Physics in Quasi-2D Materials for Spintronics Applications Topological Insulators and Graphene Ching-Tzu Chen IBM TJ Watson Research Center May 13, 2016 2016 C-SPIN Topological Spintronics Device Workshop 2016 IBM Corporation
Spintronics Electrical Properties (e.g., I, V, Q) Magnetic Properties (e.g., M, spin) Example: Spin-transfer torque MRAM (magnetoresistive random access memory) Magnetic tunnel junction Write spin-transfer torque Read tunneling magnetoresistance Building blocks: Spin generation, modulation/control, detection, transport/conduction, amplification, etc. 2
Spintronics in Quasi-2D Materials A. Spin-orbit coupling for spin generation Charge-spin conversion in topological insulators and spin-hall metals Luqiao Liu (IBM), Anthony Richardella (PSU), Ion Garate (Sherbrooke), Nitin Samarth (PSU), Yu Zhu, Jonathan Sun (IBM) [1] L. Liu, et al., Phys. Rev. B 91, 235437 (2015). [2] L. Liu, C.-T. Chen and J. Z. Sun, Nature Phys. 10, 561 (2014). B. Exchange coupling for spin modulation Strong interfacial exchange field in graphene/magneticinsulator heterostructures Peng Wei (MIT), Sunwoo Lee (Columbia), Florian Lemaitre, Lucas Pinel, Davide Cutaia, Yu Zhu (IBM), Wujoon Cha, Jim Hone (Columbia), Don Heiman (Northeastern), Ferhat Katmis, Jagadeesh Moodera (MIT) [3] P. Wei, et al., Nature Mat. (2016), doi:10.1038/nmat4603
Spin-Orbit Coupling for Spin Generation FM SH Metal Boost spin current generation efficiency: Hoffmann IEEE Trans Mag (2013) Isolate spin generation from charge current, bypassing MTJ breakdown limit Magnetic moment manipulation for in-plane moment, assume θ SH ~ 50%, I c,mtj STT I c,she ~ l t ~5 (junction size/sh metal thickness) Not yet obvious how much benefit for perpendicular moment 4
Charge-Spin Conversion in TI: Spin-Polarized Tunneling Topological surface states spinmomentum locking: * Quantify charge/spin conversion electrically * Energy dependence * Temperature dependence * Verify: symmetry * Verify: surface state vs. bulk state Method: 4-terminal spin-polarized tunneling technique * Tunneling (Inverse Edelstein effect) * Potentiometry (Edelstein effect) * Allow self-consistency check (Onsager reciprocity relationship) * Eliminate current shunting * Isolate TI from FM (CoFeB) influence 5
Charge-Spin Conversion in TI: Other Methods Potentiometry measurements: * Li, Jonker, et al., Nature Nano (2013) * Tang, KL Wang et al., Nano Lett (2014) * JS Lee, Samarth et al., PRB (2015) * Tian, YP Chen et al., Sci Rep (2015) Spin-torque FMR: * Mellnik, Ralph et al., Nature (2014) * Y. Wang, H. Yang et al., PRL (2015) Spin pumping: * Shiomi et al., Saitoh et al., PRL (2014) * Jamali, JP Wang et al., Nano Lett (2015) Spin-torque switching: * Fan, KL Wang et al., Nature Mat (2014), * Fan, KL Wang et al., Nature Nano (2016) Nature 511, 449 (2014) Ralph group 6
Spin-Polarized Tunneling in Bi 2 Se 3 : Zero Bias Tunneling configuration 7
Potentiometry Measurement in Bi 2 Se 3 : Zero Bias Potentiometry configuration (Edelstein effect) Onsager relation θ SH ~ 0.8 assuming λ sf ~ 1nm 8
Spin-Polarized Tunneling Data: Pt & Ta θ SH Pt = 0.04 0.09 θ SH Ta = 0.05 0.11 5d 8 5d 3 Liu, Chen, & Sun, Nature Phys. 10, 561 (2014) 9
Spin-Polarized Tunneling in Bi 2 Se 3 : Zero Bias Tunneling configuration θ SH ~ 0.8 assuming λ sf ~ 1nm 10
Surface State vs. Bulk State Contribution in Bi 2 Se 3 Bulk SHE: realistic bandstructure (credit: Flatte, Sahin) Fermi level Experiment: σ Bi2 Se 3 (SHC) ~(0.40 1.37) 10 3 (Ω cm) 1 assuming λ sf = 1 10 nm 1-2 orders of magnitude larger than theoretical bulk SHE value 11
Spin-Polarized Tunneling in Bi 2 Se 3 : Finite Bias Energy dependence V DC TI Liu et al., Phys. Rev. B 91, 235437 (2015) 12
Optimizing Charge-Spin Conversion via Surface State Bi 2 Se 3 (Bi 0.5 Sb 0.5 ) 2 Te 3 R H (Bi 2 Se 3 ) ~ 60 mω vs. R H ((Bi 0.5 Sb 0.5 ) 2 Te 3 ) ~ 9 Ω dv di ηp TIP J R l w η (Bi 2 Se 3 ) ~ 0.01 0.1 vs. η ((Bi 0.5 Sb 0.5 ) 2 Te 3 ) ~ (0.6 ± 0.2) dv θ SHP J ρ λ sf tanh(t/2λ di w t sf) Clearly surface-state spinmomentum locking effect θ SH (Bi 2 Se 3 ) ~ 0.8 vs. θ SH ((Bi 0.5 Sb 0.5 ) 2 Te 3 ) ~ 20 ± 5! 13
Summary A Spin-polarized tunneling study on Bi 2 Se 3 and (Bi 0.5 Sb 0.5 ) 2 Te 3 * Record-high charge-spin conversion observed in TI * Surface-state origin: spin-momentum locking * energy dependence information * RT promising Pt Ta Bi 2 Se 3 (Bi,Sb) 2 Te 3 Liu et al., Phys. Rev. B 91, 235437 (2015) Liu, Chen, & Sun, Nat. Phys. 10, 561 (2014) 14
Potential Applications Spin-orbit-torque MRAM and spin logic using TI? Q: Is technologically relevant PMA/TI practical for MRAM? SH Metal FM 1. STT in PMA-MTJ: overcome damping torque m m Beff α CoFeB ~ 0.4% 2. SOT in SH-PMA bilayer: overcome anisotropy torque large! B eff ( d )sin B B eff B ani ani cos sin d cos Hoffmann IEEE Trans Mag (2013) 15
Graphene Spintronics & Exchange Field Spin transport: small spin-orbit coupling, long spin relaxation length ( m) Spin generation: spin injection and Zeeman spin-hall effect V s V g V d graphene dielectric 2D: classical and quantum effects (e.g. QHE, QSHE, QAHE) 2D: spin control by Rashba or Exchange Field (10 100 Tesla) V bg 16
Graphene/Magnetic-Insulator: Exchange Field Graphene/EuS as model system: in-situ deposition Much better controlled stoichiometry (direct evaporation of target materials) EuS wide band-gap insulator (1.65 ev), no current shunting Large exchange splitting in bulk conduction band, ~0.36 ev (c.f. Busch, Junod, and Wachter, Phys. Lett. 12, 11 (1964)) Large magnetic moment per Eu ion, S z ~ 7μ B Expect large exchange splitting, Δ J S z EuS demonstrated to spin-polarize quasiparticles in Al and Bi 2 Se 3 P. Wei, et al., Nature Materials (2016) doi:10.1038/nmat4603 17
CVD Graphene/EuS Heterostructure: Raman, XRD, TEM X-ray diffraction (XRD): high quality single phase growth of EuS Hall-bar devices: fabrication simple EuS deposition at final step, capped with AlO x : EuS properties preserved Raman spectra: graphene structure intact Transmission electron microscopy: clean interface 18
Electrical Detection of Interfacial Exchange Field: Zeeman spin-hall effect & nonlocal transport Applied field (µ 0 H) + exchange field (B exc ) = total Zeeman field (B Z ) spin splitting (E Z ) spin-polarized electron vs. hole-like carriers near Dirac point R nl,d 1 ρ xx E Z ρ xy μ 2 Abanin et al. Science (2011) Abanin et al. PRL (2011) μ D µ 0 H ( ) couples to orbital motion: transverse spin current (Zeeman spin Hall effect) inverse spin-hall-like effect nonlocal voltage/resistance (R nl,d V nl,d /I) 19
M/M(5 K) IBM Research Zeeman Spin-Hall Effect: EuS induced enhancement T = 4.2 K R nl,d /R nl,d (5 K) 1.0 0.9 0.8 0.7 0.6 R nl /R nl (5 K) 1.0 0.9 0.8 0.7 0.6 graphene/alox 5 10 15 20 25 30 5 10 15 T c 20 25 30 T (K) T (K) 1.0 0.9 0.8 0.7 0.6 0.5 Magnetic origin of R nl,d Enhancement of R nl signal by EuS deposition Reduction of R nl signal upon EuS oxidation R nl correlates with M(T) R nl,d 1 ρ xx E Z ρ xy μ 2 μ D 20
EuS Induced Interfacial Exchange Field R nl,d / R nl,d (3.8 T) 10-1 10-2 10-3 Quantifying the EuS induced exchange field 10 0 graphene/alo T = 4.2 K X graphene/eus/alo X Graphene/EuS: x10 3 Graphene/AlOx: x10 0 1 2 3 4 Onset of exchange 0 H (T) at μ 0 H 0 ~1Tesla E Z (mev) E Z and B Z are lower-bound estimates because: We assume E Z contributed only by μ 0 H at onset 2.0 1.5 1.0 0.5 R nl,d = R 0 + β μ 0 H E z 2 graphene/eus 17.3 13.0 8.6 4.3 0.0 0.0 0 1 2 3 4 0 H (T) When μ 0 H = 3.5T, B Z >15 T β(μ 0 H) depends on mobility and is smaller in graphene/eus (<10% correction) β(μ 0 H 0 ) BZ (T) 21
Graphene/EuS: Quantum Hall regime 22
Quantum Effect: Landau Level (LL) Splitting = 0 LL splitting in R xx : R xx (k ) n C (10 11 cm -2 ) -5.5-3.7-1.8 0.0 1.8 3.7 5.5 10 n=0 8 6 4 2 data Gaussian fit Gaussian s - Gaussian s + n=-1 n=+1 0-6 -4-2 0 2 4 6 New R xy plateau at n = 0 : R xy (e 2 /h) 0.5 0.0 0.0 0 H = 3.5 T T = 4.7 K -0.5 0 H = 3.5 T T = 16 K -0.5-1 0 1 V g - V D (volt) dr xy /dv g (a.u.) dr xy /dv g (a.u.) 0-1 0 1 0-1 0 1 V g -V D (volt) = 0 LL splitting in R nl : R xx (k ) R nl (k ) R nl (k ) 9 6 3 0 4 2 0 4 2 0 I ~ 0.1 A I ~ 0.1 A I ~ 1 A n = 0 n = 0 n = 0-0.5 0.0 0.5 1.0 V g -V D (volt) 23
Quantum Effect: Spin-Polarized Chiral Edge States Graphene/EuS Graphene/AlO x Graphene/EuS 0.7 R xx, D (h/e 2 ) D. Abanin et al., PRL (2006) 0.6 0.5 0.4 0.3 R xx, D (k ) T = 4.2K 9 6 3 0Graphene/AlO 2 4 x 0 H (T) 0 1 2 3 4 0 H (T) Unique regime: Zeeman >> orbital field (µ 0 H) Chiral spin edge modes gapless modes 24
Summary B: Graphene/EuS Exchange Field Electrical detection via Zeeman SHE R nl,d / R nl,d (3.8 T) * Orders of magnitude enhancement in ZSHE R nl * R nl (T) correlates with M(T): magnetic origin * Giant B Z (> 15T) when EuS nearly polarized (spin control) * Observed unusual chiral spin edge modes in low field 10 0 graphene/alo T = 4.2 K X graphene/eus/alo X 10-1 10-2 10-3 0 1 2 3 4 0 H (T) R xx, D (h/e 2 ) 0.7 0.6 0.5 0.4 0.3 R xx, D (k ) T = 4.2K 9 6 Graphene/AlO x 3 0 2 4 0 H (T) Graphene/EuS 0 1 2 3 4 0 H (T) P. Wei, et al., Nature Mat. (2016), doi:10.1038/nmat4603. arxiv: 1510.05920 25