Post on 26-Sep-2020
transcript
Finding Unusually Strong Spin-Orbit Coupling in Post-Perovskite CaIrO3 via X-ray Magnetic Circular Dichroism!
Luke G. Marshall1!Jinguang Cheng1,2, Jianshi Zhou1, John B. Goodenough1, Daniel Haskel3, Michel van Veenendaal3,4!
Financial Support From:! Special Thanks To:!Advanced Photon Source!Beam Line 4-ID-D!Supported by the US Department of Energy, Office of Science, Office of Basic Energy Science, under contract No. DE-Ac02-06cH11357.!
1Texas Materials Institute, The University of Texas at Austin!2The Institute of Physics, The Chinese Academy of Sciences!
3Advanced Photon Source, Argonne National Laboratory!4Department of Physics, Northern Illinois University
Motivation:!
Post-Perovskite (pPv)! Perovskite (Pv)!
pPv has edge sharing octahedra along a, and corner sharing along c:!
For 5d5 Ir, we would expect metallic behavior along M-M exchanges in edge-shared direction, but pPv CaIrO3 shown to be Mott insulator. So where is the energy gap coming from?!
O1
O2 Ir
z
Octahedral Bonds:!Two Short, Four Long!
2 X Ir—O1: 1.9786(13) Å!
4 X Ir—O2: 2.0543(13) Å!Ref: J.-G. Cheng et al., PRB 83, 064401 (2011) !
(site axes)!xy!yz±izx!
" L•S!
The octahedral distortion shows that the orbital angular momentum is not quenched. Large spin-orbit coupling (SOC) can win over Ir-O-Ir " bonds.
1.978Ǻ!
2.054
Å!
C!
134.5°!Ir
O1!O2! O2!
Ref: N. A. Bogdanov et al., PRB 85, 235147 (2012) !
Ref: B.J. Kim, et al., PRL 101, 076402 (2008)!
Proposed Model:!
As SOC is taken into account, t2g states effectively correspond to L=1 states!
CaIr4+O3: LS t2g5eg
0!
Initial Conclusions:!1. pPv CaIrO3 has edge-sharing octahedra along a, and
corner-sharing along c.!
2. Octahedral site distortion forms 2 short and 4 long Ir-O bonds, which puts the hole in the degenerate yz and zx orbitals.!!Therefore, the orbital angular momentum is not quenched by the crystal field and SOC splits yz/zx the states.!
3. The observation of dominant coupling to the orbital moment in the magnetized state is consistent with an energy gap generated by strong spin-orbit coupling.!!Additional Ref: http://ssrl.slac.stanford.edu/stohr/xmcd.htm!
Experimental Techniques – XANES & XMCD:!
εF!
2p3/2 [4] L + S!2p1/2 [2] L - S!
X-ray Absorption Near Edge Spectroscopy (XANES)!
Fe!3d "* band!
M=0!
Energy!
Inte
nsity!
M=0!L3! L2!
L3!L2!
1. To explain how XANES and XMCD can be useful tools to probe SOC, we first start with a simple example of Fe-metal (no SOC). !
2. Unpolarized photons excite photoelectrons from the 2p states to empty part of 3d band while scanning x-ray energy. Resulting data shows white peaks at lower-energy L3 edge and higher-energy L2. L3 counts are twice L2 counts because of degeneracy of 2p3/2 state compared to 2p1/2. There is no magnetic field.
X-ray Magnetic Circular Dichroism (XMCD)!
M≠0 !
Energy!
Intensity
LH!M=0!
L3! L2!
3. A magnetic field applied parallel to the beam. The field
splits the 3d band as half of the electronic states are stabilized and the other half destabilized.!
4. Left-Handed (LH) circularly polarized x-rays are used. Because of the polarization, the x-rays will only “see” half of the electrons – those that have the same direction as the handedness of the beam.!
5. Since there are more states to enter, absorption intensities are enhanced.!
Energy
Inte
nsity!
L3! L2!
LH!M=0!RH!
M≠0 !
6. Then the same is repeated for Right-Handed (RH) circularly polarized light. The absorption intensities show a reduction commiserate with the smaller available bandwidth. !
Energy
Inte
nsity!
L3! L2!
LH!M=0!RH!
2p3/2 [4] L + S!2p1/2 [2] L - S!
7. It is important now to correct the sign of the L2 edge. Because of SOC in the 2p orbitals, the 2p1/2 electrons produce the opposite picture of the 3d band structure as the 2p3/2.!
Energy!
Intensity
L2 L3
8. The dichroism signal is the difference between the LH and RH signals. Above is typical XMCD for no SOC.!
Results and Analysis:!
Compared to the results for Fe that has no SOC, these results are quite striking:!
L3 dichroism is slightly smaller than L2, and the sign of the dichroism is the same.!
Is this evidence of SOC?!
M≠0! If we draw out our Mott-insulator density of states similar to above we get the picture to the left.!
In this case, we expect a larger contribution from L from the 2p photoelectrons rather than S, because of the energy gap.!
XMCD Sum Rules:!
Exchange Hamiltonian:!
€
Hexch = αLz + βSz€
IL3c − 2IL2
c
IL3c + IL2
c =4 Sz +14 Tz
3 Lz
Branching Ratio:!
€
BR =IL3IL2
=2 + r( )1− r( )
€
r =L • Snh
Modeling XMCD:!