Reduced models for domain walls in soft ferromagnetic films Lukas - - PowerPoint PPT Presentation

reduced models for domain walls in soft ferromagnetic
SMART_READER_LITE
LIVE PREVIEW

Reduced models for domain walls in soft ferromagnetic films Lukas - - PowerPoint PPT Presentation

Reduced models for domain walls in soft ferromagnetic films Lukas Dring Conference on Nonlinearity, Transport, Physics, and Patterns Fields Institute, Toronto 06/10/14 Max Planck Institute for Mathematics in the Sciences Modelling


slide-1
SLIDE 1

Reduced models for domain walls in soft ferromagnetic films

Lukas Döring Conference on Nonlinearity, Transport, Physics, and Patterns Fields Institute, Toronto 06/10/14

in the Sciences

Mathematics

Max Planck Institute for

slide-2
SLIDE 2

Modelling ferromagnetic thin films

Ω ⊂ R3 sample m: Ω → S2 magnetization “Elementary magnets”

  • Unit-length vector field

1

slide-3
SLIDE 3

Magnetization patterns in thin-film ferromagnets

  • R. Schäfer
  • J. Steiner, F. Otto

Magnetization patterns in Permalloy films Numerical simulation of domain walls

2

slide-4
SLIDE 4

Landau-Lifshitz (free) energy

Observed patterns: Local minimizers m: Ω ⊂ R3 → S2 of E(m) = d 2

|∇m|2dx Exchange energy +

  • R3|hstr|2dx

Stray-field energy

  • ∇ · (hstr + 1Ωm) = 0

∇ × hstr = 0 +Q

1 − (e · m)2dx Anisotropy energy for e ∈ S2, Q ≪ 1 −2

hext · m dx Zeeman energy

Well-accepted Non-convex Non-local

3

slide-5
SLIDE 5

Landau-Lifshitz (free) energy

Observed patterns: Local minimizers m: Ω ⊂ R3 → S2 of E(m) = d 2

|∇m|2dx Exchange energy +

  • R3|hstr|2dx

Stray-field energy

  • ∇ · (hstr + 1Ωm) = 0

∇ × hstr = 0 +Q

1 − (e · m)2dx Anisotropy energy for e ∈ S2, Q ≪ 1 −2

hext · m dx Zeeman energy

Well-accepted Non-convex Non-local

3

slide-6
SLIDE 6

Landau-Lifshitz (free) energy

Observed patterns: Local minimizers m: Ω ⊂ R3 → S2 of E(m) = d 2

|∇m|2dx Exchange energy +

  • R3|hstr|2dx

Stray-field energy

  • ∇ · (hstr + 1Ωm) = 0

∇ × hstr = 0 +Q

1 − (e · m)2dx Anisotropy energy for e ∈ S2, Q ≪ 1 −2

hext · m dx Zeeman energy

Well-accepted Non-convex Non-local

3

slide-7
SLIDE 7

Outline

Single wall in infinitely extended film Periodic domain pattern with interacting wall tails

slide-8
SLIDE 8

Wall patterns on cross-section of film

}

Film thickness 2t Material anisotropy “Easy axis” e External magnetic field hext Wall angle α

x1 x3 x2 m1 m2

e hext 2α

Anisotropy Q and external field hext determine wall angle α. Wall angle α and film thickness t determine wall type.

4

slide-9
SLIDE 9

Three wall types

Symmetric Néel wall Asymmetric Néel wall Asymmetric Bloch wall

m1 m3 m1 m3 m1 m3

m2

− sin α sin α

5

slide-10
SLIDE 10

Three wall types

Symmetric Néel wall Asymmetric Néel wall Asymmetric Bloch wall

m1 m3 m1 m3 m1 m3

m2

− sin α sin α

core

∼ d2

t

logarithmic tails

∼ t

Q

m1 m2 m1 ln x1 x1 ≈ 1

5

slide-11
SLIDE 11

Three wall types

Symmetric Néel wall Asymmetric Néel wall Asymmetric Bloch wall

m1 m3 m1 m3 m1 m3

m2

− sin α sin α

Aim: Understand transitions between wall types for Q ≪ 1

5

slide-12
SLIDE 12

Three wall types

Symmetric Néel wall Asymmetric Néel wall Asymmetric Bloch wall

m1 m3 m1 m3 m1 m3

m2

− sin α sin α

Wall angle Film thickness Wall types in Permalloy films Symmetric Néel Asymmetric Néel Cross-tie

  • Asymm. Bloch

Hubert, Schäfer: Magnetic Domains, Springer, 1998

Aim: Understand transitions between wall types for Q ≪ 1

5

slide-13
SLIDE 13

The critical regime: Optimal mix ...

min

m wall of angle π

2

E2D(m)

Otto, ’02

  • t2 ln−1

t2 d2Q,

if t2

d2 ≪ ln 1 Q,

d 2, if t2

d2 ≫ ln 1 Q.

What happens in critical regime:

t2 d2 = λ ln 1 Q?

Optimal wall profile for angle α =

m = cos θ

− sin θ

  • m

= cos θ

+ sin θ

  • asymm. “2 1

2-d” core

θ

+

m1 = cos θ m1 = cos α

long-range 1-d tails

α − θ Optimal mix:

Quantification of optimal mix difficult to access by brute-force numerics.

...of core and tails

6

slide-14
SLIDE 14

Outline

Single wall in infinitely extended film Periodic domain pattern with interacting wall tails

slide-15
SLIDE 15

Expected behavior: Large domain width...

H = 0

x1 x2

x1 ¯ m1 cos θ = H ∼ t ∼ t

core core equilibrium

w

x1 x3

...similar to one-wall case

7

slide-16
SLIDE 16

Expected behavior: Large domain width...

H = 0.2

x1 x2

x1 ¯ m1 cos θ = H ∼ t ∼ t

core core equilibrium

w

x1 x3

...similar to one-wall case

7

slide-17
SLIDE 17

Expected behavior: Large domain width...

H = 0.4

x1 x2

x1 ¯ m1 H cos θ ∼ t ∼ t

core core tail tail

∼ t

Q

∼ t

Q

equilibrium

w

x1 x3

...similar to one-wall case

7

slide-18
SLIDE 18

Expected behavior: Small domain width...

H = 0

x1 x2

x1 ¯ m1 cos θ = H ∼ t ∼ t

core core equilibrium equilibrium equilibrium

w w w

x1 x3

...leads to coalescing tails

8

slide-19
SLIDE 19

Expected behavior: Small domain width...

H = 0.2

x1 x2

x1 ¯ m1 cos θ = H ∼ t ∼ t

core core equilibrium equilibrium equilibrium

w w w

x1 x3

...leads to coalescing tails

8

slide-20
SLIDE 20

Expected behavior: Small domain width...

H = 0.4

x1 x2

x1 ¯ m1 H cos θ ∼ t ∼ t

core core tails tails tails

t

Q

t

Q

t

Q

w w w

x1 x3

...leads to coalescing tails

8

slide-21
SLIDE 21

Strongly hysteretic transition between asym. walls...

Elongated CoFeB elements 2t = 120nm, Q = 1.55 · 10−3, d = 3.86 ± 0.3nm. Origin of large jump in hard-axis magnetization?

Domain width w

x2 x1

  • C. Hengst, IFW Dresden

...due to interacting tails?

9

slide-22
SLIDE 22

Just a few building blocks...

x1 ¯ m1 cos α cos θ ∼ t ∼ t

core core tail tail

wtails wtails

domain

w

x1 x3

E2D = Exchange energy + Stray-field energy + Bulk energy ≈ d 2

  • |∇mcore

θ

|2dx + 2t2 | d

dx1|

1 2mtails

1

  • 2dx1

+ 2Qwt −

  • (mtails

1

− H)2dx1, with ( t

d )2 = λ ln 1 Q.

Interesting regime: Qwt = κλ d 2; optimal wtails = w

2 .

...combined in an optimal way

10

slide-23
SLIDE 23

Just a few building blocks...

x1 ¯ m1 cos α cos θ ∼ t ∼ t

core core tail tail

wtails wtails

domain

w

x1 x3

E2D = Exchange energy + Stray-field energy + Bulk energy ≈ d 2 |∇mcore

θ

|2dx + 2πλ (cos θ − cos α)2 + 2 κλ (cos α − H)2 , with ( t

d )2 = λ ln 1 Q.

Interesting regime:

w t = κ Q ln 1

Q ; optimal wtails = w

2 .

...combined in an optimal way

10

slide-24
SLIDE 24

Just a few building blocks...

x1 ¯ m1 cos α cos θ ∼ t ∼ t

core core tail tail

wtails wtails

domain

w

x1 x3

d −2 min

m E2D(m) ≈ min θ

  • =:Easym(θ)
  • min

m stray-field free wall of angle θ

  • |∇m|2dx

+ 2πλ min

α

  • (cos θ − cos α)2 + κ

π(cos α − H)2

as Q → 0, for ( t

d )2 = λ ln 1 Q, w = κ t Q ln 1

Q .

...combined in an optimal way

10

slide-25
SLIDE 25

Reduced model for the structure of domain walls

Theorem (κ = ∞: D., Ignat, Otto; κ < ∞: D.)

There exist critical points mQ of E2D, such that for Q → 0, λ the relative film thickness, κ the relative domain width: d −2E2D(mQ) ≈ min

θ∈[0, π

2 ]

  • Easym(θ) + 2πλ

κ π+κ(cos θ − H)2

and −

  • domain

m1,Q dx ≈ cos αopt = H +

π π+κ(cos θopt − H). ◮ Proof via Γ-conv. (minimize E2D over periodic m). ◮ Compactness requires “shifting argument” to ensure that

{mQ}Q converges to a domain wall.

11

slide-26
SLIDE 26

Stray-field free core: Néel and Bloch...

Easym(θ) := min

|∇m|2dx

  • m: Ω→S2 has wall angle θ,

with ∇·m′=0 in Ω, m3=0 on ∂Ω

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

5 10 15 20 25 30 π/4 π/2

  • Asym. Néel≈
  • Asym. Bloch

D., Ignat

4π sin2 θ + 148

35 π sin4 θ

?

wall angle θ Easym(θ)

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

...two topologically distinct wall types

deg=±1 deg = 0

m1 m3

12

slide-27
SLIDE 27

Comparison of theory and experiments

Co40Fe40B20 films (lateral width 60µm) with parameters thickness/nm 102 153 212 Q/10−3 1.36 0.93 1.16 µ0Ms = 1.48T (measured in a single film of small thickness) d = 3.86nm (from Conca et al., J. Appl. Phys., 2013) For 2t = 102nm:

0.1 0.2 0.3 0.4 0.5 0.6 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Magnetization Reduced external field w=7.5µm w=8.2µm w=9.0µm w=11.6µm w=17.4µm 0.1 0.2 0.3 0.4 0.5 0.6 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Magnetization Reduced external field w=6.0µm w=7.0µm w=8.4µm w=13.6µm

Experiments: C. Hengst

13

slide-28
SLIDE 28

Comparison of theory and experiments

Co40Fe40B20 films (lateral width 60µm) with parameters thickness/nm 102 153 212 Q/10−3 1.36 0.93 1.16 µ0Ms = 1.48T (measured in a single film of small thickness) d = 3.86nm (from Conca et al., J. Appl. Phys., 2013)

a) 30 µm b) α = 0° 45° 90° 20 µm Hdem w Haα stripe axis

Experiments: C. Hengst

13

slide-29
SLIDE 29

Further questions

Transversal (in)stability and path to cross-tie wall

x2 x1

Stability of asymmetric walls

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

Comparison of critical wall angle α∗ ≈ arccos(1 − 2

λ)

(λ ≈

t2 d2 ln 1

Q ) to experiments

α θ α∗

slide-30
SLIDE 30

Further questions

Existence of stray-field free walls under degree constraint (energy of div.-free bubbles?)

  • 1
  • 0.5

0.5 1

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2 x3 x1

Thin-film numerics with realistic wall-energy density

Van den Berg, Vatvani D., Esselborn, Ferraz-Leite, Otto

LLG evolution for unwinding walls: Fast relaxation in core – slow wall motion?