Oxide films and scanning probesJ. Aarts, Kamerlingh Onnes Laboratory, Leiden University

…problems not solved …(today)

Wanted atomic scale electronic / structure properties (local sc gap, stripes, phase separation, charge order).

Problem STM : not for insulators ; AFM : no atomic resolution

and always : clean sample surfaces

Outline

1. A nice model system : Charge order (melting) in strained thin films of Pr0.5Ca0.5MnO3

together with

Z.Q. Yang, A. Troyanovski, G.-J. v. Baarle LeidenM. Y. Wu, Y. Qin, H. W. Zandbergen HREM center, Delft

2. How STM can work (an intermezzo)Melting of the vortex lattice in a superconductor (NbSe2)

3. A roadmap for SPM on oxidesCurrent status , future prospects

1. Pr0.5Ca0.5MnO3 : a model system for charge order (melting)

Strategy

• work on thin films for flexibility (and ‘applications’) (difficulty : sample surface – no cleavage available)

• use strain to vary properties

Fabrication

• sputtered at 840 °C• high O2 pressure = slow growth ( 1 nm / min )• on SrTiO3 (a0 = 0.391 nm vs. 0.382 nm for PCMO)

a

c

b

(RE, Ca )Mn

ABO3 structure : orthorhombic Pnma = ‘3-tilt’ ; (ap2, 2ap, ap2 )

• Octahedra buckle, smaller Vcell

• Decreased Mn-O-Mn bond angle, narrower eg bandwidth, less hopping, lower TIM

)(2 OrBr

OrAr

t

Tilting due to tolerance

t < 1 :

Pr0.5Ca0.5MnO3, bulk properties

Phase diagram, Pr1-xCaxMnO3

Insulating

Charge + Orbital order : ‘CE’ – type, zig-zags

At Mn3+ - Mn4+ = 1 : 1

could have been different :

c

a

Basic properties of Pr0.5Ca0.5MnO3

Jirak et. al., PR B61 (2000)

• R(T) : ‘insulating’, with small

jump at TCO = TOO

• (T) : peak at TCO , not at TAF

• lattice parameters :

< a0 > = 0.382 nm, orthorhombic distortion at Tco = TOO

• Staggered M : onset at TAF

Question for strained films : Tco enhanced by the applied distortion?or destabilised by ‘clamping’?

Hc+Hc

-

‘Melting’ of CO by aligning Mn-core spins with a magnetic field : 1st order transition from AF-I to FM-M

x = 0.5 : needs large fields, 28 T at 5 K

Cax

x < 0.5 : CO less stable; lower fields and ‘reentrant’.

Strained films : different melting behavior ?

Pr0.5Ca0.5MnO3 ( <a0> = 0.382 nm) on SrTiO3 (a0 = 0.391 nm)

Growth : magnetron; no post-anneal, Ts = 840 oC, 3 mbar oxygen

0 40 80 1200.376

0.380

0.384

0.388 in-plane a, c

out-of-plane b

La

ttic

e p

ara

me

ter

( n

m )

Thickness ( nm )

Lattice parameter versus thickness

relaxation slow ( > 150 nm)

bulk

suggests disorder at large thickness ?

PCMO on STO

STO

PCMOb 80-nm film on STO at RT:

• clearly visible 2 ap fringes – doubling of the b-axis;

• b-axis oriented

• no remarkable defects/disorder

Transport and magnetization

102

104

106

108

1, 3, 5, 7, 9T

0 50 100 150 200 250 300 35010

1

103

105

107

109

9 T7 T

5 T0 T

T ( K )

0 100 200 300-0.5

-0.4

-0.3

-0.2

-0.1

0.0

250 K

H = 3 kOe

M (

10

-3

em

u )

T ( K )

150 nm

80nm

R(T)

0 5 10 15 20

0.0

5.0x106

1.0x107

1.5x107

2.0x107

2.5x107

80 nm

15 K

41 K

61 K

83 K

R (

)

B ( T )

R(H), 80 nm

80 nm : melting strongly hysteretic; needs 20 T at 15 K.

150 nm, melting at 15 K needs 5 T.

-6 -4 -2 0 2 4 6

-0.5

0.0

0.5 a

150nm

100K

1.8 T

M (

10

-3 e

mu

)

B ( T )

also visible in M(H); together with FM component

… which leads to the following phase diagrams

• weaker CO melting with increasing thickness / relaxation

• increasingly ‘reentrant’ – reminiscent of x < 0.5

Strain does not lead to CO-destabilization, but relaxation does

but what about Tco ?

Intermezzo – why re-entrance ?

Khomskii, Physica B 280, 325 (’00)

The high-temperature phase should be the one with higher entropy (S), but it is the CO phase (lower S).

Apparently : (1) the FM ground state is a Fermi liquid (S=0) and (2) the CO-state is not fully ordered.

which is reasonable away from Mn3+ / Mn4+ = 1 : 1

TCO from resistance

• no clear jump in R(T) but kink in ln(R) vs 1/T

• TCO > bulk value 250 K , transition width T=TCO-T*

0 100 200 300102

104

106

108

0.004 0.006 0.0086

8

10

12

14

Tco

=285K

80 nm PCMO/STOR

(

)

T ( K )

LnR

80 nm PCMO/STO

Tco

=285 K

T*=208 K

T -1 (K-1)

Observe CO and OO by HREM

[002]

[200]

[101]

[020]

[200]

[002]

[200]

[101]

View along c-axis,

[001]-type superstructure

View along b-axis,

[010]-type superstructure

at 300 K

(at 95 K)

80 nm PCMO on STO

spot at (1/2 0 0) evidence for OO

spot at (100) evidence for CO at room temperature

TCO,OO vs. film thickness

• tensile strain increases TCO/OO to above room temperature

• relaxation decreases melting fields

• SrTiO3 – + 2.5%• NdGaO3 – + 1.3%• (Sr,La)GaO4 – + 0.75%

0 40 80 120 160

200

220

240

260

280

300

320 TCO

from R(T): Open symbolsT

OO from HREM: Closed

as-grown on STO Relaxed on STO Annealed on STO On less tensile NGO, On matched SLGO as-grown on STO, Relaxed on STO

TC

O,O

O (

K)

Thickness (nm)

PCMO thin films would be interesting for STM studies :

• observe CO up to high temperatures

• study melting vs. disorder in a large field range

What about melting of charge order and stripes ?

Formation of dislocations ?

Another (model) system for STM : the vortex lattice

• Vortex imaging : coherence length versus penetration depth • Vortex matter : solid – glass – liquid related issues : elasticity, disorder, defects, vortex pinning.

dimensionality, order prm symmetry

• Imaging a solid – to – (pinned) liquid transition. the model system : single Xtal of weakly pinning NbSe2.

• Thin films : work in air by passivation. lattices in weakly pinning a-Mo70Ge30 versus strongly pinning NbN.

2. Melting of the vortex lattice in a superconductor by STM

Superconductivity elementaries

vortex core:

• is ‘normal’ : no gap in DOS in radius .• magnetic field distribution over radius .

Type II : <<

NbSe2 8 nm 265 nm

a-Mo3Ge 5 nm 750 nm

YBCO 2 nm 180 nm

Vortex lattice elementaries

A vortex contains flux 0; increasing field B leads to more vortices.

Interactions then produce a triangular lattice with

B 507.1 0a

1.5 m for B = 1 mT

49 nm for 1 T

‘decoration’ of NbSe2 at 3.6 mT and 4.2 K.

a = 0.8 m.

Magnetic field probes (Bitter-decoration, magneto-optics, scanning SQUID / Hall ) only work well when a < - typically mT – range, interactions small,

far from critical field Bc2.

STM is the best / only probe at high magnetic fields.

Current general vortex matter (B,T) phase diagram

Ideal

A-lattice

Include disorder pinning glassthermal fluctuations melting

Technique ( since H. Hess, 1989) : map current in the gap ( 0.5 mV).

NbSe2 (crystal, Tc = 7 K)

STM-image, (1.1 m)2 T = 4.2 K, B = 0.9 T t = 0.6, b = 0.35

NbSe2 is layered, passive, atomically flat (after cleaving)Ideal for constant height mode,allows fast scanning :

< 1 min / frame of (1.1 m)2

And : weakly pinning

NbSe2 – what can be new : vortices in the peak effect.

Peak : close to Bc2 a strong peak occurs in the critical current – which indicates when vortices start to move under a driving force.

in 1.75 T

It means that individual vortices can optimize their positions w.r.t defects, since inter-vortex elastic forces disappear – melting ?

Can you ‘see’ this in the vortex lattice ? Defects ? LRO ?

Not entirely trivial, close to Tc / Bc2 the signal disappears :

B = 2 T, T = 4. 28 K

Typical data around T = 4.3 K, B = 1.75 T

Blurring gets worse, needs data processing

Experiment : let T drift up slowly (5 x 10-5 K/s) and measure continuously at 1 image / min (0.3 mK).

Analyze the sequence of data.

4.30 K

1.75 T

4.44 K

4.53 K

Convolution with pattern of:

“single vortex”:

Unit cell

3x3:

Image processing

A movie of the processed data. Note T 4.47 K

Analysis : determine correlations in vortex motion between frames

i k

i k

d dK

d d

‘order prm’ : di= ri,n -ri,n+1, dk= rk,n- rk,n+1

ri = position, n = framenumber

Motion becomes uncorrelated at Tp1.

Above Tp1

Average 70 subsequent images in T-regime 4.50 K – 4.55 K

Brightness indicates probabilityof finding a vortex at a certain position :

Some vortices are strongly pinned

The picture : at Tp1, individual pinning wins from elasticity, mainly shear modulus :

2 2

66 ( ) (1 0.3 )(1 )cc B t b b b

resulting in a pinned liquid

Other superconductors - thin films ?

standard problem : clean and flat surface – only few crystals have been imaged; films (almost) never been used.

clean : in-situ cleaning ( / cleaving) + handling in vacuum; protect with passivating layer (Au ?) . The ‘wetting’

problem.flat : after cleaving; amorphous films.

amorphous superconducting films (Nb-Ge, Mo-Ge, W-Re, V-Si, …)• are weakly pinning (no grain boundaries, precipitates … )• have large penetration (no good with decoration)

a-Mo70Ge30 Tc = 7 K ; can be sputtered but oxidizes; protect with Au, continuous layer.

Au ~5 nm Mo3Ge 50 nm

Si substrate

a-Mo3Ge + Au

AFM – no Au islands

Use proximity effect

signal weak, ‘spectroscopy mode’

Optimized settings a-Mo2.7Ge, B = 0.8 T, d = 48 nm, 1.1 m2

ACF2D-FFT

Au ~5 nmMo3Ge

24 nmNbN 50 nm

Si substrate

Also for NbN, a much stronger pinner.(NbN + a-Mo3Ge + Au)

vortex positions are of the strongest pinner : NbN

Coordination number (z):

36% has z ≠ 6> 6

= 6

< 6

full positional disorder

Final result : triangular – to – square VL transition in a thin film sandwich La1.85Sr0.15CuO4 + MoGe + Au

B = 0.3 T B = 0.7 T

LSCO-film : Moschalkov (Leuven)

The transition is due to the high-Tc LSCO :

neutrons, Gilardi e.a., PRL ‘02

• A solid – to – pinned – liquid transition was observed close to the upper critical field in NbSe2.

• Thin films can be passivated (and structured). Disorder / defects can be studied, as shown with a-Mo3Ge and NbN

• STM can be an effective tool to study ordering phenomena.

Note also that for many condensed matter problems, it needs substantial dynamic range for temperature, magnetic field and conductance (+ bias voltage).

So what about oxides ?

Note the differences in possible types of experiments between smooth and rough surfaces

3. A roadmap for the oxides

What has been done by STM :

a. Bi2Sr2CaCu2O8-δ superconductorsuperconducting gap, impurity resonances, stripesatomic resolution, discussion about disorder also YBa2Cu3O7-δ , Sr2RuO4

b. La0.7Ca0.3MnO3 CMR materialphase separation, local spectroscopyno atomic resolution

c. Bi0.24Ca0.76MnO3 Charge Order atomic resolution, but not a conclusive experiment

What has been done by AFM :

d. Si(111) semiconductor (sub-) atomic resolution

a. Bi2Sr2CaCu2O8-δ

+ Zn - impurities

Pan - Nature ‘00

150 Ǻ

• ZB – anomaly

• strong scattering along gap nodes

d-wave sc; a relative success storygood metal, atomically flat surface (cleavage)

ZB map

Disorder in BSCCO - variations in gap spectra / gap width

Lang - Nature ‘02

Different for different doping Homogeneous (for optimal doping)

Hoogenboom - Phys. C ‘03

Fourier Transform STS - stripes

Direct space, 7 T

Hoffman - Science ‘02

Spatial structure around cores

FT’s at different energy

Quasiparticle interference – maps the Fermi surface

Hoffman - Science ‘02Stripes through static disorder ?

Howald - PR B ‘03

b. La0.7Ca0.3MnO3CMR and the issue of phase separation

CMR

MR

Single Xtal STM topography

Local STM spectroscopy

Different I-V characteristics

M. Fäth

Leiden

Spectroscopy on LCMO

LCMO / YBCO film, 50 K black ‘=‘ metal’

topography

dI/dV, 0 T

dI/dV, 9 T

• Surface becomes more metallic with increasing field

• Disorder is (probably) froozen

0 , 0.3 T

1 , 3 T

5 , 9 T

Small scales

Spectroscopy on LCMO - cont

LSMO thin film, T-dependence

black ‘=‘ metal’

Becker, PRL ‘02

Current picture

• phase separation probably correlates with

underlying grain structure – or twin structure

• no random percolation

• no atomic resolution or e.g. the influence of

random scatterers such as Zn in BSCCO

c. Bi0.24Ca0.76MnO3 Image charge order

BulkTCO = 250 K

Mn3+ : Mn4+ = 1 : 3

- Renner, Nature ‘02

At 300 K, ‘some terraces’ with atomic

resolution

At 146 K, doubled (a02) unit cell along [101]

Two different atomic distances

SurfaceRotated octahedra ?Surface reconstructs ?Mn3+ : Mn4+ = 1 : 1

Many insulating partsnot conclusive

General problem : a mixture of insulating and metallic parts makes STM difficult (… tip crashes …)

d. Si(111) - a possible way out, AFM ?

AFM - usually not ‘true’ atomic resolution (periodicity but not defects)

new developments in frequency-modulated mode : tuning-fork AFM

see : F. J. Giessibl, Rev. Mod. Phys. 75, 949 (2003)

noise spectrum. Ampl = 1.5 pmMeasure Δf at constant amplitude

AFM – ‘sub’-atomic resolution

Si(111)- (7x7)Giessibl, Science ‘00

Single adatom Calculation for z = 285 pm

Finally, the tuning fork tip can also be used in STM-mode

Combined AFM / STM - ideal for badly conducting surfaces

In conclusion

• STM has had limited success on oxide surfaces, mainly for well-behaved (super)conductors ( + cleavage surfaces)

• Tuning-fork AFM / STM development is very promising

Competition between strain and disorder

Thickness(nm) k1(K) TCO(K) T(K) Hc+(T)

12 >325 >2025 1889 307 67 >18

Strained 50 1342 290 64 >15Relaxed 50 1190 277 124 4Strained 80 1331 285 77 15Relaxed 80 1095 259 79 2

120 1269 284 74 9150 1114 260 72 4150 1058 250 70 <4

• Strain , activation energy k1 , Tco , Hc+ ;

Disorder weakens!

Properties of CO/OO PCMO films = Strain + disorder !

• Strain , disorder , T , Hc+ .

Strain helps!