Perovskite-type transition metal oxide interfaces

M. Matvejeff7.2.2011

• Perovskites - Chemistry and properties

• Properties of perovskite interfaces - CMR and Magnetic Tunnel Junctions (MTJs)

• Charge transfer at perovskite interfaces

Contents

Perovskites – Structure

AO + BO2 = ABO3

SrTiO3

(La,Sr)MnO3 (LSMO)

1 u.c.ABO

AO

AO

AO

BO2

BO2

Highly flexible cation stoichiometry

Wide variety of functional properties through changes in

cation stoichiometry

(La1-xSrx)MnO3 (LSMO)

Imada et al. Rev. Mod. Phys. 70

Perovskites – The Good

Highly flexible cation stoichiometry

Wide variety of functional properties through changes in

cation stoichiometry

Highly flexible oxygen content

Properties can be fine-tuned after synthesis

AO1- + BO2 = ABO3-

SrTiO3-

(La,Sr)MnO3- (LSMO)

1 u.c.

ABO

AO

AO

AO

BO2

BO2

Perovskites – The Good

The flexibility of perovskite structure and the easy tunability of the functional properties are definite bonuses as long as bulk

material is suitable for applications

For example capacitors, catalytic converters and superconductors

However, significant number of industrial applications rely on device structures consisting of several different functional

material layers, in some cases only few atomic layers in thickness

In these structures, such as field-effect transistors (FETs), the properties of the interface are often significantly more important

to the correct function of the device than the properties of the bulk material

Highly 3-dimensional structure

+Strong hybridization of 3d orbital

of the transition metal B to neighboring oxygen 2p orbitals

+Highly sensitive to small changes

in transition metal oxidation state

Properties at interfaces?

Perovskites – The Bad1 u.c.

ABO

AO

AO

AO

BO2

BO2

• Perovskites - Chemistry and properties

• Properties of perovskite interfaces - CMR and Magnetic Tunnel Junctions (MTJs)

• Charge transfer at perovskite interfaces

Contents

R. von Helmholt APL 1993

Colossal MR (CMR) in La2/3Ba1/3MnO3

Manganites exhibit CMR i.e. strong change

in resistivity under applied magnetic field

The CMR effect can be used for example for

magnetic sensor applications

As the most properties of transition metal

oxides, CMR is highly dependent on transition

metal (Mn) oxidation state

CMR in manganites

General formulaAMnO3

A = divalent and/or trivalent cation (Ca, Sr, La, Nd...)

(La,Sr)MnO3 (LSMO)

Mn3+

t2g

eg

Mn4+

t2g

eg

Itinerantelectron

Localelectrons

Electronic structure of manganites

To understand the origin of CMR

phenomenon we need to first understand the electronic structure of

manganites

Mn3+

t2g

eg

Mn4+

t2g

eg

Itinerantelectron

Localelectrons

LaMnO3

+3 +3 -2

SrMnO3

+2 +4 -2

(La,Sr)MnO3 (LSMO)

La1-xSrxMnO3

+3 +2 3...4 -2

x Mn4+

1-x Mn3+

General formulaAMnO3

A = divalent and/or trivalent cation (Ca, Sr, La, Nd...)

Chemical substitution means we’re directly playing with the average

valence of Mn

In double-exchange (DE) modelItinerant eg electron is the charge carrierwhereas the t2g electrons are localized

Mn3+

t2g

eg

Mn4+

t2g

eg

Itinerantelectron

Localelectrons

Mn3+/Mn4+-ratio (doping) has strong impact on

magnetotransport properties

LaMnO3

+3 +3 -2

SrMnO3

+2 +4 -2

La1-xSrxMnO3

+3 +2 3...4 -2

x Mn4+

1-x Mn3+

(La,Sr)MnO3 (LSMO)

Bulk CMR is not suitable for low field applications (magnetic field

required is in order of several tesla)

How to increase sensitivity?

Significantly weaker field (~coercive field of the material)

required in MTJs

FM

Insulator (t = nm-Å)

FM

Tunneling current

FM

Insulator (t = nm-Å)

FM

Tunneling current

Magnetic tunnel junction (MTJ)

What are magnetic tunnel junctions (MTJs)?

FM

Insulator

FM

Tunneling current

Tunneling current

Applied field

Junc

tion

resi

stan

ce

Applied field

Magnetization

Applied field

Magnetic field required is in order of tens to hundreds of Oe instead of several Tesla as for bulk CMR low field sensors

For maximum sensitivity RA-RAP has to be maximized

Degree of spin polarization is important!

MTJ

RA(AP) Resistance in parallel (antiparallel) configurationP1,P2 Polarizations of electrodes 1 and 2

TMR

R

P

AP

1

2

R. von Helmholt APL 1993

P. M. Tedrow and R. Meservey PRB 1973

Half-metals – Because polarization does matter…

J.-H. Park Nature 1998

Half-metallicity in bulk La0.7Sr0.3MnO3

P ~ 95-100% in low T

LSMO is a good candidate material for

MTJs

Y. Lu, APL 1996

T. Obata, APL 1999

LSMO

STO

LSMO

4.2K

Tc ~ 350K

Good TMR only at low T

TMR dissappears well below Tc

Why?

Dead layer

La0.67Sr0.33MnO3 films grown on (110) NGO (NdGaO3) and (001) LAO (LaAlO3) substrates

Clear thickness dependence in resistivity

Dead (insulating) layer forms at the interface?

How can we study this?

J. Z. Sun APL 1999

Dead layer

(2-10 u.c. LSMO – 2 u.c. STO)10-20 superstructureLSMO = La1-xSrxMnO3, 0.2 x 0.4

By changing the thickness of conducting layers (LSMO) separated by the insulator (SrTiO3) we can probe the critical thickness for transition from ferromagnetic metal (FM) to antiferromagnetic insulator (AFI)

LSMO (2-10 u.c.)

STO (2 u.c.)

LSMO (2-10 u.c.)

STO (2 u.c.)

Dead layer

M. Izumi J. Phys. Soc. Jpn. 2002

For all doping doping levels, decrease in Tc and magnetization with decreasing LSMO thickness

Decrease is faster with higher x

Samples which are closer to metal to insulator-phase diagram line loose metallicity and magnetic order already in thicker films

Y. Tokura Rep. Prog. Phys. 2006

Dead layer

Same effect also observed in M-H measurements

Also, for thinner films M-H does not saturate

This indicates competing FM and AFM interactions

FM FM+AFMFM+AFM+ ext. field!

M. Izumi J. Phys. Soc. Jpn. 2002

Dead layer

From phase diagram we see transition from FM to AF state at x ~ 0.5

Is this related to the formation of dead layer at the interface?

H. Fujishiro J. Phys. Soc. Jpn 1998

Y. Tokura Rep. Prog. Phys. 2006

So how does the dead layer actually form?

LSMO (2-10 u.c.)

STO (2 u.c.)

LSMO (2-10 u.c.)

STO (2 u.c.)

So what does actually happen at the interface layer?

Dead layer

Hole-doping at La1-xSrxMnO3-STO interface x increases The properties of the interface change

M. Izumi J. Phys. Soc. Jpn. 2002

x incre

ase

s(ch

arg

e tra

nsfe

r)STO (2 u.c.)

La0.8Sr0.2MnO3 (x = 0.2)Bulk High Tc High magnetization FM

Hole-doped LSMO (x 0.2)FM+AFM

Lower Tc/magnetization

Effect is stronger when x in the original phase is higher (already closer to critical limit of x ~ 0.5)

Why does the hole-doping occur?

x incre

ase

s

STO (2 u.c.)

La0.4Sr0.4MnO3 (x = 0.4)Bulk High Tc High magnetization FM

Hole-doped LSMO (x 0.4)

Faster decrease in properties

Y. Tokura Rep. Prog. Phys. 2006

• Perovskites - Chemistry and properties

• Properties of perovskite interfaces - CMR and Magnetic Tunnel Junctions (MTJs)

• Charge transfer at perovskite interfaces

Contents

SrTiO3

SrTiO3

LaTiO3

Sr2+ and O2-

Ti4+

(2 + x + 3*(-2) = 0)

La3+ and O2-

Ti3+

(3 + x + 3*(-2) = 0)

Let’s study the following a quantum well structure…

In theory the Ti valence changes sharply at the interface between SrTiO3 (STO) and LaTiO3 (LTO)

SrTiO3

SrTiO3

LaTiO3

LaSrTiO

Ti4+

Ti3+

Ti4+

LaSrTiO

SrTiO3

SrTiO3

Ti3+ fraction

LaTiO3

Ohtomo A. et al., Nature, 2002

Ti4+

Ti3+

Ti4+

Ti3/4+

Ti3/4+

However in practice it has been found out that Ti3+ oxidation state is not limited to the LTO

layers…

… i.e. charge transfer (transfer of electrons) occurs from LTO into STO layers forming

mixed valence interface layer

LSMO

LSMO

STO

FM

Insul.

FM

Now, our ideal TMR device the

LSMO/STO/LSMO tunnel junction

LSMO TC ~ 350 K in the bulk phase

In practice, charge transfer over the interface

Strong impact on carrier density (valence of Mn) at the interface

Instead of FM, LSMO at interface either P or AF

Formation of dead layer and TC 100 K instead of 350 K!

LSMO

LSMO

STO

FM

Insul.

FM

P/AF

P/AF

Y. Tokura Rep. Prog. Phys. 2006

Alternating AO and BO2 layers

Formula: ABO3

3D structure is the problem!

So what about structures which aren’t (fully) 3D?

Perovskite - recap1 u.c.

ABO

AO

AO

AO

BO2

BO2

Closely related to perovskite structure

Alternating AO and BO2 layers

Formula: An+1BnO3n+1

(i.e. one extra AO-layer compared to perovskites, ABO3)

Ruddlesden-Popper structure

Perovskite: 3D structure

vs

Ruddlesden-Popper (RP): 2DHigh anisotropy (ab-plane vs c-axis)

AO

Perovskite (ABO3)

1 u.c.

AO

AO

AO

BO2

BO2

AO

BO2

BO2

n = 2 RP(A3B2O7)

AO 1 formula

unit

BO2

AO

BO2

AO

AO

BO2

AO

BO2

AO

AO

BO2

AO

BO2

AO

c-ax

is

La1.4Sr1.6Mn2O7

T. Kimura & Y. Tokura, Annu. Rev. Mater. Sci., 2000

AO 1 formula

unit

BO2

AO

BO2

AO

AO

BO2

AO

BO2

AO

AO

BO2

AO

BO2

AO

Charge carriers

Cha

rge

carr

iers

A = La, SrB = Mn

Perovskite

Perovskite

RP

Perovskite 1

Perovskite 2

Weak interaction

Clean interface, little or no

modulation

Perovskite 1

Strong interaction

Modulation of interface

properties

So does it actually work?

In perovskite-type interface between (La,Sr)MnO3/(La,Sr)FeO3 electrons are transferred from Mn eg states to Fe eg states

We can study the interface electronic structure in XPS…

Kumigashira et al. APL 2004

t2g

eg

LSMO

LSFO (t = 1-7 layers)

Mn3+

t2g

eg

Mn4+

t2g

eg

Itinerantelectron

Localelectrons

… to determine the occupation of eg and t2g states

As LSFO layer thickness is increased, the charge transfer increases and eg electron occupation decreases (Mn valance increases)

LSFO

LSMO

LSFO

LSMO

Weak interaction

Clean interface

Small change in LSMO valence?

Strong interaction

Large change in LSMO valence

t2g

eg

Perovskite

RP-type interface(LSMO layer thickness = 3 u.c.)

Inte

nsi

ty [

arb

. u

nits

]

-5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0Energy relative to Ef [eV]

Mn 2p - 3d resonant valence bandh = 642 eV

LSFO/LSMO (perovskite interface) LSFO/LSMO (RP-type interface) LSMO reference

t2g

eg

Conclusions

Perovskite phases exhibit interesting functional properties in bulk form

Applications, however, are often based on device structures built from functional layers at times only few atomic layers in thickness

Interface effects arising from the 3-dimensional nature of the perovskite structure dominate the behavior of the devices

Interface effects can be, at times, partially compensated for, but this leads to expensive production processes where device properties are difficult to predict and/or control

Best solutions would be based on integrating, property-wise, 2-dimensional materials into device structures to create not only structurally but also electronically sharp interface structures