Evidence for negatively charged DX-center in Si-doped AlGaAs from persistent photoconductivity measurements

~

0038-1098/91 $3.00+.00 Pergamon Press plc

Solid State Communications,Vol. 77, No. 5, pp. 327-330, 1991. Printed in Great Britain.

EVIDENCE FOR NEGATIVELY CHARGED DX-CENTER IN Si-DOPED AIGaAs FROM PERSISTENT PHOTOCONDUCTIVITY MEASUREMENTS

I.F.L.

Dias, P.S.S.

A.G. de O l i v e i r a , GuimarAes, J . F .

J.C.

Bezerra,

R.C. Miranda

Sampaio and A.S. Chaves

D e p a r t m e n t o f P h y s i c s , UFHG, C a i x a P o s t a l 702, 30161, Belo H o r i z o n t e , (Received Oct.

MG, B r a z i l

19, 1990 by C . E . T . G o n c a l v e s da S i l v a )

Persistent photoconductivity experiments were carried out in Sidoped AlxGal_xAS at 77 K. The density of persistent free electrons was changed by a factor of 4 to 8. The electron mobilities ~ were observed to increase for increasing electron densities n and from the behavior of ~(n) the variation of the ionized scattering impurity densities Ni with n was evaluated. The densities N i show little change as compared to n. It is concluded that negatively charged centers are the only or at the ieast the dominant deep centers related to silicon impurities in AIGaAs.

A great to

the

to

many

deal

of attention

AlxGal_xAS/GaAs

those

interesting

systems

variety

of

physical

and

their

optoelectronlc devices.

applications

alloys

below about

I00 K, persistent

which

accepted

remains

the

appearance,

In

(so-called

DX)

expected center.

having

n-doped

after

arises

alloys,

from of

coexisting

from

the

the

last

behavior

and

the

not

to those impurity

dopants

resulting

impurities. is

either

DX-center

are

a very

between

its

large

and

was

in

aspects

mutually

proposed 20

charged

2d°---~ d + + DX-.

establishes existing

Ge,

S,

complexes

In this view

class:

the

neutrality

{DX)

electrons

(d) state.

properties difference and

of

ionized the

the problem

DX

resulting This

contrast

is

from

a the

negative-U

feature

which

with

other

the

all of them of the positive-U

(ionized) requires

is

(Stokes

present.

optical

model

that On

the

d +)

and thus charge

the number

equal

n of free

to the number

centers,

acceptor

the other charge

centers,

that

always

(DX+and

supposing

327

center

ciear

which

in the positive-U models one has neutral

and positive

bistabie, deep

a

it

that

distinctive

models,

and there

for of

nature

excluding 3'4'7'11-19

"reaction"

a

the

results alloy

not clear

different

has

which

the

models

negatively

model

hydrostatic

the microscopic

is still

recently

center

centers

different

clearly

the

in

DX

it affects

from

thermal

focus

are

is an intrinsic

extended

remarkable

several

(Si, Sn,

localized

d

its

electronically

(fundamental)

most

way

of

the

(less

cross-section

under

models

1 eV,

electron

small

that

GaAs

inhomogeneities 9 . However,

are

It seems now well

that the DX-center

and

The

because

the

of devices 2.

of the n-type

a

decade

(d)

of

of the DX center

a deep

with

in excludes

concentration

It

in

shallow

state and a (metastable)

shift)

appears

coexistence

hours

in an exceedingly

also

photoconductivity

and

for

10 -30 cm 2) electron-capture

pressure 5'8

It is generally

(0.1 eV

barrier

77 K3'4. Demonstration

The DX-center has received considerable

Te)

donor

than

a

below

or

established

related

a and

temperatures

effect

center

the performance

Se,

PPC

the

in

and

capture resulting

donor

unique

feature

at

for

hydrogen-like

attention truly

present,

of illumination 1.

that

of

electronic

energies I'3'4

respectively)

It is known that n-type

AIxGaI_xAS

(PPC)

ionization

paid owing

properties

high-performance

interruption

is being

heterostructures

i.e.,

impurities

hand,

neutrality

=

N.of i Ni,

are

not

n

in the negative-U results

in

n

=

328

Vol. 77, No. 5

NEGATIVELY CHARGED DX-CENTER IN Si-DOPED AIGaAs

N(d +)

-

N(DX-).

N(DX-) Thus,

and

thus

n

~

Ni

=

N(d +)

+

The

the two classes of models produce

Si-doped

experiments

AlxGal_xAS

were

performed

In four

films with x in the range

rather different predictions on the behavior of

0.26-0.29.

the mobility ~ as n is changed by any external

seml-insulating Cr doped (001) GaAs substrates.

means.

A

The behavior of ~ as a function of n as

the system was

used

is submitted to hydrostatic as

model 21 .

evidence

Further

eliminated however

analysis

the apparent

bringing

conclusive opportune

against

the

doped

thickness

question

a

more

the

comments

are

Some

the existing

big change on n. Second, resonant

hydrostatic the

with

to

studies

pressures

screening

of

conduction

p below

levels

investigated

the

ionized

Third,

(above

samples

20

1019

the

total

kbar,

0.25

was

~m

thick,

supposed

to be

of

2D

buffer all

on

layer,

samples.

intentionally

was

grown.

enough

electron

heterojunction.

layer was

finished

the

The

This

to avoid

gas

in

the

The

silicon-doped

then grown and

the structure

with

a

silicon-doped

substrate

film

AIGaAs

temperature

640°C with a flux

become

of

non-

MBE

of

GaAs

with

layer was

a

0.7

for two samples and 1.2 ~m for the other

The

for the high

cm -3 )

for

layer

by

thickness of approximately 15 nm. The thickness of

for

electronic

impurities

rather complicated (22'23) doping

about

concerning

the

band

GaAs

grown

formation

was

doped

grown

was

AIGaAs,

AIGaAs

were

the buffer

GaAs/AIGaAs

of ~(n).

because the DX in GaAs

the

predictions

thick,

problem

First, the hydrostatic pressure does not cause a

is

~m

Following

without

the

of

0.25

pressure

negative-U

samples

non-intentionally

contradiction,

point 22'23. about

the

The

alloy

was

approximately

ratio of approximately

composition

~m

two.

was

1.5.

confirmed

by

photolumlnescence, with an estimated error of less

the

than 2Z. The silicon doping

cross-section

by

is not the sum of the individual cross section

saturated n at room temperature could be done.

of

the

impurities

positive

and

not (23).

and

negative

Furthermore,

depends

ions

are

on

CV-measurements

The samples,

whether

correlated

so

level was assessed

for electronic scattering by ionized impurities

were

or

uniformly

1.32 V}

one should note that the

that

a

check

for

cooled in darkness

illuminated

mounted

in

using

the

the

to 77K,

a LED

cryostat.

(hu = Hall

relation Ni = constant holds for the negative-U

measurements were then carried out in the dark

model

using the van der Pauw geometry.

only

three DX-,

for

centers

special

are

conditions. In

presumably

fact,

stable,

the Hall electron concentration n as a function

namely

d + and d ° , and the relative densities

of the light dose.

of

the

binding

reaction

DX - --~

external

means

electrons center,

can

energy +

d +

the

Ed 2e

is

N(d+).

Ed

d° .

As

of

and

Thus,

d°

as

the

generated

shielding

decrease

increasing

of

shown

in

Figure

2.

The

~(n) was used to evaluate

by

by

The mobility ~ was observed

to increase with increasing n for all samples,

the d + and d ° centers depend on the temperature and

Figure I shows

measured

function

the variation of the

the

ionize

this

the number

I

12

Ni

I

I o

o o

3o

o

o

o

o o

1"3

does increase as n increases.

o

9

o o

In

E

this communication, we report experi-

ments on the electron concentration and mobility

p..

%

I

o

.

•

o

6

a • • "

a a

U

D

a

a

o

2~

a

D

of Si-doped AlxGal_xAs samples, the electron conv

centration being changed by nearly one magnitude by PPC.

Our results

are

order of

clearly

consistent with the posltive-U models. dicate that the PPC is based either only

c

3 4

in-

They inon ne-

gatlve-U centers or on a combination of negative-

roll

•

•

•

•

•

•

I

0

0

300

600

900

Dose ( a r b r i t r a r y units)

U centers and a relatively small amount of positive-U centers, plus, of course, the shallow donor centers.

in both

cases,

Figure I - Hall electron concentration at 77 K as a function of the accumulated light dose for the four silicon-doped AIGaAs samples

V o l . 77, No. 5

NEGATIVELY CHARGED DX-CENTER

2.5

,

, u °

329

IN Si-DOPED AIGaAs

2.5

,

8

I

1

oe

2

ou

og

2.0

2.0

D

]

O "4

°

m

."

o

o8

•

/

1

D

z 1.5

am • m m~

a Q aa

• -m

m

•

•

• o

•

•

o ,:

Io

0

°

3

°°°°

I~la a a

o

o

='"

• /,o,:~ a%

sl ° e

5 ,,

oue

."

m

10

1.5

O* 8

m~9 2

)
z

•

,°

o

oO°°°°e

•

•

o

,

I

*

4

I

I

10

I

8

~o

2-'_:/

.

,

.

.

.

,

5

0

12

.

n (1017cm-3)

10

n (1017cm-3--)1,2,3; 1016cm-3--)4)

Figure 2 - Normalized Hall mobility as a function of the electron concentration for the four samples at 77 K. The. normalizing divisor ~o is the initial Hall mobility (in the dark) for each sample.

~IEure 3 Normalized density of ionized impurities after illumination as a function of the Hall electron concentration. The normalizing divisor is the density of ionized impurities before exposure to light. The solid lines show the functions Ni/Nio = n/n ° . Note the change in the X-axis scale for sample 4.

ionized

impurities

density

N i with

the

Hall

electron density n as discussed next. The silicon-related

ionized

defects

supposed

to be by far the dominant

centers.

The potential

around an ionized defect

of charge ± e is V(r) = ± e exp(-ksr)/kr, k -I

is

the

are

scattering

Thomas-Fermi

screening

where length,

data

are

in a very

predictions. the

prediction

incorrectly discussed

clear

contrast

with

those

They are also in disagreement Ni/Nio

=

l

which

with

has

been

related to the negative-U model.

earlier

in this letter,

As

as the light

S

related the

to

Fermi

the

electron

energy

concentration mobility,

by

k2s

density =

of

states

4~e2n(EF)/k"

For

at

generates

the

a

shielding

effect

N i of such scattering centers the

in the Born approximation,

for a non

parabolic band is 21' 24

expected donor

to

(i )

additional

Here,

C

is a constant,

mCk F)

is the

V

be

small

]-

caused

exceed

(I)

and

(2)

function ~Cn) one can evaluate

the

measured

the behavior

of

any

the small was

not

indicate the

exposure change

taken

where Nio is the density to

light.

In this

in the effective

into

account.

The

figure

of

the

positive-U

to

and without it

is

increase

ionization

observed

not

in N i of

d° .

density

the observed

of

AlxGal_xAS.

These

DX-

models.

The

one

silicon

seem

the dominant

supported

Brazilian agencies FINEP,

data confirm of negatively

to

defects

in silicon doped AlxGal_xAS. This work was

total since

in n.

related

at worst,

the

centers,

the present

only,

or,

cannot

in N i is only about

solid

lines

some case,

of

deep

the existence

defects

are

In this centers

quarter

increase

the hypothesis

Ni

DX +

ionized

quarter cf the increase

charged

in

of

one of

to the relatively

DX + defects.

mass with n

the functions Ni/Nio = n/no, which are

prediction

is

neutral

expected

data

much

the

In conclusion,

Figure 3 shows Ni/Nio as a function of n

before

the about

N i as a function of the electron density.

for all four samples,

is

the

gas

the

hypotheses how

by

charged

concentration and

all

contribution

increase

positively

I * CZkF/ksl2/" equations

Ni

hoc"

2e,

and

however,

From

and

to evaluate

Another possible F(ks,kF ) = 2~{,n[1+[ 2kF 12

or

the available

"ad

+

electron

electron should

effective mass at the Fermi energy,

DX---+ d + the

some

d° ,

From

possible

of

ionize

centers

increase.

g=Cn/[Nim (kF) 2F (ks, k F) ].

reaction

to be

deep

in the

centers

in part by

CNPq and CAPES.

the

330

Vol. 77, No. 5

NEGATIVELY CHARGED DX-CENTER IN Si-DOPED AIGaAs REFERENCES

1

R.J.

Nelson,

Appl.

Phys.

Lett.

31,

351

12. E.F. Schubert and K. Ploog, Phys.

J.

13.

Appl. Phys. 67, RI (1990).

(1982)

3

D.V. Lan E and R.A. Logan, Phys. Rev. Lett.

14.

2

For a

recent review,

see P.M.

Mooney,

D.V. Lang, R.A.

A.K. Saxena,

K.L.I.

Solid-State

Kobayashl,

Logan and M. Jaros,

Phys.

Electron.

25,

127

Uchida

and

H.

Y.

15. T.N. Morgan, Phys. Rev. B 34, 2664 (1986)

Rev. B 19, 1015 (1979)

16.

5

Phys. Lett. 48, 656 (1986)

M. Tachikawa, M. Mizuta, H. Kuklmoto and S.

Minomura, Jpn. J. Appl. Phys, 24, L821 (1985) 6

30,

Nakashima, 3pn. J. Appl. Phys. 24, L 928 (1985)

39, 635 (1977) 4

Rev.

7201 (1984)

(1977)

M. Tachikawa,

T. Fujisawa, K. Kuklmoto,

H.P.

Hjalmarson

and

T.J.

Drummond,

Appl.

17. J.C.M. Hennlng and J.P.M. Ansems, Semicond. A.

Sci. Teehnol. 2, 1 (1987)

Shlbata, G. Ooml and S. Minomura, Jpn. J. Appl.

18. J.E.

Phys. 24, Lg93 (1985)

and W. Jantsch, Phys. Rev. B 38, 3276 (1988)

7 . E. Calleja,

A. Gomez and E. Munoz,

Appl.

Phys. Lett. 52, 383 (1988) g

. D.K.

Foster,

Maude,

J.C.

L. Eaves,

Dmochowskl,

J.M. Langer,

J. Raczynska

19. P.M. Mooney, G.A. Northrop, T.N. Morgan and H.G. Grfmmeiss, Phys. Rev. B 37, 8298 (1988)

Portal,

M. Nathan,

L.

Dmowski,

M. Heiblum,

T. J.J.

20. D.J. Chadl and K.J. Chang, Phys. Rev. Lett. 61, 873 (1988); Phys. Rev. B 39, 10063 (1989)

Harris and R.B. Beall, Phys. Rev. Lett. 59, 815

21. D.K. Maude, L. Eaves, T.J. Foster and V.C.

(1987)

Portal, Phys. Rev. Lett. 62, 1922 (1989)

9 . A.

Ya.

Shik and A.

Ya.

Vul.,

Soy.

Phys.

Semlcond. 8, 1085 (1975) 10.

H.J.

Quelsser

and

Phys. Rev. Lett. 62, 1923 (1989) D.E.

Theodorou,

Phys.

Rev. Lett. 43, 401 (1979) 11.

N.

Chand,

Masselink,

T.

Henderson,

R. Fisher,

22. D.J. Chadl, K.J. Chang and W. Walukiewicz,

O'Reilly,

Appl.

Phys.

Lett.

55,

1409

(1989) J.

Klem,

W.T.

Y.C. Chang and H. Morkoq,

Phys. Rev. H 30, 4481 (1984)

23. E.P.

24. R. Harrie, 69 553 (1956)

Proc.

Phys. Soc.

London Sec.

B