wrong hysteresis loop of Fe/PePt bilayer.

[phy.ash...@gmail.com]

Hi Richard,

First of all, thank you for providing the Vampire as an open-source code.

I am new to micromagnetic simulation and would like to do a test on a simple Fe(1 nm)/FePt(5 nm) bilayer. The exchange coupling between the two layers is switched of, so I expect step-like feature in the hysteresis loop (magnetization jumps twice at the coercivity of Fe and FePt). However, I did not observe this effect. I am afraid that I have set some parameters wrong. Since there is no tutorial of a hysteresis loop calculation, may I have your help to take a look at my inputs and help me to calibrate the parameters? The output magnetization is in attachment.

FePt-Fe.mat

#===================================================
# Sample vampire material file V3+
#===================================================

#---------------------------------------------------
# Number of Materials
#---------------------------------------------------
material:num-materials=2
#---------------------------------------------------
# Material 1 FePt
#---------------------------------------------------
material[1]:material-name=FePt
material[1]:damping-constant=0.2
material[1]:exchange-matrix[1]=3.0e-21
material[1]:exchange-matrix[2]=0
material[1]:atomic-spin-moment=1.458 !muB
material[1]:uniaxial-anisotropy-constant=4.94e-23
material[1]:uniaxial-anisotropy-direction= 0,0,1
#---------------------------------------------------
# Material 2 Fe
#---------------------------------------------------
material[2]:material-name=Fe
material[2]:damping-constant=0.01
material[2]:exchange-matrix[2]=4.2e-21
material[2]:exchange-matrix[1]=0
material[2]:atomic-spin-moment=2.2 !muB
material[2]:uniaxial-anisotropy-constant=0
material[2]:uniaxial-anisotropy-direction= 0,0,1

material[1]:minimum-height=0.0
material[1]:maximum-height=0.83
material[2]:minimum-height=0.83
material[2]:maximum-height=1.0


input
#------------------------------------------
# Vampire input file for M-H loop
#------------------------------------------

#------------------------------------------
# Creation attributes:
#------------------------------------------
create:crystal-structure=fcc
create:periodic-boundaries-x
create:periodic-boundaries-y


#------------------------------------------
# System Dimensions:
#------------------------------------------
dimensions:unit-cell-size-x = 3.85 !A
dimensions:unit-cell-size-y = 3.85 !A
dimensions:unit-cell-size-z = 3.73 !A
dimensions:system-size-x=4 !nm
dimensions:system-size-y=4 !nm
dimensions:system-size-z=6 !nm
#------------------------------------------
# Material Files:
#------------------------------------------
material:file=FePt-Fe.mat

#------------------------------------------
# Simulation attributes:
#------------------------------------------
sim:temperature=300
sim:equilibration-time-steps=10000
sim:loop-time-steps=10000
sim:maximum-applied-field-strength=10
sim:minimum-applied-field-strength=-10
sim:applied-field-strength-increment=0.1
sim:applied-field-unit-vector=0.00001,0,0.99999

#------------------------------------------
# Program and integrator details
#------------------------------------------
sim:program=hysteresis-loop
sim:integrator=llg-heun

#------------------------------------------
# data output
#------------------------------------------
output:applied-field-strength
output:magnetisation

screen:applied-field-strength
screen:magnetisation

config:atoms
config:atoms-output-rate=10

Thank you in advance!

Best regards,
Asher

[richard....@gmail.com]

Hi Asher,

Yes - overall this looks fine. The problem comes from the damping parameter. In general hysteric properties are 'slow' and so to get loops comparable with experiment you would ordinarily wait a long time - hundreds of nanoseconds. However, this is obviously computationally expensive, so one can accelerate the process by using critical damping (material[2]:damping-constant=1.0) so that the magnetisation relaxes quickly. This still requires reasonably long simulation times (~10 - 20 ns for the whole loop) but allows the simulation to approach the Stoner-Wohlfarth limit for the coercivity. In general it is a good idea to increase the loop-time to check the convergence of the coercivity, which should be close to 2k_u/mu_s at 0K.

All the best,

Richard

[BHANU PRAKASH PANT]

Hi Asher,

Can you please tell from where did you get the material file parametervaluesof FePt? Can you please link the source? 

Thanks.

Bhanu