Sie sind auf Seite 1von 14

Preprint DRAFT 2011

Micromagnetic Structure and Symmetry of the 90 degree Domain

Wall Suppressed by the Demagnetization Field

B.M. Tanygina1, O.V. Tychkoa


a
Kyiv Taras Shevchenko National University, Radiophysics Faculty, Glushkov av.4g, Kyiv,

Ukraine, MSP 01601


1
Corresponding author: B.M. Tanygin, 64 Vladimirskaya str., Taras Shevchenko Kyiv

National University, Radiophysics Faculty. MSP 01601, Kyiv, Ukraine.

E-mail: b.m.tanygin@gmail.com

Phone: +380-68-394-05-52

Abstract.

A magnetization distribution in the yttrium-ferrite garnet (001) plate without

unidirectional anisotropy term is numerically considered for various sample thickness. The

smooth transition (from the <110> to the <111> domains) of the domain structure in case of

the plate thickness growth have been investigated. It was shown that growth of the plate

thickness leads origin of the periodical <111> domain structure in the volume of the 90°

domain wall between <110> domains.

PACS: 75.60.Ch, 75.70.-i, 75.78.Cd

Please, cite original work as:


B.M. Tanygin, O.V. Tychko,
A magnetization distribution in a single domain wall volume,
Proceeding of the II International Conference "Electronics and Applied Physics", Kyiv, Ukraine (2006) 40-41.

e-mail: b.m.tanygin@gmail.com
Web: http://sites.google.com/site/btanygin/
Preprint DRAFT 2011

1. Introduction

It is well-known that problems of the applicability of the Bloch and Néel domain wall

(DW) model are usually solving via the micromagnetic simulation start from the work.

Usually changes of the micromagnetic structure in the DW volume or changes of the domain

structure (including domain magnetization reoritation) depend on the material sample

parameters are usually under consideration. Here we investigate both types of changes in the

same time.

An information about magnetization vector spatial distribution in a medium volume is

basic at a theoretical investigation of magnetic states and processes in a magnetically ordered

media. Reductive models that allow getting an analytical description of such distribution are

used for receiving this spatial distribution. Hence these models only can characterize a

domain structure and DW approximately, often qualitatively. One can obtain detailed

information about film media micromagnetic structure using numerical methods of

micromagnetic problem with initial and limit conditions.

2. Calculation technique

Let us consider 2D-problem numerical simulation in cross-section of a thin (100)-film.

Let Z and X axis are directed along film normal and crystallographic [100] direction

respectively. A free energy functional counting upon a unity of length along Y axis look like:

G= ∫∫ [g + ν r (α α)]d r , where ν r is the Lagrange multiplier, g = g A + g m + g K is a volume


2

energy density, ( 2
)
g A = A (∂α / ∂x ) + (∂α / ∂z ) is a an exchange energy density,
2
A is an

exchange constant, α = M / M , M is a saturation magnetization; g m = −(M ⋅ H m ) / 2 is a

demagnetization field energy density, Hm is a demagnetization field,


Preprint DRAFT 2011

( )
g K = K 1 α p α q 1 − δ pq / 2 is a magnetocrystalline anisotropy energy density, δ pq is the
2 2

Kronecker symbol, p, q = x, y, z.

Requirement of an equilibrium magnetization distribution is δG = 0 or an absence of a

mechanical moment due to rotation [M × H eff ] = 0 , where H eff = −δg / δM is an effective

field: H eff = 2 A(∂ 2 α / ∂x 2 + ∂ 2 α / ∂z 2 )/ M s + H m − 2 K1 ∑ α i 3e i / M s + β′ M , where β is an arbitrary


3

i =1

( )
real magnitude, β ′ = β + 4πQ , Q = K1 / 2πM s . The M relaxation to an equilibrium state is
2

yielded by means of its reorientation to an effective field direction (M establishment).

An order of an establishment of M orientation in finite elements (FE) can break

problem symmetry. It is assumed to use the random numbers generator for finding of FE

counters i and j for which the establishment is yielded in the given step of calculus.

x = x /δ0 , z = z /δ0 , g = g / Ms
2
Let we introduce dimensionless values and

h eff = H eff / M s , where δ 0 = A / K1 . Then, expressions are given by:

 
( )
3 3
g = −2πQ α∂ 2 α / ∂x 2 + α∂ 2 α / ∂z 2 + ∑ ∑ α i α j 1 − δ ij / 2 − αh m / 2 + β
2 2

 i =1 j =1 

 3
3
h eff = 4πQ ∂ 2 α / ∂x 2 + ∂ 2 α / ∂z 2 − ∑ e i α i  + h m + β′α ,
 i =1 

where h m = H m / M s .

So, dimensionless energy of physical space Ω = [x1 / δ 0 , x2 / δ 0 ]× [z1 / δ 0 , z 2 / δ 0 ] is

( )
I J
G = G / M s δ0 . Ω Ω ≡ U U ωij ,
2 2
Let we break into identical rectangles: where
i =1 j =1

[
ωij = [xi − ∆x / 2, xi + ∆x / 2]× z j − ∆z / 2, z j + ∆z / 2 . Now, energy is: ]
{[(
g ij = 2πQ α (i +1) j (1 − δ iI ) − αij ) + (α
2
ij − α (i −1) j (1 − δ i1 ) ) ]/ (2∆x ) +
2

[( (
+ αi ( j +1) 1 − δ jJ − αij ) ) + (α
2
ij (
− αi ( j −1) 1 − δ j1 )) ]/ (2∆z ) − ∑ ∑ α
2
3 3
i
2
( ) 
α j 1 − δ ij / 2 − αh ijm / 2 + β
2

i =1 j =1 
Preprint DRAFT 2011

and effective field:

([ ]
ij = 4πQ α (i −1) j (1 − δ i1 ) + α (i +1) j (1 − δ iI ) / (2∆x ) +
h eff

[ ( ) ( )]
3
+ αi ( j −1) 1 − δ j1 + αi ( j +1) 1 − δ jJ / (2∆z ) − ∑ e i α i + h ijm + β′α ,
3

i =1

where demagnetization field h ijm is

 
{ }
h ijm =  ∫∫ d 2 r ∑∑ ∫ 2 αij n(R ) (r − R )/ r − R d R  / (∆x ∆z ) ,
I J 2

ω k =1 l =1 ∂ ωkl 
 ij 

where n(R ) is normal to the ∂ωij . The field h ijm is given by the:

I J
h ijm = ∑∑ N̂(k − i, l − j , g )α kl ,
k =1 l =1

where g = ∆z / ∆x . Let we determine n = k −i and m=l− j with integers

n = (rkl − rij )e x / ∆x and m = (rkl − rij )e z / ∆z . Components of tensor N̂(n, m, g ) are:

N xx (n, m, g ) = K xR (n, m, g ) + K xL (n, m, g ) ;

N zx (n, m, g ) = K zR (n, m, g ) + K zL (n, m, g ) ;

N xz (n, m, g ) = K Ux (n, m, g ) + K xD (n, m, g ) ;

N zz (n, m, g ) = K Uz (n, m, g ) + K zD (n, m, g ) ,

where values K with indexes x, z and R, L, U, D are numerically equal to the x, z – th

components of the filed created by poles on the right, left, top and bottom sides of ∂ω kl

respectively. The calculations with symmetry based optimization give the following:

 FxR (n,m,g ) − FxR (n − 1,m,g ), n ≥ 1 and m ≥ 0


 R
 Fx (0 ,m,g ) − Fx (1,m,g ), n = 0 and m ≥ 0
R

K x (n, m, g ) =  L
R

 K x (− n,m,g ), n ≤ −1 and m ≥ 0
 K xR (n, − m,g ), m ≤ −1
Preprint DRAFT 2011

 FxR (n,m,g ) − FxR (n + 1,m,g ), n ≥ 1 and m ≥ 0


 R
 Fx (0,m,g ) − Fx (1,m,g ), n = 0 and m ≥ 0
R

K x (n, m, g ) =  R
L

 K x (− n,m,g ), n ≤ −1 and m ≥ 0
 K xL (n, − m,g ), m ≤ −1

 FzR (n,m,g ) − FzR (n − 1,m,g ), n ≥ 1 and m ≥ 0



− Fz (0 ,m,g ) + Fz (1,m,g ), n = 0 and m ≥ 0
R R

K z (n, m, g ) = 
R

− K z (− n,m,g ), n ≤ −1 and m ≥ 0
L

− K zR (n, − m,g ), m ≤ −1

 FzR (n,m,g ) − FzR (n + 1,m,g ), n ≥ 1 and m ≥ 0


 R
 Fz (0 ,m,g ) − Fz (1,m,g ), n = 0 and m ≥ 0
R

K z (n, m, g ) = 
L

− K z (− n,m,g ), n ≤ −1 and m ≥ 0
R

− K zL (n, − m,g ), m ≤ −1

K Ux (n, m, g ) = K zR (m, n,1 / g )

K Uz (n, m, g ) = K xR (m, n,1 / g )

K xD (n, m, g ) = K zL (m, n,1 / g )

K zD (n, m, g ) = K xL (m, n, 1 / g )

FxR(n,m,g) = f xR(n,m + 1,g) + f xR(n,m − 1,g) − 2 f xR(n,m,g)

FzR(n,m,g) = f zR(n,m + 1,g) + f zR(n,m − 1,g) − 2 f zR(n,m,g)

[ ( ) ( )]
 4nmg arctan ( gm/n ) − n 2 − g 2 m 2 log n 2+g 2 m 2 / (2 g ), n ≠ 0 and m ≠ 0
 2
gm log( gm ), n = 0 and m ≠ 0
f x (n,m,g) = 
R

( )
− n 2 log n /g , n ≠ 0 and m = 0

0, n = 0 and m = 0

[( ) ( )]
 n 2 − g 2 m 2 arctan (gm/n) + gnm log n 2+g 2 m 2 /g, n ≠ 0 and m ≠ 0
f zR(n,m,g) = 
[ ]
 − g m (sgn m )π / 2 /g , n = 0 and / or m = 0
2 2
Preprint DRAFT 2011

Symmetry in these equations corresponds to the spatial symmetry Ω . Also, there are

dipole symmetry in the 2D space Nˆ (n, m, g ) = − Nˆ (m, n, g ) for n m ≠ 0 , and expression the same

as the local Lorentz field N xx (0,0, g ) + N zz (0,0, g ) = −4π .

Let us consider case when surfaces of the sample correspond to z = z1 / δ 0 and z = z 2 / δ 0 .

Boundaries x = x1 / δ 0 and x = x2 / δ 0 are only the limit of the simulation area. They are not

real limits of the sample. The field of the surface charges on these limits should be removed:

I J
h ijm = ∑ ∑ N
k =1 l =1
{ [(x z )kI x (
ˆ (k − i, l − j , g ) − K R aˆ + K R bˆ δ + K L aˆ + K L bˆ δ α ,
z k1 ) ]}
kl

1 0 0 0
where â ≡   , b̂ ≡   .
 0 0 0 1

Independence of the simulated equilibrium magnetization distribution from the finite

element shape and their orientations is the important problem. Aharoni first proposed some

necessary conditions to prove conformity between the method discrete model and

micromagnetic approach. In describable case that conditions is given by:

G1 = G 2 = G3 = G − ∫∫ ν r (αα)d 2 r , where

[ ( ) ( ) ]
G1 = − ∫∫ K 1 α 4y + α 4z + α 2y α 2z + A ∇ 2 α x / α x d 2r ,

[ ( ) ( ) ]
G 2 = − ∫∫ K 1 α 4x + α 4z + α 2x α 2z + A ∇ 2 α y / α y d 2 r ,

[ ( ) ( ) ]
G3 = − ∫∫ K 1 α 4x + α 4y + α 2x α 2y + A ∇ 2 α z / α z d 2 r .

3. Magnetization distribution symmetry

A symmetry of initial magnetization distribution determines a final simulated magnetization

distribution. The last can be equilibrium or metastable. Therefore problem of magnetization

distribution symmetry is important. Here and hereinafter we talk about non-magnetic

symmetry classes. It means that magnetization is considered as polar vector as well.


Preprint DRAFT 2011

Let function α( x < ∆, z ≤ h f / 2 ) specifies magnetization distribution in isolated DW

volume in film with thickness h f , where ∆ → ∞ and X axis is parallel to film surface. Than

the DW boundary conditions at infinity are ( )


α x > ∆, z ≤ h f / 2 = m 1 and

( )
α x < − ∆, z ≤ h f / 2 = m 2 , where m1 and m 2 are the unit vectors along magnetization vector

in neighboring domains volumes. If external field is absent then ever isolated DW

magnetization distribution symmetry correspond to boundary condition symmetry. The first

reason of magnetization distribution symmetry is a parallel orientation of film surface planes

(symmetry transformation group S SF ). The second reason is a mutual orientation of m1 , m 2

and direction Y along which α is a constant vector (group S D ). For describing groups S SF , S D

 χˆ R 
let’s define a χ =   such as the transformation transferring б vector from point r to other
 χˆ M 

position and changing its direction: α(χˆ R r ) = χˆ M α(r ) . The group S SF consist of one

 2  
transformation that moving magnetization from one to other surface plane: S SF =  z  ,
1   

where 2 z is a reflection in plane perpendicular e z and 1 is a turn around the rotary onefold

axis. The group S D consist of transformation that relate among themselves m1 and/or

1 1   2 x   1  2 x  
m 2 vectors: S D =  ,  ,    or S D =  ,    for 180°-DW or non-180°-DW
1  2 x   2 y   1  2 y  

respectively, where 2 x and 2 y are reflections in planes those are perpendicular accordingly to

e x and e y .

All possible combinations of the transformations of a group S SF ∪ S D form a group S′′ of

magnetization distribution symmetry transformation in ever DW volume. At (m 2 + m1 ) = 0

(i.e. at 180°-DW) the group S′′ looks like:


Preprint DRAFT 2011

1 1  1  2  1  2 x 


S1′′ =  ,   ⊗  ,  z  ⊗  ,   ,
1  2 x  1 1  1  2 y 

where “⊗” means an sequel fulfilment of symmetry operations:

 χˆ R 1   χˆ R 2   χˆ R 1 ⋅ χˆ R 2 
     
 χˆ  ⊗  χˆ  =  χˆ ⋅ χˆ  . After simplification it is obtained a multitude:
 M1   M 2   M1 M 2 

1  2  1   2   2 x   1   2   1 
S1′′ =  ,  z ,  ,  ,  ,  ,  x ,   , where 1 = 2 x ⋅ 2 z and 2 z = 2 x ⋅ 2 y . By
z
 
     x   2 x   2 y   2 y   2 z   2 z 
1 1 2

means of symmetry reduction it is possible to write group for non-180°-DW):

1  2   2 x   1 
S ′2′ =  ,  z ,  ,   . Subgroups S DW of groups S1′′ or S′2′ describe magnetization
1 1   2 y   2 y 

distribution symmetry of all isolated DW with 2α =180° or 2α ≠180° respectively (tabl.1).


Preprint DRAFT 2011

Table 1. Symmetry (non-magnetic classes) of the 2D magnetic domain wall

S DW DW

1  2  1   2   2 x   1   2   1   Classical
 
 ,  ,  ,  ,  ,  ,  ,   
z z x

1 1   2 x   2 x   2 y   2 y   2 z   2 z  
1D

Bloch

180°-DW

1  2   2 x   1  
 ,  ,  ,   
z
1
     2 y   2 y  
1 1D-Brown and LaBonte DW

1  2   2 x   1 
 ,  ,  ,  
z
1
   2 x   2 y   2 y  Symmetrical LaBonte DW

1  2 z  
 ,   
1  2 x   Asymmetrical LaBonte DW

1  1 
 ,  
1  2 y  Asymmetrical Néel DW at Hubert

model

1  2   2 x   1 
An availability of all elements  ,  z ,  ,   in groups S1′′ and S′2′ is a necessary
1 1   2 y   2 y 

1  2 x   1   2   1 
condition of plane DW. An absence of all elements  ,  ,  ,  x ,   is sufficient
1  2 y   2 y   2 z   2 z 

condition of asymmetric DW.

For more percise and fast modelling an initial magnetization distribution should have

symmetry identical as a final.


Preprint DRAFT 2011

4. Results and discussion

Let we choose orts: e x || [100] , e y || [010] and e z || [001] . The (001) plate width is h f and

modeling area is: x1 = − L / 2 , x2 = L / 2 , z1 = − h f / 2 , z 2 = h f / 2 , where L = 10h f . The

direction of the α will be determined by the polar θ and azimuth angle ϕ , which are counted

from the e x and e z respectively.

Let the 90° DW divides two domains with magnetizations (θ1 = π / 4, ϕ1 = −π / 2) and

(θ 2 = π / 4, ϕ 2 = −π / 2) . The angular tolerance is ε tol = 10 −7 for our simulation. The tolerance

for the comparison of the G1 , G2 and G3 in the above-mentioned necessary conditions is

given by the 1%. The magnetization distributions in the 90° DW at different widths h f for the

parameter Q −1 = 4.62 (ferrite garnet at the temperature of the liquid Helium) are given at the

fig.1
Preprint DRAFT 2011

θ, ϕ, deg. ϕ θ, ϕ, deg.
80
100 ϕ
40 θ 50 θ
0 0

-40 -50

-80 -100

-15 -10 -5 0 5 10 15 -40 -20 0 20 40


x/δ0 x/δ0

a) b)
150
150
θ, ϕ, deg. θ, ϕ, deg.
ϕ ϕ
75 75
θ θ
0 0

-75 -75

-150 -150
-80 -40 0 40 80 -80 -40 0 40 80
x/δ0 x/δ0
c) d)
200 θ, ϕ, deg.

100 ϕ
θ
0

-100

-200
-150 -75 0 75 150
x/δ0
e)
Fig. 1. The equilibrium magnetization distributions in the volume of the 90° DW in the (001)
plate with the 2πM s / K1 = −4.62 for widths h f : a - 2.46δ 0 ; b- 7.38δ 0 ; c- 12.30δ 0 ; d-
2

14.76δ 0 ; e- 22.14δ 0 . Dashed, solid and chain-line are given for the α(x, z = − h f / 4 ) ,
( ) ( )
α x, z = h f / 2 and α x, z = h f / 4 respectively.
Preprint DRAFT 2011

In case of the growth of the plate width from the δ 0 to the 20 δ 0 transition from the Neel

DWs to Bloch DW is processing at the same time with the nucleations of the 71° DW based

domain structure with ∆α = α2 − α1 perpendicular to the (001) plate. This structure appears in

the volume of the initial DW.

The energy density of the DW is counted from the same value in the domain. In case of

the value 2πM s / K1 < −0.5 in the domains we have g A = g m = 0 and g K = K1 / 4 or


2

g K = πQ / 2 in the dimensionless form. So, we should choose constant β = πQ / 2 .The energy

density of the DW is σ = G δ 0 / h f . The components of the normalized total energies are the

( 2 2
) ( 2 2
)
following: GA = GA / M s δ 0 , Gm = Gm / M s δ 0 , GK = GK / M s δ 0( 2 2
) and σ A = G Aδ 0 / h f ,

σ m = Gmδ 0 / h f , σ K = GK δ 0 / h f . The dependences for these values are given on the fig. 2
Preprint DRAFT 2011

σA,σm,σK, arb.units
G,GA,Gm,GK arb.units
20 σA
G 0.9
16
GA 0.6
12
σm
8 0.3
Gm
4
0.0 σk
0 GK
3 6 9 12 15 3 6 9 12 15
hf /δ0, arb.units hf /δ0, arb.units

a) b)

σ, arb.units Gm/GA, arb.units


1.0
2.1
0.8
1.8

1.5 0.6

1.2 0.4

0.9 0.2
3 6 9 12 15 3 6 9 12 15
hf /δ0, arb.units hf /δ0, arb.units

c) d)

Fig 2. The energies of the equilibrium 90° DW as function of the (001) plate width.

At the width of the (001) plate higher than 13.5δ 0 , the energy GK < 0 (fig. 2 a,b). Existing

of the negative DW energy component mean that domain magnetization direction change due

to demagnetization field. In work (S.A. D’yachenko, et al., Influence of the Demagnetizing

Field on the Structure of a Bloch Wall in a (001) Plate of a Magnetically Ordered Cubic Crystal,

Physics of Solid State, 50, 1, 32-42 (2008)) it was shown that in such case model of the

isolated plane Bloch DWs lead to the infinite width of the DW. In the mentioned work, it was

called as “decay” of the DW into set of new DWs.


Preprint DRAFT 2011

If magnetization distribution has an oscillating nature then DW width (Lilley definition)

became large or limitless in extreme case.

1  1 
The magnetization distribution symmetry of this DW is S DW =  ,   . The equilibrium
1  2 y 

magnetic “charge” (magnetization divergence) distribution demonstrates on the fig. 3

Fig. 3. Example of the magnetization divergence distribution in the (001) plate cross-

section. Black corresponds to the north poles and white corresponds to the south poles.

Acknowledgements

I would like to express my sincere gratitude to my supervisor Prof. V.F. Kovalenko for

him outstanding guidance. Micromagnetic simulation has been performed on the cluster of the

Information and Computer Center of Kyiv National Taras Shevchenko University.

Das könnte Ihnen auch gefallen