ALKALI-METAL

CONTAINING

AMORPHOUS

CARBON:

REACTIVITY

AND

ELECTRONIC

STRUCTURE

M.

TOWE,

P.

REINKE,

AND

P.

OELHAFEN

Institut

fur

Physik

der

Universitat

Basel,

Klingelbergstrasse

82,

CH-4056

Basel,

Switzerland

ABSTRACT

Amorphous

hydrogen-free

carbon films

(spZ-dominated

a-C)

were deposited

under

ultrahigh

vacuum

conditions

between

room temperature

and

800'C.

These

films

served

as

matrices

for

the

in-situ

incorporation

of

alkali-metal

atoms

(Li,

Na).

In-situ sample

characterization

was performed

by

photoelectron

spectroscopy

with

both x-ray

and

ultraviolet excitation

(XPS,

UPS).

While

the

clean

metal-containing

samples

were prepared

with

metal

contents

of

about

10

at%,

a

strong

oxidation

driven

accumulation

of

metal

atoms

on

the film

surface

exceeding

50

at%

was

observed

upon

exposure

to

molecular

oxygen.

Work-function

measurements

by UPS

reflected

the

changes

within

the

electronic

structure

of

the

material.

Metal

incorporation

considerably

decreased

the

work-function,

but

only

after

oxidation

we

observed

work-functions

below

the

values

given

for

pure

alkali

metals.

INTRODUCTION

The

interaction

of

carbon based

materials

with guest

atoms,

especially

metals,

has

long

been

of

considerable

interest

in

various

fields

of

research.

The

best

known

of

these

materials

are

the

graphite intercalation

compounds

(GICs)

which

are

formed

by

the

intercalation

of

donor

or

acceptor

species

into

well oriented

graphite

[1].

As

one

example,

LiC

6

,

a

stage-one

GIC

with

only

one

graphene

layer

separating

the

lithium

layers,

sets

the

stoichiometric

limit

for

the

metal

storage

capacity

of

carbon

electrodes

in

rechargeable

lithium

batteries.

To

overcome

this

limit,

various

forms

of

disordered

carbons

have

been

employed.

The

incorporation

capacity

per

gram

of

some

of

these

materials

by

far

exceeds

the

one

of

graphite

[2].

Among

these

disordered

materials

is

amorphous

carbon

(a-C),

which

mainly

consists

of

spZ-hybridized

carbon

atoms

which

are

the

building

unit

of

graphite,

too.

In the

amorphous

material

their

arrangement

lacks

the

long

range

ordering

of

atoms

into

planar

layers which

can

be

easily

stacked.

It

is

this

disorder

which

offers

new

opportunities

for

the

interaction with

other

species.

The

current

study

is

the

application

of

carbon materials

in

fusion

research

devices

which

has

long

been

dominated

by

graphite

due

to

its

favourable

thermal

and

other

material

properties.

To

improve

energy confinement

in

fusion

plasmas,

it

is

desirable

to

minimize

vacuum

contaminations

which may

cause

radiative

energy

losses. The

main concern

is

on

oxygen

impurity

control

and

light

oxygen-gettering

elements

such

as

boron,

beryllium,

and

lithium

have

been

tested

as

constituents

of

the

first

wall

(B,

Be)

or

as

wall

conditioning

additive

(Li)

[3].

Although

the

fusion

research reactor

may

not

be

all-carbon

based

[4],

there will

still

be certain

applications

which

rely

on

a

more thorough

knowledge

of

the

interaction

of

carbon,

light

reactive

elements,

and

plasma

impurities.

In

our

study,

experiments

are

performed

with

a-C because

this

material's

structural

properties

can

be

controlled by

the

substrate

temperature

during

deposition.

It

is

hoped

that

this

will

contribute

to

the

understanding

of

the

influence

of

carbon structure

on

the

interaction

with

reactive

species.

EXPERIMENT

Amorphous

carbon

samples

were

deposited

on

silicon(100)

substrates

by

electron beam

evaporation

of

graphite.

Depositions

were

carried out

between

room

temperature

and

800'C.

Metal

atoms

were delivered

by

commercial

alkali-metal

dispensers

(SAES

Getters)

through

a

thermally

activated

chemical

reaction.

Performance

of

all

preparation

and

analysis steps

in

an

ultrahigh

vacuum setup

with

a

base

pressure

of

less

than

2.10-1°

mbar

enabled

us

to

keep

oxygen

contaminations

at

this

stage

at

about

1

to

2

at%. The alkali-metal

content

was

increased

stepwise

167

Mat.

Res.

Soc. Symp. Proc.

Vol.

593

©

2000

Materials

Research

Society

and

all

stages

of

sample

processing

were

monitored

by

subsequent

in-situ analysis

of

the

films

by

ultraviolet

and

x-ray

photoelectron

spectroscopy (UPS/XPS)

in

a

Leybold-SPECS

EA11/100MCD

spectrometer.

XPS

was

performed with

MgK

0

-radiation

of

1253.6

eV

with

a

resolution

of

about

0.9

eV.

UPS

were

excited

with

the

Hel

and

Hell

lines

with

energies

of 21.22

eV

and

40.82

eV,

respectively. Resolution

in

UPS

was

0.1

to

0.2

eV.

Electron

energies

are

referred

to

the

Au

4f

7

/

2

-

signal

of

a

clean

gold

sample

at

84.0

eV,

and

its

Fermi

edge,

respectively.

Spectra

are

presented

as

measured with neither

background

nor

satellite

subtraction.

While

XPS

mainly yields

information

about

the

elemental

composition

within

the

escape

depth

of

photoelectrons,

UPS

probes

the

electronic

structure

which

in

the

case

of

carbon materials

is

closely

the

geometric

structure.

In

addition,

our

experimental

setup

allows

the

application

of

a

bias voltage

during measurements

and

thus

the

determination

of

the

sample's

work-function

from

the

spectral

onset.

Controlled oxidation

of

the

clean

samples

was

carried

out in-situ by

exposure

to

molecular

oxygen

in

doses

of

only

several

Langmuir

to

some hundred

Langmuir

followed

by

analysis.

While

some

samples

were

annealed

in-situ

at

up

to

1

100°C

at

the

end

of

an

experimental

series,

others

were

removed

from

the

vacuum

and

stored

in

air for

later

re-analysis.

RESULTS

Carbon

and

Alkali-Metal

Amorphous

carbon

materials

as

they

are

discussed

in

this

study

consist

of

a

carbon

network

which

is

dominated

by

spZ-hybridized

atoms.

The

arrangement

of

these

atoms

in

the

network

depends

on the

substrate

temperature

during

deposition

[5].

The

thermodynamically

stable

carbon

modification

graphite

can

be

obtained

in

polycrystalline

form

at

temperatures

close

to

1000°C

and

in

a

form

with

preferential

crystallite

orientation

at

about

1300'C.

The

formation

of

graphite

requires

the

arrangement

of

sp

2

-hybridized

atoms

in

planar

graphene

sheets

which

can

be

stacked

regularly,

preferably

hexagonally.

At lower

temperatures,

such

a

regular

ordering

of

atoms

cannot

be

obtained over

large

distances,

but

there

still

exist

ordered

regions

with

graphitelike

stacking.

The

size

of

these

graphitic clusters,

which

are

embedded

in

an

amorphous

carbon

matrix,

depends

directly

on the

deposition

temperature.

Schelz

et

al.

[5]

found cluster

sizes

of

up

to

2

nm

in

room

temperature

deposited

material

and

sizes

of

up

to

20 nm

in

films

deposited

at

800'C.

In

the

following,

the

term

'amorphous

carbon'

includes

all

samples

prepared for

this

study,

although

the

ones

deposited

at

high

temperatures

may

exhibit

at

least

medium

range

order.

For

the

accomodation

of

guest

atoms within

the

carbon

network,

the

proportion

of

ordered

graphitic

regions

to

the

surrounding

amorphous

matrix

is

decisive.

As

we

discuss

in

more

detail

elsewhere

[6],

at

least

two

major

binding

environments

for

alkali-metal

guest

atoms

have

to

be

considered.

On

the

one hand,

the

ordered graphitic

regions

offer

the

opportunity

to

insert guest

atoms

into

the

spacings between adjacent

graphene

sheets thus

forming

GICs

of

limited

extension.

On

the

other

hand,

the

small

clusters

provide numerous

atoms

at

the

edges

of

graphene sheets

with

unsaturated

valences. These

edge

atoms

and

atoms

in

the

disordered

matrix which were

not

mobile

enough

to

move

into

equilibrium

bonding

sites

during deposition offer

dangling

bonds

to

available

bonding

partners.

Core

level

photoelectron spectroscopy

reveals

the

diversity

of

environments

into

which

alkali-

metal

atoms

fit.

Figure

Ia

displays

the

Nals

signals

for

metal

contents

of

0

to

15

at%

in

a

carbon

material

originally

deposited

at

550'C.

The

signal's

full

width

at

half

maximum

(FWHM)

is

large

from

the

beginning

(2.3

eV;

pure

sodium:

below

1

eV

[7])

and

increases

to

more

than

3

eV

with

higher

metal

content.

This

broadening

is

observed

for

all

samples,

but

the

higher

the

carbon

deposition

temperature,

the

larger

the

increase

and

the

final

FWHM

at

comparable

metal

content

(fig.

b).

Although

the

increase

in

peak

width

might

include

a

contribution

from

core hole

lifetime

broadening,

the

effect

is

too

pronounced

to

be

accounted

for

solely

by

this

effect.

It

is

therefore

attributed

to

the

presence

of

metal

species

in

a

variety

of

binding

environments

with

locally

different

electronic

charge

distributions

between carbon

and

metal

atoms.

Saturation

of

dangling

bonds,

for

example,

leads

to

the

formation

of

covalent

bonds,

but

the

actual

charge

distribution

168

responsible

for

the

chemical shift

still

depends

on

the local

structure

[8]

around

the

binding

site.

It

should

be

noted

for

the

present

study,

that

the

diversity

of

carbon

sites

in

the

host

material

is

matched

by

the

variety

of

electronic

charge

transfer

states

of

the

guest

atoms.

1075

binding

energy

[eV]

1070

FWHMofNals

I

a

I

in

a-C:Na

:

*

a-C

dep.

800'C

.

b)

*

a-C dep.

550'C

*

a-C

dep.

300'C

*..

*

a-C

dep.

r.t.

3.0

-

2.8

-

2.61k

2.4

2.2

....

A

5'

V

A

l ."

..U..

........

.....

.......

U

...

, A

_

A

,

I A

I , I

0 4 8 12

sodium content

[at%]

fig.

la:

Nals

signal

in

MgK,,

excitation

(1253.6

eV)

for

progressing sodium addition

to an

a-C

film

deposited

at

550'C.

Note

the

increase

in

peak

width

and

the

shift

in

position.

fig.

1b:

FWHM

of

the

Nals

signal

for

sodium

addition

to

a-C

films

deposited

at

four

different

temperatures.

Values

for

the

spectra

from fig.

la

are

included.

Lines

only

serve

as

guide

to

the

eye.

In

analogy

to

well

defined

GICs,

the

metal

containing

materials

of

this

work

exhibited

increasingly

metallic

behaviour

and

developed

a

Fermi-edge

upon

metal

addition

which

can

be

monitored

in

UPS

valence

band

spectra

(not

shown

here).

This

supports

the

assumption

of

intercalate-like

states

in

these

materials.

Accordingly,

work

functions

of

the

samples

decrease

from

the

values

of

pure

a-C

around

4.9

eV

down

to

values

of

around

3.6

eV

at

ca.

10

to

15

at%

sodium.

This

is

well

above

the

value

for

the

elements

in

polycrystalline

form

(Na:

2.36

eV[9]).

Reactivity

Towards

Oxygen

With

the

intention

of

testing

lithium-containing

materials'

oxygen

gettering

capability,

they

were

exposed

to

molecular

oxygen

in-situ.

The

effects

of

this

procedure

on

the

Lils

and

Ols

core

levels

can

be

followed

in

figure

2.

Formation

of

lithium

oxidic

species

is

the

dominant

result

of

the

reaction.

It

can be

monitored

in

the

development

of

a

new

Ols

state

at

about

530

eV,

which

is

typical

of

oxygen

atoms

in

oxides.

A

second

oxygen

state

is

found

at

higher

binding

energy

(ca.

533

eV),

which

is

present

in

traces

(less

than

1

at%)

already

on

clean

carbon

samples.

It

may

be

connected

to

hydroxidic

species

generated

with

ubiquitous

hydrogen

residues

or

may

be

due

to

not

yet

reacted

molecular

oxygen

adsorbates

[10].

In

Figure

2a,

a

broadening

of

the

Lils

state

is

observed

upon

oxidation.

This

is

again

indicative

of

the

co-existence

of

other

metal

atom

species

with

the

oxide.

169

1080

m •

m

I I

m I

I

-7

~Li/O

[at%]:

40.9/19.1

40.2/18.1

34.2

/14.4

24.9/9.9

15.9/8.0

~~ 13.1

/6.8

9.4/5.3

pristine

60

58

56 54

536

532

528

binding

energy

[eV]

binding

energy

[eV]

fig.

2:

MgK,

(1253.6 eV)

excited

spectra

of

Lils

(fig.

2a) and

Ols

(fig.

2b)

states. Lithium

and

oxygen

content

are

given

in

the

middle column,

the

difference

to

100

at%

being

carbon.

While

the

lithium

peak's

signal

to

noise

ratio

is

rather poor

due

to

the

Lils's

small

photoionization

cross

section,

two

oxygen

states

can

be

distinguished

throughout

the

procedure.

The one

at

lower

binding

energy

(ca.

530

eV)

is

due

to

the

metal oxide

and

the

one

at

about

533

eV

has

been

assigned

to

oxygen

species

such

as

suboxides or

hydroxidic oxygen

[10].

The most

obvious

result

of

the

oxidation

is

the

dramatic

change

in

surface

composition

within

the

reach

of

XPS.

The decay

of

GICs

under

the

influence

of

traces

of

oxygen

has

often

been

observed,

but

can

be

avoided

by

manipulation

under clean

vacuum

conditions

[11].

In

our

case

of

controlled

oxidation,

we

deliberately

investigated

the

product

of

this

process.

We

found

an

up

to

threefold

increase

in

lithium

atom

content

of

the

surface

layer due

to

a

reaction

driven

segregation

from

the

bulk.

Although

the

proportion

varies,

the

general

observation

is

that

finally

the

metal atom

to

oxygen

ratio

is

between

2

and

3,

thus usually exceeding

the

stoichiometric composition

of

Li2O.

The

existence

of

other

lithium

species

was

already

indicated

by

peak

widths. Possible

states

of

lithium

are

suboxides

with

a

surplus

of

lithium

per

oxygen

atom

or

atoms

covalently

bonded

to

carbon. For

example,

the

presence

of

a

variety

of

alkali-metal

oxidation

products

was

observed

in

cesium GICs

[12]

and

even

on

Cs

multilayers

[13].

With

respect

to

the

obvious mobility

of

lithium

within

the

material,

there

may even

accumulate

a

reservoir

of

unreacted

lithium

atoms

underneath

the

superficial oxide

layer

as

long

as

the

latter is

acting as

a

diffusion

barrier

for

further

oxygen

supply

from

the

outside.

The

formation

of

a

closed

oxide

layer

rather

than

a

penetration

of

the

carbon

network

by

the

oxide

has

been

confirmed

by

observations

of

the

substrate

signals under

very

thin

films

(not

shown here).

They

are

more

and

more

attenuated

in

parallel

with

the

carbon signal

as

the

reaction

progresses,

thus

indicating growing

thickness

of

the

overlayer. Furthermore,

formation

of

such

a

layer

is in

agreement

with

the

observed

pronounced

decrease

in

work

function

(figure

3).

This

effect

of

a

superficial

layer

is

due

to

the

dipole

moment

of

such

an

overlayer

when

a

preferential

orientation

can

be

assumed.

Considering

the

fact

that

monovalent

lithium

atoms should

170

preferably bind

in

terminal

positions,

insertion

of

oxygen atoms

into

lithium-surface

bonds

is

to

be

expected

at

least

for

a

certain

fraction

of

the

surface

as

it

is

oxidized.

In

this

case,

the

dipole

moment

would

contribute

to

a

lowering

of

the

work

function.

It

should

be

kept

in

mind

that

the

value

determined

by

photoemission

is

governed

by the

surface

areas

with

the

lowest

work

function

which

defines

the

edge

of

the

valence

band

spectrum.

Ik

I

..

-

4.5-

oxidation

of

a-C:M:

fig.

3:

>

-o-

a-C

dep.

800'C

(Li)

Development

of

work

functions

in

2

4.0

-M-

a-C

dep.

r.

t.

(Li)

three

carbon

samples

during

""-

a-C

dep.

r. t.

(Na)

oxidation

of

the

metal

containing

.o

3.5

films.

One

sodium

containing

sample

"Q

|is

included

(open

triangles).

=

3.0-

The

initial

values depend

on

the

__alkali-metal

content

in

the

samples

S2.5

prior

to

oxidation.

Reference values

for

polycrystalline

2.0-

1

lithium

and

sodium

are

2.93

eV

and

0

5

10

15

20

25

2.36

eV,

respectively

[9].

oxygen

content

[eV]

Annealing

of

oxidized

and

non-oxidized

samples

in-situ

at

temperatures

in

excess

of

400TC

led

to

the

removal

of

both lithium

and

oxygen.

At

a

temperature

of

800TC,

only

traces

of

both

elements

remained

while

the

pure carbon

structure's

valence

band

spectra

where

restored.

Only

with

originally

low

temperature

deposited

samples

the

graphitizing

effect

of

the

thermal

treatment

was

visible.

Storage

in

Air

Storage

of

oxidized

as

well

as

non

pre-oxidized

samples

in

air

yielded

new

spectral

core

level

features

accompanied

by

an

increase

in

carbon

content.

Features

include

a

carbonate

species

which

can

even

become

the

dominating

species

according

to

Cls

spectra.

This

binding

state

was

usually

not

found when

only oxygen

was

offered

in-situ.

It

is

therefore

presumed

that its

formation

depends

on

the

presence

of

water and/or

carbon

dioxide.

While

water

could

form

hydroxidic

species with

previously

unreacted

lithium

atoms,

carbon

dioxide

might

react

with

existing

(see

above)

or

newly

generated

hydroxide

groups

to

produce

carbonates.

A

contribution

of

adsorbed

carbon

oxides

and

hydrocarbons

to

carbon

signals

must

be

assumed.

After

storage

in

air,

work

functions

increased

compared

to

the

ones

measured

for

the

'pure'

oxide,

but

they still

remained

close

to

values

for

pure alkali-metals

(e.g.

for

Li

in

r.t.

a-C

2.8

eV

vs.

2.93

eV for

polycrystalline

Li

[9]).

This property

of

relative

air

stability

of

low

work

function

values

could

be

interesting

with respect

to

applications

for

example

in

field

emission.

CONCLUSIONS

We

investigated

the

incorporation

of

alkali-metal

atoms

into

amorphous

hydrogen-free

carbon

networks.

Results

from

in-situ

core

level

and

valence

band

photoelectron

spectroscopy

indicate

that

metal

atoms

can

be

accommodated

in

a

variety

of

states,

most

prominently

at

dangling

bonds

of

the

amorphous

matrix or

in

an

intercalated

state

in

the

spacings

between

graphene

layers

in

small

ordered

clusters.

The

resulting

materials

had

metallic

character

even

at

metal

atom

contents

of

less

than

10

at%

and

exhibited

work

functions values

between

the

ones

for

pure

a-C

and

pure metal.

Exposure

of

the

metal-containing

samples

to

molecular

oxygen resulted

in

an

oxidation

driven

accumulation

of

metal-atoms

at the

sample

surface

in

oxidic form.

The increase

in

concentration

was

up

to

threefold

and

the

concentration

of

metal

atoms

exceeded

the

one expected

for

stoichiometric

Li

2

0.

Thus,

the

oxygen

gettering

capability

of

these

materials

is

demonstrated

with

the

additional benefit

that

to

a

certain

extent

metal

atoms

from

the

bulk

can

be

delivered

to

the

171

surface through

diffusion

when required.

As

the

metal

oxide

formation

on

the

sample

surface

proceeded

and

resulted

in

a

continuous

oxide

overlayer,

work

functions

were

observed

to

fall

below

the

ones

of

the

pure

metal.

Annealing

at up

to

800'C

removed

metal

atoms

and

oxygen

and

essentially

restored

the

original

host

material.

This

may

be

interesting

from

the

point

of

view

of

recovery

and

later

'recharging'

of

a

carbon

based

material

after

a

certain

time

of

use

in

combination

with

an

oxygen

getter.

Part

of

the

results

for

oxygen

exposure

has not

yet been

confirmed

for

sodium,

and

there

is

evidence

that

a

different

mechanism

may

apply.

After

storage

in air,

a

carbonate

compound

formed

probably

by

reaction

with moisture

and

carbon

dioxide

from

the

atmosphere.

Work

functions

increased,

but

remained

close

to

the

values

given

for

pure

alkali-metals.

They

thus

represent

an

interesting

example

of

air-stable

low

work

function

materials.

Future

experiments

must

turn

to

other

reactive

gaseous

species

in

order

to

form

a

more

complete

picture

of

the

interactions

which

are

involved

in

the

observed

results.

Only

then

it

may

be

possible

to

ascribe

single

effects

to

the

influence

of

one

or

the

other species.

ACKNOWLEDGEMENTS

We

most

gratefully

acknowledge

support

of

this

work

by

Schweizerisches

Bundesamt

fiir

Bildung

und

Wissenschaft

under

project

no.

16873/56366.

We

thank

Mr.

R.

Steiner

for

his

continuous

technical

on-site

support.

REFERENCES

1.

M.S.

Dresselhaus

and

G.

Dresselhaus,

Adv.

Phys.

30,

139

(1981).

2.

J.R.

Dahn,

T.

Zheng,

Y. Liu,

and

J.S.

Xue,

Science

270, 590-593

(1995).

3.

G.L.

Jackson,

E.A.

Lazarus,

G.A.

Navratil,

R.

Bastasz,

N.H.

Brooks,

D.T.

Garnier,

K.L.

Holtrop,

J.C.

Phillips,

E.S.

Marmar,

T.S.

Taylor,

D.M.

Thomas,

W.R.

Wampler,

D.G.

Whyte,

W.P.

West,

J.

Nucl.

Mater.

241-243,

655-659

(1997).

4.

G.

Janeschitz,

G.

Janeschitz,

K.

Borrass,

G.

Federici,

Y.

Igitkhanov,

A.

Kukushkin,

H.D.

Pacher,

G.W.

Pacher,

M.

Sugihara,

J.

Nucl.

Mater.

220-222,

73-88

(1995).

5.

S.

Schelz,

T.

Richmond,

P.

Kania,

P.

Oelhafen,

and

H.-J.

Gtintherodt,

Surf.

Sci.

359,

227-236

(1996).

6.

M.

T6we,

P.

Reinke,

and

P.

Oelhafen,

to

be

submitted

7.

P.H.

Citrin,

G.K.

Wertheim,

Y.

Baer,

Phys.

Rev.

B

16,

4256-4282

(1977).

8.

M.

Nakadaira,

R.

Saito,

T.

Kimura,

G.

Dresselhaus,

and

M.S.

Dresselhaus,

J.

Mater.

Res.

12,

1367-1375

(1997).

9.

D.R.

Lide

(Ed.),

CRC

Handbook

of

Chemistry

and

Physics,

78th

ed.

(CRC

Press,

Boca

Raton

New

York,

1997).

10.

K.

Wandelt,

Surf.

Sci.

Rep.

2, 1-121

(1982).

11.

H.

Estrade-Szwarckopf

and

B.

Rousseau,

J.

Phys.

Chem.

Solids

53,

419-436

(1992).

12.

B.

Rousseau,

M.

Vayer-Besanqon,

and

H.

Estrade-Szwarckopf,

Solid

State Comm.

99,

143-147

(1996).

13.

J.

Hrbek,

Y.W.

Yang,

and

J.A.

Rodriguez,

Surf.

Sci.

296,

164-170

(1993).

172