Spin-triplet supercurrent and controllable phase states in Josephson junctions containing ferromagnetic materials

Transcription

1 pin-triplet supercurrent and controllable phase states in Josephson junctions containing ferromagnetic materials Norman Birge, Michigan tate University in collaboration with Northrop Grumman Corporation The project depicted was or is supported in part by the U.. Army Research Office and/or the Department of Energy. The information depicted does not necessarily reflect the position or the policy of the Government, and no official endorsement should be construed.

2 Collaborators: Trupti Khaire, Mazin Khasawneh, Caroline Klose, Patrick Quarterman, Hamood Arham, Kurt Boden, Yixing Wang, Eric Gingrich, imon Diesch, Kevin Werner, Alex Cramer, Bill Martinez, Joseph Glick, Bethany Niedzielski, Victor Aguilar, Josh Willard, am Edwards, Bob Klaes, Alex Madden, Thomas Bertus, Anna Osella, Reza Loloee, William P. Pratt, Jr

3 Outline Introduction: Proximity effect in /N and / systems Theoretical prediction: spin-triplet pair correlations Experimental verification using // Josephson junctions Amplitude and phase control of spin-triplet supercurrent Work toward a superconducting magnetic memory Phase control in a simpler system ummary and uture Prospects

4 Proximity effect: leakage of Cooper pairs from to N or N E E ε ε ~ -k Electron wavefunctions dephase over time ξ N = v τ = v 2πk B T k τ k 2ε Average over thermal distribution: ballistic ξ k -k -k k k k Q Q = 2 E v 2E 1 = = ex 2E ex ex v k ballistic ξ N = D N τ = D 2πk N B T diffusive D = diffusion constant ξ = D E ex diffusive

5 Proximity effect: /N vs. / ξ N ξ x ξ DN = 0.1 2πk T 1 N µ B m ξ ~ D E ex fewnm Ψ( x) = Ψ0 exp( x / ξ N ) Ψ( x) = Ψ0 cos( x / ξ )exp( x / ξ )

6 How to detect the oscillating pair correlation function? 1. Measure Tc of / bilayers as a function of d Jiang, Davidovic, Reich, Chien, PRL 74, 314 (1995). d Problem: interpretation controversial due to magnetic dead layers.

7 How to detect the oscillating pair correlation function? 2. Measure tunneling density of states in //I/N structure, as function of d Kontos, Aprili, Lesueur, Grison, PRL 86, 304 (2001) I N d

8 How to detect the oscillating pair correlation function? 3. Measure critical current of // Josephson junction, as function of d 0-state: I s = I c sin(φ 2 -φ 1 ) π-state: I s = I c sin(φ 2 -φ 1 +π) Weak : Cu 48 Ni 52 alloy, vary T d Weak : Pd 88 Ni 12, vary d Ryazanov et al., PRL 86, 2427 (2001). Kontos, Aprili 89, (2002)

9 How to detect the oscillating pair correlation function? 3. Measure critical current of // Josephson junction, as function of d 0-state: I s = I c sin(φ 2 -φ 1 ) π-state: I s = I c sin(φ 2 -φ 1 +π) More Cu 48 Ni 52 alloy d trong : Co I c R N (mv) d Co (nm) Oboznov et al., PRL 96, (2006). Robinson, Piano, Burnell, Bell, Blamire, PRL 97, (2005)

10 2001 Prediction: spin-triplet pair correlations i) are long-ranged in ii) can be induced by noncollinear magnetization E 1 2 or -k -k k k k inglet Triplet ξ = D E ex ξ T = D 2πk B T ξ << ξ Bergeret, Volkov & Efetov, PRL 86, 4096 (2001); PRL 90, (2003) Kadrigrobov, hekhter & Jonson, Europhys. Lett. 54, 394 (2001) T

11 // Josephson junction can carry spin-triplet supercurrent Bergeret, Volkov & Efetov, PRL 86, 4096 (2001) Houzet & Buzdin, PRB 76, (R) (2007): Non-collinear magnetizations convert pairs from spin-singlet to spin-triplet d ignature of spin-triplet supercurrent: Log(I c ) singlet triplet ξ = D E ex ξ T = D 2πk B T d

12 ample abrication 1. putter // 2. Pattern pillars with photo or e-beam lithography Image reversal photoresist Au Nb Nb bottom contact i ubstrate µm

13 ample abrication 1. putter // 2. Pattern pillars with photo or e-beam lithography 3. Ion mill 4. Deposit iox i O x Au Nb Nb bottom contact i ubstrate µm

14 ample abrication 1. putter // 2. Pattern pillars with photo or e-beam lithography 3. Ion mill 4. Deposit iox 5. Liftoff 6. Deposit top Nb contact i O x Nb top contact Au Nb bottom contact i ubstrate µm

15 Measurement Low sample resistance: 100 µω 10 mω Measure with QUID-based potentiometer Measure at T = 4.2 K in quick-dipper cryostat in helium storage dewar I-V characteristic of overdamped Josephson junction 5 supercurrent V (nv) 0-5 I c - I c I (ma) I c critical current

16 Problem: Large-area // junctions with strong ferromagnets distorted raunhofer patterns Nb d Co = 5 nm, 2R = 40 µm Co Nb λ L + d H ext I c (ma) R H (Oe) Random raunhofer pattern due to complex domain configuration

17 olution: Co/Ru/Co synthetic antiferromagnet cancels flux and restores raunhofer pattern d Co = 13 nm w = 20 µm 0.06 d Co = 5 nm w = 40 µm Ic(mA) I c (ma) H (Oe) H (Oe)

18 How to generate spin-triplet supercurrent I c R N (nv) without, with, = PdNi, d = 4 nm Nb Co Ru Co Nb Cu Cu Cu Cu D Co (nm) ix D Co = 20 nm and vary d Khaire, Khasawneh, Pratt, & NOB, Phys. Rev. Lett. 104, (2010)

19 Control amplitude of triplet with d, d 100 Ni Nb I C R N (nv) 10 1 PdNi Co Ru Co Cu Cu Cu Cu 0.1 CuNi Nb d ' (nm) Khasawneh, Khaire, Klose, Pratt & NOB, upercond. ci. Technol. 24, (2011).

20 Microscopic mechanism for triplet generation (M. Eschrig) E ,0 ψ = = = 1 = ,0 ( ) iqx iqx ( e e ) [( ) cos( Qx) + i( + ) sin( Qx) ] z cos rotate basis: ( Qx) + 1,0 sin( Qx) ( θ ) z ( θ ) sin sin 1,0 = 1,1 + cos( θ ) 1,0 1, 1 z θ θ 2 2 long-range triplet components in 2 short-range triplet component θ -k -k k k 1,1 Q θ 1, 1 k = k k θ = = θ θ

21 Optimization of triplet generation M. Houzet and A. I. Buzdin, PRB 76, R Ni I C R N (nv) d x (nm) PdNi CuNi Triplet contribution to the critical current is observed only for d X ( )ξ increase ~ sin 2 (Qx); decrease ~ e x /ξ E > E > Ni ex ξ < ξ < ξ Ni PdNi ex PdNi E CuNi ex CuNi

22 Earlier observation of triplet: Keizer, Goennenwein, Klapwijk, Miao, Xiao, Gupta, Nature 439, 825 (2006) µm CrO 2 is a half metal E E k Long-range propagation of supercurrent, but large sample-to-sample variations in I c. Reproduced in 2010 by J. Aarts: M.. Anwar et al., Phys. Rev. B 82, (2010)

23 pin-triplet supercurrents observed by others in Robinson et al., cience 329, 59 (2010) 2. prungmann et al., PRB 82, (2010) Heusler alloy Cu 2 MnAl

24 pin-triplet supercurrents observed by others in Anwar et al., PRB 82, (2010) 4. Wang et al., Nature Phys. 6, 389 (2010)

25 irst step toward control of supercurrent: magnetize the samples (Expect I c to drop by reducing non-collinear magnetization?) Log(I c ) Nb singlet: triplet: Co Ru Co Cu Cu inhomogeneous M Nb Cu Cu homogeneous M d

26 I c increases! 1000 I c R N (nv) 100 = = Ni d Ni = 1 nm H applied (Oe) Magnetizing field Why does I c increase? Klose et al., Phys.Rev. Lett. 108, (2012)

27 Co/Ru/Co undergoes spin-flop transition pin-flop transition optimizes 90 angle between M Co and M Ni and aligns Ni magnetizations Magnetic configuration confirmed at NIT by EMPA (McMorran & Unguris) and PNR (Ginley & Borchers); see Klose et al., Phys.Rev. Lett. 108, (2012)

28 Other probes of spin-triplet correlations: T c of // trilayers ingh, Voltan, Lahabi, Aarts, PRX 5, (2015).

29 Next step: Control spin-triplet supercurrent by controlling magnetic states On/Off Phase Change Off (inglet only) On Triplet 0 π Rotate M by 90 Requires only one junction Need two external field coils Rotate M by 180 Measure with QUID (interference) Need one external field coil

30 Desire single-domain magnets: abricate submicron // Josephson junctions Nb Cu Cu Nb side view ma-n 2401 negative e- beam resist Junctions patterned by Ar ion milling side view H app 0.50 µm H applied top view M 1.26 µm top view M

31 raunhofer pattern for junction with = Ni 73 e 21 Mo 6 Nb Cu Cu Nb side view H app d NieMo = 1.0 nm M top view Airy pattern is shifted in opposite direction of M layer is single-domain in remnant state Niedzielski, Gingrich, Loloee, Pratt, & Birge, ut 28, (2015).

32 Nb Cu Cu Nb = NieMo I c vs NieMo thickness Data fit to: 0 - π transition occurs at d NieMo = 2.25 nm I c decays rapidly with d NieMo : ξ 1 = 0.48 nm Niedzielski, Gingrich, Loloee, Pratt, & Birge, ut 28, (2015).

33 Next step: Control spin-triplet supercurrent by controlling magnetic states On/Off Phase Change Off (inglet only) On Triplet 0 π Rotate M by 90 Requires only one junction Need two external field coils Rotate M by 180 Measure with QUID (interference) Need one external field coil

34 Cartoon Representation ( Py = Nie) Orthogonal magnetizations High state Non-orthogonal magnetizations Low state Martinez, Pratt & Birge, PRL 116, (2016)

35 Rotate Py Magnetization to turn off H H lowly increase H to rotate M Py Measure at 0-field Martinez, Pratt & Birge, PRL 116, (2016)

36 Rotate Py back to turn on H H lowly increase H to rotate M py back again Measure at 0-field Martinez, Pratt & Birge, PRL 116, (2016)

37 Turn on and off repeatedly

38 Next step: Control spin-triplet supercurrent by controlling magnetic states On/Off Phase Change or or Off (inglet only) On Triplet 0 π Rotate M by 90 Requires only one junction Need two external field coils Rotate M by 180 Measure with QUID (interference) Need one external field coil

39 Controllable 0-π switching with spin-triplet supercurrent A { Cu Cu Ru Cu Cu All units in nm Ni(1.6) [Co(0.3)/Pd(0.9)] n Ru(0.95) [Pd(0.9)/Co(0.3)] n Nie(1.25) [Co/Pd] multilayers have strong perpendicular magnetic anisotropy. Central Ru layer creates a synthetic antiferromagnet, to reduce stray field at domain walls Ni(1.6) has fixed in-plane magnetization direction Nie(1.25) is in-plane free layer Glick et al. (in preparation) 39

40 QUID design & measurement protocol QUIDs have one ellipse with aspect ratio = 2 and one hex bit with aspect ratio = 3, both have area = 0.5 μm 2. Measurement protocol: Initialize bits at Oe, then remove trapped flux Apply set fields H set = +5 Oe, +10 Oe, +15 Oe, etc. until magnetic state switches After each set field, measure I c (I flux ) Apply set fields H set = -5 Oe, -10 Oe, etc. to return to initial state All measurements are performed at H set = 0. Keep H set small enough so that only the elliptical bit switches (minor loop). H set M Py,1 H flux M Py, 2 40 I flux I sample

41 Controllable 0-π switching with spin-triplet supercurrent Data will be re-inserted after publication (Glick et al. in preparation) 41

42 This work is not just of academic interest!

43

44 A superconducting computer needs memory Josephson Magnetic Random Access Memory (JMRAM) Anna Y. Herr & Quentin P. Herr, U Patent 8,270,209 (2012) Northrop Grumman Corporation Memory cell is QUID loop One junction has two stable states for 0 and 1 Magnetic states are written using standard MRAM techniques

45 Recall: I c of // Josephson junction oscillates with d 0-state: I s = I c sin(φ) π-state: I s = I c sin(φ+π) Weak : Cu 48 Ni 52 alloy d trong : Co I c R N (mv) d Co (nm) Ryazanov et al., PRL 86, 2427 (2001); Oboznov et al., PRL 96, (2006). Robinson, Piano, Burnell, Bell, Blamire, PRL 97, (2005)

46 JMRAM memory cell junction //: I c vs d pin-valve memory element 2 N 1 P state: Cooper pair phase shifts add π state AP state: Cooper pair phase shifts subtract 0 state / 1 /N/ 2 /: fixed d 1 & d 2 AP P Control phase rather than amplitude: JJ acts as passive phase shifter; no need for large I c R N 0-state π-state x ξ 1 φ = ± 1 x ξ 2 2 Example: eofanov et al., Nature Phys. 6, 593 (2010) φ for fixed 1 layer

47 Demonstration of 0 - π switching of / 1 /N/ 2 / spin-valve Josephson junctions side view Nb Cu 1 Cu 2 Cu Nb 1 = fixed layer: Ni (1.2 nm) 2 = free layer: Nie (1.5 nm) AP 0-state P π-state x ξ 1 φ = ± 1 φ for one layer x ξ 2 2 H in witch magnetization with in-plane field JJ s with different aspect ratios switch at different applied magnetic fields 0.50 µm H in M top view 1.26 µm On-chip current line couples magnetic flux into QUIDs

48 Major loop data show switching of both JJs 0-π 0-0 π-π 0-0 π-0 π-π witching ields: +30 Oe, +50 Oe, -35 Oe, -100 Oe Gingrich, Niedzielski, Glick, et al., Nat. Phys. 12, 564 (2016) 48

49 Data cuts for the four magnetic states I c+, I c- I c ave = (I c+ - I c- )/2 π - π 0 - π 0-0 π - 0 Gingrich, Niedzielski, Glick, et al., Nat. Phys. 12, 564 (2016) 49

50 I c (Φ) curves have tilted ratchet shape when loop inductances and/or critical currents are asymmetric L 1 I c1 Voltage (µv) I c- I c+ L 2 I c Current (µa) I c+ (Φ) and I c- (Φ) oscillations are asymmetric when L 1 L 2 & I c1 I c2 I c+ and I c- shift by equal amounts and in opposite directions along the Φ axis Analyze I c+ and I c- peak shifts to extract JJ phase shifts Ic Ic+ Φ Ic- 50

51 Quantitative fits to QUID modulation data for the four magnetic states I c+, I c- I c ave = (I c+ - I c- )/2 π - π 0 - π 0-0 π - 0 Gingrich, Niedzielski, Glick, et al., Nat. Phys. 12, 564 (2016) 51

52 Quantitative Analysis Consistently Assigns the Inductance and Critical Currents of Each tate state I c1 (ma) I c2 (ma) L 1 (ph) L 2 (ph) π - π π π ave σ asthenry simulations: L 1 7 ph, L 2 13 ph itting parameters from independent fits of 4 magnetic states are highly consistent Exception: critical current of JJ #2 changes slightly in π state when JJ #1 switches from π to 0 state Gingrich, Niedzielski, Glick, et al., Nat. Phys. 12, 564 (2016) 52

53 What needs to be done Memory Optimize performance of magnetic materials Lower M sat lower E switch Reduce extrinsic sources of anisotropy in thin films ind better material for fixed layer (Ni has issues) Minimize underlayer roughness Develop read/write electronics & interface to Q logic Make the rest of the computer! 53

54 ummary Experimental observation of long-range supercurrent in // Josephson junctions: requires three layers can control either supercurrent amplitude or junction ground-state phase by changing the magnetic state Magnetic Josephson junctions may be useful for ultra-low-power cryogenic memory phase control achieved with a spin-valve JJ