Étude de certaines propriétés de couches de SiO2 sur support de silicium

HAL Id: jpa-00205767

https://hal.archives-ouvertes.fr/jpa-00205767

Submitted on 1 Jan 1964

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.

Étude de certaines propriétés de couches de SiO2 sur support de silicium

Akos Revesz, Karl H. Zaininger

To cite this version:

Akos Revesz, Karl H. Zaininger. Étude de certaines propriétés de couches de SiO2 sur support de

silicium. Journal de Physique, 1964, 25 (1-2), pp.66-69. �10.1051/jphys:01964002501-206600�. �jpa-

00205767�

66.

ÉTUDE DE CERTAINES PROPRIÉTÉS DE COUCHES DE SiO2

SUR SUPPORT DE SILICIUM

Par AKOS REVESZ et KARL H. ZAININGER,

RCA Laboratories, ^{Princeton New} Jersey, ^{U. S. A.}

Résumé.

Des couches de SiO2 ^sur support ^{de Si ont été} produites soit par oxydation ^thei- mique ^dans la vapeur d’eau, soit par dépôt ^à partir ^d’acide hydrofluosilicique, soit par oxydation anodique

électrolytes

aqueux. Elles ont été étudiées par ellipsométrie. L’indice de réfrac- tion

été utilisé pour calculer ^le coefficient de remplissage. ^On

^{étudié la} cinétique de croissance par oxydation dans la vapeur. L’oxydation anodique ^dans

^{sel fondu}

^donné

^film composé.

Abstract.

Films of SiO2

Si, produced by ^thermal oxidation in steam, by deposition ^from hydrofluosilicic acid, ^and by anodic oxidation in non-aqueous electrolytes,

^examined by ellipsometry. The index of refraction

used to calculate the coefficient of compactness. ^The growth kinetics of the steam oxidation

examined. Anodic oxidation in

molten salt resulted

in a composite ^film.

LE

PHYSIQUE TOME 25, JANVIER-FEVRIER 1964,

Films of Si02 ^{on Si can be} produced by ^various

methods. They ^are conveniently categorized ^into

the following groups : (1) Thermal oxidation in various ambients. (2) Anodic oxidation in diffe- rent electrolytes. (3) ^Chemical deposition

methods. The properties ^{of the} resulting SiO 2

films depend ^{upon the method} of formation.

Archer [1] found that there is a considerable dif- ference in the index of refraction and density ^of

films obtained by thermal oxidation in dry oxygen, in high pressure steam, anodic oxidation in an

aqueous electrolytes, ^{and thermal} decomposition

of tetraethyl-ortho-silicate. ^Deal [2] ^{observed si-} gnificant changes ^{in the} density ^{of films} produced by thermal oxidation ⁱⁿ dry oxygen, ^wet oxygen, and atmospheric ^steam. Politycki ^{and Fuchs} [3]

determined by ^electron microscopy ^that films pro- duced by anodic oxidation in different electrolytes

had various densities of pores.

The purpose ^of this investigation ^{was to} study by ellipsometry properties ^of Si02 ^films prepared by

various methods. Ellipsometry allows the deter- mination of the index of refraction and thickness of these films. The refractive index was used to calculate a coefficient of compactness which char- acterizes the structure of the oxide film. The thickness determinations were utilized for the’

study ^{of the} growth ^{kinetics of one} of the oxi- dation methods.

The ellipsometer ^{used is a} spectrometer ^with analyzer, polarizer, ^and quarter ^wave plate, ^all

mounted in divided circles, ^{and a} monochromatic

light ^{source of 5 461 A} wavelength. The illumi- nated sample ^{area was} 0.2 cm2. A photomul- tiplier microphotometer ^{was used as a detector}

and extinction settings ^{were obtained} by employing

a method of successive approximation. During

all experiments ^the angle of incidence was kept

constant at 70.00°. For the relation between A and tan § (i.e. ^the phase difference and amplitude

ratio of the two components ^{of the} elliptically polarized light), and the index of refraction, n, and the thickness, t, ^{of the} films, ^{we used the} graphical representation of the results of a computer ^solution

for transparent ^{films on} silicon, ^as published by

Archer [1].

The experimentally determined values of the index of refraction allowed us to get a measure- of the compactness ^{of the structure} of the films _pro- duced by using ^the equation [4]

where 8

coefficient of compactness,

n1 == index of refraction of material exa-

mined,

n2 = index of refraction of reference mate- rial.

Because all the films examined are amorphous,

the obvious choice for the reference material would

seem to be fused Si02. However, if this is done,

then the 3-values of some of the films are larger

than unity, ^which is, ^of course, meaningless. ^If

we chose a-cristobalite as a reference then all the 8-values are less than unity. This choice is per- missible because the molar refraction of «-cristo- balite is essentially ^{the same as} that of fused Si02, indicating that their polarizabilities ^{are the same}

in spite ^of the difference in their index of refraction and density ^{values. This} is, ^of course, ^{a conse-} quence of the well-known structural and thermo-

dynamical similarities between these two modifi- cations of Si02. ^{On the basis of this} argument

fused SiO 2 ^can be characterized as «-cristobalite

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01964002501-206600

67

having ^a coefficient of compactness ^{of 0.954} using

the following ^values [5] :

The coeflicient ^of compactness ^can also be deter- mined by ^the density ratio if the polarizibilities ^are

the same :

where d,

density of material examined, d2

density of reference material.

Density determinations, however, ^{are much more}

cumbersome and less accurate than measurements of the index of refraction. Therefore, ^{it is} pref-

erable ^to determine 8 from"Eq. (1). If 8 and 8’

differ significantly, ^{then the} polarizabilities ^are

not the same and some change ^{in the structure}

must be present.

Silicon wafers having ^a (111) orientation were

oxidized in atmospheric ^{steam at 900 °C and}

1000 OC. It was found that there was no diffe-

rence in the index of refraction of the Si02 ^films

obtained at the two different temperatures. ^The

mean value of n is 1.445 and the distribution is characterized by ^{a standard deviation of} 0.017,

based on 11 samples. The coefficient of com-

pactness ^{is 0.930.}

Infrared absorption measurements revealed that the 0 H- concentration in these oxide films is 2 X 10-2 moles per mole of Si02 [6]. However,

their index of refraction is smaller than that of fused Si02 ^{with the same OH"} concentration indi-

cating ^that other factors (pinholes etc.) ^{are re-} sponsible ^{for this.}

A graph of the thickness of the oxide films vs

the time of oxidation on a logarithmic ^{scale is}

shown in figure ^{1. From the} diagram ^{it can be}

seen that there are two regions which have dif-

PIG. 1.

Thermal oxidation of Si in atmospheric ^steam ..,at ^{1 000 °C.}

ferent growth ^{kinetics. Above an} oxide thickness of 3 000 A ^the growth ^is diffusion controlled as

given by ^the slope of 0.5. This is in agreement

with the results of other workers [2, 7]. Below ^an

oxide thickness of 2 000 A the slope ^{of the} straight

line is 1.2 which shows that the kinetics in this thickness range is ^not diffusion controlled. This is in agreement ^{with the} generally accepted ^theories

of oxide growth [8] ^{which state that diffusion is}

the controlling ^mechanism only ^{if the oxide}

thickness is larger than the space charge region.

Films on mechanically polished silicon wafers have been obtained by deposition ^{of silicon dioxide}

from hydrofluosilicic ^acid supersaturated ^{with sili-}

ca [9]. The average index of refraction of these films is 1.439 with a standard deviation of 0.015 based on six samples. This results in a coefficient of compactness 8

^0.916 indicating ^{that the}

films produced by this process ^are less dense than those produced by atmospheric ^{steam oxidation.}

The thickness of the investigated ^films covered the range from 900 A to 4 000 A. For several of these

samples ^{the oxide thickness was} determined as a

function of the position ^{on the} sample by multiple

beam interferometry [10]. For this purpose, the

SiO, ^{was removed} except ^{for an area} of 1 cm2 in the center of the sample ^{and the thickness was then}

determined along ^the edges. Evaluation of these data shows that there ^{is a} statistically significant

local variation in the .oxide thickness. Thus, for a specimen ^{with an} average oxide thickness of about 3 300 A,this variation is + ¹⁰⁰ A. ^Our ellipsometric

thickness measurement did not reveal these varia- tions because it only gives the average oxide thickness for the illuminated sample ^area.

The value of 8’ using ^an average density ^{of 2.1}

is 0 . 95. This differs somewhat from the value of a. However, ^{both the error and the} spread ^of

the density determination are such that it is not

justified ^{to draw the} conclusion that the polari- zability ^has changed.

Chemically polished silicon wafers were oxidized

anodically ^{in a} non-aqueous electrolyte consisting

of 0 . 04 ^M KNO3 ⁱⁿ N-methylacetamide [11]. ^The

index of refraction of the films produced by ^this

process is 1.480 resulting ^{in a} coefficient of com-

pactness ^0.990. For films produced by ^anodic

oxidation of silicon in an aqueous electrolyte ^we

calculate 8

0.774, using n

^1.362 [1]. ^From

this comparison ^{it is} immediately obvious that the

electrolyte ^{has a very} strong ^{influence on} the prop- erties of the resulting oxide film. This led us to

an investigation of oxide films produced by ^the

anodic oxidation of silicon in ^a molten salt.

The electrolyte consisted of 34 O/o ^of LiN03 ^and

66 % ^of KN03 ^at 160 °C. Curves of voltage ^vs

time for the constant current case as well as curves

of current vs time in the constant voltage ^case

indicate that insulating films have been grown.

The resistivity of these films at the operating ^tem- perature ^is generally ^{in the} order of 1011 ohm-cm.

Since little is known concerning ^the behavior of Si

in molten nitrates, ^some specimens ^{have been}

immersed for various times (1-15 minutes) ^{in the} electrolyte ^{without an} applied voltage, ^{and subse-} quently ^examined.

Reflection electron diffraction experiments per- formed on dipped ^and oxidized wafers showed that the oxide films are amorphous. (It ^{was also}

found that these films are soluble in HF.) Dipped wafers, ^{on the other} hand, ^{did not exhibit} any

amorphous ^film (thicker ^{than about 50} A) ^and

showed only ^the single crystal pattern ^{of silicon.}

The specimens ^were investigated by ellipsometry

and their representative points ^are plotted ^{in the}

proper section of the A vs § plane ^{and shown in} figure ^{2. The} points ^{do not fall on a} single ^curve

of constant index of refraction but rather cover a

FIG. 2.

Section of A vs § plane with lines of constant index of refraction

evaluated for films

silicon.

Shown are the ideal and the best experimental point ^for film free silicon. The points ^which

^{inside the dashed} rectangle represent samples which have either been

simply immersed in the molten salt

which have been

anodically ^{oxidized and} then treated with HF. Shown

are also points corresponding ^{to oxidized} samples.

range of n from 1.7 to 2.7. These values ^are

quite unrealistic for Si02 films. Based on previous experience [1] it is safe to assume that films grown

by ^{the same} process have almost constant index of refraction. Applying ^this assumption ^{to the}

ease under discussion and drawing ^{a line of con-}

stant index of refraction by connecting ^our oxperi-

mental points ^{we see that this curve} approaches ^a point ^{which is} quite ^different from that charac- teristic of film free silicon.

Points representing ^immersed specimens ^{as well}

as oxidized wafers which have undergone ^{a sub-} sequent ^{HF treatment also do not} cluster in the

region ^otherwise characteristic of film free silicon.

They ^{fall on curves for a film on} silicon with an

index of refraction between 3.4 and 3.6. The thickness of these films, ^{as estimated} by ellipso- metry, ^{is about 90 A. HF treatment of the immer-}

sed wafers did not change ^the ellipsometric ^results, indicating that these films are not soluble in HF.

Both the high ^{index of} refraction as well as the data obtained from electron diffraction expriments

showed that these films are not Si0 2 ^{but are similar}

to the stain films reported by ^Archer [12]. ^Howe-

ver, it should be noted that in our case there is no

possibility of formation of any silicon-hydrogen compound. ^For several oxidized specimens ^the

oxide thickness was measured by multiple ^beam interferometry. ^The resulting ^{values were} always larger than those obtained by ellipsometry ^based

on the ideal Si-SiO2 ^interface.

From all these observations we draw the con-

clusion that if silicon is anodically ^{oxidized in this}

molten salt, ^the resulting ^structure consists of a

thin film characterized by ^a high ^{index of refrac-}

tion covered by ^the SiO.. ^{For this reason} the gra-

phical ^{results as} evaluated ^{for the ideal case of} transparent ^{films on silicon are not} applicable ^here

and the index of refraction and the coefficient of

compactness ^{of these oxide films were not deter-}

mined. Work is in progress ^to accurately ^measure

the optical properties ^{of this} intermediate film and

to solve the ellipsometry equation ^{for this case so}

that the properties of the oxide films can then be determined.

Acknowledgments.

The authors would like to thank R. J. Archer for very stimulating discussions

as well as for supplying ^the large version of his

published A vs § graph, ^{and R. J.} Evans, ^who performed the interferometric investigations ^and

assisted in the experimental ^work.

Discussion

M. PLISKIN.

What refractive index did you

get ^{for the} dry oxygen oxidation ?

In view of the fact that you showed different rates for steam oxidation, ^{for what} thickness did you determine the given refractive index of 1.445.

This refractive index of 1.445 is significantly ^less

than that which we have determined for thick Si0 2

films. In our studies films of several thousand

angstroms produced by ^{steam at 1000 OC at atmo-} spheric pressure ^{are the same as those} produced by dry oxygen except ^for hydrogen ^{bonded SiOH} groups which appear ^to be concentrated near the surface and are easily ^removed by subsequent dry

oxygen oxidation ^at 1000 OC. Recently ^{B. Deal}

in the Electrochemical Society Journal has shown

that with sufficient precautions ^the density ^{of films}

produced by ^{steam oxidation are} essentially ^the

69

same as that produced by dry oxygen oxidation.

Réponse : ^The presence ^of different controlling mechanisms-depending ^{on the} thickness of the oxide film

does not mean that the oxide has ^a

structure varying ^{as a function of} the thickness.

The measured values of SiO 2 ^films grown in steam are essentially ^{the same and are} independent

of the thickness.

LITERATURE

[1] ^ARCHER (R. J.), ^J. Opt. ^Soc. Amer., 1962, 52, ^970.

[2] ^DEAL (B. E.), J. Electrochem. Soc., 1963, 110, ^527.

[3] ^POLITYCKI (A.) ^{and FucHs} (E.), ^Z. Naturforsch., 1959, 14a,271.

[4] ^BÖTCHER (C. ^J. F.), Theory of Electric Polarisation,

p. 415, Elsevier, ^New York, 1952.

[5] Gmelin’s Handbuch der anorganischen Chemie, ^Teil B, System ^Nummer 15, Silizium, 1959.

[6] ^REVESZ (A. G.) and ZAININGER (K. H.), ^RCA Rev.,

to be published.

[7] ^ATALLA (M. M.), Properties ^{of Elemental and Com-} pound Semiconductors, vol. 5, Metallurgical Society Conferences, p. 163, Interscience Publishers, ^New York, ^1960.

[8] ^GRIMLEY (T, B.), ⁱⁿ Chemistry ^{of the Solid} State, W. E. Garner editor, p. 336, ^Academic Press,

New York, 1955.

[9]

^THOMSEN (S. M.) ^{and NICOLL} (F. H.), ^{U. S. Pat.}

2,505,629.

b. THOMSEN (S. M.), ^J. Amer. Chem. Soc., 1952, 74,

1690.

[10] ^TOLANSKY (S.), ^Surface Microtopography, ^Inter-

science Publishers, Inc., ^New York, 1960.

[11] ^SCHMIDT (P. F.) and MICHEL (W.), J. Electrochem.

Soc., 1957, 104, ^230.

[12] ^ARCHER (R. J.), ^J. Phys. ^Chem. Solids, 1960, 14, ^104.

OPTICAL ABSORPTION IN VERY THIN DIELECTRIC FILMS AND ITS ORIGIN

By ^{ROBERT C.} PLUMB,

Worcester Polytechnic Institute, Worcester, Massachusetts, ^{U. S. A.}

Résumé.

^Le comportement optique de couches assez épaisses ^d’un diélectrique transparent déposé ^{sur des surfaces} métalliques peut être décrit avec précision par

modèle idéal

admet- tant qu’elles ^sont homogènes ^{et à faces} planes ^et parallèles. ^{Les couches} diélectriques d’épaisseur

inférieure à 100 Å

comportent différemment. On constate la présence ^d’une absorption ^anormale

dans les couches minces diélectriques. On propose

explication ^{de cette} absorption

^moyen

de la double couche électrique.

Abstract.

Although ^the optical behavior of rather thick transparent dielectric films

metal surfaces

be accurately ^described by

idealized model considering ^the system

plane parallel-

sided homogeneous phases ^{with a} discontinuous boundary ^{between them,} deviations

encoun-

tered when the dielectric phase ^is thinner than 100 Å. Evidence is presented that there is

lous absorption in the thin dielectric films. An explanation ^{of this} absorption ^{in terms of the}

electric double layer ^is proposed.

JOURNAL DE

PHYSIQUE

25, JANVIER-FÉVRIER 1964,

Introduction.

It is certainly ^an idealization to consider that two solid phases ^{in contact are des-}

cribable as two plane parallel-sided homogeneous phases ^{with a} discontinuous boundary ^between

them. Of course there _are, in most systems ^one encounters, deviations from the simple plane parallel-sided model caused by ^the topographical

structure of the two phases ^{in contact} but, ^under

realizable circumstances, ^these topographical ^devia-

tions from ideality may be minimized. When ^the

topography ^is such that the interface is indeed

planar, ^{it is still an} idealization to consider that two phases ^{in contact are} homogeneous ^{and in}

contact through ^a discontinuous boundary ^because phases ⁱⁿ contact come to ele9trochemical equi,

librium with each other by transfer of particles

across the interfacial boundary. ^This equili-

bration produces ^many physical ^effects including

contact potentials ^{and Voltaic cell} potentials.

This paper will be concerned with effects of the

equilibration ^between phases ^{upon the} optical pro-

perties ^{of an} interface and will include a theoretical discussion of the phenomenon ^and experimental

studies of the optical ^effects.

One experimental indication of the existence of interfacial optical effects has come about through

observations on the application of the Drude equa- tions to the study of thin films on metal surfaces.

The Drude linear equations permit ^{one to calcu-}

late ^the index of refraction, n and the thickness