HAL Id: jpa-00220910
https://hal.archives-ouvertes.fr/jpa-00220910
Submitted on 1 Jan 1981
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.
PARAMETERS CONTROLLING THE DEPOSITION OF AMORPHOUS AND MICROCRYSTALLINE
SILICON IN Si/H DISCHARGE PLASMAS
S. Vepÿek, Z. Iqbal, H. Oswald, F. Sarott, J. Wagner
To cite this version:
S. Vepÿek, Z. Iqbal, H. Oswald, F. Sarott, J. Wagner. PARAMETERS CONTROLLING THE DEPO- SITION OF AMORPHOUS AND MICROCRYSTALLINE SILICON IN Si/H DISCHARGE PLAS- MAS. Journal de Physique Colloques, 1981, 42 (C4), pp.C4-251-C4-255. �10.1051/jphyscol:1981453�.
�jpa-00220910�
JOURNAL DE PHYSIQUE
CoZZoque C4, supplgment au nOIO, Tome 42, octobre 1981 page C4-251
PARAMETERS CONTROLLING THE DEPOSITION OF AMORPHOUS AND MICROCRYSTALLINE SILICON IN S ~ / H DISCHARGE PLASNAS
S. vep?ek, Z. Iqbal, H.R. Oswald, F.A. Sarott and J.J. Wagner
Institute of Inorganic Chemistry, University of Ziirich, Winterthurerstrasse 190, 8057 Ziirich, Switzerland
Abstract: During deposition of silicon fron Si/H-low pressure discharqe plasmas the decisive parameter controlling the formation of either an amorphous or crystalline phase is the departure of the system from chemical eauilibrium, which in turn is related to discharge parameters: Kinetic data are presented illustrating semiquantitative&y these relations. A lower limit of the crystal- lite size of - 3 0 A exists for stress freepc-Si films, below which the microcrystalline phase is unstable with respect to the amorph- ous one which has a different topology.
1. Introduction. During the last year, mlcrocrystalline silicon (MC-Si) has received increased attention because of its 3ossible applications such as contact interlayers in solar cells f1,2]. The preparation of thin filns ofpc-Si and -Ge via chemical transport In a hydrogen plasma was reported in 1968 by vep?ek and ?lareEek [ 3 ] . More recentlyfic-Si has been obtained via deposition from silane diluted to a few mole% by hydrogen in a discharge operating at a higher nower level [1,2,51. We shall show that the latter method is equivalent to chemical transport since, for both, the deposition takes place close to chemical eauilibrium, whereas a-Si is ty~ically obtained under con- ditlons far away from equilibrium. It is also known that the presence of a high concentration of dopants, such as P and B can cause crystal- lization due to their catalytic effect on the formation of nuclei with the diamond lattice (e-g. [ I ] ) . The present paper deals with pure
(undoped), hydrogenated silicon. Details on sample preparation and properties of the deposited films were given in [ 3 , 6 ] .
2. Parameters controlling the chemistry of the Si/H-system. Consider a complex, heterogeneous reaction
where r(T
,. .
. ) , r ( T,. .
. ) are the rates of the forward and reverse reaction PespectivelFr T is the temperature of neutral gas, n and T~are the electron concent9ation and temperature, respectively,
%
is the total pressure and Vb is the difference between the electric potential of the silicon sample and of the plasma-
called the "bias". Here, unless stated otherwise, we restrict ourselves to the case T s T andv g - 5
to -1 0 volts (floatins ?robe).
r (. . .
and E I .. .
) are cgmpffcatedArticle published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1981453
JOURNAL DE PHYSIQUE
and essentially unknown functions of the plasma parameters. In anexp- eriment, they are controlled hyTn,and the discharge current (for details see [7] and below). Chenlcal equilibrium in a non-isothermal plasma is achieved when r =
r
( [ 7,81).
Starting f ron a non-equilibriumstate,the Chemical Equilibrium (Ch.E.) is approached at a rate charac- terized by the relaxation timer. Since the mechanism of reactionl (1) is unknown, we define 'Y as the time at which [SiHx], = [SiF ,1 ( I -e )
.
The departure of the system from Ch.E. depends on the r a t h t / T I where t is the residence time of the educts and products %?€he reactiogeEbne. With tres/y the system is close to Ch.E. Finally, a high reaction rate means a smallr an6 vice versa.
Two extreme, non-equilibrium cases of reaction (1) are the etch- ing of silicon in a H-plasma (from the left to the right hand side of eq. (1)[6a]) and the deposition of a-Si from silane (reverse process, eq. (1 ) [B]).In the preparation of M c-Si films via chemical transport, the forward reaction dominates in the charge and the reverse one in the deposition zone, but the system is close to Ch.E. in the latter case [3,6,7].
The rate of the forward reaction (1) increases with increasing discharge current, i, ("power density") [6a]
.
Eith increasing temper- ature T (constant i),it increases to a maximum at T N 40-80°C (T of the maxlmum depends on i) and nabove^
100°C it decregses again [6a?.At p = 0.13mbar, i = 360mA and discharge tube radius R = 2.5cmr the etcp q t e at the maximum amounts to 16A0/s, i.e.
-
8x10'~ Si-atomscm s
.
Using a Wrede-Harteck gauge [9], we estimated a degree of dissociation o f 5 3 % , and from the Langmuir characteristic of the Si- sample used as a probe we measured an ion current density of O.lmA/cm.Thus, the fluxl,jlens~t&ys face areN1x10 cm s a n d ~ 5 x l O cm s of H-ato
T3 "3
DL?
sitive ions towards the sur-,
respectively. Conse- quently, the forward reaction (1) is due to H-atoms and the most prob- able mechanism is the step-wise additicn of H-atoms, i-e.:This reaction competes with the hydrogen recombination
which can take place in the gas phase as well as at the Si-surface.
With increasing temperature above*lOO°C reaction (3) leads to an overall decrease of the coverage of the Si-surface with chemisorbed hydrogen resulting in the (measured) decrease of the rate of reaction
2 Limited data are available on the role of electronic and vibra- tional processes in the dissociation of SiH species, but the rela- tively small change of the deposition rate $r = r-r) with a nega- tive bias up to -900V (Fig. 1 ) suggests that rgsgtion (3) dominates.
Fiq. 1: Dependence of the depo- sition rate, r on the bias,
of the subflppate. (p=lmbar
,
rb;26mRr T =260°C, P=2_5cn).
40 Here, V'
=vdep - ,
yV is the*'wall p8te~&?~rf~khich can be ac- curately obtained from the Lang-
2 0
muir characteristic. V.., is also "
called the "floating potential"
and is 5 to 10V neqative with +lo0 0
-
300-
600-
900vb respect to Vplasma.
Under electron bombardment (positive bias V', region of the expo- nential rise of Langmuir characteristic) the dep$sition rate strongly decreases (Fig. 1). This indicates the increasing role of theelectron stimulated chemisorption of hydrogen [lo] followed by the formation of volatile silane species. Thus, electron bombardment accelerates the forward reaction (l), i.e. the plasma etching.
2.1 Deposition of k(c-Si by chemical transport. The results givenabove show that, for V 5 -5 to -lOV the chemical transport reaction (1) is basically contro!?led by the temperature, T and discharge current. i t
(a less appropriate parameter is "power density" [Ill). A higher Tn and lower i favours deposition, i.e. the transport direction is TI4T2
(TI < T2
,
i definition [7] ) :and i +i
Si (s) + x/mF (4)
Using mass spectrometry and a col- lisionless sampling technique, we have measured the relaxation time of reac- tion (4) (starting from the L.H.S.) as
1 .5 a function of T n and i [I21
.
Figure 2shows, as an example, the dependence of the relaxation time on discharge cur-
1 0 rent density in a range typical for the
%' ( s ) deposition (cf. similarity principles
for glow discharges). Further measure-
0 . 5 ments have shown that, at a constant i,
Tincreases with increasing temperature i. e.
0 o l o 30 so
'X
(T2)Ii
> 'X (TI )Ii and1/nu2 ( m ~ i c r n ~ )
W L ~ )
iT
> %(i2)Lnn
Fig. 2. Dependence of the re- Chemical transport typically oper- laxation time of reaction (4) ates in the r a n g e r 3 0.2
-
0.5s. Thus, on discharge current density in a discharge tube of 2R=5cm and(p=0.33 mbar, T &200°C, R=0.8 length of the charge zone of 3lOcm, cm)
.
C.T.R. deffotes the typ- chemical equilibrium is nearly estab- ical range for chemical trans- li3hgq if the flow rate is 6 5 0 nbar port. cm s (t es 35s-7).
Under these con-ditions tffe equilibrium concentration of SiH-- (9) species is about 1 % [ 1 21
.
2.2 Comparison with the deposition o f s c - ~ i from SiH,/H,
-
mixtures.The typical conditions are 63%SiH4/S,97%H2 at the gas'in et and a power density of about 0.1-O.2w/cm [I
,
21. These conditions :onpare with that for the chemical equilibrium given in Sect. 2.1 (In the above case, the power density equals the product of the current density and the axial field strength ofwl0-15V/cm, but the discharge current den- sity is a more appropriate discharge parameter for comparison 1111.) From the limited data on the apparatus given in [lb] we estimate t t3 2.5s and from the R.F. power density and Fig. 2 we guess 7 SO .2-6?2s i.e. t
7 .
Thus, the deposition of ~ c - S i from silane diluted with H in gES"high power R.F. discharge" takes place under conditions cfose to chemical equilibrium in reaction (I).2.3 Comparison with deposition of a-Si from silane. Typically, one uses pure silane and "low E.F. power". Assuming the same discharge volume as in Sect. 2.2, we estimate for the ;low power" deposition of a-Si [Ib] an R.F. power density of-O.OlW/cm and tre 23 0.1s. At this power density, the relaxation time, estimated f rom2ex?rapolation of the data in Fig. 2 to a current density of 6lmA/cm
,
is r r I s . Thus,C4-254 JOURNAL DE PHYSIQUE
tr
2 ~ 7 .
Very limited experimental data of other authors suggest that a-51 is obtained whenever the condition t c 7 is fulfilled. The "low power" used in such cases further means tff% the forward reaction (I), i.e. the plasma etching, is negligible, and also the total rate of the heterogeneous recombination of B atoms, eq,(3), is slow because of the low H-
concentration. Thus, a-Si is obtained if the deposition takes place far away from chemical equilibrium where only a relatively small power is dissipated into the surface of the growing film.3. The surface processes controlling the formation of either crystal- line or amorphous deposits. Ostwald's well known "law of stages"
reflects the fact that the formation of nuclei of the stable. crvstal- . ..
line lattice requires a higher activation energy than the formation of the nuclei of the metastable, amorphous phase [131. The reason is that the most stable structure of small clusters, which is basically gov- erned by the minimization of the surface energy, is generally incom- patible with the structure of the stable crystal lattice [14,16]. It is further known that the radial distribution function of a-Si cannot be approximated by a microcrystalline model; rather, a certain distrib- ution of the dihedral angle which is incompatible with the diamond lat- tice has to be assumed 1171. These aspects will be discussed in more detail elsewhere [18]. Here we shall briefly summarize the most impor- tant results.
The obvious conditions for the deposition ofpc-Si are suffic- iently high rates of the forward reaction (1) and of the heterogeneous recombination (3). The forward etch reaction facilitates conditions.
where the less stable, small nuclei are preferentially etched and large, more stable ones are growing at their expense ("mineralization effect"
-
see, e.g. [8,11,19]). The heterogeneous recombination (3) represents a process by which a large amount of energy is being pumped into the vibrational degrees of freedom of the nuclei. For example, under conditions given in Sect. 2 an$ a deposition rate o f w l ~ / s , more than*do3 recombination events take place on the surtace for every de- posited Si-atom. Evidently, the power dissipated is orders of magni- tude higher than that due to the ion recombination (cf. the corres- ponding flux densities, Sect. 2) and, unlike the latter, the power due to the recombination of atoms is dissipated directly into the vibra- tional degrees of freedom whereas most of the recombination energy of ions due to an Auger process goes into the electronic states of the solid and into the kinetic energy of the ejected electron. The energy quantum dissipated into the solid by one H+H = H2 recombination event equals s R - E (diss. ),
wliich amounts to ~l eV (where E (diss. ) is the dis- sociation energy of H and the energy accomodation coefficient [29]).Consequently the exci$ation of the phonons of the nuclei is a highly non-equilibrium process (4. Ediss>> kg, 8 is the Debye temperature) and the equilibration takes place relatively slowly via anharmonic phonon- phonon interactions.
The excellent control of the deposition achieved by the chemical transport allowed us to study the stability of AC-Si as a function of the crystallite size 1211. Figure 3 shows that in stress free, hydro- g e n a t e d ~ ~ - s i films, which are essentially free of any amorphous com- ponent [6d], the lattice constant, d l itcreases with decreasing crys- tallite size, a. In the limit of a = 3 0 A thehc-Si becones unstable and discontinuously transforms into the amorphous phase. The critical lattice dilatation (which is related via a to the surface energy of the small crystallites) is given by eq. ( 5 )
ad 2
ZV
(mole kB (-1 do crit. = d ~ ( a ) - T & S ( ~ )where V(mo1e) and B are the molar volume and bulk modulus of c-Si, and
Fig. 3. Lat-tice dilatation, ~ d / d ,
,
versus the crystallite size, a , for stress free, MC-Si films.
$ E (a) and 6 S (a) are the excess en-
References
[la] tlatsuda A., Yamasaki S., Nakaqawa K., Okushi H . , Tanaka K., Izima S.
,
Ilatsumura 11. and Yamamoto H., Jap. J. Appl. Phys.E(
1980)L305.[lbl Matsuda A., Yamasaki S., Okushi H., Izima S., Tanaka K., Yanamoto H.
,
Jlatsumura !I., ibid (1 981 ) (in press).
121 Spear W.E., Willeke G., Le Comber P.G. and Fitzgerald A.G., Pro-
c e e d i n g s of t h i s C o n f e r e n c e
[3] vep;ek S. and ~ a r e E e k V., Solid State Electronics
fi
(1968) 683.[4] Chittick P.C., Alexander J.H. and Sterling H.F., J. Electrochem.
Soc.
116
(1969) 77.[5] Usui S. and Kikuchi M., ,T. Non--Crfst. Solids
2
(1979) 1.[6a] Kebb A.P. and ~ e p g e k S., Chern. Phys. Letters
62
(1979) 173.[6b] Iqbal Z., TJebb A.P. and vep?ek S., Apgl. Phys. Lett.
36
(1980) 163.[6cl ~ e p g e k S., Iqbal Z . , Oswald H . R . and Webb A.P., J. Phys. C
2
(1981) 295.
[6d] Iqbal 2 .
,
~e$ek S.,
Vebb A.P. and Capezzuto E, Solid State Commun.37 (1981) 993.
[71 E p ? e k S., Chimia
Ifl
(1589) 989.[a]
vep?ek S., in: Topics in Current Chemistry Vol. 56 (Springer-lrer- lag, Berlin 1975).[91 Schotzau H.J. and vep;ek S., Appl. Phys. 7 (1975) 271.
[lo] Winters H., in: Plasma Chenistr", Vol.111, eds. vep?ek S. and Ven:gopalan (Springer-Verlag, Berlin 1980).
[Ill Veprek S., Pure Appl. Chen.
48
(1976) 163.[I21 ITagner J.J. and \?eP;ek S. (to be nublished).
[I31 Hirth J.P. and Pound G . ? ! . , Condensation and Evaporation (Perqamon, Oxford 1963).
[I41 Allpress J.G. and Sanders J.V., Aust. J. Phys. 23 (1970) 23.
[I51 Hoare Y.B. and Pal P., Adv. in Phys.
24
(1975) 3 5 .[I61 Takasu Y. and Bradshaw A.?I., in: Chem. Phys. of Solids and their Surfaces, Vol. 7 (Chem. Soc., London 1978).
[I71 Grigorovici I?., in: Amorphous and Liquid Semiconductors, ed. J.
Tauc (Plenum Press, London 1974).
[I 81 vep;ek S., Iqbal Z. and Sarott F. -A., Phil. Elag. .- in yess
[ 19 ] Schafer H., Chemical Transport P.eactions, (File?), New York 1964).
[20] bleltn G.A. and Ptadix F..J., Trans. Farad. Soc.
67
(1971) 198; 2711.[21] Veprek S., Iqbal Z . , Oswald H.P.., Sarott F.-A., Nagner J.J. and Webb A.P., Solid St. Commun. (1981) (in press).
=
do ( X I 1.2 0.80.4
o h
ergy and entropy of a-Si [I81
.
a - s i u c - S i
-
Equation (5) explains why there isa lower limit for the crystallite size of fic-Si. It is also under-
-
standable that a large activation.8
energy is necessary in order to-
transform an "amorphonic" nucleus contai2ing more than 1000 Si atoms(a S30A) into the smallest, stable
I crystalline nucleus. This energy is
o 40 8 o 120 provided by the heterogeneous
c r y s t a ~ l i t e ~ ~ z
(i)
e recombination discussed above (see[I81 for further details).
