Surface Physics:

Structure and Composition & How to study them!

Course structure
1. Why surfaces are important & how can we study them?
2. Surface science techniques - phenomenology & determination
3. Surface crystallography & structural / chemical determination

Lecturing schedule

Comments:

week

Tues 10-11
PS128

Wed 11-12
B2.02

Thurs 9-10
F1.11

Fri 12-1
B2.02

CFM1

CFM2

CFM3

CFM4

CFM5

CFM6

10

CFM7

CFM8

CFM9

PowerPoint: - hand-outs; research & past paper; problem workshop; lab tour
Texts:

Woodruff & Delchar, Modern Techniques of Surface Science, CUP:

Zangwill, Surface Physics, CUP: McCash, Surface Chemistry, Oxford
Briggs & Seah Practical Surface Analysis ,Vols: 1 & 2 , Wiley

Prof Chris McConville

P431

1
c.f.mcconville@warwick.ac.uk

SURFACE PHYSICS
Why are surfaces interesting?
Fundamental: a surface is a special kind of defect the most
extreme kind of defect - in a perfect 3-D periodic solid with
different geometrical (atomic) and electronic structure
Practical:
1. all gas-solid and liquid-solid interactions occur at the
surface. e.g. corrosion, adhesion, wear, heterogeneous
catalysis (surface reactions, chemistry), electronic junctions
2. the surface chemistry (compound formation) and electronic
structure of solid-solid interfaces can dominate the performance
of a reaction or the operation of an electronic device
3. surfaces and interfaces can also be modified by adsorption
(segregation) from the bulk - e.g. grain boundary segregation
and intergranular brittle fracture or by alloying at the surface

grain boundary segregation and intergranular brittle fracture!!

a 2nd world war Liberty ship

3

What is difficult about studying surfaces?

Theory: surfaces break the 3-D periodicity commonly
exploited in describing many properties of solids.

Experiment:
1. Surface Sensitivity
need to detect very small amounts of material (i.e. very few atoms)

e.g 1 ML (monolayer) 1019 atoms m-2 1015 atoms cm-2

say surface probe is 1 mm2, so in 1% of 1 ML have 1011 atoms

for carbon (m = 2 x 10-23 g) this is equivalent to 2 x 10-12 g

(c.f. wet chemistry - detects 10-4 g)
4

What is difficult about studying surfaces?

Experiment:
2. Surface Specificity

Need to detect these small amounts of material (very few atoms)

in the presence of the underlying bulk solid.
e.g. 1 mm thick sample has 5 x 106 atomic layers
so 1% of a monolayer is 1 part in 5 x 108 of the total no. of atoms

Need to use a very, very local clean environment

to ensure the surface stays clean
ULTRA-HIGH VACUUM (UHV)
5

3. The need for ultra-high vacuum (UHV)

consider the rate of arrival of molecules at a surface from the surrounding gas
kinetic theory of gases: rate of arrival of molecules of
no. density n /unit. vol., average velocity ca, is r = n ca
now c2rms = 3 kBT/m & ca = (8/3)1/2crms
(kB - Boltzmanns constant, T - absolute temp., m - molecular mass)
and P = nkBT (cf P = RT/V) so r = P/(2kBTm)1/2
substitute kB = 1.38 x 10-23 J.K-1 and m = 1.66 x 10-27M kg

gives r = 2.63x1024 P/(MT)1/2 molecules m-2

take M=28 (N2, CO), T = 300 K and convert P (in Pa) to p in mbar (1 mbar = 100 Pa)
so r = 2.87 x 1024 p molecules m-2
1 ML 1019 molecules m-2 so with unity sticking factor the monolayer time is
= 3.48 x 10-6/p s or 3.48/p s
e.g. p = 1 mbar,
= 3.5 s
p = 3.5 x 10-6 mbar, = 1 s
p = 3.5 x 10-10 mbar, = 104 s or 3 hrs

MORAL : need UHV for

realistic experimental
timescales on clean and
created surfaces 7

How to achieve UHV?

1. Use clean (oil-free) pumps
e.g. Titanium sublimation pumps (molecule trapping on walls)

Ion pumps (Ti+ ions spiral in magnetic field & capture molecules)
Turbo-molecular pumps (high speed fans)
Might previously have (or even still!) used diffusion pumps
(heated polyphenylether oil accelerated through vents)

2. Bake - out chamber :

Removes weakly-adsorbed gas molecules from walls of chambers

which act as virtual leaks reduces pressure by x100

Turbo
Molecular
Pump
(high speed fan,
oil free)

Diffusion
Pump
(low vapour
pressure oil)

How to produce a clean surface in UHV?

1. Cleavage - need brittle crystal, generally crystal has only one
cleavage plane, cannot re-clean surface

2. Heating (annealing) to high temperature - desorb adsorbed

species (refractory materials, some oxides
3. Chemical cleaning - heat the sample in a partial pressure of gas
(O2, H* )
e.g. Cads + O2 CO/ CO2 and Oads + 2H* H2O
xCads + yH2 CH + CH2 + CH3 + CH4
4. Ion bombardment - Ar+, Ne+ ~500-5000 eV to remove surface atoms
+ annealing - to heal damage & re-order the surface region (BUT
creates problems of surface segregation of bulk impurities on
annealing compounds (i.e. with 2 or more elements) can be an
10
issue if components sputter / desorb at different rates.

What kind of surface?

Simple - low Miller index single
crystal surfaces

e.g. face centred cubic (fcc)

General - especially vicinal

surfaces (low index + steps &
kinks)

Can do the same for body-centred

cubic (bcc), hexagonal closepacked (hcp), wurtzite (wz), and
zinc-blende (zb) structures11

wz

zb

12

Examples of high-index (stepped) surfaces

fcc(410)

step atoms shaded for clarity

Primitive unit mesh

Centred-primitive unit mesh

Terrace:

(100)

Step

<010>
13

Surface structure why the interest?

Surface relaxation & atomic reconstruction
atomic rearrangement for energy minimisation
Adsorption where do adsorbed atoms and
molecules sit?
active sites in heterogeneous catalysis

Adsorption bond lengths

implications for bonding and reactivity

14

Surface structural phenomena - a brief overview

Clean surface (e.g. metal): surface relaxation
(i.e. the outermost layer spacing changes)

WHY?
Electron charge depletion in surface layer
due to spill over into vacuum & smoothing typically causes outer layer contraction.
Damped oscillatory layer spacing changes
with depth due to propagation of charge
fluctuations.

Semiconductor surface reconstruction

WHY? Covalently-bonded networks (e.g. Si, Ge, GaAs, InSb,.) -

surface leads to dangling bonds several surface

reconstructions possible to reduce no. of dangling bonds (on
the polar surfaces) and depending on surface composition.

Leads to depletion / accumulation of charge at the surface

15

Surface Physics:
Structure and Composition & How to study them!
Course structure
1. Why surfaces are important & how can we study them?
2. Surface science techniques - phenomenology & determination
3. Surface crystallography & structural / chemical determination

Lecturing schedule

Comments:

week

Tues 10-11
PS128

Wed 11-12
B2.02

Thurs 9-10
F1.11

Fri 12-1
B2.02

CFM1

CFM2

CFM3

CFM4

CFM5

CFM6

10

CFM7

CFM8

CFM9

PowerPoint: - hand-outs; research & past paper; problem workshop; lab tour
Texts:

Woodruff & Delchar, Modern Techniques of Surface Science, CUP:

Zangwill, Surface Physics, CUP: McCash, Surface Chemistry, Oxford
Briggs & Seah Practical Surface Analysis ,Vols: 1 & 2 , Wiley

Prof Chris McConville

P431

16
c.f.mcconville@warwick.ac.uk

Si(100)-(2x1)
Ideal termination has 2 dangling bonds per
surface Si atom - surface atoms pair (form
dimers) to reduce this to 1 dangling bond
per surface Si atom
NB: dimers are asymmetric (buckled)

Si(111)-(7x7)
Ideal termination has 1 dangling
bonds per surface Si atom reduced by:
1. surface dimer formation removes these dangling bonds
2. adatoms bond to groups of 3 Si
surface atoms (reduce 3 dangling
bonds to 1
3. stacking fault appears in 1/2 of
surface unit mesh

17

Si(100) (2x1) surface

Si(111) (7x7) surface

LEED

18

Surface structural phenomena - a brief overview

Ionic solids (e.g. oxides or III-V semiconductors)
- instability of polar surfaces
WHY? dipole interaction energy becomes infinite so
reconstruct to remove dipole interactions
(e.g. nano-facets to non-polar orientation)
e.g. NiO(111)

(100) nano-facet

MgO(111) - microfacets

reconstructed
ideal bulk-terminated
19

Some other metal surface reconstructions

(111)
nanofacets

zig-zag rows

Examples of clean metal surface missing row and close packed

20

Atomic adsorption on metals - usually highest-coordination site

The so-called clock reconstruction of a Ni(100) surface following

the adsorption of atomic C or N ( but not O).

21

Molecular adsorbates commonly form local directional bonds

Si(100)(2x1)-OH+H
H2O OHa + Ha

Cu(110) + glycine
2NH2CH2COOH 2NH2CH2COOa + H2

Need techniques that will allow this

detailed information to be determined
22
(with sufficient accuracy!)

Surface Structure & Surface Crystallography

NB: a surface is a 3-D object but has only 2-D periodicity
Layer(s) of different
species from solid or
gas phase or bulk
Outermost atomic
layers which differ
due to presence of
termination of solid

Terminology

adsorbate
selvedge

surface

substrate
Structure identical
to that of infinite
bulk solid

23

Surface Structure & Surface Crystallography

Classification of structures
Identify both translational and point symmetry operations

Classify structural types by relationship of surface and

substrate periodicities
Devise a convenient notation relating surface
periodicity (unknown) to substrate periodicity (known)

NB: a surface is a 3-D object hence;

Surface = Adsorbate + Selvedge (+ Substrate)

- but has only 2-D periodicity

and remember - ALL symmetry properties considered
24
apply to the 3D object

Classify the 2-D periodic surface according to:

(1) their periodicity; 2-D Bravais nets

25

Classify the 2-D periodic surface according to:

(2) their point group symmetry; Ten 2-D point groups

Elements: 1,2,3,4,6 - fold rotation &

mirror - all operations act within a
plane parallel to the surface

Stereograms of the ten two-dimensional point groups. On the left are

shown the equivalent positions, on the right the symmetry operations. The
names follow the full and abbreviated 'International' notation.

26

Classify the 2-D periodic surface according to

(3) their space group symmetry; 2D space groups

17 - 2D Space Groups

combine: Bravais nets &

p.g. operations + glide

27

also we classify structures according to the relationship

between the surface and substrate periodicity
substrate primitive translation vectors (2-D) a and b
surface primitive translation vectors (2-D) a and b
hence can write

a=G11a + G12b

and

b = G21a + G22b

where Gij are four coefficients which form a matrix G

so

and

Note: the area of substrate mesh is |a x b| so detG is the ratio

of the areas of the surface and substrate primitive unit meshes
28

Classification:- several possibilities

a) det.G is integral and all matrix components are integral
the two meshes are simply related - mesh of surface+substrate is the
same as the mesh of the surface alone i.e. same translational symmetry
e.g. in 1-D

b) det.G is a rational fraction (or det.G is integral and some matrix

components are rational fractions)
the two meshes are rationally related - structures are commensurate

e.g. in 1-D

so now the mesh of surface + substrate is a, b

det P
detP and detQ have the det G
det Q
smallest integral values
29

Classification:
c) det.G is irrational
incommensurate structure - implies surface atoms do not see the
corrugated potential of the substrate (may occur for adsorbed layers
with very strong adsorbate-adsorbate interactions)

Nomenclature
Most general: - use matrix notation - G
More convenient: - use Wood notation
e.g. adsorbate A on {hkl} surface of material X
primitive translation vector lengths related by |a|=p|a| , |b|=q|b|
surface mesh rotated by
structure is X{hkl}(pxq)R-A

e.g Ni{100}(2x2)-O and

Ni{111}(3x3)R30-O

NB: - cannot use Wood notation when the included angles of

the surface and substrate mesh differ

30

Examples of surface meshes and

their notation

( 2 2 ) R45

( 3 3) R30

c(2x2)

( 2 2)

(2 1)
31

Diffraction from surfaces how to make the signal

surface specific
Use low energy electrons diffracted signal because it only
comes from near-surface atoms but WHY?

Two contributing processes

1. inelastic electron scattering
inelastic scattering removes signal from any technique
which only detects electrons which have not lost energy
such as elastic scattering (diffraction)!
Mechanisms
a. electron-hole pair excitations - low energy excitations dominant at low (<10-20 eV) electron energies

b. plasmon excitations (cooperative longitudinal oscillations

of valence electron gas) - energies 10-20 eV
c. core ionisation events - energies 50 eV upwards
32

Theoretical calculation of inelastic scattering in jellium with a charge

density appropriate for Al (devised by Quinn in 1962)

eh

plasmon

Inelastic
scattering
mean-freepath

(Angstrom)

Electron energy above EF (eV)

plasmon scattering dominates for electron energies above about 30

33 eV

Experimental determination from electron spectroscopy of electron

attenuation lengths in many different materials (Seah & Dench, 1979)

e-

Attenuation
Length
(Angstroms)
d

B
detect attenuation
of electrons
emitted from B as
function of
thickness of
overlayer film of A

Electron energy above EF (eV)

I I 0 exp( d / )
34

2. elastic electron scattering

Elastic scattering can also

contribute to attenuation length
measurements by increasing the
average escape distance

Part of the incident flux is elastically (back-)

scattered out of the crystal by each atomic
layer so incident flux is attenuated as it
penetrates the solid due to elastic
scattering as well as inelastic scattering

35

Quantitative surface structure determination

(Electron) Diffraction at Surfaces & the Reciprocal Net
k

First, recall the situation in 3-D

periodic solids
2
2
k
'

k
conservation of energy

conservation of reduced electron wavevector (momentum) k ' k g hkl

where g hkl ha * kb * lc * is a reciprocal lattice vector

bc
ca
b* 2
V
V
In a 2-D periodic system
a* 2

conservation of energy

k' k
2

c* 2
2

a b
V

so k ' //

V a.b c
2

k ' perp k // k perp

2

conservation of reduced electron wavevector parallel to surface

k ' // k // ghk
bn
a* 2
A

where g hk ha * kb *

na
b* 2
A

is a reciprocal net vector

A a.b n

n is a unit surface
36
normal

Ewald Sphere Construction

a convenient representation of these conservation equations

3-D case
1. Draw the vector k to the origin of the
reciprocal lattice
2. Draw a sphere, radius |k| centred on the
start of the vector k

3. Diffracted beams k correspond to the

vectors from the centre of the sphere to the
intersections of the sphere with reciprocal
lattice points

NB - the sphere ensures energy conservation, the reciprocal lattice

points ensure reduced momentum conservation
37

Ewald Sphere Construction

a convenient representation of these conservation equations

2-D case
1. Draw the vector k to the origin of the
reciprocal net
2. Draw a sphere, radius |k| centred on the
start of the vector k

3. Diffracted beams k correspond to the

vectors from the centre of the sphere to the
intersections of the sphere with reciprocal net
rods drawn through reciprocal net points and
perpendicular to the surface
NB - the rods show that electron momentum transfer
perpendicular to the surface is a continuous variable - but the
parallel component is discrete

38

NB: - a key practical difference between 2-D and 3-D

what happens when one changes the incident energy (wavelength)? change the sphere radius
3-D case: some beams vanish,
new beams appear

2-D case: beams simply change

angle but are continuously visible

g20
39

Low Energy Electron Diffraction - LEED

typical energies 30-300 eV - WHY?
Wavelength interatomic distances

p2
h
h
E so
de Broglie
and
2m
2mE
p
with E in eV, in ngstroms, we have

150.4
E

e.g. if E=150.4 eV, = 1.0

Surface specificity - this energy range gives highest elastic
and inelastic electron scattering cross-sections

40

Practical implementation: NB: - only elastically scattered

electrons are coherent and so
form diffraction pattern

LEED optics
Field-free space

Retarding-field region
Acceleration region - electrons
hit fluorescent screen
high-pass filter

~+5 keV

all in ultra-high
vacuum

41

LEED pattern is a projection

of the reciprocal net with a
magnification determined by
the electron energy (and thus
Ewald sphere radius)

42

Low energy electron diffraction (LEED)

Retarding field analyser (RFA)

Low energy electron diffraction (LEED)

Low energy electron diffraction (LEED)

Cu(100)

Cu(111)
E0 ~ 60 eV for
all 3 surfaces

Cu(110)

LEED pattern is simply a

projection of the reciprocal
net with a magnification
factor determined by the
electron energy (and thus
Ewald sphere radius)

Si(100) (2x1) surface

Si(111) (7x7) surface

LEED

47

Clean GaAs(001)
Surfaces

 from the
diffraction
pattern
determine the
reciprocal net
and hence
invert to
obtain real
surface net.

NB:
Nomenclature for
diffracted beams
- indexed relative
to substrate
reciprocal net

Real space

Diffraction pattern (k-space)

square
(1x1)

centred
rectangle
(1x1)

square
c(2x2)

square
(2x1)
49

Overlayer Structures
with LEED
(4x2)

c(4x2)

Two domains of
the (4x2) rotated
90o (a) and 45o
(b) relative to each
other

Determining the surface net from the diffraction pattern

- the problem of domains when the point group symmetry of the surface
net is lower than the point group symmetry of the substrate

rectangular
on square

rectangular
on
hexagonal

51

Multiple domains can also have an effect when point group

symmetry of the surface structure is lower than that of the substrate

both domains have

same diffracted beam
locations, but different
(2-fold symmetric)
relative intensities.
Sum of two is 4-fold
symmetric

General conclusion the effect of domains ensures that the point group
symmetry of the surface diffraction pattern is always the same as that of
the substrate.
52

Reflection High Energy Electron Diffraction (RHEED)

(LEEDs high energy cousin)

In LEED, low energy electrons used

- provide large elastic scattering cross-section for back-scattered electrons
- keep the penetration depth of the electrons short
In RHEED, another solution is used
- provide large elastic scattering cross-section for forward-scattered electrons
- keep penetration depth small by using grazing incidence
- High energy electrons (>15 keV) used with grazing incidence geometry (< 2o)
- Surface sensitivity due to low component of momentum normal to surface (few )

Kinematic Basis of RHEED

No translational symmetry in surface normal
- surface is 2D periodic
- 5 Bravais lattices; (square, rectangular, centred rectangular, hexagonal, oblique)
2D unit mesh (cell) defined by lattice vectors a1 and a2

ki

kf
ki||

kf||

Conservation of energy (E = 2k2/2m):

|ki| = |kf|
Conservation of parallel momentum (Laue condition):
Dk|| = kf|| ki|| = G(h,k)

h
h

p mv

2
1
2 p
E mv
2
2m

h
2mE

convert to and E to eV

150
E

G(h,k) = ha1* + ka2* is the 2D reciprocal lattice vector

symmetry of reciprocal 2D lattice and real space 2D lattice are the same

Kinematic Basis of RHEED

Ewald sphere construction
for 2D diffraction
k
k

Ewald sphere

f
k

f
G (h,k)

diffraction

(00)

reciprocal lattice

|ki| = |kf|

Dk|| = kf|| ki|| = G(h,k)

rods

No translational symmetry in the surface normal (2D)

- reciprocal lattice rods which are perpendicular to the surface
Ewald sphere is constant energy sphere of radius ki
- diffraction occurs when sphere intersects a reciprocal lattice rod (at kf)
streaked RHEED patterns generally observed from flat surfaces
- due to thermal broadening of the lattice rods and surface imperfections

pattern

Theory similar to LEED:but the incident wavevector IkI is now very large

56

RHEED Geometry

57

Surface Reconstruction
Semiconductor surfaces are characterised by dangling bonds (unsatisfied bonds) due
to lower coordination of surface atoms compared to bulk
Many semiconductor surfaces (polar surfaces) undergo a reconstruction to reduce
number of dangling bonds and minimise the surface energy

new larger 2D unit cell

Formation of surface dimers is common - change of periodicity:
e.g. GaAs(001)-(2x4)
x2

x1

Ga
As
Other examples include:

Si(001) - (2x1) and Si(111) - (7x7)

GaAs(001) - c(4x4), c(8x2), (4x2), (2x3),

InSb(001) - c(4x4), (1x3), c(8x2), (1x1).

Origin of RHEED Patterns

e.g. GaAs(001)

RHEED patterns along different azimuths

GaAs(001)-(2x4) Surface Structure

unit cell for
ideal terminated
surface

order
streaks
side
view

plan
view

[110]

order
streaks

[110]

streaked diffraction pattern

(2x4) unit cell

4th layer Ga
3rd layer As
2nd layer Ga
1st layer As

[110]
[110]

2x periodicity due to As dimer formation

4x periodicity due to existence of 2 dimers
and by 2 missing dimers
reconstructed surface characterised by (2x4)
unit cell

Clean GaAs(001)
Surfaces

Reflection High Energy Electron Diffraction (RHEED)

Ga

As2/As4

In-situ monitor of
MBE growth
RHEED pattern
electron gun

GaAs

High energy electrons (>15 keV up to 35 keV) used with grazing incidence
geometry
Ideal geometry for in-situ monitoring of growth

Diffraction pattern obtained during growth provides information regarding

surface structure (2D unit cell) and surface quality during deposition
Intensity of diffraction features can be monitored and provide calibration of
growth rates RHEED intensity oscillation technique

Surface reconstructions observed during GaAs(001)

growth by molecular beam epitaxy (MBE)

Growth diagram obtained in-situ by

RHEED
Strong dependence on substrate
temperature and incident flux ratio
(BEP = beam equivalent pressure)
14 different surface reconstructions - both
As and Ga terminated
Most common growth surface is (2x4) As
terminated surface
~ 500-600 C and As/Ga flux ratios < 1:1

Classic Thin Film Growth Modes

3D island growth

2D layer-by-layer

Frank van der Merwe (FvdM)

Dg = gf + gi - gs

Volmer-Weber (VW)

2D3D growth

Stranski-Krastanov (SK)

gs,f = surface energies of substrate and epilayer

gi = interfacial energy

Dg < 0; 2D growth Dg > 0; 3D growth

2D growth occurs when the atoms of the deposit material are more strongly
attracted to the substrate than to themselves
3D growth occurs when the deposit atoms are more strongly bound to each
other than to the substrate
2D-3D growth occurs in lattice mismatched heteroepitaxial systems: increased
strain leads to an increase in gi as the film thickness is increased
e.g. Ge/Si (~ 4%), InAs/GaAs (~7%)

Monitoring 2D Layer-by-Layer Growth: RHEED Intensity

Oscillations
Ga flux on

Ga

= 0 ML

Ga flux off

intensity (arb. units)

= 1.25 ML
= 0.25 ML
= 1.5 ML
= 0.5 ML
= 1.75 ML

10

20

30

40

50

= 0.75 ML

time/s

measure intensity of specular beam spot

as a function of growth time
oscillatory behaviour due to 2D layer-bylayer growth mode - 2D island nucleation
and coalescence model
provides accurate calibration of growth
rate by measurement of Ga

= 2.0 ML
= 1.0 ML

Semiconductor Quantum Dots (QDs) formed by

Self-Assembly
Formed during lattice mismatched heteroepitaxial growth

Initial 2D layer formed

so called wetting layer

Deposition

Stranski-Krastanov growth:
strain plays an important role

Substrate

2D wetting layer

Self-assembled
Coherent (dislocation free)
3D islands occur at some
critical coverage crit
Growth mode transition easily
monitored in MBE using RHEED
Important factors; size, shape,
density, composition (alloying
can be important)

3D islands
Examples:
InAs / GaAs
Ge / Si
InAs / InP
InSb / GaAs
InSb / Si

7.2%
4.1%
4.0%
14.6%
20.2%

Using RHEED to Monitor S-K Growth

Example: InAs/GaAs(001) growth by MBE: 7.2% lattice mismatch
Growth mode transition easily monitored
through abrupt change in RHEED pattern

= 0 ML

Chevrons

= 1 ML
> critical

[110]

2 ML

Spots
critical ~ 1.8 ML
[110]

[110]

GaAs(001)

2D wetting layer

c(4x4)

(1x3)
[110]

Streaked diffraction pattern

characterises 2D growth mode
Surface reconstruction changes from
c(4x4) to (1x3) during InAs growth

transmission diffraction features

3D islands (QDs)

Chevrons can provide information

on facet formation and shape

Summary of electron diffraction techniques:

LEED

Typically 30 -150 eV used to maximize the surface sensitivity.

Electrons elastically scattered-information on surface periodicity /
reconstruction.
Typically used to monitor static structures and determine the structure of
clean and adsorbate covered surfaces.
Can be used in I-V (I-E) mode in conjunction with multiple scattering
calculations for a full structural analysis approach

RHEED

Typically uses 15 keV electrons at very grazing incidence to maximize

the electron path length in the near surface.
Used to monitor growing surfaces (geometry in separate azimuths)
Used for calibrating deposition rates (RHEED oscillations)
Can detect 2D to 3D transitions in growth e.g. layer-by-layer to quantum
dot growth
Can also be used in higher pressure environments (<10-5 mbar) e.g. during
growth by pulsed layer deposition (PLD) in a high oxygen atmosphere.

Surface structure determination using LEED

From diffraction pattern can determine the reciprocal net and
hence invert to obtain the real surface net.
Try to extract any additional symmetry information from the
diffraction pattern (point group, space group) - if possible!
Can (potentially) determine atomic positions within the unit net by
analysing the diffraction beam intensities
In bulk structure determination by X-ray diffraction - measure intensities
of many diffracted beams can use Fourier transform of the intensities
to get some direct information on the structure
Surface structure determination using LEED I-E
Dkperp is a continuous variable - measure intensity-energy
spectra of diffracted beams, i.e. I-E plots
effect of complex scattering factors and multiple scattering
precludes use of Fourier transforms - use trial-and-error method
69
with multiple scattering simulations

Fundamentals of ESCA
(Electron spectroscopy for chemical analysis)
Basic principle:- the binding energies of atomic core levels
are characteristic of the atom or species present on the
surface - use this to identify the elemental species
Basic technique:Use electrons or photons as the incident probe to eject electrons
(cf: photoelectric effect)

Detect the electrons emitted from the surface

Measure their kinetic energy (and the emission direction)
Qualitative and quantitative elemental determination
Why does it work?
Exploit the surface sensitivity of low energy electrons

General classification of electron spectroscopy methods

Basics: exploitation of the photoelectric effect

What is the response of an atom to the loss of an electron?
1. Atom becomes a positive ion.
2. Coulombic screening of the nuclear charge is reduced
3. Energy levels adjust to new screening and chemical environment

K L M

KE = h - BE

The photoelectric process

Ejected Photoelectron
Incident X-ray
Evacuum

XPS spectral lines are

identified by the shell from
which the electron was
ejected (1s, 2s, 2p, etc.).

Conduction Band

EF

The ejected photoelectron

has kinetic energy:

L2 , L3

KE = h BE

Valence Band

2p
2s

1s

L1

Incident photons are weakly attenuated but

electron emission signal only detected from
close to surface XPS information
surface specific

Following this process, the atomic

energy levels will relax in response to
the core hole before de-excitation,
by the release energy, possibly by the
emission of an Auger electron or an
X-ray photon.

Surface sensitivity and specificity

Electron Mean Free Path

Mean Free Path (nm)

10

Au

Al
Au
Au Ag

Ag
Ag
Au

Au

Ag

0.5
0.3

10

Be
Be
Ni Be
Ag
P
Fe
W
Ag
Ag
Mo Be

50

Au
Ag
Be
Ag
Ag
C
Mo

100

500

Au
C
Be

Mo
W

1000 2000

Electron Kinetic Energy (eV)

Measure attenuation lengths of electrons through known layer thickness

Universal curve of inelastic mean-free path () vs. KE
Minimum of 0.5 - 1 nm for KE of 50 -1000 eV - Maximum surface sensitivity

X-ray photoelectron spectroscopy (XPS)

h
Sample
Analyser /
Detector

e-

Ek
h

Ek
Vacuum level

sa

sp
Fermi level

Ebf

Conduction
band
Core level

kinetic energy of the

peaks in the photoelectron
energy spectrum identifies
the emitting atom

Ek = h Eb sp

Available photon energies for XPS

deep core levels

Accessible core levels

essentially for all elements
and (low) photoelectron
kinetic energies
i.e. those which
correspond to short
attenuation lengths

valence levels

XPS - h values used - 1487 eV (Al K ) & 1254 eV (Mg K )

Photoionisation cross-sections for XPS

Essentially all elements
have one (or more) core
levels with similarly large
photoionisation crosssection (hence similar
sensitivity).

uppermost
factor of 10 in
cross-section
XPS cross-sections:
photoionisation cross-sections at
hu =1500 eV (cf. Al K)

X-ray photoelectron spectrometer

X-ray photoelectron spectrometer

Computer
System
Hemispherical Energy Analyzer
Outer Sphere

Magnetic Shield

Analyzer Control

Inner Sphere

Electron
Optics

Lenses for Energy

Adjustment
(Retardation)

Multi-Channel Plate
Electron Multiplier

Lenses for Analysis

Area Definition

Position Address
Converter

Resistive Anode
Encoder
Position Computer

X-ray
Source

Position Sensitive
Detector (PSD)
Sample
5 4 .7

What is the photoelectron binding energy measured in XPS?

The simple view:

KE = h - Eb

Koopmans theorem

Eb is the one-electron BE

but this is NEVER observed!

The apparent (photoelectron) binding energy is the difference between

the energy in the initial state and the energy in the final state
initial state - neutral ground state atom + photon
final state - core-ionised atom + energetic (KE) electron
What is the time-scale of the emission process? - how does
the core-ionised atom respond?
Free atom - adiabatic approximation (fully relaxed)
KE = hu - Eb + Ea

Ea = intra-atomic relaxation energy

This would be fine if photoemission was a slow process in reality it is

rapid so valid to use the Sudden approximation
i.e. Final state is one where electron is in an excited bound state of atom
or is ejected into the continuum of unbound states above Evacuum

What is the photoelectron binding energy measured in XPS?

Free atom - adiabatic approximation (fully relaxed)
KE = hu - Eb + Ea
Ea = intra-atomic relaxation energy

Free atom - sudden approximation

shake-up - other electrons in excited bound states
shake-off - other electrons in continuum (for atoms
= multiple ionisation)
Solid - adiabatic approximation (fully relaxed)
KE = hu - Eb + Ea + Er
Er = inter-atomic relaxation energy

Solid - sudden approximation

shake-up and shake-off - but metals have a
continuum of states above Ef - so even excitation
to states below vacuum level are shake-off-like

What is the photoelectron binding energy measured in XPS?

What are Auger electrons?

How does the core hole created by the photoionisation get
refilled? - two alternative processes
KE = EA - EB - EC

h = EA - EB

EC

EC

EB

EB

binding
energy

EA

EA

branching ratio favours

Auger process for small
BE (< 10 keV) - i.e. low Z
elements i.e. Z < 33
NB: Auger electrons also
have energies that are
characteristic of the
emitting atom

X-ray emission

Auger electron emission

(c.f. lab X-ray tube)

(radiationless decay)

Mg K = 1253 eV
Zn-Cu-Co alloy

XPS spectra show

BOTH photoemission &
Auger electron peaks
Enhances the spectral
finger-print of the atom
(element) - several peaks
with characteristic relative
intensities on staircase
background.

Al K = 1486 eV
Photoemission intensity

Cu(100) sample

Distinguish photoemission
and Auger peaks by
changing photon energy

KEphotoemission= h - EA
KEAuger = EA - EB EC
inelasticallyscattered
background

XPS & Auger peak notation

photoemission peaks: quantum numbers n (number), l
(letter: l =0 is s, l =1 is p, l =2 is d, etc), j (number)
e.g. 2s, 2p1/2, 3d5/2 etc.

Auger electron peaks: use X-ray shell notation

for the three energy levels EA, EB, EC
so 1s=K, 2s=L1, 2p3/2 & 2p1/2 = L2, L3
3d5/2 & 3d3/2 =M4, M5
Photoemission intensity

e.g. 2s core hole, 2p3/2

& 2p1/2 final holes
is an L1L2L3 Auger peak
NB: L1L2L3 L1L3L2

Chemical state information in core level spectroscopy especially XPS

Recall - basis of XPS is that core level
binding energies are characteristic of
the atomic species and so can be
used for element
identification/quantification
However this is not the
whole story ..

kinetic energy of the

peaks in the photoelectron
energy spectrum identifies
the emitting atom

Chemical effects in XPS

How can the photoelectron binding energy of an atomic core level be influenced
by the surrounding atoms (chemical bonding = electronic structure)?
initial state effect - change in one-electron binding energy Eb

due to change in valence electron environment

final state effect - change in inter-atomic relaxation energy Er

combined effect leads to an experimentally-observed

chemical shift
NB: - the two contributory effects are NOT separable experimentally
Use the chemical shift as a spectral fingerprint of different
bonding environments

Auger Electron Spectroscopy (AES) is a different technique

Use incident energetic electrons to produce core ionisation
all peaks are due to Auger electron emission

No equivalent of photoemission peak because in electron

ionisation incident electron does not give up all its energy
Surface specificity due to inelastic & elastic scattering of outgoing
electrons - essentially same as XPS

Practical problem: use of incident electrons creates a large

background of inelastically-scattered and secondary electrons - so
signal-to-background ratio is poor
Solution: use electronic differentiation of spectrum to suppress
background and allow amplification
Consequence: peak in N(E)
becomes double feature in
dN(E)/dE (or N(E))
E

typical Auger electron spectra of Si(001)

incident
electrons
3-10 keV

eelectron
energy
analyser

NB: unlike with photons, the

electron ionisation cross-section
peaks around Ep = 3 x B.E.

AES - an application - characterising layer-by layer growth

Ag
V

knees in the AES

amplitude vs. exposure
show monolayer
completions and
provide an absolute
coverage calibration

AES - an application - identifying grain

boundary segregation in intergranular fracture

200 m

0.45 ML of P found
had segregated to
grain boundaries

Turbine rotor failure - Hinkley Point power station, UK.

Quantification of surface composition

XPS:- Measured intensity of emission depends on:
Surface composition (how much of an element)

Photoionisation cross-section
Instrumental factors (analyser efficiency, sensitivity etc.)

Electron attenuation length

Depth distribution of elements
many of these factors can be accounted for by using standard
reference spectra for relative peak intensities

AES:- Also need to think about:

Auger branching ratio
Contribution to ionisation from backscattered electrons
depends of nature of substrate

XPS

vs.

AES

Incident X-ray photons - difficult

to focus, low flux

Incident electrons - can be high

flux, easy to focus and scan

Low flux density - gives low

damage

High flux density - can produce

damage

Good signal-to-background
ratio

Signal-to-background ratio less

good because of secondary e-s

Difficult to scan (raster) so

imaging is NOT routine (but
is done)

Ease of focus allows good spatial

resolution & imaging.

Tends to be higher cost

Can be achieved with a LEED optics

Both can be used in conjunction with ion beams for depth profiling

XPS of a dilute III-N-V alloy GaN0.104As0.896

Wide Scan XPS of GaN0.104As0.896
3500
Intensity (arb. units)

Ga (LMM)

3000
2500

As (LMM)

N 1s

300

400

2000
1500

Ga 3d Ga 3p Ga 3s
As 3d

1000

As 3p As 3s

500
0

100

200

Binding Energy (eV)

500

Quantitative determination of N content in Ga1-xNxAs by XPS

N 1s Ga LMM
GaNAs
GaAs

x = 0.104

GaNAs - GaAs
394

x = 0.083

396
398
400
402
Binding Energy (eV)

404

x = 0.073

396

398

400

402

Binding Energy (eV)

Al Ka XPS (hv = 1486.6 eV)

N 1s core-level at 397.6 eV,
and the Ga LMM Auger
peaks at 398.9 eV and
centred at 393.7 eV

404

394

396

398

400

402

404

Binding Energy (eV)

Difference spectra from subtracting

a normalised GaAs spectrum from
the GaNAs spectra with 10.4% N
The N 1s peak at 397.6 eV is now
clearly resolved.

Band off-set determination using XPS: ZnO/AlN

T.D. Veal, P.D.C. King, S.A. Hatfield, L.R. Bailey & C.F. McConville, Applied Physics Letts. 93, 202108 (2008)

XPS of stainless steel treated with anti-wear additive

zinc dialkyldithiophosphate (ZnDTP)

Analysis of carbon fibre based polymer composite

material by XPS
XPS analysis identified the functional
groups present on composite surface.
Fluorinated
SideofChain
Polymerinterface
- Angle
Chemical
nature
fiber-polymer
will influence
its properties
.
Resolved
C 1s.
33%

e-

49%
6

CF3

3% 18%

(5CF2)2
3

CH2
O

C=O

[1CH2

C ]n

Woven carbon
fiber composite

CH3

Analysis of materials for solar energy collection

XPS Depth Profiling - The amorphous-SiC/SnO2 interface
Profile indicates a reduction of the SnO2 at the
interface during deposition. Such a reduction
would effect the collectors efficiency.

Photo-voltaic Collector

SnO2
Sn

Solar Energy

Conductive Oxide- SnO2

p-type a-SiC
a-Si

Depth
500

496

492

488

484

Binding Energy, eV

480

XPS analysis of pigment from Egyptian Mummy

Pb3O4

Egyptian Mummy
2nd Century AD
(World Heritage Museum
University of Illinois)
PbO2
C
O
150

145

140

135

130

Binding Energy (eV)

Pb Pb

N
Ca
Na
Cl

500

400

300

200

Binding Energy (eV)

Pb

100

XPS analysis showed

that the pigment used
on the mummy
wrapping was Pb3O4
rather than Fe2O3

The impact of XPS?

X-ray Photoelectron Spectroscopy (XPS),
also known as Electron Spectroscopy for
Chemical Analysis (ESCA) is a widely used
technique to investigate the chemical
composition of surfaces.
Kai Siegbahn (1918 - 2007 )
1981 Nobel Prize in Physics

for his contribution to the development

of high-resolution electron spectroscopy "
Phys. Rev. 105, 1676, (1957)

Instinct says beer. Reason says Carlsberg

Instinct says Surface Analysis. Reason says XPS

Electron scattering methods for surface structure determination

LEED - remote electron source - incident
plane waves - interference - diffraction.
NB - much of interference comes from
substrate layers

Alternative approach - use photoemission from a core level of a surface

(adsorbate) atom as the source of the electrons which interfere

Photoelectron Diffraction detect photoelectrons

(Surface) Extended X-ray Absorption

Fine structure - SEXAFS - detect
103
photoabsorption

Surface structure & composition using scattered ions

Factors governing ion-surface interaction
e.g. Energy, mass, charge exchange, etc.
Surface structure & composition
Low energy ion scattering (LEIS)
Coaxial impact collision ion scattering spectroscopy (CAICISS)
Sub-surface structure & composition

Medium energy ion scattering (MEIS)

Secondary ion mass spectrometry (SIMS)
Combinations of techniques

First UK MEIS data

Factors governing ion-surface interaction

Energy:

Mass:

LEIS:
MEIS:
RBS:

0.5 - 10 keV
50 - 400 keV
>1 MeV

SIMS:

250 eV - 2 keV

Hydrogen (H+)
Helium (He+) or Neon (Ne+)
Alkali metals (e.g. Li+)
Oxygen (O+) and Caesium (Cs+)

Momentum transfer: (e.g. sputtering)

Charge transfer:

(e.g. neutralisation)

Ion scattering spectroscopy: basic principles

Binary collision model in ion
scattering

Two-body billiard ball collision:

Conversation of energy
E0 = E1 + E2
E0 = M1v02/2 etc.
Conservation of momentum
M1v0 = M1v1cos1 + M2v2cos2
M1v1sin1 + M2v2sin2

E1
1
2
2
1/ 2 2

[cos

(
A

sin

)
] Why is the binary collision model
1
1
2
E0 (1 A)
adequate (ignore solid)?

E2
4A
2

cos
2
2
E0 (1 A)
where A is the ratio of the
masses A = M2/M1

Duration of collision is short

(cf. vibrational timescale)
Energy transfer large (cf
binding energy of the atom
106
to the solid)

Low energy ion scattering (LEIS)

1-10 keV He+ , Ne+ or Li+

Fixed scattering angle

(usually as small as
possible or 90)

Compositional information
from ion energy losses
Surface structure determination
via shadowing & blocking cones

Low energy ion scattering (LEIS)

1-10 keV He+ , Ne+ or Li+

Fixed scattering angle

(usually as small as
possible or 90)

Compositional information
from ion energy losses
Surface structure determination
via shadowing & blocking cones

Low energy ion scattering spectroscopy - element

identification through scattered ion energy spectra
Typical 1 keV He+ LEIS spectrum
scattering angle 90o so

E1 A 1

E0 A 1

Note: peaks get closer as M2

increases
- mass resolution best for large
scattering angles and small A
values
Surface specificity & shadow cones

1 keV He+ ions

NB: shadow cone widths are close to

inter-atomic spacing at low energies
Focussing effect at edge of cones
(O atom)

109

Several different energy regimes of ISS

in addition to LEIS - Low energy (~500-5000 eV, He+, Ne+..) also have

MEIS - Medium energy 50-400 keV, H+, He+ ...

HEIS - High energy

1-2 MeV, H+, He+ ...

At these high energies:

shadow cones much narrower

ions penetrate below surface layer

Shadow cone width (as characterised here by 5o
impact parameter) decreases with :-

increasing energy
decreasing ion mass (nuclear charge)
decreasing scatterer atom mass (nuclear charge)
MEIS & HEIS are not intrinsically
surface specific

110

Surface structure determination using ion scattering

General principle is to exploit elastic shadowing
Detailed approach different for wide shadow cones (LEIS) and narrow
shadow cones (MEIS, HEIS)

First consider LEIS (say 1-5 keV - mainly He+)

NB - wide shadow cones - limited
sub-surface penetration
1 keV He+ ions

(O atom)

use different incident directions to

illuminate/shadow subsurface atoms
i.e. exploit incident ion shadowing
111

Multiple scattering in LEIS - a source of structural information

scattering from atomic chain

pseudo-double
scattering

treat as succession of binary collisions

At grazing incidence skimming

trajectories lead to minimum and
maximum possible scattering
angles which are related to the
atom spacing in the plane of
incidence

pseudo-single
scattering

BUT neutralisation may

suppress scattered ion signal
from these trajectories
112

Influence of charge exchange on multiple scattering in LEIS

Li+

especially
scattering achieve charge
equilibrium when
close to the
surface
final charge
state determined
by outgoing
trajectory only
(point of
equilibrium loss)
no suppression
of multiple
scattering signal

mechanisms

especially He+ scattering

- the longer the trajectory
close to the surface, the
more neutralisation
trajectory-dependent
neutralisation probability

suppresses multiple
scattering signal
resonant
charge
exchange
(ionisation &
neutralisation)

Auger
neutralisation

113

Newtonian scattering of a classical particle?

30

60

90

120
0

30

60

90

120

and at the atomic scale .

E1
E0

(A 1)2
(A + 1)2

i.e. Life gets simpler if the total

scattering angle is 180 !!
NB:- Remember A is the ratio of the masses of the scattered & target atoms

Shadowing & Blocking

Shadowing cone
formation for calculated
ion trajectories at grazing
incidence (here shown at

c the critical angle)

i.e. top layer specific!

Blocking cone formation
shown by ion trajectories
starting from a point source
(in the third layer). Note the
blocking cones are broader.

CAICISS
(Coaxial Impact Collision Ion Scattering Spectroscopy)

(c)

(d)
Time of flight
Sample
angle (polar/azi)

Pulsed ion beam

(b)
MCP detector

(a)
Beam chopping
and steering

Time of flight

He or Ne

Ion beam
Source (Neilson)

Conceived by M. Aono et al. RIKEN, Japan

CAICISS @ Warwick

Analysing CAICISS data

Schematic showing how the bond direction and bond lengths of neighbouring
atoms A and B are determined at the same time using CAICISS

Structural changes to Si(111) - ( 3x 3) - Ag

caused by atomic hydrogen (H*)
Proposed model for this
metal-semiconductor system
(1)

Ag / (3 x 3)-Ag : Epitaxial growth

of Ag on Si (111)

(2)

H / (3 x 3)-Ag : Real-time
monitoring of structural change

(3 x 3)

Adsorption
of H at R.T.

Desorption
of H

(1x1)

(LEED)

Interaction of H with Si (111) - ( 3 x 3) - Ag

Series of time-of-flight CAICISS spectra each showing two

peaks (Ag and Si) for (a) exposure to H* at R.T. and
(b) following the H* desorption sequence

CAICISS of Si (111) - ( 3 x 3) - Ag atomic hydrogen (H*)

Separation of ions and neutrals

using a synchronised pulsed
voltage.

At 2 keV the probability of ion

neutralisation for sub-surface
scattering events is ~100%

Crystalline Ag and an Ag atom of the

(3 x 3) structure have the same
neutralisation efficiency for He+ ions.

Hence the intensity ratio in the Ag

(ion) peak is ~0.25 implying that the
Ag clusters are 4 atomic layers
(~7) high.

Medium energy ion scattering (MEIS)

Surface relaxation by shadowing and

blocking

MEIS/HEIS - much narrower shadow cones - must use

specific incidence directions for surface specificity

2 MeV He+
scattering from
W(100)

sub-surface
shadowing
(surface-specific
signal) in
crystallographic
(channelling)
directions

surface peak

incidence
direction

124

Subsurface scattering
elastic + inelastic

Channelling in MEIS

MEIS - more precise structural information

from double-alignment experiments
scattered
ion
blocking

incident ion
shadowing

Measure blocking
curves of
elastically
scattering ion flux
(surface peak) as
function of
outgoing direction bulk and surface
blocking dips are
displaced for a
relaxed surface

126

101 keV

H+

ions from Ni(110)

Analysis of MEIS spectra

MBE grown & ion implanted samples

Annealing a single Sb d layer : monitoring diffusion

Difference spectra indicate Sb in lattice

sites.
Lattice site occupation high (~75-80%) even
after 450C anneal.
At 640C non-lattice site Sb appears along
with surface segregated material, although
substitutional Sb remains at the same depth
with no apparent broadening.
By 800C very little sub-surface Sb
remains and only a peak at the surface
can be seen.

MEIS from an Al-Pd-Mn quasi-crystal

Intensity (counts)

4200

3800

3400

3000
0

60

120

180

Azimuthal Angle (deg)

LEED pattern at 75 eV showing

the 5-fold icosahedral symmetry
of a Al70Pd21Mn9 quasi-crystal.

MEIS data showing local structure

including 5-fold symmetry at 72 intervals

Absolute scattering yields in MEIS - provide quantitative

information on the number of displaced atoms induced by
adsorption
blocking
curves

theory:
e.g. Cu(111)+SCH3 pseudo(100) reconstruction - how
many reconstructed Cu
layers?

2-layer
reconstruction
1-layer
reconstruction
clean surface

131

Daresbury MEIS Facility

Scattering
chamber
Preparation
chamber

Experimental station

500kV power supply

Electrostatic
lens
Collimator

Dipole magnet

Accelerator tube

Ion source
platform

Secondary Ion Mass Spectrometry (SIMS)

Near Surface Processes in SIMS

Several processes :- removal of the matrix & impurity incorporation

:- implantation of the incident ion beam
:- charge exchange & neutralisation
:- Altered layer forms prior to steady-state erosion
Also, ionisation is influenced by the electronic state of the surface, so
secondary ion yields can vary by 102 - 103 for different elements.
As dopant profiles in semiconductors get shallower, the need for thinner
altered layers increases, hence the need for SIMS at ultra-low energies
(i.e. < 250 eV)

Summary - surface structure determination by ion scattering

Key underlying physics is the use of the shadow cone
LEIS - shadow cone is wide (atomic separation), low subsurface penetration
multiple scattering useful for interatomic determination in
surface plane - but these trajectories may have enhanced
neutralisation probability (especially for He+)
MEIS/HEIS - narrow shadow cones, strong sub-surface
penetration possible in non-shadowing incident directions
absolute scattering yields equate to number of illuminated
layers - so can give structural information from sub-surface
shadowing

use of ingoing shadowing and outgoing blocking

(double alignment) enhances specificity of method
SIMS erodes the surface and analyses the emitted ions to
determine the composition

135

Scanning probe microscopy (SPM) techniques:

Scanning tunnelling microscopy (STM)
Metal surfaces & atomic manipulation
Semiconductor surfaces & epitaxial growth
Atomic force microscopy (AFM)
Conducting & non-conducting surfaces
Biological surfaces
Other applications
Scanning ... force, magnetic, capacitance, electrochemical ... microscopy

Scanning Probe Microscopy

Scanning probe microscopy (SPM)
Covers a range of imaging from several 100 m to 100 pm
Surfaces can be resolved with atomic resolution (STM & AFM)
Resolves structural features, steps, defects, adsorbates in vacuum,
air or liquid
Impact of SPM
SPM has become an essential tool in nanoscience and nanotechnology
Local experiments on single atoms or molecules can be performed
Force measurements of single chemical bonds, biological systems or
optical spectra of single molecules can be performed
Local probe can manipulate materials on the atomic and molecular
scale - build artificial structures on the atomic scale

Introduction

Scanning Tunneling Microscope (STM) was invented by Gerd Binnig and

Heinrich Rohrer at IBM Zurich in 1981 (Nobel Prize in Physics in 1986).

Binnig also invented the Atomic Force Microscope (AFM) at Stanford University,
with Quate and Gerber, in 1986.

All other applications and techniques flow from these two design
concepts.

The Scanning Tunneling Microscope (STM)

The STM is an electron microscope that uses a sharp (single atom)
tip to attain atomic resolution images of the surface electron density
since you cant actually see atoms.

The Scanning Tunneling Microscope (STM)

The STM is an electron microscope that uses a sharp (single atom)
tip to attain atomic resolution images of the surface electron density
since you cant actually see atoms.

Si(111) (7x7) surface reconstruction

Since you are measuring the electronic states, images of the same surface can vary!

Filled states image

Empty states image

Si(111) (7x7) surface reconstruction

Since you are measuring the electronic states, images of the same surface can vary!

Filled states image

49 atom unit cell model for (7x7)

Si(111) (7x7) surface reconstruction

Since you are measuring the electronic states, images of the same surface can vary!

Filled states image

High resolution image

BUT why not see the atomic positions with STM?

Principle of scanning tunnelling microscopy (STM)
Bring conducting (usually W) tip to
within atomic dimensions of surface.
Measure current tunnelling through
vacuum gap between tip and surface - this
depends exponentially on separation
Scan tip parallel to surface - current
will vary (exponentially) with separation
- either map the current as a function
of position or - more usually - map the
height variations applied to the tip to
maintain constant tunnelling current E
Atomic-scale images of the (filled or
unfilled) electronic states at the
surface

tip
surface

unfilled
states

filled states
144

Scanning Tunnelling Microscopy (STM)

Requirements:
z
computer

y
x

Operates in vacuum or air (or

solution)

tunnelling
Current
amplifier

substrate tip w

Ultra-high vacuum allows atomic

resolution

set tunnelling
voltage

Conducting substrate (metal, oxide

or semiconductor)

display

Provides real-space information

Need good vibrational isolation (!)

Practical:
Sharp metal tip (W or Pt/Ir) is brought close to conducting substrate
Motion of tip is controlled by 3 piezoelectric drives (x,y,z)
Bias voltage is applied between the substrate and tip
Quantum tunneling occurs and current flows between the two

Scanning Tunnelling Microscopy (STM)

Requirements:
z
computer

y
x

Operates in vacuum or air (or

solution)

tunnelling
Current
amplifier

substrate tip w

Ultra-high vacuum allows atomic

resolution

set tunnelling
voltage

Conducting substrate (metal, oxide

or semiconductor)

display

Provides real-space information

Need good vibrational isolation (!)

Practical:
Sharp metal tip (W or Pt/Ir) is brought close to conducting substrate
Motion of tip is controlled by 3 piezoelectric drives (x,y,z)
Bias voltage is applied between the substrate and tip
Quantum tunneling occurs and current flows between the two

Quantum mechanical tunneling

Tunnelling current, I ~ exp(-2kw)

where k = (2m)1/2/h , = work function
change in barrier width of 1 leads to x10 change in tunneling current

STM: basic principles

Treat sample and tip as metals
Simple 1D potential model; barrier width = w

Electron in state yn with energy

En lying between EF-eV and EF
has a chance to tunnel into the
tip. The contribution to the tunnel
current is proportional to;

|y n (o) |2 e2k 'w

EF
EF - eV

Sample
tip

Total current I is proportional to

the number of states within the
energy interval, eV;

Ef
I~
|y n (o) |2 e2k 'w
En E f eV

w
Z=0

k ' 2m

Z=w

For small V the summation can be written in terms of the local density of states
E
at the Fermi level
1
2
( z, E )
|y n ( z ) |

En E

Hence current is proportional to

V (0, E f )e2k 'w ~ V (0, E f )e1.025w

i.e. tunnel current is determined by the LDOS of the sample at the Fermi energy

STM modes of Operation

current

small w, large
current

lateral
distance

z-piezo
voltage

lateral
distance

Constant height - tip is

scanned in x,y plane and
remains at constant
height in the z-direction.
Variation in tunneling
current is measured.
Good for very flat
surfaces.
Constant current tunneling current is kept
constant by changing the
height of the tip and the zpiezo voltage is plotted
versus lateral position. Most
common method.

Calibration of tip shape care needed!

How does the image

formed relate to what
you are actually
looking at?

(a good moto whenever

interpreting SPM images!)

STM imaging different materials

Metals - valence electrons generally highly delocalised - strong effect
due to smearing out (Smoluchowski smoothing) - STM images of clean
metals show very weak corrugations (< 0.1 ) - mainly above atoms.
Usually requires low temperatures for high resolution.

Semiconductors - valence electrons more strongly localised and directional

between atoms (covalent bonds) - STM images of clean semiconductors
show strong corrugations (up to 1 or more) - protrusions especially above
dangling bonds

Compounds & adsorbate-covered surfaces - atoms of different elements

may appear quite different due to electronic effects. Some atoms may even
image as dips, not protrusions

Insulators - cannot image with STM

Look again @ STM from Si(111)-(7x7) surface

typical STM image

STM is an electronic probe, so for

surface structural applications there
are key problems:

Do the protrusions correspond to

atomic positions?

Do you see all atoms in the same

way? (chemical effects?)

Are the relative heights and lateral

positions reliable (quantitative)?

NB: valence states at the surface

of a covalently-bonded solid are
localised, so STM shows large
corrugations
Si(111)(7x7) structure

152

STM from elemental metal surfaces

Generally corrugations are very
weak but protrusions are above
surface atoms - image appearance
essentially independent of bias
voltage - but best contrast at low
voltage, very close tip-sample
separation

Strong bias voltage effects (and

hence misleading images) do occur
in special cases - e.g. bcc (110) see simulations on W(110)
153

STM from an adsorbate on a metal - C on Ni(100)

Isolated C
atoms imaged
as deep dips

Ni(100)(2x2)-C p4g
clock reconstruction

low coverage

clock
154

STM from adsorbates on Pt(111) - simulations

P. Sautet, Surf. Sci. 374 (1997) 406155

Added row structure formed by O reaction with Cu(110)

O atoms

Cu(110)-(2x1)-O
outermost layer Cu
atoms
lower layer(s) Cu
atoms

153 x 128
STM image
156

Added row structure formed by O reaction with Cu(110)

Cu atoms are removed
from the atomic steps to
create the added rows so the steps recede as
the added rows grow
153 x 128
STM image

Cu-O-Cu-O chains
are added to the
surface and move
together

235 x 256
STM images increasing O2
exposure

F. Besenbacher, Rep. Prog. Phys.

59 (1996) 1737

157

Clean metal surfaces - Cu(111)

Low temperature STM (>10K)
Observe Friedel oscillations

Molecular imaging with STM

Cyclopentene molecules on a
Ag(110) surface at 80K

Most famous STM image September 28th 1989

Xe atoms on a Ni(110) surface at 4K

Positioning Atoms with an STM

D.M. Eigler & E.K. Schweizer, Nature 344, 534 (1990)

Fe atoms on the surface of Cu(111) at 10K

M.F. Crommie, C.P. Lutz, D.M. Eigler, Science 262 (1993) 218

Molecular manipulation
with STM

P. Zeppenfeld, C. P. Lutz and D. M. Eigler, Ultramicroscopy 4244, 128 (1992)

In-situ STM-MBE system

MBE

STM
GaAs (001)

RHEED
Ga/In

As2/As4
pattern

MBE

electron gun

Quench grown
sample in vacuum
GaAs

Ga/In

As2/As4

- Ga, In, As, Sb, Si solid sources

- N plasma source
- atomic H source

in-situ monitoring of growth by reflection high

energy electron diffraction (RHEED)
atomic scale snapshots of growth by scanning
tunneling microscopy (STM)
- 1 cm2 substrates
- rapid quenching

STM imaging of MBE grown GaAs(110) surface

(110) surface is non-polar and there is no surface reconstruction
200 nm x 200 nm

Top view

unit cell

0.4 nm

Side View

0.57 nm
0.2 nm

Filled states STM image of GaAs(110)

- lone pair on surface As atoms

Example of voltage dependent imaging

Non-polar GaAs(110) surface

+1.9 V

sample bias

Plan view of surface

-1.9 V

Atom selective imaging:

Charge transfer from Ga to As atoms in surface

(1x1) unit cell

Ga dangling bonds empty, As dangling bonds full

Unoccupied state density - Ga (empty states imaging)
Occupied state density - As (filled states imaging)

InSb(001)-c(4x4) surface
Same symmetry and periodicity as for
GaAs(001) c(4x4) surface

High resolution filled states image

(bias = -2.5 V, tunnel current = 1 nA)

GaAs(001) c(4x4)

10

Improved image resolution relative to GaAs.

InSb lattice constant > GaAs lattice constant

Semiconductor Heteroepitaxy
Optoelectronic devices require
combinations of different materials
with different band gaps
Semiconductor heteroepitaxial
growth allows band gap engineering
Relative lattice constants of different
materials play important role in
growth behaviour and quality
low lattice mismatch (typically < 1%)

AlAs/GaAs, CdTe/InSb, InAs/GaSb

% misfit between two different materials:

= (ae as)/as x 100

Ge/Si, InAs/GaAs, InSb/GaAs

ae = lattice parameter of epilayer

as = lattice parameter of substrate

high lattice mismatch (typically > 3%)

strain
dislocations
growth mode changes

STM evidence for 2D Layer-By-Layer growth mode

GaAs(111)A homoepitaxy - 2ML GaAs growth
Direct evidence for 2D island nucleation, coalescence
and completion of each layer
1.25 ML

0.5 ML

1.5 ML

0.75 ML

1.75 ML

Surface step density measured from STM images

Surface step density

0.25 ML

0.5

1.5

2.5

Coverage (ML)

Oscillation in step density correlates with

measured RHEED intensity oscillations
1.0 ML

2.0 ML

1000

Effects of Lattice Mismatch on growth mode

E.g. InAs growth on different low index GaAs substrates ( = 0.0756)
(001)

(110)

(111)A

Plan view STM images

GaAs(001) - 2.5 ML InAs

Stranski-Krastanov
growth
strain relief - 3D islands
(Quantum dots)

GaAs(110) and (111)A - 5 ML InAs

Strain relief - dislocations

2D layer by layer growth

Stranski-Krastanov growth & QD formation with STM

InAs growth on GaAs(001):- 7.2% lattice mismatch
(a)

(c)

(b)

(d)

(a) 1.4 ML: InAs 2D wetting layer

Different grey levels correspond
to step edges (0.3nm) of 2D
wetting layer
(b) 1.7 ML: Initial stages of 3D
island formation. Coexistence of
small and large 3D islands
randomly nucleated on surface
(c) 2.0 ML: Evolution of 3D islands
and rapid increase in number
density
(d) 2.7 ML: Coalescence of 3D
islands into larger 3D islands

Shape of quantum dots from high resolution STM imaging

InAs/GaAs(001)

Atomic scale STM imaging reveals crystallographic facets and the shapes of the 3D
islands
Number of facets formed, e.g. {137}, {111}, {110}
Shape and island aspect ratio depend on growth conditions

Atomic Force Microscopy (AFM)

Reference force

Detector

Mirror
Laser

Cantilever
Tip

Does not require electrically

conducting substrate

Sample
z
y

PZT
x

Tip-sample force interaction used to

probe surface

Scan output
Computer
Z-value

Normally used in ambient

atmosphere and not UHV

Feedback amplifier

Sample mounted on piezoelectric tube (PZT)

Tip (~ 2 m long, 20 nm diameter) mounted on end of cantilever (e.g. silicon nitride)
Displacement monitored using a laser beam (interference shifts)

Light focused onto back of cantilever - collected by photo-detector

Measured cantilever deflections used to produce topographic map of surface
Constant height mode - scanner height is fixed, measure spatial variation of
cantilever deflection (used for fast scanning and real time measurements)

Constant force mode - cantilever deflection used with a feedback circuit that moves
scanner up and down in z-direction so keeps cantilever deflection (and force) constant

Tip-Sample Interactions in AFM

Several different forces contribute to cantilever deflection

Repulsive:
- short distances (few Angstroms)
- due to overlap of electron clouds associated
with atoms in tip and sample
contact regime
Attractive:
- long distances (10-100 Angstroms)
- long range van der Waals interaction

non-contact regime
Contact AFM mode
- tip makes soft physical contact with sample (forces ~ 10-8 - 10-6 N)
- repulsive forces cause cantilever to bend to accommodate topographic changes
- stiff cantilevers allows nano-patterning via surface deformations
Non-contact AFM mode
- cantilever vibrates near sample surface close to its resonant frequency
- detect change in frequency or amplitude of tip oscillation
- minimal contact and low forces (~10-12N) allows study of soft samples

Examples of AFM images from Semiconductors

Si/Ge heterostructure (30% Ge)
MBE (550C)
Roughness (rms)
15.856 nm

InAs/GaAs density & distribution

of quantum dots grown by MBE
40
30
20
10
0
10

CVD (550C)
Roughness (rms)
4.200 nm

Cross-hatch pattern due

to misfit dislocations

20

30

40

50

60

UHV-AFM image
Si(111)(7x7) surface

(Scan size:13nm x 13nm, Cantilever: Piezo

Resistive type, Frequency shift:-33Hz)

Provides information on large area surface morphology on near atomic scale

High resolution biological imaging with AFM

"Nano-Alps" AFM image of a monolayer of

a bacterial surface-layer protein (Bacillus
sphaericus CCM2177).
The layer exhibits square lattice symmetry
with a lattice constant of 13.1nm. The 50
nm image was obtained in contact mode
under water.

AFM reconstruction based on averaging over

100 plaque particles each for the luminal and
cytoplasmic face of an asymmetric unit
membrane (AUM) of urine bladder epithelium.
The luminal side (left) of particles protrudes
about 6.5nm relative to the lipid bilayer and the
cytoplasmic face (right) at only 0.5nm; hence the
name ''asymmetric unit membrane.
Averages are based on contact mode AFM
imaging in buffer solution. These particles form
2-D crystalline plaques in situ. Centre-to-centre
distance is 16nm

Course Summary: the topics

Why, how, what surfaces? Surface sensitivity &

specificity.

Surface structures - 2D symmetry & phenomenology

Diffraction from surfaces, reciprocal net, electron
diffraction (LEED) and RHEED

Chemical structure of surface determination using

electron and photoelectron spectroscopy - XPS & AES including chemical state specificity used for structural
determination e.g. photoelectron diffraction

Ion scattering methods of surface structure

determination (LEIS, MEIS, SIMS)

Scanning probe microscopy STM and AFM

177

Richard Feynman - 1959