Note: Descriptions are shown in the official language in which they were submitted.
3L~4~:8~9
PHD. 83.068
The invention relates to a method of manufact-
uring optical fibres in which glass :layers are deposited
on the inner wall of a glass tube heated.at a temperature
between llO0 and 1300C by passing a reactive gas mixture
through the glass tube a-t a pressure between l and 30 mbar
while a plasma is reciprocated strokewise in the interior
of the glass tube after which the glass tube, after a suf-
ficient number of glass .layers has been deposited, is to
col:lapse so as to form.a solid preform from which optical
fibres are drawn.
"Glass tube" is to be understood to mean in this
connection.a suhstrate tube or claddlng tube which consists
of.amo.rphous silica wh:ich either is made synthetically or
by melting quartz crys.tals (fused silica, fused quartz)
which:amorphous silic.a may be doped, or which tube consists
both.of synthetically made;amorphous silica:and of amor-
phous silica made by melting quartz crystals (fused silica,
fused quartz) which may.also be doped.
The manufacture of optical fibres, according to
the.aho~e-described method is known from US Patent Speci-
fications Re 30 635:and 43 14 833. This method of manufac-
turing is:referred to in the.art.as "non-isothermal plasma
CVD-method" (non-isothermal PCVD-method). In this method,
glas.s layers.are deposited directly from the gas phase on
the inner wall of the glass tube (heterogeneous reaction).
The formation of glass soot in the gas phase is avoided;
this is described in de.tail in particular in US Patent
Specification 43 14 833.
Another method for -the manufacture of high-
quality optical fibres is the so-called MCVD method.
The essential difference between the PCVD method
.and the MCVD method.resides in the nature and-manner in
which the chemical reactions necessary for the deposition
PHD 83.068 -2- 14.6.1984
of the glass components are induced. Whereas the MCVD-
method utilizes thermal excitations, the PCVD-method uses
electron impact excitation. In the MCVD-method primarily
fine dust particles (glass soot) are formed, the deposition
of which in the temperature and gravity field can occur
only uniformly in and on rotating substrate .In contrast
herewith the formation of fine dust particles does not
occur in the electron impact excitation of the PCVD-method;
the gaseous reaction products are rather present in a mole-
cular form and can reach the inner wall of the normallyused SiO2-tube via a rapid diffusion and can condense there;
a sintering is not necessary afterwards. For this reason
it is also possible in the PCVD-method to deposit glass in
both directions of movement of the plasma. In contrast here-
with in the MCVD-method only when the thermal energy source
moves in the sarne direction as the reactive gases glass
particles can be deposited.
It could be demonstrated ("Deposition of SiO2 with
low Impurity Conten-t by oxidation of SiC14 in a non Iso-
thermal Plasma" J. Koenings, D. Kippers, H.Lydtin, H. Wilson,Proceedings of the 5th Int. Conf. on CVD (1975) I. 270-
280) that with -the PCVD-method in well controlled labora-
tory experiments a uniform inner wall coating of a station-
ary tube is possible, i.e. the extremely small influence
of gravity In atomic and molecular speciesmay be neglected.
On transfer of the laboratory data to pilot and production
equipments, however, certain -types of profile imperfections
which in particular showed no cylinder symmetry occurred in
the manufactured fibres. It is clear that the bandwidth
value of such fibres should be far below the theroretically
possible value. Although said profile imperfections accord-
ing to the known PCVD methods with stationary cladding tube
can principally be avoided this requires a comparatively
high expenditure of apparatuses
For example, the deposition behaviour of the
index-varying doping material and hence the shape of the
refractive index profile depend on the temperature (P.Bach-
mann, P. Geittner, H. Wilson, Proc. of the 8th ECOC-Confe-
..
., .
~2a~
PHD 83.068 -3- 14.6.1984
rence, Cannes, ~.516 (1982). Non-uniform temperature dis-
tributions over the circumference of the tube during the
deposition automatically lead to peripheral profile imper-
fections. A non-uniform temperature distribution can be
caused inter alia by convection currents and inhomogeneous
heat dissipation over the circumference of the furnaces
used for heating the tube. In order -to achieve a uniform
temperature distribution, a high expenditure of apparatuses
would generally be necessary.
The energy distribution in the plasma can in
practice also be asymmetric - for example via a non-centric
arrangement of the tube in the microwave resonator - which
also produces deposition asymmetries. Furthermore in the
known PCVD me-thod a hardly noticeable bending of the tube
occurs, which involves difficulties upon collapsing.
Such non-centric arrangements and the bending could be
avoided by positioning -the tube very exactly in the reso-
nator and suppor-ting the tube during the whole deposition.
However, this would further complicate the technical invest-
ment of apparatuses as well as the problems regarding tem-
pera-ture distribution.
It is the object of -the inven-tion to avoid the
formation of peripheral radial and axial profile imperfec-
tions in PCVD optical fibres in a simple manner
According to -the invention this object is achiev-
ed in -that in a method of the type mentioned in the opening
paragraph the heated glass tube is continuously rotated
during the deposition of the glass layers and the direct-
ion of rotation is reversed every time the direction of
movemen-t of the plasma is reversed.
The glass tube at each reversal of the direction
Y of movement of the plasma at the same time is rotated over
,, an ~6~e smaller than or equal to 180 around its longitu-
dinal axis. The result thereof is that failures caused by
an asymmetrical energy distribution in the plasma are com-
pensated.
Advantageously, the rotation is carried out at
at least 0.25 rotations per s-troke. At smaller rotation
8~
- PHD 83.068 -4- 14.6.1984
veloci-ties a compensation of a non-uniform temperature
distribution would hardly occur. The upper limit of the
rotation velocity is limited substantially only by the
performance of the necessarily required vacuum-tight ro-
tating lead-throughs.
The effect of the measures according to the in-
vention used separately or in combination is substan-tially
based on the fact that peripheral imperfections built in
in radial and/or axial direction during the deposition of
lD successive individual layers are varied as completely as
possible in their relative position with respect to each
o-ther and in such a manner that such imperfections over
the deposition length compensate each other.
As already stated, rotation is carried out in
the MCVD-method continuously so as to eliminate the effects
of gravity. In the PCVD me-thod on the contrary, difficul-
ties of a different nature are introduced with a continuous
rotation in fact imperfections would not be compensated at
certain points along the deposition length.
The disturbing effects of non-uniform tempera-
ture distributions are removed in a simple manner by the
way of rota-tion according -to -the invention, since herewith
the temperature on the circumference of the tube can be
uniformly adjusted to an average value due to the thermal
inertia of the -tube.
All in all it could not be expected that all these
manifold problems could be obviated by the comparatively
simple way of rotation according to the invention.
As already mentioned hereinbefore, refractive
index profiles which are subs-tantially free from peripheral
radial and axial dis-turbances can also be achieved by means
of the PCVD-method without rotation. However, higher order
disturbing effects can be obviated in a simple manner by
the method according to the invention, i.e. improvements
in the miniature structure of the optical fibres are achiev-
ed. These improvements are desirable in particular to
achieve bandwidths over 1 GHz.km.
The invention will be described in greater detail
2~
PHD 83-068 -5- 1~.6.1g84
with reference to a drawing and a comparative example.
In the drawing
ig. 1 shows an optimum parabolic refractive index
profile manufactured according to the PCVD method without
rotation and
Fig. 2 shows a refractive index profile manufac-
tured according to the PCVD method according to the inven-
tion.
In both figures the variation of the refractive
index dependent on the radius of` the preform is shown.
In the first part of the example it was tried
to produce an optimum parabolic refractive index profile
in the core of a preform according to the PCVD method
without rotation. The deposition of the material occurred
under the following experimental conditions:
SiO2-substrate tube having an inside diameter
of 15.0 mm and an outside diameter of 18.0 mm was used.
The length of deposition was adjusted at approximately
45 cm. The average pressure in the deposition area was
approximately 15 mbar, the wall temperature of the sub-
strate tube was approximately 1150 to 1250C. A microwave
resonator having a power consumption of 500 W was recipro-
cated over the desposi-tion area at a rate of 8 m/min. The
overall gas flow was kept constant at 800 sccm (= cm3 of
gas per minute under normal conditions: 0 C, I bar) during
the whole deposition lasting approximately 120 minutes;
so the overall number of layers was 2100.
The reaction gas flows Q during deposition were
supplied as follows: the oxygen flow remained constant at
Qo2 = 7 sccm. In the op-tical cladding area the SiCl4 and
GeC14 flows were kept constant at QSiCl = 110 sccm and
QGeCl = 0 sccm; in the core area, QSiCl4
110 to 100 sccm and QGecl was varied from 0 to 18 sccm
so that (with an approximately constant chloride gas flow
of approximately 110 to 120 sccm) a parabolic refractive
index profile was achieved. The deposition rate for the
doped core material in the present conditions was approxi-
... .
PHD 83-068 -6- 14.6.1984
mately 0.30 to 0.35 g/min.
Fig. 1 shows the refractive index profile thus
obtained in the collapsed preform (York Technology, P101-
Analyzer-Plot, Position 300 mma O degree; 961 points in
10 /um steps). As is clearly seen, -the profile in particu-
lar in the proximity of the centre - beside the central dip-
shows a pronounced asymmetry A of approximately Z /0 (re-
lated to the average refractive index difference in the
centre), which impedes the adjustmen-t of higher bandwidths
and hence is undesired. The bandwidth ox the fibre drawn
from the preform in the present example was 800 MHz.km (with
a transmission wavelength of 900 nm). Further a geometrical
asymmetry of approximately + 1.5 % is found.
In the second part of the example a further pre-
lS form was manufactured with identical deposition conditionsin which this time, however, the substrate tube was rotated
during the deposition. The rotation of the tube was made
possible by means of two rotation leadthroughs which were
substantially gas-tight (leakage rates in the range from
10 4 to 10 5 mbar.l.s 1) and were driven synchronously.
In the present example the rotation frequency was 25 rpm
which, with the adjus-ted stroke frequency of approxima-tely
16,6 strokes/min. (coating length 45 cm with a resonator
speed of 8 m/min) corresponds to 1.5 ro-tations per layer.
The angle rotation at each reversal of the direction of
plasma movement was 180 ; at the same time the direction
of rotation was reversed.
Fig. 2 shows the absolutely rotationally symme-
trical refractive index profile of the preform resulting
in this case (measuring conditions as in fig. 1). The band-
width of the fibre drawn from said preform was approximate-
ly 1600 MHz.km at 900 nm, so was approximately double as
high as in the case of the first part of the example with-
out rotation during the coating. Optical and geometrical
asymmetries amount to approximately a few tenths of a
percent.
