Heat treatment effect on p type Zn doped InP substrates

HAL Id: jpa-00245664

https://hal.archives-ouvertes.fr/jpa-00245664

Submitted on 1 Jan 1987

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL, est

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Heat treatment effect on p type Zn doped InP substrates

A. Dhouib, B. Maloumbi, C. Martinez, L. Gouskov, D. Bayaa, T. Bretagnon, R. Coquille

To cite this version:

A. Dhouib, B. Maloumbi, C. Martinez, L. Gouskov, D. Bayaa, et al.. Heat treatment effect on p type

Zn doped InP substrates. Revue de Physique Appliquée, Société française de physique / EDP, 1987,

22 (10), pp.1159-1168. �10.1051/rphysap:0198700220100115900�. �jpa-00245664�

1159

Heat treatment effect

on

p type ^Zn doped ^{InP substrates}

A.

Dhouib,

^B.

Maloumbi,

C. Martinez

(*),

^L.

^Gouskov,

^D.

Bayaa (1),

^T.

Bretagnon (1)

and R.

Coquillé (2)

Centre

d’Electronique

^de Montpellier

(CEM)

^{associé au CNRS}

(USA 391),

Université des Sciences et

Techniques

^du Languedoc, place ^E. Bataillon, ³⁴⁰⁶⁰

Montpellier

Cedex, ^France

(1) Groupe

d’Etudes des Semi-conducteurs

(GES)

^{associé au} CNRS, Université des Sciences et

Techniques

^du

Languedoc, p1.

^E. Bataillon, ³⁴⁰⁶⁰

Montpellier

Cedex, ^France

(2)

Centre National d’Etudes des Télécommunications

(CNET),

^{route de} Trégastel, BP 40, 22301 Lannion

Cedex,

France

(Reçu

^{le 15}

septembre

1986, révisé le 2 mars 1987, accepté ^{le 5} juin

1987)

Résumé. ²⁰¹⁴ Au cours du processus d’élaboration de

photodiodes

^{InP n+}

/p

pour la conversion d’énergie ^solaire

par diffusion de soufre à 700 C dans des substrats d’InP

dopés

Zn, ^les

propriétés électriques

^et

photoélectriques

des substrats sont fortement modifiées. Les mêmes effets sont observés après ^un

simple

^recuit

des substrats à la même température. ^La modification la plus ^{notable est une forte} augmentation de la densité de trous

(plus

d’un ordre de

grandeur)

^{uniforme en}

profondeur.

^Une

dégradation superficielle

^des paramètres

photoélectriques, longueur

de diffusion L et durée de vie 03C4 des porteurs ^a aussi été observée. Les

pièges profonds

détectés dans les substrats après traitement

thermique

^{ne sont} pas

typiques

^{de ce} traitement.

Abstract. ²⁰¹⁴

During

the process of elaboration of n+

/p

^InP

photodiodes

for solar energy conversion, ^{we have}

observed that the 700 C sulfur diffusion into Zn

doped

InP substrates

strongly

modifies the electrical and

photoelectrical properties

of these substrates. A

simple annealing

^{at the same} temperature produces nearly

equivalent

effects. The most relevant modification is a strong,

depth independent

^hole

density

^increase

(more

than one order of

magnitude).

^A

degradation

^{of the}

photoelectrical

parameters : ^diffusion

length

^L and carrier lifetime 03C4 has also been detected near the surface of the heat treated substrates. The

deep

traps measured from

deep

level transient spectroscopy

(DLTS)

^{are not}

typical

of the heat treatment.

Revue Phys.

Appl.

²²

(1987)

^{1159-1168 OCTOBRE} 1987, ^PAGE

Classification

Physics

^Abstracts

8140G - 8140R

1. Introduction.

InP has received

increasing

^{attention in the} past few years because it is ^an attractive material for

high speed circuits,

^infrared

(IR) photodetection

^and

solar cells.

Very

^efficient devices have

already

^been

realized in these

investigation

^fields

[1-3]. Among

the various steps involved in device

processing,

^heat

treatments are

required

^{for contact}

formation,

^im-

purity diffusion, implantation annealing.

^{The III-V}

compounds

^are very sensitive to these heat treat- ments which may

strongly modify

^not

only

^the

surface

properties

^{but also the} characteristics of the bulk

starting

material. An

investigation

of the heat

treatment effects is therefore necessary in order ^to obtain a better control of the process parameters.

During

^the elaboration of n+

/p

^InP

photodiodes

for solar energy conversion

by

^{sulfur diffusion at}

700 C into Zn

doped

p type substrates we have observed a strong modification of the

doping

^level

into the substrate. This paper describes this effect.

In order to check the influence of sulfur on this

modification,

^the substrates have also been submit- ted to

simple

^{700 C}

annealing

processes.

In

InP,

^{Zn is a}

commonly

^used p type

dopant

^and

some

investigators

^have

already

noticed that such substrates were affected

by

^heat treatments

[4-5]

giving

^{rise to}

doping

level modification and traps

introduction ;

the models

they

^have

proposed

^to

explain

these modifications will be

compared

^{to our}

results.

Besides the main effect of hole

density

increase,

we have also measured the

photoelectrical

^par-

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/rphysap:0198700220100115900

ameters

(diffusion length

^{and carrier}

lifetime)

^and

the traps into the heat treated substrates.

2.

Experimental.

The

investigated samples

^were p type ^Zn

doped

LEC _{grown InP}

crystals.

^The

crystals

^{50 and 158}

come from CNET

(*),

^the

crystal

^{148 was}

provided by Crystacom.

2.1 HEAT TREATMENTS. - Heat treatments were

performed

^{at 700 C in a sealed}

quartz ampoule. A phosphorus

overpressure ^was insured

by,

^{a red P}

charge

^{or InP}

powder.

^This

type

^{of treatment is} very similar to thermal

implantation annealing

^condition.

As

explained

ⁱⁿ the introduction two

types

^{of heat} treated

samples

^{have been}

investigated :

- annealed

samples (only

^P

charge

^{in the am-}

poule)

- sulfur diffused

n+ /p ^{samples ;}

^{in this case,}

In2S3

^{was added to the P}

charge (In2S3 -

1 mg ; P : 1.5 mg is the ⁵

cm3 ampoule).

The time of heat treatment : t was varied between 1/2 h and 90 h.

2.2 INVESTIGATION METHODS. - In order to inves-

tigate

^{the substrate}

modifications,

^various characteri-

zations have been made :

- Hall effect measurements

(Van

^{der Pauw}

method)

^{before and after the heat}

treatment

^with

and without sulfur diffusion.. ^.

-

Electrical

and

photoelectrical

measurements

on

n+ /p

sulfur diffused

junctions

^{and on semi-}

transparent

^Au/InP

Schottky

^diodes.

-

Deep

level transient spectroscopy

(DLTS)

measurements.

The annealed

samples

^{will be} identified

by

^the

reference number of the substrate followed

by

^the

letter T in the case of

simple

^{heat treatment and S in}

the case of sulfur diffusion. The _as-grown substrate will be denoted A. The

Schottky

diodes have been made at various

depths

^{into the} heat treated sub- strate,

they

^{are noted Sch 0}

(on

^the

surface)

^{Sch 1}

(1 >m deep),

^{Sch 3}

(3 >m deep),

^{Sch 10}

(10 03BCm deep).

3. Results and discussion.

3.1 FREE HOLE DENSITY VARIATION. ^- The com-

parison

of the hole carrier

density

^and

mobility

values before and after treatments is

given

in table I.

Figure

¹ presents the variations of the hole

density (p =1/RH q, RH :

^Hall

constant)

^versus

103/T

measured on

samples

⁵⁰

A,

⁵⁰

T,

¹⁴⁸

A,

^{148 T.}

Table I. ^- Electrical characterized

of

various Zn

doped

^InP substrates submitted to 700 C heat treatments - p :

(cm-3)-Jlp : ^(cm2fV.s).

CNET : Centre National d’Etude des Télécommunications, ^France.

1161

Fig.

^1.

- - 1

^{as a} function of

103 T

for the

samples

50 A, ⁵⁰ T, ¹⁴⁸ A, 148 T. Heat treatment time t = 12 h.

- The most relevant effect observed on heat treated

samples

for any ^{heat treatment time} is the

strong

p increase

(greater

^{than one order of} mag-

nitude).

Hole densities deduced from Hall measure- ments on the as-heat-treated substrates and after a

controled

etching

^{of some} tenths of microns of the

same substrates ^are

identical, showing

that the hole

density

is uniform into the substrate.

The hole

density

increase varies between some

1014

and

1016 cm- 3

for the less

doped crystal ;

^{3 to}

9 x

1016 cm- 3

for the

crystal

⁵⁰

(original density :

1.1 x

1015 cm- 3) ;

^{3 x}

1016 cm- 3

for the

crystal

¹⁵⁸

(original density :

^{1.8 x}

^{10 17} cm- 3).

- C-V measurements on

Schottky

^{diodes or S}

diffused

junctions

^{confirm the hole}

increase ;

^{the C-}

V densities exhibit some weaker values than

1/RH q ,

^J

values, they

^{are similar on}

Schottky

diodes Sch

0,

³

and 10.

- Before heat treatment, Hall effect ^measure-

ments on various

samples

^{of the}

crystal

^{148 have}

demonstrated a

good homogeneity

^{of this}

strongly compensated crystal.

^{After heat} treatment, hole densities at 300 K vary from

1014 cm- 3

to 1.6 x

1016 cm- 3.

The C-V measurements of

Schottky

diodes made on this resistive substrate have been made at low

frequency (103-104 Hz).

A discussion of these electrical results can be made

taking

^into account two

complementary

^infor-

mations.

a)

^SIMS

profiles

^{of Zn have been made}

^by

^H.

Gauneau on the substrate 50

[6]

^{and on} the substrate 148.

They

^{are ffat and} identical before and after heat treatment :

Nzn

^{= 7 x}

1016 cm- 3 (crystal 50), Nzn ^{=1016 cm-’} (crystal 148).

b)

^{The Zn} acceptor activation energy has been deduced from

photoluminescence

measurements.

Ea

^{= 47 meV in}

agreement

with the determination of E. Kubota

[7]

^{and E. W. Williams}

[8].

^No

deeper

level appears after ^{heat treatment.}

- Model

of

^{Zn out}

di f fu,sion.

^{- A Zn out-}

diffusion model has been

proposed by

^CCD

Wong [5]

^to describe the heat treatment effect on Zn

doped

InP

crystals

^{on the as} grown substrate : Zn atoms are

present

^{in the lattice as} substitional

acceptor Zns

and interstitial

compensating

^donor

Zni ; during

heat treatment, the

Zn.,

^{atoms remain}

unchanged

whereas the

Zni

^{suffer an} out-diffusion and evapo-

rate at the surface

leading

^{to a} decrease of compen- sation. CCD

Wong

based his model on the _exper- imental observations of a Zn loss into the heat treated substrates

(SIMS measurements)

^{and on the}

existence of a carrier concentration

profile (deduced

from electrical

measurements).

This model

desagrees

^{with our results because :}

1)

^The

predicted

^{decrease of Zn atoms has not}

been observed in our SIMS measurements.

2)

^The

predicted inhomogeneous ,

^{hole densities}

into the

substrate,

for moderate treatment times at

700 C,

^{have not been observed in electrical}

measurements.

- Model

of

Zn activation. ^- The

density

^{of Zn}

atoms

being unchanged

^{after heat} treatment, ^{it can} be

thought

that the increase of carriers

might

^{be due}

to an electrical activation of Zn : interstitial Zn

(Zni) turning

substitutional Zn

(Zn,,).

In order to check this

hypothesis,

^{the Hall effect}

measurements have been

compared

^{to the hole}

density

deduced from the relation :

valid in the case of a

compensated

p type ^semicon- ductor :

Na is

^the acceptor

density : Na

⁼

NZns

^{if Zn}

is the

only

acceptor.

Na

is the donor

density : Nd

⁼

Nzi

⁺

Ndr ;

the residual donors

Ndr

^{into InP}

are

generally

sulfur and silicon. SIMS

profiles

^meas-

urements have been made on the

crystal

^{50 before}

and after heat treatment, ^the sulfur

density (5

^x

1015

cm-

3)

^was

^flat, independent

^{of the heat treat-}

ment

(the

^{sulfur out diffusion described}

by

^B.V.

Dutt et al.

[9]

^{was not}

apparent),

^{therefore we have}

taken

Ndr

^{= Cst. = 5 x}

1015 cm- 3

for the three

samples. Ny

^{is the}

equivalent

^state

density

^{of the}

valence band

(for my*

^{= 0.85 mo,}

N y (300 K) =

1.93 x

1019 CM- 3, Nv (77 K)

^{= 2.56 x}

¹⁰¹⁸ cm- 3 ) ; g

represents ^the

degeneracy

^{factor of a shallow} accep- tor

level,

^{in the} valence band structure of InP

With the

preceeding hypothesis, NZni

^and

NZns ^obey

the two relations

Table II presents the values of

NZn, p

^and

NZns

^and

Nz.,

^{for the}

^crystals

^{50 and} 148 before and after heat treatment.

2022 For the

sample

^{50 T the} relations 2 and 3 cannot be valid because

they

^{lead to}

which is

impossible.

2022 For the other

samples

^the

calculated p = f (T )

variations

(relation (1)) completely desagree

^with

the

experimental results, especially

^the saturation of carrier

density

^{near 300 K}

expected

^{from the shallow}

Zn level is not observed. This fact indicates also that the model of electrical activation of Zn is not valid.

- In conclusion neither the Zn out diffusion model nor the Zn activation model can account for

our

experimental

observations. It seems that the

presence of other

deeper

acceptors ^{would be neces-} sary ^to

explain

both results before and after heat treatment. Such levels would not

participate

^to

radiative recombinations as indicated

by

^luminesc-

ence measurements so an

arbitrary

^{choice of their}

energy level and

density

^{to account for}

the p

⁼

f (T)

measurements would be unrealistic.

3.2 CARRIER LIFETIME PROFILE. ^- This parameter has been deduced from the forward I V character- istics of S diffused

n+ /p

^mesa

^diodes,

^{in the}

polarization

range of

generation-recombination

^con-

duction.

Figure

² presents ^{the room}

temperature

^forward

1 V characteristics for diffused

junctions

^{realized on}

CNET-50 substrate : diodes

S3

^and

S32

^{are shallow}

junctions (xj

¹

^{03BCm) ;}

^diodes

^S31

^and

^S36

^are

^deeper

(xj

> 1

03BCm).

^The

ideality

coefficient n is constant

n ⁼ 2 in the whole bias

voltage

range for shallow diodes whereas it is

equal

^to

unity (diffusion

^conduc-

tion)

^at

high injection

^for

deeper

^diodes

showing

^the

better

quality

of these last ones.

The forward current

density

^can then be written :

in which

J0 GR

^and

Jo D

^are

respectively

the preexpo- nial terms of G-R and diffusion currents. In the

voltage

range in which n ⁼

2, generation-recombi-

nation

(G-R)

^{was confirmed}

by

^a temperature

study : figure 3(a)

^{shows the forward I-V character-}

istics of the shallow diode

S32

^{between 300 and}

353 K.

Figure

3b shows the variation Ôf

I0 G-R

^vs.

103/ T ;

^{the deduced} activation energy value is about 0.7 eV

(half

energy gap

(Eg ) ^value).

Therefore the G-R lifetime TG_R ^can be deduced from these I-V measurements

[10]

^{and are}

reported

^on table III and

figure

4. The TG_R values ^are

strongly dependent

^on

junction depth, they

^{increase with}

_xj.

^The

good quality

^of

deep junction

^{is confirmed}

by

^{the domin-}

ant diffusion current observed at

high injection (J >

¹

mAlcm2).

^{On our best diodes this}

quality

^was

associated to a low dark reverse current :

jo.9 V. _ ^{10 - 7 AIcm2, VB :}

^breakdown

^{voltage equal}

Table II. ^- Substitutional and interstitial Zn densities

Zns, Zni

^deduced

from

SIMS and 300 K Hall

effect

measurements in the case

of

^{a Zn} electrical activation model.

1163

Fig.

^{2. - Room} temperature forward I-V characteristics for mesa sulfur diffused

n+ /p

InP diodes with various

junction depths.

Substrate 50

to 14 V in

agreement

^{with the}

empirical

^relation

given by

^Sze

[10].

Similar results have been obtained

by comparing

the diodes

S47 (xj

^{= 0.8}

^03BCm)

^and

^S49 (xj

^{= 2.15}

^03BCm)

made

by diffusing

^{S into the substrate 148.}

Figure

⁵

presents

^{their room} temperature ^{forward character-}

istics,

^the

ideality

coefficients observed on various

mesa diodes

S47

vary from 1.66 to 2 whereas

n = 1 is

only present

^{on the diode}

S49

^{in the}

high voltage

range. The TG.R values in this substrate 148

are very similar ^to those determined on the substrate 50

(see

^Tabl.

III)

^for

equivalent junction depth ; they

^decrease

by

^{one order of}

magnitude

^{when the}

junction depth

increases from 0.8 to 2.15 _F£m. The increase of

JoD in

this substrate with respect ^{to the}

Fig.

^{3. -}

a)

Forward I-V characteristics at various tem-

peratures ^Diode S32.

b)

IOG.R ^versus

103 / T

^{for diode} 532.

substrate 50 can be related to the

higher doping

^level

of this last one.

The forward characteristics of these 700 C sulfur diffused

junctions

^{can be}

compared

^to those of A.

Yamamoto

[11]

fabricated at 625 C on a p type substrate

(p =1016 cm- 3 ).

^{On mesa diodes}

xj =

0.3 _03BCm, thèse authors have observed n ⁼

1.29,

J0 = 1.23 10-13 a/cm2.

3.3 DIFFUSION LENGTH PROFILE. ^- The detailed

photoelectrical study

^of substrate 50 has been made

Table III. ^- Electrical parameters

of

^some

n+/p sulfur diffuse

^InP

junctions.

Fig.

^{4. -} Generation-recombination lifetime bG-R ^{as a}

function of

junction depth

in sulfur diffused

n+ /p

^InP

diodes

(substrate

^{50 : e,} substrate 148 :

1 ).

and allows to deduce the

Ln profile

in this heat treated substrate.

- Schottky

^{diodes. - The}

Ln profile

^{has been}

deduced from the

analysis

^{of the}

efficiency

^{of semi-}

transparent

^Au/heat treated p InP

Schottky

^diodes

Fig.

^{5. -} 300 K forward 1 V characteristics of sulfur diffused n+

/p

^{InP 148 diodes}

1165

realized at various

depths

^{into the}

substrate,

^ob-

tained

by

controlled

etching using

^{0.5 % Br-}

methanol

(etching

^{rate - 0.7}

>m/min).

The evapo- rated

gold

^{dots were 80 A} thick with a diameter of 360 _tjbm. In a

Schottky ^diode,

the internal collection

efficiency

TJ can be written :

the first term of the sum

represents

the contribution of the neutral

substrate,

the second the contribution of the space

charge region ; a

^{is the}

absorption

coefficient. This relation shows that for a

given

^a

value, Ln

^{can be} deduced from the variation of 77 versus w

(that

^{is to} say ^versus the reverse

voltage).

The modulated

light (À

^{= 0.85 lim, a =}

1.85 x

104 cm-1,

^{P = 5}

>W)

is delivered

by

^{a GaAs}

light emitting

^diode

(LED).

The modulated

(1000 Hz)

^induced

photocurrent

^{is measured}

through

^a load with a lock-in

amplifier. Figure 6

. presents ^the

experimental

^{relative values}

n 1 (- V)/ï?i(0)

^{as a} function of w and the calculated

curves

giving

^{the best} agreement ^with

experiment.

The

Ln

^{values as a} function of

depth

^are

reported

^on

figure

^{7a. On the same}

figure

^the TG-R

profile

^{is also}

presented.

^{Two remarks can be made :}

1)

^The Tn values deduced from

Ln

^and

Jo D

^values

for

equivalent junction depths (see

^Tab.

III)

^{for the}

substrate 50 agree with TG_R values. This result ^can be related to the fact that S diffusion does not

degrade

the substrate material.

2)

The variations of

Ln

^{and T versus}

_Xj

^are

similar, indicating

^{that the}

mobility

^of

minority

carrier is

homogeneous

into the heat treated substrate.

Fig.

^{6. -}

Expérimental

and calculated variations of

~i(-v) ~i(0) = f(w)

^{for the four}

^Schottky

^diodes

(À

^{= 0.85}

03BCm).

Fig.

^{7a. -} Ln ^and

J UO-R ^proûles

in the substrate 50 for a

depth

⁵ >m.

Fig. ^{7b. -} Ln values deduced from EBIC measurements

on diode

S».

^,

EBIC measurements have allowed to deduce the

Ln

^values

deeper

into the substrate. The determi- nation are

reported

ⁱⁿ

figure

7b. The heat treatment seems to

degrade

^the

photoelectrical

parameter

Ln

^{more than 10 J.1.m} far from the surface. The bulk value

(Ln

^{= 12}

J.1.m)

is normal for untreated p InP of

equivalent doping

^level

[12].

- S

diffused

ⁿ⁺

/p

^{diodes. - The}

efficiency

^{of an}

homojunction

^{is due to the}

photoresponse

^{of three}

regions :

the front diffused n+

region,

^the space

charge region

^localized in the less

doped

p

region (the doping

^{level in the n diffused}

layer

^{is as}

high

^as

2 x

1018 cm- 3),

and the base. In the

photon

energy range 1.27-1.35

eV,

^{in which a is}

weak,

^{the contri-} bution of the base is dominant and _{il is}

equivalent

to :

Figure

⁸

presents

the variation

of ni i 1 (V

⁼

0 )

^{as a}

function

a -1

for the shallow

junction S32 . a -1 ^being

deduced from the data

given by Seraphin [13].

^The

Fig.

^{8. -}

11 i 1= f (« -’)

^{for the}

homojunction S32.

. :

expérimental points

^°

: linear variations for Ln ^{= Cst}

---- : calculated variation for the Ln

profile

^of

figure

^7.

experimental points

^{do not} follow the linear

expected

variation in the case of a constant

Ln

^value

(full line).

^An agreement ^between

experiment

^{and calcu-}

lation has been obtained

by taking

^{into account the}

Ln profile

ⁱⁿ

figure 7 ;

^a numerical resolution of

continuity equation

for electrons was in this case

necessary ^and the calculated

rl î 1

^{variation is re-}

ported

ⁱⁿ

figure

⁸

(broken line).

^This

spectral

^re-

sponse allowed ^to

investigate

the diffusion

length profile

¹⁵ f.Lm far from the

junction.

The

study

^{of the relative}

variation q i (- V) / "., i (0)

as a function of w for À ⁼ 0.85 >m has allowed to

precise

^the

Ln profile

^{in a} restricted range ^near the space

charge edge. Figure

⁹

presents

the exper- imental determinations and the calculated variation

(broken line) taking

^{into account the}

Ln profile

ⁱⁿ

figure

^{7. The}

good agreement

^between

experiment

and calculation confirms the

validity

^{of the diffusion}

length profile : Ln

^{= 0.3} )JLm ^near the surface and

Ln

> 4 >m ^{at a}

depth

^{of 3} 03BCm.

3.4 DEEP LEVEL TRANSIENT SPECTROSCOPY

(DLTS)

MEASUREMENT. - The

traps

will be de- noted

HTI.

^The

type

^of measurement have been

performed

^{in order to find a}

possible

correlation between the carrier

density

and the presence of

deep

Fig. 9. 2013 7?i(- V)

⁼

^W

^for

homojunction 32’

. :

expérimental points

: constant diffusion length Ln ^model

---- : calculated variation with the Ln profile ^of

figure

^7.

_

traps. ^{The behaviour of}

crystals

^{50 and 158 and that}

of

crystal

^{148 are}

strongly

^different.

- Substrates 50 and 158. ^- DLTS measurements

on heat treated substrate 158 reveal two hole traps

HTT1

^and

HTT2

^with

respective energies : E,

⁺ 250 meV and

E,

^{+ 475 meV and}

capture

^cross sections : 1.5 x

10-17 CM2

and 2 x

1O-12 cm2.

On heat treated substrate

50, HTT1

^is

only

present.

These

traps

^{are not detected into as} grown

crystals

so we can conclude that

they

^have been introduced

by

^heat treatment. We have

investigated

^{in more}

detail the

profile

^{of the}

HTT1 trap using

double DLTS

(DDLTS)

measurements.

Figure

10 shows that this

trap desappears

^{at a}

depth

^{of 0.2} ktm.

Analog

behaviour was observed on

HTT2 trap.

Measurements

on

deep Schottky

^diodes

confirm,

^{that no}

trap

^is

present

in the bulk of the heat treated substrates. In S diffused

diodes,

in which the traps ^are observed

deeper

^{into the}

crystals (xj

> 0.2

p,m ),

^{no additional}

trap

has been detected. This observation shows

that,

S diffusion does not introduce

trap.

The _energy level of

HTT2

is similar to that of the

trap

^observed

by

^C.C.D.

Wong [5] : (Et

^{= 434}

meV,

1167

Fig. ^{10. -} Deep ^level

Hrr, profile

into heat treated

crystal ^50.

Ut ⁼ 3.2 x

10-14 cm2) :

^{but its} capture ^{cross section} is very much

higher: uHm

^{= 2 x}

^{10-12 cm2.}

HTT2

^{is also different from the}

deep

^hole

trap (522 meV) [14, 15]

^{observed on Si}

implanted

^and

annealed n+

/p

^{InP 158}

junctions.

The DLTS results

on

simply

annealed p InP confirms that the 522 meV

trap

^was introduced

by implantation.

- Substrate 148. ^- DLTS measurements have not been

possible

^{on the}

highly

^resistant

crystal

148 A.

On the heat treat

crystals

^four traps ^{are detected.}

Figure

¹¹

presents

^the

typical

^{DLTS scan, and the}

parameters ^{of the observed} traps.

The difference between the

magnitudes

^{of the}

HTT4

traps ^on

samples

^{Sch10 and S49 does not}

correspond

^{to a net} difference in traps

densities,

^{it is} due to the greater ^hole

density

into the S49

sample (see

^Tabl.

I)

^as

AC/C

^oc

Nt/2 (Na - Nd ).

The four traps ^are

present

^{both at the surface and}

in the bulk of all the

samples indicating

^that

they

^are

probably

^native defects. The traps

HTT4

^{could be}

compared

^{to the hole} trap

(Et

^{= 480}

^meV,

U 00

=10-13 CM2)

^observed

by

G. Bremond

[16]

in p type InP and the

trap HTT5

^{to the} trap ^detected

by

^SS

Li et al.

[17]

^{into Zn}

doped

InP Some authors

[18]

have

reported

^that substitutional Zn could be as-

sociated to native

defect,

^{the lack of DLTS measure-}

ment on as grown 148 substrate does ^not allow to attribute the hole increase to a dissociation of such a

complex.

The absence of

deep

traps ^{in the bulk of as} grown ^{and heat} treated

crystals

50 and 158 does not allow to attribute the hole increase ^to this

phenomenon.

Fig.

^{11. - DLTS scan of heat} treated 148

crystals

^as

observed on various devices, table of the traps parameters.

4. Conclusion.

Electrical and

photoelectrical parameters (hole

^den-

sity

p, lifetime ^T, diffusion

length Ln)

^{in Zn}

doped

InP substrates are

strongly

^modified

by

^{a 700 C heat}

treatment

during

time of half an hour or several hours. The hole

density

^increase

(more

^{than one}

order of

magnitude)

^is

homogeneous

into the sub- strate. The Zn

exodiffusion,

^Zn electrical activation and the dissociation of a Zn

complex

^{failed to}

account for this increase. The variations of ^T and L

as a function of

depth

^{into the heat treated}

samples

indicate a strong

degradation

^{of the material near}

the surface. Such a behâviour

impedes

the realization of efficient shallow diodes for solar energy ^conver- sion

involving

^{heat treatment}

(T

⁼

700 °C )

^{on this}

type of InP substrate.

References

[1]

KAWAKAMI, T., OKAMURA, M., Electron. Lett. 15

(1979)

^502.

[2]

KIM, ^O. K., FORREST, ^S. R., BONNIER, W. A., SMITH, ^R. G.,

Appl. Phys.

^{Lett. 39}

(1981)

^402.

[3]

YAMAGUCHI, M., UEMURA, C., YAMAMOTO, A., SHIBUKA, A., Jpn. ^J.

Appl. Phys.

²³

(1984)

^302.

[4]

TSUBAKI, K., SUGIYAMA, K., Jpn. ^J.

Appl. Phys.

¹⁹

(1980)

^1789.

[5]

WONG, ^{C. C.} D., BUBE, ^R. H., ^J.

Appl. Phys.

⁵⁵

(1984)

^3804.

[6]

GAUNEAU, M., CHAPLAIN, R., RUPERT, A., RAO, E. V. K., DUHAMEL, N., ^J.

Appl. Phys.

⁵⁷

(1985)

^1029.

[7]

KUBOTA, E., OHMORI, Y., SUGII, K., J.

Appl. Phys.

55

(1984)

^3779.

[8]

WILLIAMS, E. W., ELDER, W., ASHES, M. G., WEBB, M., MULLIN, ^J. B., TRAUGHAN, ^B. S., TUFTON, ^P. J., J. Electrochem. Soc. 120

(1973)

1741.

[9]

DUTT, ^B. V., CHIN, ^A. K., CAMLIBEL, I., BONNER, W. A., J.

Appl. Phys.

⁵⁶

(1984)

^1630.

[10]

SZE, S. M.,

Physics of

Semiconductor Devices, ^2nd Ed.

(John

Wiley ^and Sons, ^New

York)

1981, p. 104.

[11]

YAMAMOTO, A., YAMAGUCHI, M., UEMURA, C.,

Appl. Phys.

^{Lett. 47}

(1985)

^975.

[12]

GOUSKOV, L., LUQUET, H., SOONCKINDT, L., OEMRY, A., BOUSTANI, M., NGUYEN, P. M., ^J.

Appl. Phys.

⁵³

(1982)

^7014.

[13]

SERAPHIN, ^B. O., BENNETT, ^H. E., Semiconductors and Semimetals V3, ^{ed. R.} C. Williard Sons and A. C. Beer

(Ac.

^Press

London)

1967, p. 530.

[14] CONJEAUD,

A. L., ORSAL, B., DHOUIB, A., ALABEDRA, R., GOUSKOV, L., ^J.

Appl. Phys.

59

(1986)

^1707.

[15] GOUSKOV,

L., CONJEAUD, A. L., DHOUIB, A., FAVENNEC, ^P. N., SALVI, M., L’HARIDON, H., BASTIDE, G., BAYAA, D., ^{Solar Cells 11}

(1984)

343.

[16]

BREMOND, G., ^{Thèse de}

Docteur-Ingénieur,

^INSA Lyon

(1981).

[17]

LI, ^S. S., WANG, ^W. L., SHABAN, ^E. H., ^{Solid State}

Commun. 51

(1984)

^15.

[18]

RAO, ^{E. V.} K., DJAMEI, M., DUHAMEL, N., ^Proc.

of

13th Int.

Conf. Defect in

Semiconductors, ^ed.

L. C.

Kimeling

^{J. M.} Parsey

(Bell

^Lab. Murray Hill, ^New

Jersey)

1984,

p. 1123.