Sei sulla pagina 1di 11

J. Phys. Chem. Solids

Pergamon

Press

1960.

Vol.

1.5. pp.

324-334.

Printed

in Great

Britain.

REACTIONS

OF

LITHIUM

AS A DONOR

AND

AN

ACCEPTOR

IN

ZnO

 

J.

J.

LANDER

 

Bell Telephone

Laboratories,

Incorporated,

Murray

Hill,

New

Jersey

(Receiwed 7 Junua~y 1960)

Abetract-Reactions of oxidized lithium with ZnO have been investigated by means of conductivity measurements. Donors are formed in ZnO if the atmosphere surrounding the crystal is reducing. The donor centre is probably the molecular ion LiO’. Oxidizing conditions favor an acceptor center presumably obtained by displacement of a lattice zinc atom by lithium. Solubilities and diffusion coefficients were measured and some results are reported for the kinetics and equilibria of the dis- placement reaction. High temperatures favor displacement. In one atmosphere of oxygen the Fermi level, which is normally near the top of the 3.0 eV band gap, can be pushed down to the centre of the band gap. Impractically high oxygen pressures would be required to produce strongly P-type material, The precipitation of lithium on dislocations and their decoration were also studied.

 

1.

INTRODUC’I’ION

 

IT HAS

been found

in earlier

work that reactions

of

ZnO

with

hydrogen,(r*z)

zinc,@)

and

indium(4)

produce

n-type

conductivity.

Since p-type

material

has

not

been

reported

and would have properties

of

considerable

interest,

a study

has

been

made

of reactions

which

might

be

expected

to

produce

such

material.

The

substitution

of

an alkali

atom

for

a lattice

zinc

atom

should

yield

an

acceptor

center

and

reactions

with

lithium

are the

subject

of

this

report.

Some

properties

of

sodium

re-

actions

with

ZnO

were

also studied

and found

to

be similar to those of lithium.

 
 

A qualitative

understanding

of the

reactions

of

lithium

with

zinc

oxide

was

obtained

from the

following

experiments

carried

 

out

on

single

crystals

of needle form with

diameters

of the order

of

O-01 cm

and

initial

conductivity

of

the

order

of

0.1

(ohm cm)-1

(donor

concentration

about

4 x lOI5 per ems) at room temperature.

 
 

(1)

When

crystals

of ZnO are exposed

at moder-

ate

temperatures

(300°C

to

600°C)

to

Li

vapor

at low pressures

(10-S mm of Hg) or dipped

in

dilute

solutions

(0.1

per

cent)

of

Li

in

Zn

or Hg,

reaction

occurs.

Conductivity

is

observed

to

in-

crease

at

a

rate

which

depends

on

temperature,

activity

of

lithium

and size

of

the

crystal.

High

values may be obtained.

It is apparent

that

in this

stage

of the

reaction

donors

diffuse into

ZnO,

 

though

they

may

be

interstitial zinc.(a)

 

(2) After prolonged

reaction

under

the

above

conditions

it

is found

on cooling

that

the

crystals

have

darkened

and lost strength

(some

disinte-

grate)

and that they contain

a high

concentration

of lithium

(greater

than

1019 atoms

per

ems).

 

Thus under these conditions:

 

(4

Lithium

has diffused into the crystal.

 

(b)

The

resulting

n-type

solutions

are too con-

centrated for accurate study by semicon-

ductor

techniques.

(4

The

reaction

proceeds

to

the

internal

pre-

cipitation

of

lithium

which produces

 

darkening

and severe

strains

in the

crystal.

The

major

part

of

this

precipitation

pre-

sumably takes place during cooling.

 

(3)

If reaction is interrupted

in a preliminary

stage and the crystal

is quenched

to room tempera-

ture

it is found

that:

 

(a)

The

donor concentration

is maintained

(the

donor

is a “hydrogen-like”

centre

with

an

ionization energy near 0.050 eV).c5)

(b)

The crystal remains strong.

 

(4)

If the crystal

is etched

(so that

there

will be

no

subsequent

reaction with surface contamina-

tion)

and heated

in

air

at

300°C

or

higher,

324

REACTIONS

OF

LITHIUM

AS

DONOR

AND

ACCEPTOR

IN

ZnO

325

conductivity decreases and ultimately may reach a value as low as 10-9 (ohm cm)-1 at 300°C but p-type conductivity is not observed. This low conductivity once attained is not changed much by heat treatment in air at higher temperatures

(800°C).

It is concluded that under these conditions:

(4

(b)

(4

(4

Lithium at low concentrations reacts further to produce acceptor centers.

This reaction proceeds readily at tempera- tures above 300°C. The acceptor centers are probably im- mobile at moderate temperatures. The diffusion rate of lithium in the donor form is relatively high.

2. MODEL

In order to account for the above results and to set up a basis for quantitative work the following assumptions are made :

(1) The donor and mobile form of lithium in zinc oxide is interstitial lithium, which may form a molecular ion with oxygen. (2) Interstitial lithium becomes an acceptor by displacing a lattice zinc ion. Since the maximum charge on the lithium ion is + 1 the center has a charge of -1 if surrounding oxygen ions have their normal charge. The charged acceptor center will be described by the symbol Li-, according to a convention which indicates only the excess charge of the center. (3) Local reversibility of this displacement re- action can be attained at moderate temperatures. This assumption will be valid for a sufficiently large crystal. (4) The displaced zinc ion becomes an inter- stitial and as such must play an important role in subsequent kinetics and equilibria. THOMA# has studied the behavior of interstitial zinc in ZnO and has reported values for the solubility as a function of temperature and zinc pressure and for the diffusion coefficient. According to this model the minimum system of solute species in an n-type crystal contains L$, Li-, Zn,’ and e-.* The simplest reactions whrch can be written to represent the system in equili- brium with its enviroment (assumed gaseous) are:

* The subscripts i and s will be dropped in succeeding equations.

(1)

The

solution

of

lithium

at

pressure

PLY

in the gaseous environment to produce inter-

sitial lithium in the solid

Li(u)

*

Li(s)

Kr

=

[LiJ/PLi

(1)

(2) The ionization of interstitial lithium

Li(s)

*

Li+ + e-

&

=

[Li+][f7]/[Li]

(2)

(3)

The

solution of zinc at pressure Pzn in the

environment

to produce interstitial

zinc.

G(w)

*

Zn(s)

&

=

[Znl/Pzn

(4)

The ionization of interstitial zinc

(3)

Zn(s) ?+ Zn+ + e-

K4

=

[Zn+][e-l/[zn] (4)

In addition there is the homogeneous reaction in which interstitial lithium displaces a lattice zinc ion to produce the acceptor center

Zn(Z)+ Li,+ + e-

*

Znl

+ Li;

KS =

pi-][Zn+]/[Li+][e-]

(5)

where Zn(Z) is a lattice zinc atom. This will be called the displacement reaction and it is similar to reactions proposed for copper and nickel in silicon and germanium. (637)

Note that the equilibrium

of

Reaction

(5)

is

shifted to the right by the removal of interstitial

zinc, for example by oxidation at the surface. This results in an increasing relative concentration of acceptors.

In addition one expects the production

of ion

pairs and more complex centers under appro- priate conditions. For example the molecule Liz0 may be formed within the crystal by the reaction of Li+, Li- and an oxygen atom. The energy of the electrons in this molecule may lie below the top of the valence band. Similarly a variety of reactions may occur between foreign imperfections and lithium.

3. THE

SOLUBIUTY

AND

DIFFUSION

CO-

EFFICIENTOFlNTERSTITIALLITHIUMIN

zno

It was noted above that the reaction between ZnO and lithium even in dilute solution in Zn and at the lowest practical temperatures, produces lithium concentrations which are too high to be studied by semiconductor techniques. The activ- ity of lithium in the external phase must be

326

J.

J.

LANDER

reduced drastically to obtain solutions of the proper dilution. Consider a crystal of ZnO coated with the stable product of reaction between LisO and ZnO, for example LisZnOs, and heated in zinc vapor. Since zinc is much less active than lithium, the reaction

LisZnOs+

Zn(rv)*

2ZnO + ZLi(0)

KS = &fPZ*

03

solubility of lithium in ZnO coated with LisZnOz(?) were obtained as a function of temperature and zinc pressure from conductivity data obtained in the apparatus shown in Fig. 1. The part most relevant to the experiment is the closed quartz

should provide a source of lithium with low activity. The dependence of interstitial lithium concen- tration in ZnO on zinc pressure, obtained from equations (6), (1) and (2) and their equilibrium constants, is given by

LisZnOs + Zn(et) *

2Li+ + 2e-+

2ZnO

K7 =

[Lif]2[e-]s/Pz,

= K,K,Ke

2

2

(7)

When interstitial lithium is the dominating solute species [Li+] = [e-l and, since practically all inter- stitial lithium is ionized at reaction temperatures, lithium concentration should vary with Pq&ds Similarly if oxygen pressure is varied one obtains from

ZnO * Zn+*O2 Ka = Ps,1J2Pz, (8)

ZnO

SINGLE

CRYSTAL

OVENS

and (7), a dependence of interstitial lithium con- centration on the reciprocal one-eighth power of oxygen pressure. One can thus hope to control the activity of lithium over a very wide range with- out adding a new component to the system. This scheme was found to be practical, though the coatings applied were actually Li evaporated in thin films onto crystals and subsequently oxi- died, or LiOH (which dissociates to the oxide when heated to about 3009C in a sufficiently dry atmosphere free of COs). The latter method is the most convenient and was introduced by D. G. COMAS. Very similar results were obtained with each. Note that equation (7) defines a solubility pro- duct for Li+ and +Y-,but does not define their con- centrations unless other interacting solute species are present in negligible quantity. In particular

the concentration of the acceptor form, Li-, must be negligible. This was found to be the case at tem- peratures below about 400°C if the pressure of zinc is high.

The

diffusion coefficient of

lithium

and

the

THERMOCOUPLE

FIG. 1. Apparatus

for

studying

changes

in conductivity

of ZnO

in Zn vapor

and coated

with LiOH.

tube containing zinc and crystal surrounded by separate ovens. The crystal is held by four spring tungsten leads (forming a four probe circuit) running through a four holed rod of alumina up to a four lead stem. A vacuum of the order of 10-s was maintained above the ceramic during most of the tests. Temperatures were obtained from two Pt/Pt-Rh thermocouples pressed at appro- priate spots against the quartz tube by sprung tungsten rings.

REACTIONS

OF

LITHIUM

AS

DONOR

AND

ACCEPTOR

IN

ZnO

327

The parameters measured in these experiments

will be confined for the moment

to the

rates of

were conductivity,

time, temperature

of zinc and

solution.

temperature

of the crystal.

Conductivity,

u, was

The diffusion of lithium into the hexagonal

reduced to carrier concentration equation

by means of the

shaped needles of ZnO was observed to obey simple cylindrical diffusion theory and diffusion coefficients were calculated from the equation

[e-l

=

u/ep =

9.6 x lO%T3/3

(9)

 
  • D =

O.O60~2,/t(l/2)

( cm2/sec)

(10)

Data for the mobility, CL,are those reported by HUTSON@)and e is the electronic charge. Donors

where

~0 is half the average diameter and t(1/2) is

 

TEMPERATURE

IN

DEGREES

CENTIGRADE

 

TEMPERATURE

IN

DEGREES

CENTIGRADE

I.0

1.2

1.4

1.8

2.0

2.2

2.4

 

1000/T

K

1000/T

OK

FIG. 2. Diffusion

coeflkienta

of Li+

and

Zn+ in ZnO

as a

FIG.

3.

Solubility of Li+ and Zn+ in ZnO

as a function

of

 

function

of temperature.

 

temperature.

 

are nearly all ionized in the range of measurement and conditions under which electron concentra-

tion

[e-l

is

equal

to

donor

concentration

pi+]

have been mentioned above. Assuming that these conditions have been met, diffusion coefficients may be obtained from rates of increase of electron concentration, and solubility from the saturation values. Under some conditions decreasing con- ductivity may not be a measure of the dilfusion of lithium out of the sample and therefore attention

the time taken to reach 50 per cent saturation. Data are given in Fig. 2 together with the curve obtained by THOMAS(~)for interstitial zinc which will be used later. The curve drawn through the experimental points for Lif has the equation

  • 20 =

O-040 exp( - 0.98/H)

(11)

The factor of 2 indicates that the ambipolar nature of the diffusion process has been taken into account. The results are not greatly different from those

328

J.

J.

LANDER

obtained for H+ which difhrses about ten times faster than Li+ in the 300°C to 400°C range (and

is far less soluble at the same pressure).

Several tests with sodium were also made, The

results indicated a diffusion coefficient about one order of magnitude less than that of lithium.

Values for the solubility of lithium are given in Fig. 3. Most of the data were obtained with the

source of zinc vapor at 300°C. Additional data for

the conditions zinc source at 260” and at 380”

yield a one-quarter power dependence on zinc

pressure within experimental error. The equation

for the concentration of lithium (zinc source at

Tz*o)

is

[Li+]

=

(K7P~~)~/~/[tr]

=

1.3 x 1024 exp( -0*86/KT)

(12)

Data for the vapor pressure of zinc over solid zinc

can be obtained from the equation

Pzn =

1.5 x 109 exp( -

1*365/W)

(13)

which is assumed sufficiently

accurate through a

moderate range above the melting point of zinc

(420°C).

THOMAS@)gives the equation

[Zn+] =

3.4 x 1020 exp( -0*65e/KT)

= K3K4Pznl[e-]

(14)

for the solubility of interstitial zinc when ZnO

and Zn are at the same temperature.

It

has been proposed@) that hydrogen reacts

with lattice oxygen in zinc oxide to form the

molecular ion OH*. It is expected that the bond

will be stable and that the odd electron, being

spread over a large volume in a medium of high

dielectric constant, will be easily ionized. Diffu-

sion then proceeds by jumps from oxygen to

oxygen with alternate breaking and formation of

bonds. Similar molecular ions are expected in the

case of lithium and zinc, the ion LiO- probably

being much more stable than ZnO=. The values

of the diffusion coefficients (cmQ/sec) for these

donors at 500°C are

2DLif= 2x10-s,

2DHf

=

7x

10-s

2DZn+ =

2x

10-7

which is the order of the expected stability of the

bonds. The formation of molecular ions will also

be an important factor in solubility.

Attempts were made to observe an infrared

absorption to be associated with vibration of the

  • GO- ion. None was found, but the fundamental

mode should be at about lop, which is a region

obscured by strong harmonics of lattice vibrations of the zinc oxide, and therefore a sensitive test

cannot be made.

4. KINETICS OF THE DISPLACEMENT

REACTION

It was found that at higher temperatures and

lower zinc pressures ~splacement of lattice zinc

ions by interstitial lithium is no longer negligible. The resulting rate phenomena (an example is

given in Fig. 5) have properties which are not

observed in simpler diffusion processes. They are produced by the replacement of interstitial lithium

by interstitial zinc. The purpose of this section is

to illustrate these properties with a special case

which has a relatively simple solution. The results will be used in a qualitative way in descriptions of

observed phenomena. Consider the displacement reaction:

Zn(P)+Li++e-?t

Li-+Zn+

Kr, =

[Li-][Zn*]/[Li+][e-]

(5)

The mobile atomic species at moderate tempera-

tures are Li+ and Zn+. Changes in their concen-

trations in a region of a crystal are determined by

their concentration gradients and by the change in

concentration of Li- produced by displacement.

Neglecting field effects, they are described (in a

one dimensional case) by the equations

-=

a[Li+]

9[Li+]l

Dr.i+----- -

a[Li-]

(15)

at

--

ax2

at

a[Zn+]

=

Dz_i

__ a2[Zn+] + _ a[Li-]

 

-

116)

 

at

ax2

at

In the special

case DL~+= Dznf= D

the sum

function

 

a[Li+ + Zn+]

=

D as[Li+ + Zn+]

 
 

at

ax2

(17)

has well known solutions.

Dzn+is in fact

much

larger than

DJ.,~+,

except

at very high

tempera-

tures.

REACTIONS

OF

LITHIUM

AS

DONOR

AND

ACCEPTOR

IN

ZnO

329

The surface boundary conditions are deter- A qualitative picture of the following character-

mined

by

the

zinc

pressure.

 

Assuming

that

surface

reactions

are

not

rate

limiting

a

fixed

zinc

pressure

results

in the

conditions

 
 

[Li+]

=

[Li+]s

x

=

0,

all

t

(18)

[Zn+]

=

[Zn+]s

x

=

0,

all

t

(19)

Furthermore

it

is

assumed

that

the

crystal

is

initially empty.

 

Reaction (5), which describes the displacement

process,

imposes

an

internal

 

constraint

on

the

system.

If

local

 

equilibrium

is

not

attained

two

rate constants

are required to describe the process,

one

for

the

forward

and one for the reverse re-

action.

It

was

found

that

the

reaction

can

be

observed

at temperatures

below 3OO”C, therefore

the

assumption

 

of

local

equilibrium

at

higher

temperatures

seems

fair

and

one

notes

that

its

validity

increases

with

increasing

crystal

size.

The

neutrality

condition

 

[Li+]

f

[Zn+]

=

[Li-]

+

[e-]

(20)

together with equations (15) and (16) ,the assumed

boundary conditions, and the internal equilibrium

described by KS completely define the system. The neutrality condition together with the ex-

pression for Ks generally result in quadratic forms for the dependent variable. In one particular case

a simple form is obtained. If Ks = 1 (displacement

strongly favored) the result is that everywhere

[Li+]

=

[Li-]

(21)

DL~+ as[Li+]

 
 

2

a9

(22)

 

all concentrations

as

 

diffusion

 

4.

perpendicular

to

the

axis

 

length

with

radius

rs.

species

are

plotted

as

a

 

=

1,

[Zn+]a

=

O-05,

 

Note that if every-

In this case equation (15) becomes

 

a[Li+]

 

-=-_-

 

at

and it is possible

 

a function

of time

processes.

 

They

are

for

diffusion

of

a

cylinder

of

Total

amounts

of

each

to evaluate

using data for simple

Solutions for this special case are given in Fig.

infinite

function of (Dt/r$)VJ. The assumed boundary

conditions are [Li++Zn+]e

with the crystal initially empty.

where

[Li+]

=

[Li-]

then the neutrality condition

requires that [Zn+] = [e-l everywhere.

istic properties of the approach to equilibrium

when displacement is not negligible is suggested by

the data of Fig.

4:

(1)

Since

the

solubility

of

Li+

is much

larger

than that of Zn+, the production of Zn+ by dis-

placement may result in total concentrations

of Zn+

higher than the equilibrium value. In-the example chosen the concentration of interstitial zinc ana

therefore of electrons increases to a maximum

which is considerably higher than the final value.

 

O

 

0

0.2

0’4

0.6

0.8

I.0

1.2

1.4

 

(Dt&+

 

FIG.

4.

Kinetics

of the displacement

reaction

discussed

 

in Section

4.

Since

Dzn+

is

in

fact much

larger

than DJ,~+ the

maximum value of Fig. 4 is an exaggeration. A

ratio

 

Dzn+/&i+

 

equal

to

10

is

approximately

correct

for

the

range

400”

to

600°C.

As

a first

approximation,

the value of the difference between

the maximum

and the

final value

should

be

de-

creased by d(

10).

In some cases during the initial

stages of the reaction [Zn+] will be much higher

than [LP]

near the center of the crystal.

 

(2) Displacement

impedes

the attainment

of the

final state. When displacement

is negligible, the rate

is determined

 

by the

diffusion rate

of Lif.

When

displacement

is strongly

favored

the rate

is much

slower

because

the

displaced

zinc

concentration

must

 

hunt

for the

boundary

value.

 

For

example,

in the case of Fig. 4, the time to attain 90 per cent

of

the

equilibrium

electron

concentration

is

in-

creased by a factor of about five over that for no displacement.

330

 

J.

J.

5.TH.EMEAsUREME

 

NT

OF

KS,

THJ3

EQUILI-

BRIUM

 

CONSTANT

FOR

THE

DISPLACEMENT

 
 

REACTION

 
 

Values

of

K5

=

[Li-][Zn+]/[Li+][e-]

are

easily

obtained

if

zinc

pressure

 

is

adjusted

to

a range

which

yields

an equilibrium

ratio

[Li-]/[Li+]

near

unity. [e-l is then obtained by a measurement of

conductivity

and

the value

of [Zn+]

from

the

data

of THOMAS. This

is

found

to be practical

through-

out

a considerable

range

 

of interest

because

the

ratio

 

[Li-]/[Li+]

is

near

unity

over

a

very

wide

range of zinc pressure.

 
 

6.0

 

0

0

2

 

,&k5

6

FIG.

5.

Change

in

conductivity

of

ZnO

coated

with

LiOH

and

held

at

504°C

in

Zn

vapor

from

source at

 

450°C.

 
 

As

an

example,

consider

 

the

data

of

Fig.

5

where

conductivity

is plotted

against

time

for

a

crystal

coated

with

LiOH

 

and

held

at

504°C

in

zinc

vapor

from

a source

at 450°C.

The

diameter

of

the

crystal

was

6.0x

10-s cm.

Compare

the

maximum

in

conductivity

 

with

that

in

the

curve

for

electron

concentration

in

Fig.

4. Presumably

 

the concentration

of Iithium

continues

to

increase

after

the

maximum

and

the

degree

of compensa-

tion

also

increases.

Suppose

 

the

ratio

[Li-J/[Li+]

is large,

say

O-9. The

 

equilibrium

concentration

 

of electrons

obtained

from

the

conductivity

 

value

of

2-O (ohm

cm)-1

and

an

estimated

mobility

 

(at

504°C)

equal

to

30

is

4 x 1017 per

cm3. According

to

the

data

of Thomas

 

[Zn+][e-]

N 1 x

1Osa under

these

 

conditions.

 

Thus

 

K5(504”C)

e

0.9 x 1 x 10=/16x

10s’

=

0@057

 
 

(23)

LANDER

With

a crystal

at

600°C

and

the

zinc source

at

600°C the value

of

[Zn+][e-]

is

3.6

 

x los3. During

lithium

doping

and

a final

conductiv-

ity indicating

a maximum a large amount

of compensation

was

observed.

The

final

conductivity

 

was

4,

which

corresponds

to

an

electron

concentration

equal

to

about

9 x 1017. Assuming

again

 

that

the

ratio

[Li-]/[Li+]

is about

0.9,

then

K5(6OO”C) N

0.9 x 3.6 x

loss/81

x

1034 =

0.0045

The

above values

of Ks

are

to

be

taken

as approxi-

mations.

Results

of repeated

tests

 

scattered

by

a

factor

of about

four,

but

the

cause

of scatter

was

not determined.

 
 

A consequence

of the

assumptions

 

made

about

the

lithium

reactions

is

that

the

equilibrium

total

 

concentration

of

lithium

is

independent

of

zinc

pressure.

This

is easily

seen

in

the

region

where

compensationis

nearly complete. Then

 

[Li+] N [Li-]

[Zn+]

N Ks[e-]

and

equations

(1)

to

(5)

and

the

neutrality

condition

yield

[Li+] 21 K&s/K&5,

which

is independent

of zinc

pressure.

This

pre-

diction

has been tested

by spectrochemical

analyses

of

crystals

doped

with

lithium

through

a

very

 

wide range

of zinc pressures.

The

results

are given

in Table

1.

 

Again considerable

scatter

is observed

in

the

re-

sults,

nevertheless

samples

treated

in

air

contain

about

as much

lithium

as those

treated

at the

same

temperature

in high

zinc

pressures,

and

the result

is in agreement

with

the

model.

The

cause

of the

scatter

has

not

been

determined.

 

It

is probably

not

due to variation

in initial

purity.

The

crystals

were from

the same lot and

had initial

conductivi-

ties of about O-2, corresponding

to donor

concentra-

tions

of about

1 x

10-16.

 
 

6.

DIFFUSION

 

OF

Li+

OUT

OF

ZnO

Consider

a crystal

which

has been saturated

with

 

Lif

at

a moderate

temperature

(300-4OO”C)

and

which

is in equilibrium

 

with

zinc from

a source

at

about

the same temperature.

Since

compensation

does

not

dominate

the

conductivity

behavior,

conductivity

is

a measure

of the

concentration

of

Li+.

If

the

zinc

pressure

is abruptly

 

decreased

by

about

one

order

of

magnitude

conductivity

is

observed

to

decrease

to

a

new

quasi-equilibrium

value.

The

initial

and later values

are proportional

very

nearly

to

the

reciprocal

of

the

one-quarter

power

of the ratios

of the zinc pressures,

aa theory

REACTIONS

OF

LITHIUM

AS

DONOR

AND

ACCEPTOR

IN

ZnO

Table 1

331

Sample

No.

J(6)

J(1)

JUO)

J(2)

“:,:;

;I;;

J(7)

Doping

Temperature

(“C)

I

/

390

500

500

500

500

600

600

600

600

Pressure

of

Zn (mm

of Hg)

0.09

1.5

1.5

0.1

1O-28 (air)

13

1O-2o

0.1

(air)

10-20

(air)

Total

(atoms

Lithium

per ems)

2.9

x

4.8

x

6.5

1.4

x

x

101’

101’

101’

101*

7.0

x

1.4

x

101’ I

10’9

2.1

1.1

xx 101s

1018

3.2

x

10

Average

3 x

101

a x

107

5 x

101s

predicts. A subsequent slow and small decrease in

conductivity which is produced by the formation of Li- ends at a true equilibrium value.

Diffusion coefficients calculated from the first

change in conductivity agree very well with those

given for Li+ in Fig. 2. The diffusion rate of Lif out of ZnO can not be

obtained simply from the change in conductivity produced by an abrupt lowering of temperature.

The reason will be given in the next section.

non-interacting

line dislocations leads to the ex-

pression

x

=

1/0*36&r

(cm)

7. INTERNAL

NUCLEATION,

PRECIPiTATION

AND

DECORATION

An extreme case of internal precipitation was

described in the introduction. It was observed that

crystals of ZnO containing very high concentra-

tions of Li+ (B 1 x 101s) disintegrate on cooling.

A similar process, which does not weaken the

crystal seriously, has been observed where concen-

trations are lower. The basis for the interpretation

of the effect of precipitation on conductivity in

zinc oxide has been discussed by THOMAS@)who

studied the precipitation of indium.

Data are given in Fig.

6 for the change in con-

ductivity of a crystal which was saturated with

Li+ at 440°C (with the Zn vapor also at 44O”C),

quenched to room temperature, then heated to the

indicated temperatures in air. The changes in

conductivity were observed at the temperatures

indicated and the run at 317°C is a continuation

of that at 286”C, etc. The lines do not meet at a

point at t = 0 because of the correction for the

change of mobility with temperature. Note the change of time scale with change in temperature.

Theory for nucleation and precipitation along

286°-0

327’-30

363’446

10

40

56

20

50

66

TIME

30

60

76

40

70

66

IN

MINUTES

50

60

96

I

3

90

FIG. 6. Change

in conductivity

in ZnO with precipitation

of Li at temperatures

indicated.

and a linear relation between log CIand time. x is the length of the cylinder of precipitated material

(total length of dislocations), r is the time required

332

J.

J.

for (T to drop

to crso/2*72and D is the

diffusion

co-

efficient

for

the

rate

limiting

transport

process.

The

data of Fig.

6

fit

a linear

log

o

vs.

t relation

very well.

Two

numbers

can

be

obtained

from

these data; the activation

energy for the rate limiting

process

and the total length

of dislocations.

A com-

parison

of the

temperature

coefficient

of the

time

constants

with that

for the diffusion

coefficients

of

Li+

indicates

that

the

rate

limiting

factor

is

the

diffusion

rate of Li+. A total length of line dis-

locations

was therefore

calculated

using the values

of D for Li+. The results are given in Table 2.

Temp.

((3

286

327

363

Table 2

7

(sec.)

7600

1700

810

D

(cm2/sec)

5.0

x 10-11

2.2

x 10-10

6.4

x lo-10

X

(cm)

2.3

x 106

2.4

x lo6

1*7x106

LANDER

concentration

of

about

1 x

1018

(atoms/ems),

quenching

to room

temperature,

then

heating

the

crystal

in

air

to

about

700°C

for

a

few

minutes.

During this treatment

the

conductivity

at 700°C

dropped

to

2-O x 10-5

and the value

at 400°C

was

found

to

be

1.2

x

10-8.

The

activation

energy

computed

from

this

temperature

dependence

is

about

1.5

eV

which

is half

the

band

gap for

this

temperature

range

..

The

concentration

of electron8

calculated

from

the

conductivity

 

at

400°C

is

1.4

x

10s.

The

formula

for

intrinsic

ionization

(neglecting

holes) yields

 
 

[e-l

=

NC exp( -

3*0/2KT)

 
 

=

1 x

1020x

lo-‘1.2

=

6x

108

(24)

where

the value of Ne, the density

of states

in

the

conduction

 

band,

is

that

given

by

HUTsON@).

Thus

material

which

is

practically

intrinsic

has

been

produced

by doping

with lithium.

 

The value for x corresponds to a high dislocation

density

and

is

similar

to

that

observed

by

THOMAS(~)

for some crystals in a more extensive

study

of nucleation

of

an

InsOs+ZnO

phase

in

ZnO. The precipitation

of

1 x

1018

atoms

of

metallic

Li (corresponding

to the initial conductiv-

ity

equal

to

7)

as cylinders

 

with

a total

length

of

2 x

10s

cm

yields

cylinders

with average diameter

equal

to

about

150 A,

and

average

distance

apart

 

equal to about 60,000 A.

Precipitation of lithium from much higher con-

centrations

obtained by doping in the neighbor-

hood

of 700°C

result8 in a darkening

which

is due

 

to decoration

of the dislocations.

This is observable

under the high power of a microscope.

The decora-

ted dislocations

were seen, in the crystals studied,

to

be fine

dark lines

running

perpendicular

to the

axis of the needle

shaped

crystals

with hexagonal

cross

section.

Their

density

varied

by

a factor

of

about

three

and

counts

gave

an average

value

of

about

2 x

10s

cm

per

cm3 in

agreement

with

the

values

given in Table

2. Crystals

decorated

in this

fashion are extremely fragile.

 
 

8.

INTRINSIC

BEHAVIOR

The

lowest conductivity

so far observed

in ZnO

was

obtained

by

doping

a crystal

with

Li+

to

a

9. CONCENTRATION

OF SOLUTE SFECIES

AS

 

A

FUNCTION

 

OF

ZINC

PRESSURE

 

The

solute

species

of

the

system

have

been

assumed

to

be

Li+,

Li-,

Zn+

and e-.

In

a p-type

range

e+ must

also be

considered. Their concen-

trations

are determined

by the pressures

of oxygen

and zinc