Transcript Ultrafast Manipulation of the Magnetization, 2009

Ultrafast Manipulation of the
Magnetization
J. Stöhr
Sara Gamble and H. C. Siegmann,
SLAC, Stanford
A. Kashuba
Bogolyubov Institute for Theoretical Physics, Kiev, Ukraine
Drivers of Modern Magnetism Research: Smaller and Faster
The ultrafast
technology
gap
“for the discovery of giant magnetoresistance”
“pinned”
“0”
“1”
GMR “reads” the magnetization state
from “reading” to “writing” information
?
“0”
“1”
The big questions:
• What are possible switching methods?
• What are the physical processes (and intermediate states)?
• What limits the speed of switching?
Conceptual methods of magnetization switching
Optical pulse
Lattice shock
t ~ 1 ps
electrons move in
femtoseconds
Electrons
Phonons
atoms move in
picoseconds
exchange
or
spin-orbit
t ~ 100 ps
?
Spin
field pulses
or
spin currents into magnetic element
Creation of large, ultrafast magnetic fields
Use field pulse created by a moving electron bunch
Origin of the fast switching idea…
What is the pattern written by a lightning bolt in magnetic rock?
100 kA in a flash
of a few microseconds
Magnetization follows the field lines
The world’s biggest lightning bolt
Stanford Linear Accelerator Center - SLAC
3km, 30GeV
Magnetic writing with SLAC Linac beam
thin Co film on Si wafer
premagnetized
5 ps
1nC or 1010 electrons
C. H. Back et al., Science 285, 864 (1999)
In-Plane Magnetization: Pattern development
• Magnetic field intensity is large
• Precisely known field size
no circles around beam !
very different from lightning
pattern
540o
Rotation angles:
720o
180o
360o
The pattern written by a picosecond beam field
M
initial magnetization
direction of sample
beam
damage
Max. torque
T=MxH
Min. torque = 0
Fast switching occurs when H ┴ M
Ballistic Switching – From nano to picoseconds
Patent issued December 21, 2000: R. Allenspach, Ch. Back and H. C. Siegmann
end of
field pulse
M
Relaxation into “down” direction governed
by slow spin-lattice relaxation (100 ps)
- but process is deterministic !
Precise timing for a=180o reduces time
Toward femtosecond switching
Experiments with sub-ps bunches
• reduce bunch length from FWHM
t = 5 ps →
140 fs
• keep beam energy and charge fixed (~1010 electrons or 1 nC)
• fields B ~ charge / t
and E = c B are increased by factor of 35
• our fields have unprecedented strength in materials science:
B-field: 60 Tesla
E-field: 20 GV/m or 2V / Angstrom
How does a relativistic e-beam interact with a material ?
note E and B fields are defined
within and outside e-bunch
Magnetic pattern is severely distorted for short bunch
10 nm Co70Fe30
140 fs
on MgO (110)
15 layers Fe
on GaAs (110)
5 ps
damaged
area
Magnetic pattern is severely distorted
--- does not follow circular B-field symmetry
Calculation of pattern with Landau-Lifshitz-Gilbert theory
known magnetic properties of film, known length, strength, radial dependence of fields
B-field only
B-field and E-field
Consider effect of giant E-field of beam
magnetocrystalline anisotropy caused by anisotropic atomic positions
“bonding fields” distort valence charge – static effect
Beam field E ~ 1010 V / m = 1 V / Å comparable to “bonding fields”
leads to ultrafast distortion of valence charge - all electronic dynamic effect
all new “magnetoelectronic anisotropy” – ultrafast !
Magneto-electronic anisotropy is strong ~ E 2
352 or about 1000 times stronger than with previous 5 ps pulses
B-field torque
E-field torque
Practical Realization of E-field switching
• Giant accelerator is impractical
• Want to produce pure E-field effect – no B-field effect
• Field pulse needs to be fast
How about photons ?
We know effect is ~E2
Linear B-field effect cancels over a full cycle
SLAC e-beam pulse corresponds to THz half-cycle pulse
100 fs
10 THz
red: SLAC pulse
black: THz half cycle pulse
true “EM wave”
• Need strong THz radiation - not readily available
• Presently only produced by accelerators
• Laser generated THz about 100 times weaker at present
Can sample handle intense THz pulse ?
Heating of sample would be problem….
Compare beam impact region for different pulse lengths
same sample: 10 nm Co70Fe30 on MgO (110)
Magnetic image
Topological image by means of SEMPA microscopy
Pulse length: 4 ps
beam damage
Pulse length: 140 fs
35 times shorter pulse &
stronger fields cause
no heating, no damage !
If there is an E-field - why is there no heating?
strong E field should cause current flow - severe Joule heating
Potential of a regular linear lattice
Co bandwidth DV ~ 3eV
a
Potential along E field direction
E ~ 1010 V/m
a = 0.25 nm
DV = e E a ~ 2.5 eV
Offset of “bands” ~ bandwidth
potential gradient leads to breakup of conduction path
no current flow due to field – no heating
Summary
material behave very strange in extreme fields !
• Unusual E and B field effects
• No apparent heating or damage by beam
• Extreme THz science just starting….