Pat t erned magnet ic st ruct ures f rom f undament al micromagnet - - PowerPoint PPT Presentation

Olivier Fruchart - Laboratoire Louis Néel, Grenoble, France. Olivier Fruchart - Laboratoire Louis Néel, Grenoble, France.

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned magnet ic st ruct ures f rom f undament al micromagnet ism t o micron-scale applicat ions

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.2 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.2 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Table of cont ent

Micromagnetism (fundamental) The background Magnetostatics The fundamental issues of micromagnetism Coherent reversal Domains and walls Characteristic length scales Multidomains : theory ( ) and real life( ) Applications for ‘large’ microstructures Magnetic recording heads (general) Magnetic recording heads (alditech : tapes)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.3 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.3 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > ref erences

Micromagnetism =

Continous media theory describing the magnetization distribution inside samples

! Classical theory !Atomic structure of matter is ignored !Analytical as well as numerical approach Magnet ic domains, A. Hubert and R. Schäf er, Springer Verlag, 1998.

Practical although rigourous approach to micromagnetism. More imaging.

An int roduct ion t o t he t heory of f erromagnet ism, A. Aharoni, Clarendon Press, 2001.

A more mathematical approach. More historical (math.) concepts.

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.4 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.4 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > exchange

Exchange energy Ferromagnetic order comes from quantum mechanics Pauli exclusion principle + Electrostatic forces Spins do not ignore each other

2 1 2 , 1 ex

.S

S

J e − =

Exchange energy For ferromagnetic substances : parallel alignement is favored

! Magnetic moment, M(T), etc.

( ) ( )

2 2 ex

/ x

θ A θ A e ∂ ∂ = ∇ ≈

for 1D situation

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.5 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.5 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > magnet ocryst alline anisot ropy

Magnetocrystalline anisotropy energy

Atom nucleus (crystal structure) Electronic cloud

Spin-orbit coupling ! the energy of both spin and orbital moment depends on orientation

Series development on an angular basis:

(Derived f rom slide of A. Thiaville – CNRS/ Or say)

...

4 4 2 2 mc

+ + =

z z

m K m K e

Uniaxial

... ) (

2 2 2 2 2 2 4 mc

+ + + =

x z z y y x

m m m m m m K e

Cubic

… Anisotropy energy Alignement of magnetization is favored along given axes of the crystal Normalized magnetization components

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.6 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.6 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Zeeman energy

Zeeman energy External magnetic field

(applied by magnets, earth, etc.)

H M .

S Z

µ e − =

Zeeman energy

Analogy : a compass needle in the earth’s magnetic field

Alignement of magnetization is favored parallel to the external field

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.7 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.7 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > dipolar energy

Dipolar energy

Magnetic moments (spin or orbital) are assimilated to microscopic currents

" they create long-range dipolar fields H " What is the effect of these fields ? The dipolar energy is the Zeeman energy of the sample in the dipolar field Hd created by all its spins

d S d

. 2 1

H M

µ e − =

(per unit volume)

( )

1 2 2 1 1 2 2 1 1,2

. . 2 1 . .

H µ H µ H µ H µ

+ − = − = − = µ µ µ E

1 2 Mutual energy should be counted only once ! Local dipolar energy

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.8 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.8 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > dipolar energy

Cone of alignment 1 2

θ

      − =

) . ).( . ( 3 . 4

2 1 2 2 1 3 1,2

r µ r µ µ µ

r r π µ E

Parallel alignment is favored for Antiparallel alignment is favored for

° ≈ <

74 . 54

C

θ θ ° ≈ >

74 . 54

C

θ θ

3 / 1 ) ( cos2 =

C

θ

[ ]

θ µ µ π µ θ

2 2 1 3 1,2

cos 3 1 4 ) ( − = r E

1/ r3 decay: the dipolar interaction is long ranged Mutual energy of two magnetic dipoles :

Let us assume two magnetic dipoles with vertical direction, either ‘up’ or ‘down’ :

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.9 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.9 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > dipolar energy

Cone of alignment How to use the ‘cone of alignment’ to predict the effect of dipolar fields ? Situation 1 : M perpendicular

Most of the spins are in the antiparallel cone

" not favorable

Situation 2 : M parallel

Most of the spins are in the parallel cone

" favorable

The favored magnetization direction is along the long axis of the sample

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.10 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.10 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > magnet ost at ics laws

Electrostatics / Magnetostatics parallel

ε ρ div =

E

Electrostatics

2 3

4 ) ( ) (

PM πε P d P ρ M

PM

u E

∫∫∫

=

2 3

4 )] ( [ div ) (

PM π P d P M

PM

u M H

∫∫∫

− =

M H

div div

− = =

B

div

M H B

+ = µ

Magnetostatics ‘Magnetic charge’ ‘Electric charge’

For a finite size sample:

after integration over the entire space, a new term arises due to the magnetization discontinuity at the sample’s surface:

dS PM π P PM π P d P M

PM

∫∫ ∫∫∫

+ − =

surface s sample' 2 sample 2 3 d

4 ). ( 4 )] ( [ div ) (

n M u M H

‘Volume charges’ ‘Surface charges’

z M y M x M

z y x

∂ ∂ + ∂ ∂ + ∂ ∂ =

M

div with : The dipolar field coming from a sample can be calculated from these ‘magnetic charges’

d S d

. 2 1

H M

µ e − =

Local dipolar energy Maxwell’s equations :

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.11 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.11 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > st ray- and demagnet izing f ields S N

+ + + + +

• Stray field =

Field created

• utside the sample

Long-range : dipole-like

(I mages f rom A. Thiaville – CNRS/ Or say)

M + + + + +

• Magnetic charges

Demagnetizing field =

Field created inside the sample

(acting from the sample on itself)

+ + + + +

• S

N

d

H

Note: a free dipole aligns itself parallel to the stray field H of the magnet

Example

Let us assume a uniformly magnetized prism body :

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.12 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.12 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > demagnet izing f ields

Demagnetizing fields How to use the ‘surface charges’ model to predict the effect of dipolar fields ? Situation 1 : M perpendicular

Many surface charges : high dipolar fields

" not favorable

Situation 2 : M parallel

Few surface charges : low dipolar fields

" favorable

The favored magnetization direction is along the long axis of the sample + + + + + + + + + + + + + + + + + + + + + +

• +

+ + +

• -

« Shape anisotropy »

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.13 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.13 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > shape ef f ect

( )

2 2 2 d

2 1

z z y y x x

M N M N M N µ e + + =

, 1

≥ = + +

i z y x

N N N N

Hypothesis : Uniformly magnetized body with arbitrary shape

It can be shown that : With :

Notes and consequences :

2 S sample d max d

2 1 . 1 max

M µ τ d e V e = =

∫∫∫

The ‘shape’ energy is uniaxial This is the ‘Shape anisotropy energy’ Ni is higher along short sample directions Effective anisotropy energy:

d mc eff

K K K + =

See validity for real samples, later in the course

M

Even if M is assumed to be uniform in the system, Hd is in general not uniform, except for special shapes.

! see examples

(see analogy with magnetocrystalline…)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.14 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.14 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > shape ef f ect examples

+ + + + + + + + + + + + + + + + + + + + + +

• 

     = =

2 S d d

2 1 5 .

M µ e e

z y

d =

x

e

Infinite cylinder : + + + + + + + + + + + + + + + + + + + + + +

• Infinite thin films :

2 S d

2 1

M µ ez =

d d

= =

y x

e e

In thin films (or portions of thin films) the magnetization usually lies in the plane of the film « Shape anisotropy »

z y x

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.15 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.15 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > general equat ion

General micromagnetic equations

Must be minimized locally (Lagrange minimization)

) ( ) ( ) ( ) ( ) (

d Z mc ex

r r r r r e e e e e + + + =

Any spin interacts with all other spins in the sample: dipolar term is non-linear and non-local.

No general solution. Only extremely simple problems can be solved analytically

(historical approach) Since about 10 years: ‘small-scale’ problems (<1µm) have become tractable with computers " Numerical micromagnetics

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.16 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.16 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > coherent reversal

Coherent reversal

• L. Néel, Compt e rendu Acad. Sciences 224, 1550 (1947)
• E. C. St oner and E. P. Wohlf art h,
• Phil. Tr ans. Royal. Soc. London A240, 599 (1948)

I EEE Trans. Magn. 27(4), 3469 (1991) : reprint

[ ]

) cos( sin

H ext S 2 eff tot

θ θ H M µ θ K V E − − =

θH θ M

H

Approximation:

Cte = = M

r m

) (

(Extremely strong !)

d mc eff

K K K + =

Question to answer:

What happens to a piece of ferromagnet when an external field is applied antiparallel to its magnetization ? (The sample is fixed in position and orientation in the external field : different from a compass needle)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.17 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.17 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > coherent reversal

• 90°

0° 90° 180° 270°

H = 0

H = 0.2 Ha H = 0.7 Ha H = Ha

H

[ ]

) cos( sin

H ext S 2 eff tot

θ θ H M µ θ K V E − − =

30 60 90 120 150 180 210 240 270 300 330

E a s y a x i s E a s y a x i s Hard axis Hard axis

( )

2 / 3 H 3 / 2 H 3 / 2 r

cos sin 1

θ θ H + =

) ( H

r θ

H

Predicted switching field: Stoner-Wohlfarth ‘astroïd’ ‘Hysteresis loops’

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.18 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.18 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > coherent reversal

Switching field (polar plot)

• 1

1

• 1.5
• 1
• 0.5

0.5 1 1.5

M h

0° 10° 30° 45° 70° 90°

Hysteresis loop

… but very different jump height !

0.2 0.4 0.6 0.8 1 0° 30° 60° 90° 120° 210° 240° 270° 300° 330°

hsw

H a r d H a r d E a s y E a s y

Easy and hard axis have identical reversal fields…

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.19 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.19 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > coherent reversal in clust ers

Coherent reversal model dates back to 1947, but the first experimental proof came in 1997!

DPM, CNRS, Lyon, France : LASER vaporization and inert gas condensation source

• M. J amet , V. Dupuis, M. Negrier, J . Tuaillon, A. P

erez

Example of experimental evidence of coherent reversal

HRTEM along a [110] direction fcc - structure, faceting

! Tiny, model system

• M. J amet et al., Phys. Rev. Let t ., 86, 4676 (2001)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.20 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.20 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > coherent reversal in clust ers

Experimental evidence for coherent reversal

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3 µ0 Hz(T) µ0 Hy(T)

T B - 14 K 0.04K 1K 2K 4K 8K 12K

Experiment

• 1.5
• 1
• 0.5

0.5 1 1.5

• 1.5
• 1
• 0.5

0.5 1 1.5 Hz Hy

E0 0.9 E 0 0.8 E 0 0.7 E 0 0.6 E 0 0.5 E 0 0.4 E 0 0.3 E 0 0.2 E 0 0.1 E 0

Theory

Back to anisotropy well

! Good agreement with Stoner-Wohlfarth model

Agreement was found in the late 90’s, only in tiny samples : not relevant for ‘real’ systems

(see temperature dependance later in the lectures)

• M. J amet et al., Phys. Rev. Let t ., 86, 4676 (2001)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.21 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.21 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > domains in real samples

In most samples there are magnetic domains

++++

• Note magnetic poles

annihilation ! In-plane domains in FeSi(100) crystal – Kerr microscopy Do domains contradict ferromagnetic theory (exchange) ?

What occurs at the boundary between two domains ?

! How domains will influenced by microscale structures ?

I mages: book by Hubert & Schäf er

Perpendicular M component In-plane M component

• n Co(1000) crystal – SEMPA

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.22 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.22 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Bloch wall

Bloch domain wall profile calculated by variational technique

( )

, sin /

2 2

> + = K θ K dx θ d A e

exchange anisotropy

J/m

3

J/m

?

= θ π θ =

Variational calculation (Lagrange equation) :

      ∂ ∂ = ∂ ∂ dx θ d e η d d θ e

2 2

2 cos sin 2

x θ A θ θ K ∂ ∂ =               = K A x x θ

/ exp Atan 2 ) (

• 10.0
• 7.5
• 5.0
• 2.5

0.0 2.5 5.0 7.5 10.0 0.00 0.25 0.50 0.75 1.00

Asymptotic width:

K A π λ

/

B =

(Bloch wall width) (Bloch wall profile) Dimensional analysis: exchange against anisotropy

nm 1

B =

λ

nm 100

B =

λ

Hard (magnets) Soft

See again magnetization reversal What happens in the middle ? Try to develop a basic model to describe the transition region.

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.23 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.23 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > exchange lengt h

Competition of dipolar and exchange energy

( )

2 ex

θ A e ∇ =

J/m

3

J/m

Exchange energy

2 S sample d max d

2 1 . 1 max

M µ τ d e V e = =

∫∫∫

Dipolar energy

Determine a length scale characteristic of dipolar/ exchange competition

) /( 2 ) (

2 S ex

M µ A π λ =

(Exchange length) Dimensional analysis: exchange against dipolar Situation 2 : M parallel

Few surface charges : low dipolar fields " favorable

+ + + +

• -

D

If

ex

10λ

D ≥

Significant magnetization deviations appear at the cylinder length

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.24 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.24 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > lengt h scales

What if dipolar and anisotropy compete

θ K e

2 an

sin

=

3

J/m

Anisotropy

3

J/m

2 S sample d max d

2 1 . 1 max

M µ τ d e V e = =

∫∫∫

Dipolar energy Quality factor :

2 S

/ 2

M µ K Q =

1

ff Q

1

pp Q

Hard (permanent magnet)

Anisotropy dominates over dipolar

B

λ

is the most relevant length scale

Soft material

Dipolar dominates over anisotropy

ex

λ

is the most relevant length scale

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.25 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.25 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > use of lengt h scales

Use of characteristic magnetic lengthscales : guess qualitatively the system’s behavior.

Is the magnetization in-plane or out-of-plane ? Do we expect some domains ? Does the size of the system play a role ? Is the system hard or soft ? Zeeman energy was not discussed in the previous slides

" more definitions of length scales

Dimensional approach only: Exact numerical values depend on the exact sample geometry All four energies may play a role simultaneously

Real life is more complicated : Magnetic imaging, analytical and numerical calculation have to be used to unravel the complexity of micromagnetics

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.26 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.26 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: t heory

div

=

M

.

=

n M

• H. A. M. Van den Berg, J . Magn. Magn. Mat er. 44, 207 (1984)

Hypotheses: Infinitely soft material (K=0)

2D geometry (neglect thickness) Size >> all magnetic length scales (wall width) Zero external magnetic field (no surface charges)

Z =

e

mc =

e

ex →

 e

d =

e

« Flux closure »

Looking for a solution with : (no volume charges)

EXAMPLES See walls: charge free See volume: charge free

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.27 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.27 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: t heory

Extension for non-zero field :

The domains with magnetization parallel to the applied field are favored

• P. Bryant et al., Appl. Phys. Let t . 54, 78 (1989)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.28 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.28 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: experiment s

I n the f ollowing, many pictures taken f rom Hubert’s book

Zero field : agreement with Van den Berg’s model Material : Ni80Fe20 ‘Permalloy’, Py.

View details

Longitudinal applied field The domains with magnetization parallel to the applied field are favored

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.29 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.29 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: experiment s det ail

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.30 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.30 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: demagnet izat ion process

First experiment Second experiment

Oscillating demagnetizing field parallel to hard axis Oscillating demagnetizing field parallel to easy axis

Questions:

Why more fragmentation with H applied along hard axis ? Compatibility with Van den Berg’s model ? Why leaf and ellipsoid not affected ?

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.31 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.31 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: demagnet izat ion process

Compatibility with Van den Berg’s model ? YES

Artificial separation # Still compatible with model

div

=

M

.

=

n M

(Analogy with fluid dynamics)

Why more fragmentation with H applied along hard axis ? + + + + + + + +

• Many magnetic poles at saturation :

spins start to rotate at many places simultaneously

! Fragmentation of domains

More poles

Why leaf and ellipsoid not affected ?

Not completely clear !

Experimental finding : pointed ends stabilize (‘pin’) magnetization

Many poles Few poles Local energy minimum: magnetization is pinned

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.32 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.32 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Van den Berg model: weak anisot ropy

Large dots

"many degres of freedom "many possible states "history is important "even slight perturbations

can influence the dot (anisotropy, defects, etc.). Easy axis of weak magnetocrystalline anisotropy Easy axis of weak magnetocrystalline anisotropy

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.33 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.33 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Applicat ions f or well-above-micron-scale st ruct ures

Such structures have been studied for decades.

History

As seen above, possibility to have complex and unpredictable magnetization states:

" Great care must be taken for the design of the dot:

Ground state ? Magnetization reversal and field response ? Defects ?

Sensitivity.

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.34 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.34 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads

Longitudinal recording media density (Hard disks)

With in-plane magnetization

court esy of D.Weller – Seagat e,

• N. Dempsey, LLN-CNRS

June 2001

0.1 1 10 100 1000 10000

~ 1 Tbit/in

2

50-100 Gbit/in

2

Seagate IBM

Self Organized Magnetic Arrays / Patterned Media / Perpendicular Recording

50-100 Tbit/in

2 Single Particle Superparamagnetic Limit

LABORATORY DEMOS Hitachi Fujitsu Seagate IBM

Historical 60% CGR line

(Gbit/in

2)

Availability Year Density 1990 1995 2000 2005 2010

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.35 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.35 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads

Top view Longitudinal recording : principle Tens of microns read/ write head Media cross-section

View of state-of-the-art hard disk media 35Gbit/in2

I mages court esy of J .-P. Nozières, LLN-CNRS,

• S. Wang, St anf ord Universit y

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.36 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.36 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads

Domain pattern in a recording head Explain why a slight perpendicular magnetocrystalline anisotropy is used

(interesting during read-out) The yokes are nearly saturated during the writing process… Less noise (‘pop-corn’) Linear response (non-pinned spins)

This type of heads if not anymore used for read-out in HDD

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.37 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.37 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads (t apes)

Jean- Baptiste Albertini

Juin 2001

Top View of HSS Hea d

40-turns copper sole noid marks for gap depth m easu rem en t and micromachining mag neti c circuit tape side

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.38 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.38 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads (t apes)

Jean- Bap tiste Albertini

Juin 2001

Poles and Gap with Azimuth

sili con sup erstrate mag neti c quad rila yer silic on subst rate gap with 20° azi muth VIEW FROM TAPE SIDE silicon dioxide

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.39 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.39 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads (t apes) Jean- Baptiste Albertini

Juin 2001

Solenoid Coil : Electrical Connections

cross section

• il

10 µm 8 µm copp er 2 µm SiO 2

TOP VIEW

{

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.40 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.40 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Micromagnet ism > Magnet ic recording heads (t apes)

Jean- Baptiste Albertini

Juin 2001

High performance

Extrem e m iniatu rizati on Narrow track width Goo d un iformity Mass production Hig h yiel d

New features

GMR Har d material Integrate d azim uth Multiple heads in a chip

Prot ec ted by 16 wo rldwid e patent s

Silicon Tec hnolog y : 2000 heads

/wa fer

Olivier Fruchart - Laboratoire Louis Néel, Grenoble, France. Olivier Fruchart - Laboratoire Louis Néel, Grenoble, France.

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned magnet ic st ruct ures : current and prospect ive applicat ions

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.2 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.2 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t able of cont ent

From flux-closure to near single-domain structures (fundamental)

Applications of near single-domain structures (list)

Magnetic sensors Shape anisotropy (fundamental) Magnetic Random Access Memory (MRAM) Challenges of dot switching control (fundamental) Magnetic recording media Magnetic computing Conclusion on patterned magnetic structures

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.3 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.3 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es

Macroscopic and microscopic features

Macroscopic : Flux-closure configuration Microscopic : vortex, not domain wall

• P. Bryant et al., Appl. Phys. Let t . 54, 78 (1989)

Prediction for infinitely large disk :

• M. Schneider et al., Appl.Phys.Let t .77(18), 2909 (2000)

15nm thick permalloy disks ; Lorentz microscopy

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.4 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.4 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es >

size ef f ect

Macroscopic and microscopic features

• M. Schneider et al., Appl.Phys.Let t .77(18),

2909 (2000)

15nm thick permalloy disks

Do we expect the vortex state down to very small size ?

NO

Vortex state should disappear at least below 5-10 times λex ! 25-50nm diameter

Exchange and dipolar compete

! λex is a relevant length scale

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.5 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.5 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es >

size and t hickness ef f ect s

5 10 15 20 100 200 300 400 500

Thickness (nm)

Vortex state Single domain state

Single domain state becomes favorable well above λex Dipolar energy is higher for thicker dots

(think in terms of magnetic charges)

How do reversal loops look like ?

300nm/ 10nm 100nm/ 10nm

‘Phase diagram’ of single domain versus flux-closure

R.P. Cowburn, J .Phys.D:Appl.Phys.33, R1–R16 (2000)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.6 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.6 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es >

3D dot s

300K

Applied field µ0H (T) Magnetization M/MS

Numerical calculations

(J.C. Toussaint)

Ilot 3D compact

500 nm

Hsat // [001]

MFM measurements (Y. Samson, DRFMC)

• 1.5 -1.0 -0.5 0.0

0.5 1.0 1.5

• 1.0
• 0.5

0.0 0.5 1.0 // [001] // [1-10] // [110]

[001] [1-10] (110)

L~550nm, w~350nm, h~65nm

Magnetometry, microscopy, and calculations in a sub-micron-size 3D dot

P.-O.J ubert et al., Phys.Rev.B 64, 115419 (2001)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.7 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.7 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es >

3D dot s

Length = 550nm 3D dots Length = 550nm

Van den Berg’s model is still valid although size is < micron Reason : High thickness " High dipolar fields " flux closure more favorable

The size is not the only factor determining the single domain/ flux closure state. Anisotropy (even weak) and thickness are also key parameters

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.8 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.8 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > t owards near single-domain st at es >

recipes

What recipes can we use to be sure to fabricate single-domain dots ? Moderate thickness Moderate lateral size

(Critical size can be increased using shape dipolar anisotropy, see later)

Soft materials Moderate thickness Moderate magnetocrystalline anisotropy (even for Q<<1) Arbitrary size Semi-soft materials Note from fundamental studies:

If devices require only one dot per chip is needed, then micron-size or well-above micron dimensions are sufficient

# High resolution lithography not required # lower cost.

These devices would make use of: moderate thickness AND moderate magnetocrystalline anisotropy (or other tricks)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.9 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.9 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

applicat ions

Applications of micron-size single-domain dots

(current and prospects)

Magnetic field sensors

Resistance depends on magnetization orientation.

Do not need magnetization reversal: use only susceptibility

# ‘easy’ to achieve low noise

Magnetic memories (MRAM)

Resistance depends on magnetization orientation.

# non volatile

Magnetic recording Magnetic computing

Patterned media with one bit per dot ?

# extremely high density

Replace wires by chains of magnetic dots

# extremely low consumption

For all these applications:

Thin films and elements $ high integration $ new devices and low prices

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.10 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.10 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet ic sensor s

Some physical effects can be used for field sensing… Applications: Current sensing

Current

Conductor

Precise, fast, robust.

• Hard disks heads
• Tape heads
• Magnetic sensors: position, rotation, etc.

magnetic sensor sensitive axis field time left-to-right field time right-to-left

Direction sensing for vehicles driving over magnetic sensor.

Movement detection

The earth’s magnetic field can be approxi- mated by a dipole field

Positioning

I mages f rom Honeywell

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.11 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.11 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet ic sensor s

Example of sensors : cars

Slide J . Bangert (Siemens – Erlangen) f or Workshop in Aspet (France, 2001)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.12 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.12 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet ic sensor s >

AMR

Field sensing effect : anisotropic magnetoresistance Spin-orbit coupling

• 100
• 80
• 60
• 40
• 20

20 40 60 80 100 45 135 180 225 270 315 360 90 Output (%Full Scale) Direction (degree) X Y

Design: very simple (1 single-domain structure) Magnitude: ~1%

Electrical current

Two crossed sensors for 2D direction sensing

! electronic compass

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.13 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.13 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet ic sensor s >

GMR

Giant Magneto-Resistance (GMR)

Design: more complex: multilayers + small size (for high resistance) Magnitude: <20%

Currently in use in hard disks heads

A.Bart hélémy et al.;Handbook of magnet ic Mat erials, vol.12, 1, Ed. K.H.J .Bushow (1999), Elsevier. Slide A. Bart hélémy (CNRS-Thalès, France)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.14 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.14 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet ic sensor s >

TMR

Tunneling Magneto-Resistance (TMR)

Co NiFe

eV

E E

F F

F1 F2 Insulator

Tunneling effect Design: more complex: multilayers + high insulator resistance (electronics !) Magnitude: <40% Not in commercial use at the moment

J .Moodera et al.; Ann. Rev. Mat . Sci. 29, 381 (1999)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.15 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.15 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

magnet izat ion reversal

All these effects (AMR, GRM, TMR) and their applications to field sensors will be studied in great detail in the lecture of Pr. Manon. I will concentrate on micromagnetics issues in the following Some other applications require to switch the dot magnetization See application table of content

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.16 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.16 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape anisot ropy

Conventional uniaxial shape effect Reminder : hysteresis loops

for magnetocrystalline anisotropy

Along hard axis Along easy axis

Reminder : for stricly single-domain particles,

) cos( sin

H ext S 2 d tot

θ θ H M µ θ K E − − =

Kd is the averaged dipolar energy (shape anisotropy)

Note: Switching field is lower than anisotropy field

R.P. Cowburn, J .Phys.D:Appl.Phys.33, R1–R16 (2000)

In real world, finite-size dots are

not strictly single-domain. However…

• 1

1

Field (Oe)

H H 600 400 200

• 200
• 400
• 600

Normalized magnetization

! ‘Shape anisotropy’ has some significance (not 100%…) ! Understanding of coercivity is very phenomenological

250x10nm

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.17 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.17 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape anisot ropy

Shape anisotropy increases with dot thickness

100 200 300 400 500 600 700

H

(a)

100 200 300 400 500 600 700

Major axis (nm) H

600 500 400 300 200 100

15nm 10nm 5nm 10nm 5nm Saturation Switching

What should we remember about shape anisotropy ?

Shape anisotropy =

Phenomenological description only Predicts reasonably well:

• Easy / Hard axis directions
• Saturation field and susceptibility along

hard axis

Can be used qualitatively:

• Switching field order of magnitude

Always remember that is not strictly valid !

R.P. Cowburn, J .Phys.D:Appl.Phys.33, R1–R16 (2000)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.18 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.18 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

MRAMs

MRAM’s: principle of operation

Magnetic Random Access Memory

GMR or TMR junction

Read/ Write Read/ Write

Single junction (1bit memory)

‘1’ ‘0’

TMR: under intense development : IBM, Siemens, Japan. Mbit-Gbits prospects.

GMR: product by Honeywell (64 kb?) for space applications (radiation insensitive)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.19 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.19 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

MRAMs

MRAM’s: write obstacle H H H

2

Write with current-induced magnetic field On-chip array of cells for mass storage

# Roughly speaking: all dots must switch under √2H,

but none should switch under H ! Is it feasible ?

# What factors influence magnetization reversal (Switching field) ? # Back to fundamental studies !

NO in general : one observes a very broad distribution of switching fields !

(From cell to cell, and even cell irreproducibility)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.20 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.20 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape det ails : end domains

End domains arise in order to avoid surface magnetic charges ‘C state’ ‘C state’ ‘S state’ ‘S state’ At least 8 nearly equivalent ground-states for a dot !

!how does a dot reverse its magnetization ? !Sensitive to any dissymetry like defects, temperature, stray fields, etc. !Rectangles are not be the best shape for single domaine devices…

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.21 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.21 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape det ails : end pinning ef f ect s

Magnetization is pinned at sharp ends Numerical micromagnetic calculation

}

J .G. Zhu

Two ground-states each Eight ground-states

!GOOD: Better reproducibility !BAD: Higher switching field

(I mages court esy of J . Milt at – CNRS, Orsay, France)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.22 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.22 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape det ails : end pinning ef f ect s

Magnetization is pinned at sharp ends

Permalloy (soft)

Experiments

Similar

Experimentally confirmed

!GOOD: Better reproducibility !BAD: Higher switching field

K.J . Kirk et al., J .Magn.Soc.J apan, 21 (7), (1997)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.23 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.23 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

shape det ails : roughness

Roughness influences the dot ground state and reversed state Numerical micromagnetic calculation

!Source of noise !!! !Decrease of switching field

(not controlable)

J .G. Deak et al., J .Magn.Magn.Mat er.213, 25(2000)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.24 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.24 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

conf igurat ional anisot ropy

Configurational anisotropy : deviations from single-domain

( )

2 2 2 d

2 1

z z y y x x

M N M N M N µ e + + =

θ VK E

2 d tot

sin

=

2 10 4 6

∞

R.P. Cowburn, J .Phys.D:Appl.Phys.33, R1–R16 (2000)

However, in real samples magnetization is

never perfectly uniform

Num.Calc. (100nm)

Corrective energy terms, that depend upon

the average magnetization angle

! higher order contribution to the

anisotropy ? ‘Configurational anisotropy’ Strictly speaking, ‘shape anisotropy’ is of second order: 2D:

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.25 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.25 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

conf igurat ional anisot ropy: experiment s…

500 Oe 400 Oe 300 Oe 200 Oe

Polar plot of experimental configurational anisotropy with various symmetry Color code: strength of anisotropy in a given direction Radius: size of measured pattern Direction: direction of measurement Configurational anisotropy may be used to stabilize stable configuration

R.P. Cowburn, J .Phys.D:Appl.Phys.33, R1–R16 (2000)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.26 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.26 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > single domain >

conclusion on sensor s and memories

What should we remember about the design of small near-single-domain magnetic elements for memories ?

(dots that require to be magnetically switched forth and back)

The subject is not trivial ! Approaches need to be three-fold to give a complete view: * experiments * analytical * numerical calculation Single domain favored by : * small lateral size * moderate thickness * even weak in-plane magnetocrystalline anisotropy Coercivity is obtained from : * magnetocrystaline anisotropy * shape anisotropy * configurational anisotropy

* other means not mentioned here…

Switching control is better achieved for : * structures with pointed ends * low edge roughness

Back to list of applications

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.27 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.27 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

areal densit y

85 90 95 100 105 110 115

Availability Year

0.01 0.1 1 10 100 1000 10000

Areal Density, Gbits/in2

IBM Advanced Storage Roadmap

Superparamagnetic Effect

1 Gbit/in2 Demo 3 Gbits/in2 Demo 5 Gbits/in2 Demo 12.1 Gbits/in2

20.3 Gbits/in2 Deskstar 40GV Ultrastar36LZX Enhanced Magnetic

Disk Drive

Advanced Storage

Technology/Holography

2000 05 10 15

Microdrive 10K RPM Integrated Head/Suspension Giant MR Head/Pico Slider Ramp Load/Unload, Glass Substrates No-ID MR Head/Nano-slider PRML Data Channel Thin Film/High Coercivity Disks

Small Form Factor

Lab Demos 3.5 Inch FF 2.5 Inch FF >10 Inch FF

ED GROCHOWSKI at ALMADEN

advrdmp20a.prz

35.3 Gbits/in2

Travelstar30GT

Increase of areal density of Hard Disk Drives (HDD), as viewed by IBM

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.28 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.28 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

bit size

1 2 3 4 5

2 4 6 8 10 12

Trackwidth, " m Bit length, " m

(1990)

0.09 Gbit/in2 20:1

(1992)

0.2 Gbit/in2 19:1 (1994) 0.5 Gbit/in2 18:1

(1998)

5 Gbit/in2 14:1

(2000)

20 Gbit/in2 12:1

12 "m 7.8 "m 4.8 "m 1.3 "m 0.62

0.60 0.41 0.27 0.094

A Decade of Shrinking Bit Cell

b i t c e l l 2 a . p r z

Ed Grochowski at Almaden

0.052 0.17

(2002)

80 Gbit/in2 8:1

(1996)

1.3 Gbit/in2 17:1

2.9 "m

0.03

0.25

Size of magnetic bits written on IBM Hard Disk Drives (HDD)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.29 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.29 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >principle a D T W

Stable magnetic recording requires :

• Ferromagnetic grains (nano-magnets)
• Many grains per bit (high signal to noise ratio)

Commercially shiped : 10-20 Gbit/in2 Demonstrated (summer 2001) : 100 Gbit/in2 Top view

(I mages court esy of D. Weller – Seagat e

• C. Chappert – I EF-Orsay)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.30 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.30 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

superparamagnet ic limit substrate lubricant CrV NiAl Co-alloy Carbon

Side view

a D

Top view

Current HHD: longitudinal granular media

(weak coupling between grains ; in-plane magnetization)

High S/ N ratio needed for read-out: number of grains per bit must remain high (102-103) Increase media density % shrink grains size % face superparamagnetism

Keep number nt

• f grains

consta E

B

=KV

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.31 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.31 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

superparamagnet ic limit

Anisotropy barrier E B~KV

UP DOWN

~25kT

Phenomenological description of thermal activation : Néel-Brown theory

Brown, Phys.Rev.130, 1677 (1963)

τ t e t P

/ ) (

− =

Probability of not having switched

T k E e τ τ

B B /

=

Mean reversal time

s 10

9

−

≈ τ

) / ln(

τ t T k E

B B =

Barrier for not switching during time t

Lab Observer : t=1s Magnetic recording : t>>109s

T k E

B B

25

≈

T k E

B B

60

≈

Blocking temperature

B B B

k E T

25 /

=

Back to roadmap

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.32 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.32 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media

• S. Chou et al.' proposition (ca 1994) :

1 bit ~ 1 grains Requirements : ultra smooth media planari zation

Patterne d magneti c média

Electrodeposition of Ni into arrays of holes

• in PMMA resist
• in a SiO 2 layer

2 orders or magnitude gain on superparamagnetism

(Slide court esy of C. Chappert – I EF-Orsay)

This prospect has triggered a lot of work in the past five years

# sub-micrometer-sized single domain dots (see previous section…)

However: how to meet the above two requirements ?

(many other issues ; we focus here on magn.)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.33 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.33 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

growt h on t emplat e

Preparation of a topographic pattern on Si (well established processes)

Preparation of arrays of resist dots or lines by either :

• electron beam lithography (CEA/LETI) : resolution~50nm
• r
• nano-imprint using a mold made by e-beam

(A.Lebib, Y.Chen, CNRS/L2M) : resolution 30nm

Si

Removal of resist resist dots RIE etching Smallest size by nano-imprint: 30 nm dots with edge to edge spacing of 30nm.

(Slide court esy of B. Diény – CEA/ SpinTech-Grenoble)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.34 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.34 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

growt h on t emplat e

Magnetic materials overgrowth

C

• Pt

Deposition of magnetic material (sputtering or MBE) Advantages :

• Etching of Si very well controlled in microelectronics,
• Possibility of very high aspect ratio,
• Substrate preparation independent of the choice of the magnetic materials,
• No processing after deposition of the magnetic materials,
• Possible large scale preparation by nano-imprint (180Gbit/in2 demo).

(Slide court esy of B. Diény – CEA/ SpinTech-Grenoble)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.35 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.35 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

growt h on t emplat e

Magnetic state : single domain (‘bits’)

Pt 20 nm/( Co 0.5 nm/Pt 1.8 nm)4 multilayer with perpendicular magnetic anisotropy :

400 nm square dots (spacing of 100 nm, height of 47 nm)

Local checkerboard patterns most stable from magnetostatic point of view 8 µm x 8 µm image

MFM: magnetic pattern, as-deposited

GOOD: magnetic properties more homogeneous (less edge defects) BAD: topography

(Slide court esy of B. Diény – CEA/ SpinTech-Grenoble)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.36 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.36 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

ion beam

Ion beam irradiation patterning : initial media with perpendicular anisotropy

(Slide court esy of A. Mougin – LPS-Orsay)

Strongly anisotropic spin-orbit coupling

! Perpendicular anisotropy

despite thin film shape effect

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.37 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.37 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

ion beam

Ion beam irradiation patterning : irradiation physical effect

(Slide court esy of A. Mougin – LPS-Orsay)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.38 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.38 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

ion beam

Ion beam irradiation patterning : irradiation magnetic effect

(Slide court esy of A. Mougin – LPS-Orsay)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.39 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.39 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

ion beam

Ion beam irradiation patterning : flat patterned magnetic media Replication lithography

possible

! Mass production

GOOD: magnetic properties more homogeneous (less edge defects) GOOD: flat topography

(Slide court esy of A. Mougin – LPS-Orsay)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.40 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.40 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic recording >

pat t erned media >

conclusion

In principle interesting:

* More homogeneous properties * flat topography possible * one grain per bit : no more superparamagnetism

However many unsolved issues:

* how to read/write * demo density of cenventional media is now 100Gbit/in2 (with read/write !!!), whereas reasonnable demo with patterned media is 100Gbit/in2.

! Patterned media came to late ? Obsolete idea ?…

(see roadmap…)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.41 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.41 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic comput ing

Very prospective subject… A working magnetic logic device Transmission line: replace copper wires by macrospin dots

Dots are superparamagnetic if isolated, but stabilized by dipolar interactions in a chain

... ...

HCK

Input pin AND-gate

• Number of dots per chain

= 70

• Dot diameter

= 110nm

• Dot pitch

= 135nm

• Dot thickness

= 10nm

Cowbur n et al., New J . Phys. 1, 16.1 (1999)

ROOM TEMPERATURE

Cowbur n et al. Science 287, 1466 (2000)

(Slide court esy of R.P. Cowburn – Durham, UK)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.42 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.42 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > magnet ic comput ing

A working magnetic logic device All logic functions are possible : AND, OR, … AND gate

Higher density on chips than conventional semiconductor processors 10-3 less power dissipation (propagation is not dissipative !).

Cowbur n et al. Science 287, 1466 (2000)

Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.43 ] Olivier Fruchart - LLN-CNRS. [ 09/10/2001 / p.43 ]

Slides on- line: http:/ / ln-w 3 .polycnrs-gre.fr/ them es/ couches/ ext/

Pat t erned > Conclusion

Field of patterned magnetic structure… Extremely rich from the fundamental point of view.

50 years old subject Renewed recently (demand, technology) Computer simulations from the 90’s

Extremely promising from the technological point of view

Sensors : sensitivity and integration

$ $ $

Storage : memories, Hard disks, tapes, etc. Prospects: magnetic processing ? Spintronics ?