selection rules and an opposite spin-orbit splitting between K and K’ valleys [3,4]. Thus, the manipulation of optical orientation can be used to create a spin valley polarization. These spin valley polarizations have been shown to have potentially long lifetimes, on the order of nanoseconds [5–10]. These properties, in addition to their two-dimensional nature, make TMDs attractive materials for the formation of spintronic heterostructures [11–13]. While encapsulated and/or exfoliated monolayers of these materials can have more pristine optical properties and exhibit longer spin valley lifetimes [13–14], monolayers grown via processes such as MOCVD are important to understand due to the scalability they offer compared to that of mechanical exfoliation [15].
In this paper, we utilize time-resolved Kerr microscopy (TRKM) with micron scale spatial resolution to investigate the spatial dependence of spin valley polarization in monolayer MOCVD-grown WSe2 and observe a non-uniform Kerr rotation amplitude. Multiple triangular flakes are examined to determine the repeatability of the observed pattern. Spatially resolved microphotoluminescence (micro PL) and reflectivity measurements are then used to investigate the origin of the nontrivial spatial pattern in the WSe2 flakes.
Lastly, a 4-f setup is used to steer the position of the pump laser beam on the sample relative to the probe beam. This allows pump-probe spatial overlap scans to be performed at multiple time delays to determine what role, if any, diffusion plays on the time scale of the spin valley lifetime.
2. Methods
The WSe2 monolayer flakes are grown using metalorganic chemical vapor deposition (CVD) on Si/SiO2 substrates. As seen in Fig. 1(a), a typical grain has a triangular or six-sided star-shaped pattern and ranges from 5 to 10 microns across. The samples are then mounted in a closed-cycle helium cryostat. For the remainder of this paper, measurements are conducted with the sample cooled to 10 K unless noted otherwise.
Fig. 1. (a) Optical image of multiple flakes on the sample used in this research. (b) TRKM data (points) with fit for bi-exponential decay (solid line). Negative delay data represents a 12-13 ns delay after the previous pump pulse due to the 13.16 ns repetition rate of the laser. (c) Photoluminescence spectrum taken over 100s with the sample held at 10K (black). The pump used is a 533nm diode laser. No excitonic side-peaks are seen in this sample. Also shown is the amplitude of long-lived spin valley signal extracted from TRKM measurements as a function of the pump/probe laser wavelength (blue). Finally, a white light absorption spectrum is shown with a peak near 730 nm (red). This was calculated by taking two spectra of the white light source, one reflected from a flake and the other from the substrate. The on-flake spectrum was then subtracted from and normalized by the off-flake. Artifacts in the PL spectrum at 805 and 828 nm were removed.
The TRKM measurements are performed using a mode-locked Ti: Sapphire laser with a repetition rate of 76 MHz. This laser is split into two beams, a pump path modulated between left- and right-circular polarization at 50 kHz by a photo-elastic modulator and a probe path modulated by an optical chopper (789 Hz). The relative time delay between the two beams is adjusted by a mechanical delay line in the pump path before they are recombined and directed through a 100x microscope objective to a ∼2 μm spot on the sample (determined through spatially-resolved reflectivity measurements on a calibration sample). Neutral density filters are used to reduce laser powers to 0.6 mW for the pump and 0.1 mW for the probe upon reaching the objective. Valley polarization is generated by the circularly-polarized pump pulses [4,16] and measured through the Kerr rotation of the linearly-polarized probe pulses. This rotation is measured by directing the reflected beam to a photodiode bridge through a Wollaston prism, separating the horizontal and vertical components and taking their difference. The largest Kerr rotation angles measured on the samples shown in this paper are approximately 5 microradians. The pump and probe beams are each balanced previous to measurements using half-wave plates for the most efficient lock-in detection. A representative TRKM plot with a bi-exponential fit can be seen in Fig. 1(b). In addition to the time delay, the position of the pump laser on the sample can be scanned relative to the probe spot, allowing pump-probe overlap scans to be performed. This, combined with changing the time delay allows us to view any changes that the shape and/or position of the created spin valley packet may undergo over time. Kerr microscopy measurements are initially taken at a range of wavelengths to determine the optimal wavelength for continued study (Fig. 1(c)) and temperature-dependent measurements show a substantial decrease in lifetime as temperature rises, in agreement with previous studies [5–7,17].
In addition to the wavelength dependence of the Kerr signal, the photoluminescence spectrum and white light absorption measurements are also shown in Fig. 1(c). The micro PL spectrum shows a broad peak centered around 770 nm with none of the excitonic side peaks often seen in TMDs [6,7,17]. This may be caused by differences in the MOCVD growth technique which can cause the photoluminescence to be mediated by localized defects [18]. Due to this, it should be noted that the PL and TRKM peaks are offset by 60 meV in these samples, with the Kerr signal disappearing completely more than 15 meV (∼7 nm) away from the Kerr rotation peak at 742.5 nm. In contrast to the PL spectrum, however, white light absorption measurements show a broad feature with a peak near 730 nm that the Kerr signal falls within. This absorption peak/Kerr signal is likely caused by the A exciton resonance which has been shown near these wavelengths [19].
3. Results
We report data using multiple flakes from a single WSe2 growth. Following from Fig. 1(c), measurements are taken with pump and probe beams at 742.5 nm. Using stepping piezoelectric stages located within the cryostat, TRKM measurements are repeated at multiple points traveling across the first flake, Flake A, which is approx. 10 μm in diameter (Fig. 2(a)). The individual scans show a consistent bi-exponential decay with an initial fast decay of time constant t1 ∼100ps and a longer decay with t2 ∼3ns. Previous studies have also shown multi-exponential decays providing the best fitting for TRKR measurements on TMDs [5–7]. Figure 1(b) shows one such scan overlaid with the best fit bi-exponential curve.
Fig. 2. (a) Kerr rotation measured as a function of position across the flake and pump-probe delay time. Contour lines are included as a guide to the eye. (b) Spatial dependence of polarization lifetimes extracted from bi-exponential fits across the first 10 μm WSe2 flake. Lifetime of shorter-lived spin signal is multiplied by a factor of 20 for comparison. (c) Spatial dependence of the amplitudes of the shorter lifetime (red) and longer lifetime (black) decays. (d), (e) Spatial dependence of the amplitudes of the shorter (red) and longer (black) lifetime decays across another flake of similar 10 μm size, Flake B, (d) and a smaller, 7 μm flake, Flake C (e).
Looking closer at Fig. 2(a), we see a nontrivial spatial dependence of the TRKM signal while scanning across Flake A. The amplitude of the Kerr signal has a noticeable drop in the center of the flake at later time delays. A similar nontrivial spatial dependence has been observed in similar monolayer TMDs, WS2[6] and MoS2[19]. To further investigate this phenomenon in our samples, we plot the fitting parameters t1 and t2, as well as the amplitudes of the two exponential decays, A1 and A2, respectively (Figs. 2(b) and 2(c)). Two contrasting features can be seen from this. First, there is no noticeable spatial dependence of the decay time across this flake; secondly, the amplitudes have a peak near the flake’s edges and a dip near the middle. Continued measurements show this drop to be small but repeatable across this flake.
To ensure this is a consistent phenomenon and is not unique to a single flake, measurements are repeated on two additional flakes: a second flake of similar size (Flake B) and a third with a smaller width of ∼7 μm (Flake C). The results of these measurements can be seen in Figs. 2(d) and 2(e), respectively. In Flake B, we once again see an amplitude drop in the center of the flake, more noticeably for the longer lifetime than the shorter lifetime decay. However, the smaller Flake C does not show this same few micron dip in its spin valley amplitudes. This is possibly due to Flake C not being large enough for the offending area to exist at its center.
There are several possible effects that could cause the nontrivial spatial variation of the Kerr signal that we observe. The effects that we will focus on here are the potential spatial inhomogeneity of the optical properties of the flake, which could change the efficiency of optical spin pumping and/or detection, and diffusion of the optically-pumped spin valley polarization. Previous studies have shown a strong spatial inhomogeneity of the PL in other TMD samples [20,21], with patterns similar to what is seen in our spatially-resolved Kerr rotation data. This would provide evidence that the flakes have stronger optical coupling on the edges than in the center. The flakes studied here exhibit a six-sided star pattern, suggesting the existence of grain boundaries and defects on the edges of the flakes. However, if these defects played a large role in affecting the spin valley polarization, we would also expect to observe a change in decay time as the beam travels across the flake. The absence of spatial variation in the decay time shows that there is no inhomogeneity of the dominant spin relaxation mechanisms on the micron scale. Similarly, a measurement of the reflectivity at the pump-probe wavelength could provide a simple explanation for the spatially-dependent signal strength. Lastly, spin valley polarization diffusion away from our optical spot occurring on a comparable nanosecond time scale would cause a reduction in signal and thus the fit lifetime. Because an exponential fit is being performed on the data, any change in signal over time will be accounted for by only the amplitude and lifetime parameters. This includes spins that may leave the measurement area through diffusion. This would be more noticeable in the center of the flake, since diffusion near the edges could reflect from the boundary of the flake and thus keep a greater number of spins within the area that the probe beam measures.
First, we perform spatially-resolved photoluminescence measurements to compare with the Kerr data. Once again, Flake A is stepped through the beam path to obtain spatially resolved data on individual flakes, this time using a 533 nm pump laser for PL excitation. Figure 3(a) shows the integrated peak normalized by the background counts for the PL spectra across the first flake measured, and we note that all three flakes showed similar results from the photoluminescence measurements. Examining our spatial PL data next to the TRKM magnitude data shown for the same flake in Fig. 3(b), we see that, in contrast to the variation observed across the flake in the Kerr amplitude measurement, the PL signal reaches a maximum once the beam is fully on the flake then remain constant until moving off the flake. We note that the PL amplitude changes more gradually near the flake edges compared to the Kerr amplitude. Whereas the Kerr amplitude is measured using a pump-probe technique and depends on the spatial overlap of the pump and probe beams, photoluminescence is generated from a larger volume of the sample, based on where the excitation light is absorbed and carrier diffusion. We utilize a 50 μm pinhole between a collimated lens pair in the collection path to improve the spatial resolution. The reflectivity data in Fig. 3(c) also shows a similar homogeneity across the center of the flake. The reflectivity data was found by calculating the sum of the two diodes of the photodiode bridge, as compared to the difference which supplied the Kerr rotation. The PL and reflectivity data both indicate that the spatial dependence of the Kerr magnitude is not due to some change in the optical properties between the different areas of a flake.
Fig. 3. (a) Spatial dependence of the integrated intensity over the PL peak from 760 to 780 nm normalized by off-peak background signal. (b) Spatial dependence of the long-lived spin signal’s amplitude across the first flake. (c) Reflectivity data taken from a piezo scan across the same flake, calculated using the total probe intensity reflected from the sample and directed to the diode bridge.
Lastly, the 4-f setup shown in Fig. 4(a) is used to perform pump-probe overlap scans to determine whether spin valley diffusion is occurring on the same time scale as the Kerr signal. This configuration utilizes a pair of planes that are four focal lengths apart where any light directed through the plane and into the first lens will exit the final plane at a position and angle reflected across the shared optical axis. By placing a motorized steering mirror which the pump beam strikes on the optical axis at the first plane and the microscope objective that focuses light on our sample at the final plane, the pump beam can be directed to enter the objective across a wide range of angles without any change in the laser power incident on the sample that would otherwise be caused by clipping on the objective. Once through the objective, this range of angles translates to a range of positions on the sample that are then used to perform pump-probe spatial overlap scans by scanning the pump spot across the probe which has been centered on a flake. These scans are then performed at a variety of time delays and fit using a Gaussian (Fig. 4(b)). The fitting of these scans provides a measure of the width of the pump-polarized spin valley packet over time (Fig. 4(c)). As shown, there is no significant diffusion of the spin valley signal on the nanosecond time scale. Thus, it is determined that diffusion of spins away from the probe laser spot has negligible effect on the magnitude of the spin signal seen in the delay scans.
Fig. 4. (a) Diagram of the 4-f setup necessary to perform pump-probe overlap scans on these length scales. The light crossing the focal planes at the steering mirror and microscope objective is identical (up to a reflection across the optical axis), allowing the pump beam to be steered in a wide range of angles without sacrificing any coupling of the beam into microscope objective, thus avoiding a change in pump power when scanning off-axis. (b) Two pump-probe overlap scans with the time delay fixed at 2 and 12 ns, with Gaussian fits to determine the width of the spin valley packet as a function of delay time. (c) Gaussian widths of the best-fit curves to multiple overlap scans such as those shown in (b) plotted as a function of delay time.
4. Conclusion
We have experimentally shown a nontrivial spatial dependence of the spin valley polarization in MOCVD-grown monolayer WSe2, where, in large enough flakes, the excited polarization has increased magnitude near the edges. This occurs for both components of the bi-exponential fit with decay time constants of 3 ns and 80 ps. Interestingly, this is independent from the lifetime of the polarization which remains relatively constant across each flake. Spatially resolved microphotoluminescence and reflectivity measurements revealed constant behavior across each flake, implying that this does not result from inhomogeneity in the optical properties of WSe2 near the edges of flakes. Utilizing the 4-f setup, pump-probe overlap scans were also employed at a range of delay times within the repetition rate of the laser and showed that there was no significant diffusion of the polarization over the time scales investigated in this study, eliminating diffusion as the cause of the increased spin valley polarization near the edges. These findings show that time-resolved Kerr rotation microscopy can be a useful tool for characterizing inhomogeneity of spin valley polarization in monolayer WSe2 and other TMDs.
Funding
Division of Materials Science and Engineering, Basic Energy Sciences, U.S. Department of Energy (DE-SC0016206).
Acknowledgments
The authors would like to thank Fauzia Mujid and Jiwoong Park for providing the MOCVD-grown TMD samples.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. K. F. Mak, C. Lee, J. Hone, J. Shan, and T. F. Heinz, “Atomically Thin MoS2: A New Direct-Gap Semiconductor,” Phys. Rev. Lett. 105(13), 136805 (2010). [CrossRef]
2. A. Splendiani, L. Sun, Y. Zhang, J. Kim, C. Y. Chim, G. Galli, and F. Wang, “Emerging Photoluminescence in Monolayer MoS2,” Nano Lett. 10(4), 1271–1275 (2010). [CrossRef]
3. D. Xiao, G. B. Liu, W. Feng, X. Xu, and W. Yao, “Coupled Spin and Valley Physics in Monolayers of MoS2 and Other Group-VI Dichalcogenides,” Phys. Rev. Lett. 108(19), 196802 (2012). [CrossRef]
4. T. Cao, G. Wang, W. Han, H. Ye, C. Zhu, J. Shi, Q. Niu, P. Tan, E. Wang, B. Liu, and J. Feng, “Valley-selective circular dichroism of monolayer molybdenum disulphide,” Nat. Commun. 3(1), 887 (2012). [CrossRef]
5. X. Song, S. Xie, K. Kang, J. Park, and V. Sih, “Long-Lived Hole Spin/Valley Polarization Probed by Kerr Rotation in Monolayer WSe2,” Nano Lett. 16(8), 5010–5014 (2016). [CrossRef]
6. E. J. McCormick, M. J. Newburger, Y. K. Luo, K. M. McCreary, S. Singh, I. B. Martin, E. J. Cichewicz Jr, B. T. Jonker, and R. K. Kawakami, “Imaging spin dynamics in monolayer WS2 by time-resolved Kerr rotation microscopy,” 2D Mater. 5(1), 011010 (2017). [CrossRef]
7. F. Volmer, S. Pissinger, M. Ersfeld, S. Kuhlen, C. Stampfer, and B. Beschoten, “Intervalley dark trion states with spin lifetimes of 150 ns in WSe2,” Phys. Rev. B 95(23), 235408 (2017). [CrossRef]
8. C. Jin, J. Kim, K. Wu, B. Chen, E. S. Barnard, J. Suh, Z. Shi, S. G. Drapcho, J. Wu, P. J. Schuck, S. Tongay, and F. Wang, “On Optical Dipole Moment and Radiative Recombination Lifetime of Excitons in WSe2,” Adv. Funct. Mater. 27(19), 1601741 (2017). [CrossRef]
9. L. Yang, N. A. Sinitsyn, W. Chen, J. Yuan, J. Zhang, J. Lou, and S. A. Crooker, “Long-lived nanosecond spin relaxation and spin coherence of electrons in monolayer MoS2 and WS2,” Nat. Phys. 11(10), 830–834 (2015). [CrossRef]
10. P. Dey, L. Yang, C. Robert, G. Wang, B. Urbaszek, X. Marie, and S. A. Crooker, “Gate-Controlled Spin-Valley Locking of Resident Carriers in WSe2 Monolayers,” Phys. Rev. Lett. 119(13), 137401 (2017). [CrossRef]
11. P. Rivera, K. L. Seyler, H. Yu, J. R. Schaibley, J. Yan, D. G. Mandrus, W. Yao, and X. Xu, “Valley-polarized exciton dynamics in a 2D semiconductor heterostructures,” Science 351(6274), 688–691 (2016). [CrossRef]
12. M. Y. Li, C. H. Chen, Y. Shi, and L. J. Li, “Heterostructures based on two-dimensional layered materials and their potential applications,” Mater. Today 19(6), 322–335 (2016). [CrossRef]
13. J. Kim, C. Jin, B. Chen, H. Cai, T. Zhao, P. Lee, S. Kahn, K. Watanabe, T. Taniguchi, S. Tongay, M. F. Crommie, and F. Wang, “Observation of ultralong valley lifetime in WSe2/MoS2 heterostructures,” Sci. Adv. 3(7), e1700518 (2017). [CrossRef]
14. J. Li, M. Goryca, K. Yumigeta, H. Li, S. Tongay, and S. A. Crooker, “Valley relaxation of resident electrons and holes in a monolayer semiconductor: Dependence on carrier density and the role of substrate-induced disorder,” Phys. Rev. Mater. 5(4), 044001 (2021). [CrossRef]
15. K. Kang, S. Xie, L. Huang, Y. Han, P. Y. Huang, K. F. Mak, C. J. Kim, D. Muller, and J. Park, “High-mobility three-atom-thick semiconducting films with wafer-scale homogeneity,” Nature 520(7549), 656–660 (2015). [CrossRef]
16. K. F. Mak, K. He, J. Shan, and T. F. Heinz, “Control of valley polarization in monolayer MoS2 by optical helicity,” Nat. Nanotechnol. 7(8), 494–498 (2012). [CrossRef]
17. T. Yan, J. Ye, X. Qiao, P. Tan, and X. Zhang, “Exciton valley dynamics in monolayer WSe2 probed by the two-color ultrafast Kerr rotation,” Phys. Chem. Chem. Phys. 19(4), 3176–3181 (2017). [CrossRef]
18. Z. Wu, W. Zhao, J. Jiang, T. Zhen, Y. You, J. Lu, and Z. Ni, “Defect Activated Photoluminescence in WSe2 Monolayer,” J. Phys. Chem. C 121(22), 12294–12299 (2017). [CrossRef]
19. J. Huang, T. B. Hoang, T. Ming, J. Kong, and M. H. Mikkelsen, “Temporal and spatial valley dynamics in two-dimensional semiconductors probed via Kerr rotation,” Phys. Rev. B 95(7), 075428 (2017). [CrossRef]
20. H. R. Gutiérrez, N. Perea-López, A. L. Elías, A. Berkdemir, B. Wang, R. Lv, F. López-Urías, V. H. Crespi, H. Terrones, and M. Terrones, “Extraordinary Room-Temperature Photoluminescence in Triangular WS2 Monolayers,” Nano Lett. 13(8), 3447–3454 (2013). [CrossRef]
21. W. Bao, N. J. Borys, C. Ko, J. Suh, W. Fan, A. Thron, Y. Zhang, A. Buyanin, J. Zhang, S. Cabrini, P. D. Ashby, A. Weber-Bargioni, S. Tongay, S. Aloni, D. F. Ogletree, J. Wu, M. B. Salmeron, and P. J. Schuck, “Visualizing nanoscale excitonic relaxation properties of disordered edges and grain boundaries in monolayer molybdenum disulfide,” Nat. Commun. 6(1), 7993 (2015). [CrossRef]
