The influence of the chlorine-hydrogen ratio in the gas phase
on the stability of the {113} faces of silicon in Si-H-Cl CVD
Citation for published version (APA):
Gardeniers, J. G. E., Mooren, M. M. W., Croon, de, M. H. J. M., & Giling, L. J. (1990). The influence of the chlorine-hydrogen ratio in the gas phase on the stability of the {113} faces of silicon in Si-H-Cl CVD. Journal of Crystal Growth, 102(1-2), 233-244. https://doi.org/10.1016/0022-0248(90)90906-2
DOI:
10.1016/0022-0248(90)90906-2
Document status and date: Published: 01/01/1990
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
Journal of Crystal Growth 102 (1990) 233—244 233 North-Holland
THE INFLUENCE OF THE CHLORINE-HYDROGEN RATIO IN THE GAS PHASE ON THE STABILITY OF THE (113} FACES OF SILICON IN Si-H-Cl CVD
J.G.E. GARDENIERS, M.M.W. MOOREN, M.H.J.M DE CROON and L.J. GILING
Department of Experimental Solid State Physics III, Research Institute for Materials, University of Nijmegen, Toernooiveld,
6525ED N~/megen,The Netherlands
Received 3 July 1989: manuscript received in final form 23 January 1990
The orientation dependence of silicon crystal growth in the Si—H—Cl CVD system has been studied as a function of the chlorine—hydrogen ratio of the gas phase. This was done by the use of hemispherical single crystal substrates. As was reported before, the stability of faces with the indices (hhk}h<k is dependent on temperature: above a certain critical temperature flat (113) and 337) faces are found on the hemispheres, while below this temperature only macroscopic steps appear in positions corresponding to these faces. In this study it is found that the above-mentioned critical temperature is strongly dependent on the chlorine—hydrogen ratio in the gas phase. It will be demonstrated that this “chemical roughening” effect is caused by the competitive adsorption of chlorine and hydrogen. From the experimental dependencies values for the heat of chemisorption of chlorine and hydrogen can be derived of —370±70kJ mol~and —316±5kJ moL’, respectively. These values are in good agreement with literature values of
Si—Cl and Si—H bond strengths.
I. Introduction it from the well-known “thermal roughening”
temperature (see, e.g., ref. [3]) of a crystal face.
In a previous paper [1] we reported on the In ref. [2] it was calculated that the adsorption
orientation dependence of the crystal growth of of hydrogen might explain the observed trend as a
silicon as a function of temperature in the Si—H—Cl function of temperature, however, the absolute
system. Among other things it was found that the temperaturebelow which the (113) faces no longer
stability of faces with indices (hhk}h <,< (h~ 0) grow flat, was calculated to be approximately 400
increases at higher temperatures, an effect which K below the observed value. In order to study the was explained by the dependence of the surface adsorption effects in more detail, in this paper the
tension on the adsorption of hydrogen, as demon- influence of variations in the chlorine—hydrogen
strated by a plot of surface tension versus the ratio of the gas phase on the stability of the
angle ~ in the [110] zone [2]. From this plot it was (hhk}h<k faces will be investigated.
derived that the step free energy on the (113) As was demonstrated in ref. [2], the effect of
faces decreases with increasing hydrogen coverage, fundamental gas phase parameters on the stability A certain critical coverage exists, at which this of crystal faces in the chemical vapour deposition
step free energy becomes zero, which means that (CVD) of silicon can be studied very nicely by the
at and above this critical coverage the (113) faces use of hemispherical substrates (see also ref. [4]) in
no longer have the tendency to grow flat. In view an experimental system where crystal growth is
of the fact that the surface coverage increases at carried out at near-equilibrium conditions, as was
lower temperatures, it could thus be concluded described by Bloem et al. [5].It was shown before
that the (113) faces will not grow flat below a that equilibrium calculations can be used to
de-certain critical temperature. In the following we scribe the processes that take place in this system
shall call this temperature the “chemical roughen- [5—9].The gas phase composition which follows
ing” temperature of the (113) faces, to distinguish from these equilibrium calculations will be the
234 J. G.E. Gardeniers et at./ Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CVD
basis of our discussion of the adsorption effects. It tion of all growth experiments was 2 h. The radius
will be shown that the above-mentioned tempera- of the hemispheres was 3.00 ±0.05 mm.
ture effect on the stability of (113) most probably It was found that the growth rate on the planar
is caused by the competitive adsorption of chlo- (001) substrates (2°-off in on of the (110)
direc-rine and hydrogen. tions), which were used to obtain near-equilibrium
conditions at the position of the hemispheres [I], was in good correspondence with the “solubility”
2 Experimental curves [5] taken from equilibrium calculations,
which confirmed that these calculations had given a good description of the near-equilibrium crystal
The experimental procedure was essentially the growth process
same as described before [1]. In this study we will concentrate on the changes in crystal habit as a
consequence of variations in the chlorine—hydro- 3 Results
gen ratio of the gas phase, together with variations
in growth temperature. As a definition of the At a constant (Cl/H) ratio the morphology of
chlorine—hydrogen ratio (Cl/H) of the gas phase the hemispheres at angles from approximately 20°
we will use. to 35° from the (Ill) face in the (hhk}h<k part
of the [110] zone was different for different
tem-~ z,p(Sitem-~Htem-~,CL) peratures. This is shown in figs. la—Ic. Three
(Cl/H) = . (1) different morphologies can be distinguished. At
~ p (Si H Cl low temperatures morphology I is observed: only
i=1 macroscopic steps appear (fig. Ia). These steps
reach from the (111) faces to the nearest {001}
In this expression m stands for the total number faces. No (113) or (337) faces can be
dis-of gas phase species present in the system, and tinguished. The positions where they should occur
p(Si~.H~.CL)for the partial pressure of the gase- are indicated in the figure.
ous species Si~H1,Cl,obtained from equilibrium On the other hand, at high temperatures in
calculations [5,101. these regions at approximately 23.5° and 29.5°
In the experiments the chlorine hydrogen ratio from the (111) faces the flat faces (337) and
of the gas phase was varied from 0.005 to 0.12 by (113) appear, respectively. This will be called
a change in the relative amounts of mass flow of morphology III (fig. Ic). When this morphology
the input gases SiH2C12, HCI and H2. The super- occurs, in the part of the (hhk}h<k region
be-saturation (see ref. [1] for a definition) in these tween the (113} faces and the nearest (001) face
experiments was always 0.1 or lower, except for macroscopic steps either are only slightly visible
the experiments at a (Cl/H) ratio of 0.005, where or do not appear at all.
in order to obtain mean thicknesses of several Morphology II is an intermediate case between
microns on the hemispheres, it was necessary to morphologies I and III (see the discussion section).
use supersaturations of up to 1.0; it was noticed It is observed either when the amount of silicon
that in order to have decisive information on the deposited on the hemisphere is relatively low (e.g. stability of faces, epitaxial layers of at least several for very short growth experiments or very low microns are required. The thicknesses of the grown supersaturations), or at conditions where the (337)
layers, averaged over the surface of the hemi- and (113) faces obviously are not very stable. In
spheres, normally varied from approximately 3 fig. lb these faces can be seen as very small bands.
~em to approximately 50 ~sm. Their position, which can be derived from the
The total mass flow of the gases was always 400 narrowing of the range of macroscopic steps, is
SCCM, leading to a gas velocity of approximately indicated by arrows (the two thin white lines are
J.G.E. Gardeniers et aL/Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CVD 235
In a following series of experiments the stabil- takes place at a somewhat higher temperature (10 ity of these two faces has been studied as a func- K at most) for the (337) faces than for the (113)
lion of the (Cl/H) ratio between 0.005 and 0.15 faces.
and as a function of temperature between 1190 In fig. 2 the experimental data of Van den
and 1480 K. The temperature regions where mor- Brekel and Nishizawa are indicated. These authors
phologies I, II and III are present turn out to be have investigated the orientation dependence of
dependent on the (Cl/H) ratio, as is shown in fig. silicon CVD, using cylindrical [11] or hemispheri-2. The dotted line is used to indicate the transition cal [12] substrates. They report the presence of flat
from morphology I to morphology II. A more (113) and (112) faces. Except for the indices of
detailed analysis reveals that for experiments with the latter faces (in our opinion these are not
conditions close to the dotted line in fig. 2, the correct, the faces should be indexed (337), see ref.
transition from morphology I to morphology II [1]), their findings fit well with our observations.
80 from
(
0 0 1 )-face 45° 100 j~m 40° 35° (113) 30° — 25°— ‘-~~(337) 20° 15° 10°a
edge of (ill )-faceFig. 1. Morphology of the parts of the hemisphere which can be described with the crystallographic indices {hhk)5<5. (a) Morphology I: macroscopic steps, at low temperatures; (b) morphology II: very small (113) and (337) faces, at intermediate temperatures; (c) morphology III: large (113) and (337) faces, at high temperatures. Note: in (a) and (b), the surface is curved in the
236 J.G.E. Gardcnicr~~‘rat Influence of Cl—H rotor on it~biluio/ (11$,’ /a~ci ‘I Si in5i /1—CI C VD 1
2 5i.äiij
~ —~ _~
c~1~
~
~ ..‘i ..~ ~c
1:1g.1.(continued).log(Cl/H) Unfortunately, as indicated in fig. 2, they have not
-1.0 examined growth temperatures below 1400 K, and
for that reason have not observed the same
tern-0 ~DQ’v .SS~--.~ perature effects as we have. i~I
/ ~Vi
—1 5 4. Discussion
/ V
/ 4.1. General aspects of growth forms
-2.0 0 .v•w •
Very generally it can be said that after crystal
growth on hemispheres three different categories
0/ V w• of morphological features may show up:
-2.5 ./...,..,...,.., (I) flat faces, i.e. surfaces which are not curved in
1150.0 1250.0 1350.0 1450.0 any direction (see fig. Ic. morphology III);
Temperature ~n K (2) ranges of macroscopic steps. i.e. surfaces which
Fig. 2. Existence regions of morphologies I. II and Ill as a are curved in one direction (see fig. Ia.
mor-function of chlorine-hydrogen ratio and temperature. Open phology I)’
circles: morphology I: closed circles: morphology III; trian- ‘ . .
(3) surfaces which are curved in all directions.
gles. morphology 11, open square. experimental conditions of .
Van den Brekel [111 and hatched rectangle: experimental con- A way to interpret these morphologies is to
J.G.E. Gardeniers et al. /Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CUD 237
- -- ~. A very low surface tension
[13].
In our case the~
crystallographic direction of the terrace edges is
- - A [110], which is the direction of the strongest
pen-- - -- ‘~-~—— — — odic bond chain (PBC) in the crystal bulk
struc---- -‘ ture of silicon [14].
- Similar reasoning as above can be used to
inter-- - - pret the other morphologies in fig. 1. Morphology
Fig. 3. Illustration of the definition of angles in the Wulff plot. III is thus the result of cusped minima in both
A indicates the reference plane, A’ and A” are planes tilted io
y(cr) and y(/3) for both the (113) and (337)
orien-angles a and$ with the reference plane, respectively. . . .
tations. In our opinion this is also the case in morphology II, however, for one of the two func-function of the crystallographic orientation. This tions the minima are only very shallow, leading to
way of interpretation is only allowed when the only very small flat regions with orientation (113)
morphological features to be discussed are equi- and (337). Thus morphology II can be considered
libriumstructures. This can of course never be the an intermediate case between I and III, where in
case when one is considering a crystal growth the sequence Ill—Il—I the minima in one of the
situation. Nevertheless, as we are working at mod- functions y(a) or y(/3) disappear.
crate supersaturations (“near-equilibrium” growth, In a previous paper [2] we have given a possible
see ref. [5]), we will assume that the lowest growth explanation for the disappearance of cusped
rates will occur for those crystallographic orienta- minima in the y-plot. In this discussion we will tions which correspond to cusped minima in the use the theoretical results of that paper to explain
polar plot of y (the “Wulff plot”). With this the effect of the (Cl/H) ratio on the stability of
assumption above categories of growth morpholo- the (hhk
}
faces. To do this we will firstinvesti-gies can be explained in terms of surface tension. gate the changes that occur in the gas phase
equi-Suppose we use the angles i and
/~,
which are librium composition as a result of a change in thedefined in two perpendicular planes, to describe (Cl/H) ratio.
the deviation of the crystallographic orientation of
some part of the hemisphere from a predefined 4.2. Gas phase equilibrium composition as afunction
reference orientation (see fig. 3). Then it can be of the chlorine—hydrogen ratio said that if in the Wulff plot both y(a) and y(/3)
have a cusped minimum for a certain crystallo- In fig. 4 the result of a gas phase equilibrium
graphic orientation, this orientation will appear as calculation [5—10,15]is presented. In the calcula-a flcalcula-at fcalcula-ace on the hemisphere calcula-after growth. On the tion the following gaseous species are included: Si,
other hand, if neither of the functions y(s) and Si7, Si3, SiH, SiH2, SiH3, SiH4, Si2H6, SiC1,
y(J.~)has a cusped minimum for a certain range of SiC17, SiCl3, SiCl4, SiHCI, SiHC13, SiH2C12, orientations, this will manifest itself as a region on SiH3C1, H, H2, HC1, Cl, and Cl2. The gas phase is the hemisphere which is continuously curved in all in equilibrium with solid silicon. Thermochemical
directions. If only in one of the two functions data were taken from refs. [16—19].The figure
y(t) and y(/3) a cusped minimum is present, this shows the partial pressures of the main gas phase
will lead to a region on the hemisphere, which is species in the Si—H—Cl system as a function of
curved in only one direction. (Cl/H), at atmospheric pressure and 1350 K.
We consider morphology I in fig. la to be an As can be seen in this figure, the main
conse-example of the latter morphology. The formation quence of an increase in (Cl/H) is that
chlorine-of the macroscopic steps which can be seen in this containing species become more important with
figure most probably is caused by the tendency of respect to species which do not contain chlorine.
the growing crystal to maintain an as low as This also implies that the total amount of silicon
possible surface tension, which can be achieved by in the gas phase (the “solubility”, see e.g. ref. [5]) the formation of large terraces of orientations with increases with an increase in (Cl/H).
238 J.G.E. Gardeniers et al. /Influence of Cl—H ratio on stability of (113)faces of Si in S/—H—CICVD
log(p/atrn) angle ~ in the [110] zone: below a critical
cover-o age ~ in this plot an inward pointing cusp exists
- at the angle ~1i3 corresponding to the (113) face.
above this O~an outward pointing cusp at ~l13
SiHC13 appears. Or, in other words: at low coverages a
positive step free energy (see, e.g., ref. [3]) exists
HCl~— SiCI S1H9C12 on the (113) face, while at high coverages the step free energy becomes negative [21. It is well known
SiHCI
that the surface coverage will increase at lower
SiH3CI temperatures, so that the above is consistent with
H _________ the observation that due to the high coverage, the
-6 SIH4 (113} faces are absent from the (near-)equilihrium
_____ S[1-12 (growth) from the silicon crystal below a certain
critical temperatureT~,the “chemical roughening”
8
/
Cl SiCI temperature of the (113) faces.- / ~ ~S(H. In ref. [2] the above described phenomenon was
5111 explained by the adsorption of hydrogen.
How-Si ever, with the use of reasonable values of the heat
10 ‘SjCl and entropy of hydrogen adsorption, the chemical
—~//sici3 .V~l2 roughening temperature of the (113
}
faces was5 4 3 calculated to be 920 K. This is much lower than
log((’l/II) the experimentally observed value of 1340 K at
Fig. 4. Partial pressures of Si—H—Cl gas phase species in the (Cl/H) ratio of 0.06, which was used in ref.
equilibrium wiih solid silicon, as a function of the (Cl/H) ratio [1]. Although the thermochemical data for H
ad-at a temperad-ature of 1350 K and a total pressure of 1 ad-atm. sorption are not known very accurately, this dis-crepancy led us to the opinion that in addition to H other species in the gas phase might be
respon-The gas phase compositions extracted from sible for the destabilizing effect on (113}. This
equilibrium calculations, as e.g. presented in fig. 4, idea can now be supported by the experimental
will be used in the following discussion on the data in fig. 2: it can be seen that the chemical
influence of adsorption on the stability of the roughening temperature depends on the (Cl/H)
(hhk)/,<k faces. In ref. [21 it was derived that ratio. From fig. 4 it is also clear that in the range
when the crystal surface is free from adsorbates, of experimental (Cl/H) values the partial pressure flat (113) faces are expected, because in the Wuiff of atomic hydrogen does not depend on the (Cl/I-I)
plot cusped minima are present for these orienta- ratio, so the observed changes in morphology as
tions. Unfortunately, with the aid of the structural discussed in this paper for the
{
hhk},~
<~ facesmodels in ref. [2], no cusped minima could be can not be caused by the adsorption of hydrogen
found for the (337} faces, so we are not able to alone. It is therefore plausible to assume that a
discuss the stability of the latter faces quantita- chlorine-containing adsorbate is involved in the tively, but the rules derived for the (113} faces shift of 7T~~of these faces.
qualitatively also apply for (337). To examine this in more detail the following
strategy will be pursued: first it will be assumed
4.3. The influence of adsorption on the occurrence of that only chlorine atoms are present on the surface
the (hhk}h <k faces of the silicon crystal. It will be investigated whether
this assumption can explain the observed effects
In ref [21 the influence of temperature on the and whether the adsorption parameters that can
appearance of the (113) faces was explained by be derived from the experimental dependencies
J.G.E. Gardeniers et al. /Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CUD 239
thermochemical data for silicon-chlorides and is used to indicate the orientation dependent
estimated adsorption data. Next the case will be parameters in eq. (5).
considered where both chlorine and hydrogen The surface tension of a crystal face which is
atoms are chemisorbed on the silicon crystal only slightly misoriented from a flat face at angle
surfaces, and it will be examined whether or not 4)~can be calculated with the aid of: this gives a satisfactory description of the data in
fig. 2. Finally the importance of the adsorption of ~
[4)1
= ‘[
4)oI
+ (ysiep/d)I
4)—4)oI’
(6)silicon-containing species will be investigated.
where Ystep is the free energy of a step on the face at 4)~and d the height of this step.
4.3.1. Chlorine adsorption Eq. (6) can be rewritten in the following form:
Suppose that chlorine atoms are the only
ad-sorbates present on the silicon surface. An adsorp- ‘Ystep — y[4)] — ~[4)o] (7)
tion equilibrium to describe this case would be:
‘~T”
I 4)
—I
Clgas +surface site Cladsorbed (2) For 4) larger than but very close to ~ the term on
the right of the equality sign can be replaced by
with equilibrium constant K~1,which is defined in dy/d4).
the conventional way. For Langmuir adsorption Application of the foregoing to eq. (5) leads to:
the chlorine coverage of the silicon surface, 0c~is
given by the equation: Ysiep[4)l —
Ysiep.*[4)l
— dq[4)] kT ln(1/O~), (8)d — d d4)
0ci= Kctpçi~q/(1+ Kcipcieq), (3)
In words: at a certain coverage of adsorbates the where 6~is the fraction of surface sites covered step free energy on a surface at an angle 4) is equal by chlorine atoms, and PCi,eq the equilibrium par- to the step free energy on this surface, when no
tial pressure of chlorine atoms, as e.g. given in fig. adsorption takes place, minus a term proportional 4. When the equilibrium constant K~1in eq. (3) is to the variation of the adsorption site density with
written in a form that contains the thermody- orientation and proportional to the parameter T
namic parameters /.~HC°iand ~S~°1,which are the ln(1/9~).
change in standard enthalpy and entropy for equi- O~will always be positive and less than one,
librium reaction (2), respectively, the following which leads to a positive term T
1n(1/8~).Ad-equation is obtained: sorption will therefore either lower or raise the
/ \ ~‘~° ~ step free energy on a particular face, dependent on
(4) whether the term dq[4)]/d4) is positive or
nega-InI Pci,eq(1 —)) = — R1~~+ R tive, respectively. In ref. [2] it was shown that
dq[4)]/d4) is positive for steps on the (113) face
where R is the gas constant, which are parallel to the [110] direction, which is
It is possible to use the data in fig. 2 to fit the the direction of the most stable PBC in the silicon thermodynamical parameters in eq. (4) in the fol- crystal, and inclined in, e.g., the [3321 or in the lowing way: in ref. [2] an equation was derived to [332] direction. Therefore for these steps the step
describe the effect of adsorption on surface ten- free energy will decrease with an increase in T
sion y[4)]: ln(1/9~).Most probably the same will hold for
y[4] = ~~[4)]—kTq[4)] ln(1/9*), (5) steps on the (337) face which are parallel to the
[110] PBC direction. For steps on the (113) face where y,1,[4)] is the surface tension of the ad- inclined in the [110] or in the [110] direction and sorbate-free surface, q[4)] the adsorption site den- parallel to e.g. the [332] direction (i.e. steps per-sity, O,~,the fraction of free sites and 4) the angle of pendicular to the stable PBC directions) the term the surface with a predefined reference plane; [4)] dq[4)]/d4) is also positive, so for these steps also
240 1. G.E. Gardeniers et al./ Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CUD
the step free energy will decrease with increasing I ~
in
---—T ln(I/O~).
pc
1 1-~~ciIn ref. [2] it was derived that if for steps on the 22
(113) face which are parallel to the [110] direction 21
the term T ln(1/8~)becomes larger than 2250 K,
/
the step free energy on this face will no longer be 21)
positive, and these faces will no longer grow flat. 19
i.e. the faces have become “chemically roughened”. Although in ref. [2] this was only discussed for the
case of hydrogen adsorption, the above require- t7
ment for the term T ln(1/~~)on the (113) faces
applies for every species which adsorbs in adsorp- 16
lion sites similar to those for H. ___________________________________
If the transition from morphology II to mor- 0.70 0.75 0.80 0.85
phology I in fig. 2 is interpreted as a chemical l000/T in K~
roughening phenomenon, it thus follows that at Fig. 5. Fitof the data in fig. 2 according io eq. (4). The closed
circles correspond to transition temperatures of 1210. 1250.
the dotted line in this figure the following relation 1310. 1340 and 1350 K. for (Cl/H) values of 0.005, 001, 0.02,
should hold: 0.06 and 0.12. respectively (see text for the meaning of the
solid and dashed lines).
~
]n(l/O~)= 2250 K. (9)Considering that in the case of Cl adsorption
= 1— O~,we can thus calculate O~as a func- = 380 ± 20 kJ molt,
tion of T from the data in fig. 2. To obtain the a value which is valid for temperatures from 1210
parameters ~ and L~S~?J.we have to plot the to 1350 K. According to equilibrium reaction (2).
term on the left-hand side of the quality sign in ~H~8 should be approximately equal to the
eq. (4) versus 7~i’.The slope of the resulting strength of a Si—Cl bond on the silicon surface. In
straight line will then correspond to the value of table 2, literature values of Si—Cl bond
dissocia-—~H~//R and the intercept of the line with the tion energies are given for several silicon-chloride
vertical axis to ~S~1/R. Such a plot is shown in compounds. With statistical thermodynamical
fig. 5. where values of Pcl.eq from equilibrium
calculations as, e.g.. shown in fig. 4 are used. Table I
Two different linear least-squares fits of the Entropy changes during adsorption of chloride atoms
points in fig. 5 were made: Entropy Entropy changes Entropy changes
(i) the dashed line in the figure is a fit through the contribution at 298 K at 1300 K
five points without further assumptions. This leads (J K ‘ mol~‘) (J K’ mol ‘)
to the values: Loss Gain Loss Gain
= —440 ±20 kJ moL~, Translation 153 0 184 0
Rotation 0 0 0 0
= — 180 ±30 J K— mol Vibration ~ 0 20—35 0 54—71
Electronic‘~ Il 0 Il 0
(ii) With the use of statistical-thermodynamical
arguments (see, e.g.. ref. [15]) it is possible to Neit 137 ±S (33±9
estimate the entropy change for the adsorption of .c Estimated vibrations (see rcfs [15,20—23]):a stretch mode
chlorine atoms. This is shown in table I. If the with a frequency of 300—550 cm ‘. and two bend modes
calculated value of ~S~?1at 1300 K. —133 ±9 J with frequencies of 100—200 cm 1(550cm ‘is the value of
K mol - divided by R,is used as a fixed point the Si—Cl stretch frequency in the SiH~Clmolecule accord-ing to ref. [24]).
in a linear least-squares fit of fig. 5 (this value Si The adsorbate-free silicon surface site and the adsorbed
should occur for 77/ = 0). the solid line results, chloride atom are assumed to he electronically doubly
J.G.E. Gardeniers et aL/Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CUD 241 Table 2 librium reaction (2), also the following adsorption
Silicon—chloride bond strengths at 298 K equilibrium has to be considered:
Bond Bond strength References
(kJ mol~) Hgas + surface site Hadsorhed 10
CI3Si—Cl 448+8 with equilibrium constant KH. The coverage of
Cl Si—Cl2 276289±8 [251[26] the silicon surface with species X (i.e. H or Cl) isnow given by the equation:
ClSi—Cl 477 [25,27]
SiCI 381 [25,27] K~PX.eq 11
(CH3)35i—Cl 473 [25] °X= 1+ KH PH eq + K~1Pci.eq’
H3Si—Cl 527 [281
H2CISi—C1 Again at the critical temperatures T In(1/O~)will
(CH1)H2Si_Cl 479 [29] be equal to 2250 K (see the previous section);
however, in this case O,~,= 1 — — 1— °toia’
From this relation and equation (11) it then fol-lows that:
methods it is possible to calculate that L1H~°1will K ~ — °toiai — K (12)
only be approximately 5 kJ mol1 higher for the Cl (i.e9 — I — 0ioia HPH .eq’
temperatures examined in our experiments.
There-fore it can be concluded that both values of This equation can be rewritten in a form that
as deduced from fig. 5 are in agreement resembles equation (4):
with the literature values of Si—Cl bond strengths
iota)
as given in table 2. ln ~ — — KHPHeq — In
It can be calculated that for steps on the (113) blat
face parallel to the [332] direction (i.e. perpendicu- ~H°
~
lar to the PBCs), the step free energy will become = — ~ + —~—. (13)
negative when the term T ln(1/9~)exceeds 31500
K. This value is much larger than the above-men- To study whether the assumption of competitive
tioned value for steps parallel to the [110] direc- adsorption of chlorine and hydrogen adsorption
lion, which is understandable because in order to can explain the experimental results, we evaluated
form these steps, the stable PBCs should be broken this equation in the following way: PH.eq was
up, which requires high amounts of energy. The obtained from equilibrium calculations (see
high value of the term T ln(1/O~)implies that on above), and several different values of the
adsorp-the (113) faces, most probably for all adsorp-the growth tion heat of hydrogen, ~ were considered; in
conditions used in this study, the step free energy this evaluation the adsorption entropy of
hydro-in the (332) directions will be larger than zero. gen, L~S~was assumed to be —125 J K1 mol~
This is consistent with the observation that in all [15]in the temperature range considered, i.e. from
our growth experiments either macroscopic steps 1210 to 1350 K. After the calculation of the term
parallel to the stable PBC directions or flat faces KH PHeq in eq. (13) with these data, we fitted the
are observed at the positions of (113) on the resulting parameters in the same way as was done
hemispheres. in the previous section for eq. (4). The values of
L~H~°1and L~S~°1obtained from these linear
least-4.3.2. Competitive adsorption of hydrogen and chlo- squares fits of eq. (13) are shown in figs. 6 and7,
rine respectively.
It may be expected that next to chlorine, also It is assumed that the values of 1iH,~1and i.1H~
hydrogen will be present on the surface of the that best fit the data of fig. 2 are those which
242 J. G.E. Gardeniers et al. / Influence of Cl—H ratio on stability of (1/3) faces of Si in SE—H—Cl CVD
.1555. 1 Table 3
Hnr,1 — — — — —- Silicon—hydrogen bond strengths at 298 K
Bond Bond strength References (kJmoI~) 100.0. 1. H1Si—H 378 [27] .SSftO - 384 [35] HSi—H 268 [27] 303 [35] HSi—H 351 [27] 317 [35] Si-H 293 [27] 285 [35]
250(1241).))___________________________2611.)) 28)).)) 3(1)1.0I 320(11 34(1.1)1 (CH3)H2Si—H(CH3)3Si—HCI1Si—H 378374382 [27][27][27]
k.1no,l~ Si(1 11)—H 300—340 [301
Poly-Si—H 318 [31]
Fig. 6. Chlorine adsorption heat derived from a fit of the data
308, 354 [32] in fig. 2 according to eq. (13), for different values of the
Amorphous-Si—H 270—360 [33]
hydrogen adsorption heat (see text).
328 [34]
±9 J K~ mol~in table 1. In fig. 7 this
theoreti-The corresponding value of ~1H,~/follows from fig. cal value is indicated by the horizontal solid line,
6: while the horizontal dashed lines indicate the
un-certainties in this value. From the intersection of ~ H~°1= —370 ±70 kJ mol— the theoretical and the “experimental” lines we
obtain: If it is assumed that ZIH~1does not change more
than a few percent over the temperature range
= —316 ±5 kJ mol~. 298-1400 K (see above), it can be seen that the
values of ~H~1298and ~H~298are in agreement with the literature values of Si—Cl and Si—H bond strengths in tables 2 and 3, respectively.
2(0)II
~ 4.3.3. Adsorption of silicon compounds
.55/ i,, I 1- K
-To obtain a complete view of the influence of
15)))) gas phase composition on the stability of the
(hhk)h<k faces, in addition to the adsorption equilibmi (2) and (10), the following series of
equilibria has to be included:
Si~H,,Cl,5_’+ surface site S~rHCI..~ .,d,,rbrd (14)
1000. Of all the silicon compounds which are present in
the gas phase at equilibrium, we will assume that
500 only radicals are able to chemisorb on the silicon
21(c) 26(c) 2800 ((1(1(1 320.0 :1100 surface; of these SiC]2’ SiC13, SiHCI, and SiH2
is k.1 ,,,,,l’
are the most abundant in the equilibrium gas
Fig. 7 Chlorine adsorption entropy derived from a fit of the phase mixtures (see fig. 4). As a very rough
esti-data in fig. 2 according to eq. (13), for different values of the mate of the heat of adsorption of these species, we
J.G.E. Gardeniers et al. /Influence of Cl—H ratio on stability of (113) faces of Si in Si—H—Cl CUD 243 log & 2250 K, it can be calculated that in the
tempera-- ture range examined in the experiments (1200—
1450 K) the (113) faces become rough if O,~,
-1 ~ becomes lower than approximately 0.2. It can be
seen from fig. 8 that at low (Cl/H) ratios the coverage of the silicon surface with Cl is low, so
-2
CI
~— that the chemical roughening temperature in thiscase is almost completely determined by H ad-sorption. At higher (Cl/H) ratios the coverage of
/.
//
Cl increases, while the H coverage remains the~ / same or decreases somewhat because of the
corn-‘~
//
// petition with Cl. So at the higher (Cl/H) ratios///Si
-°“° the chemical roughening temperature will bede-~ /// ___ termined by the adsorption of both Cl and H. As
// / the total coverage Ototal at higher (Cl/H) ratios
// log (Cl/H) becomes higher, chemical roughening will start to
-6 I occur for 0iotai = 1 — 0~ 0.8. At still higher
- - .. . (Cl/H) ratios the growth temperature has to be
Fig. 8. Surface coverage 0 of Cl, H andsilicon species as a ,
function of (Cl/H). For every adsorbate three lines are drawn, raised in order to ensure the stability of the (113)
corresponding to three different temperatures, viz. 1250. 1350 faces. This is exactly the trend observed in fig. 2.
and 1450 K. Finally we want to make some remarks on the derived heats of adsorption of Cl and H. Com-parison of the derived values with those in tables 2
which is minus half the sublimation enthalpy of and 3 shows that although the agreement is
satis-silicon, and ~S~°~5 —170 J K’ mol_i. To have factory, the value of~H~1 is somewhat lower than
some idea of the surface coverage, we summarized most values in table 2. A reason for this may be
the equilibrium partial pressures of these species that at the relatively high coverages which we are
and used the before-mentioned thermodynamic dealing with (see above), adsorbate interactions
data and the Langmuir isotherm. The resulting will become substantial. More specifically, it may
coverages as a function of the (Cl/H) ratio, to- be expected that the large electronegativity of the
gether with those of hydrogen and chlorine, are Cl atoms with respect to Si will lead to a small
shown in fig. 8 for three different temperatures. negative charge on the adsorbed Cl, leading to a
As can be seen, the surface coverage of growth repulsive interaction between these adsorbates.
species is always much less than that of chlorine This repulsion will lower the heat of adsorption,
and hydrogen, even at extremely high chlorine— which might explain the observed difference.
hydrogen ratios of, e.g., 0.3 (i.e. log(Cl/H)=
—0.5). Therefore we conclude that in above
dis-cussion on the effect of adsorption on the stability 5. Summary
of the (113) faces, the contribution of silicon
species can be neglected. In this paper the orientation dependence of
silicon crystal growth in the Si—H—Cl CVD
sys-4.3.4. Concluding remarks tern was studied as a function of the
chlorine—hy-It was shown in this section that the shift of the drogen ratio of the gas phase. This was done by
“chemical roughening” temperature of the (113) the use of hemispherical single crystal substrates.
faces with the (Cl/H) ratio is caused by the corn- As was reported before [1], above a certain
petitive adsorption of Cl and H. As it was derived critical temperature flat (113) and (337) faces are
that chemical roughening of these faces occurs found on the hemispheres, while below this
posi-244 J.G.E. Gardeniers et al./ Influence of Cl—H ratio on stability of (113)faces ofSiin Si—H—C/CUD
tions corresponding to these faces. This critical [9] J.W. Medernach and P. Ho, in: Proc. 10th Intern. Conf.
temperature is strongly dependent on the chlo- on CVD, Honolulu. Hawaii. 1987. Ed. G.W. C’ullen
(Elec-rine—hydrogen ratio of the gas phase. It was dem- [10] B. Nolang. in: Proc. 5th European Conf. on (VI). Up-trochem. Soc.. Pennington, NJ, 1987).
onstrated that this so-called “chemical roughen- psala. 1985. p 107.
ing” effect is caused by the competitive adsorption [11] C.H.J. van den Brekel. J. Crystal Growth 23 (1974) 259.
of chlorine and hydrogen. From the experimental [12] J. Nishizawa. T. Terasaki and M. Shimho. J. (‘rystal
dependencies, values for the heat and entropy Growth 13/14 (1972) 297.
changes for chemisorption of chlorine and hydro- [13][14] P. Hartmann, Z. Krist. 121 (1965) 78.C. Herring. Phys. Rev. 82(1951) 87.
gen can be derived. These values are in good [15] U. Giling. H.H.C de Moor. W.PJ.H. Jacobs and A.A
agreement with literature values of Si—Cl and Si—H Saaman. J. Crystal Growth 78 (1986) 303.
bond strengths and entropy values derived with [16] JANAF Thermochemical Tables, J. Phys. (‘hem. Ref.
the use of statistical-thermodynamical methods. Data 11(1982) Suppl. p. 695.
[17] V.P. Glushko. V.A. Medvedev et al., Eds., Termicheskie Konstanty Veshchesty. Part 1(1965).
[18] JANAF Thermochemical Tahles. J. Phys. Chem. Ref.
Acknowledgements Data 14 (1986) Suppl. I.
[19] J.M. Jasinski. B.S. Meyerson and BA. Scott, Ann. Rev. Phys. Chem. 38 (1987) 109.
The authors want to thank Dr. R. van Meerten [20] (‘-A. Chang. High Temp. Sci. 17(1984)445.
for stimulating discussions and Mr. W. Maas and [21] G. Herzherg. Molecular Spectra and Molecular Structure,
Mr. P. van der Heyden for experimental contribu- Vols. I and 11 (Van Nostrand. Princeton. NJ, 1950).
lions. One of the authors (J.G.E. Gardeniers) [22] M. Seel and PB. Bagus, Phys. Rev. B28 (1983) 2023.
acknowledges the support of the Dutch Founda- [23] SM. Mohapatra, B.N. Dcv. K.C. Mishra. N. Sahoo. W.MGibson and T.P. Das, Phys. Rev. B38 (1988) 12556.
tion for Chemical Research (NWO/SON). [24] C. Newman. J.K.O. Loane, SR. Polo and M.K. Wilson, J.
(‘hem. Phys. 25 (1945)855.
[25]J. Ling-Fai Wang, J.L. Margrave and J.I.. Franklin, J.
Chem. Phys. 61(1974)1357.
References [26] R. Walsh, J. Cheni. Soc.. Faraday Trans. I, 79 (1983) 2233.
[I] JOE. Gardeniers, W.E.J.R. Maas, R.Z.C. van Meerten [27] R. Walsh. Ace. Chem. Res. 14 (1981) 246.
and Li. Giling, J. Crystal Growth 96 (1989) 821. [28] ME. Coltrin, R.J. Kee and J.A. Miller, i. Electrochem. [2] J.G.E. Gardeniers, W.E.J.R. Maas, R.Z.C. van Meerten Soc. 133 (1986) 1206.
and L.J. Giling, J. Crystal Growth 96 (1989) 832. [29]TN.Bell, K.A. Perkins and PG. Perkins, J. (‘hem. Soc.. [3] P. Bennema and J.P. van der Eerden. in: Morphologyof Faraday Trans. 1. 77 (1981) 1779.
Crystals. Ed. I. Sunagawa (Terrapub. Tokyo, 1987) p. 1. [30] OS. Schulze and M. Henzler. Surface Sci. 124 (1983) 336. [4] L. Hollan andC.Schiller, J. Crystal Growth 13/14 (1972) [31] K.W. Frese. Jr., Surface Sci. 182 (1987) 85.
319 [32] P. Gupta. V.L. Colvin and SM. George. Phys. Rev. B37 [5] J. Bloem. Y.S. Oei, H.H.C. de Moor, J.H.L. Hanssen and (1988) 8234.
L.J. Giling,J. Electrochem. Soc. 132 (1985) 1973. [33] D.C. Allan. iD. Joannopoulos and W.B. Pollard, Phys. [6] L.P. Hunt and E. Sirtl, J. Electrochem. Soc. 119 (1972) Rev. B25 (1982) 1065.
1741. [34] K. Zellama. P. Germain, S. Squeland, B. Bourdon, J.
]7[ E. Sirtl, L.P. Hunt and D.H. Sawyer. J. Electrochem. Soc. Fontenille and R. Danielou, Phys. Rev. B23 (1981) 6648. 121 (1974) 919. [35] P. Ho, ME. Coltrin, iS. Binkley and CF. Melius, J. Phys. [8] UP. Hunt, J. Electrochem. Soc. 135 (1988) 206. Chem. 89 (1985) 4647.