вход по аккаунту


Chemical Reactions in the Manufacture of Waveguides for Long Distance Optical Data Transmission.

код для вставкиСкачать
Chemical Reactions in the Manufacture of Waveguides
for Long Distance Optical Data Transmission
By Michael Binnewies,” Marion Jerzembeck, and Andreas Kornick
Dedicared io Professor Hans Georg von Schnering on the occasion of his 60th birthday
One of the modern high technologies which has advanced enormously in the last few years is
glass fiber technology. This is used in the manufacture of glass fibers for lighting purposes and
for the optical transfer of analog and digital data with a high transfer density. The technical
demands made on the glass fibers required for data transfer, the optical waveguides, are
extremely high and are already fulfilled to a large extent by industry. At present about four
million kilometers of fiber, worth ca. 800 million DM are produced worldwide (10% in the
Federal Republic of Germany). Numerous chemical processes take place during the manufacture of optical waveguides. However, in contrast both to the high and constantly growing
demands on the quality and to the increasing production volume of such fibers, little is in fact
known about the reactions involved. The present article will attempt to develop a picture of
the multifarious reactions occurring in the course of this technical process on the basis of
literature data and our own studies.
1. Introduction
The transmission of analog, and in particular digital signals at a high spatial and temporal information density has
been of rapidly growing importance in the past few years.
Classical metallic conductors are subject to physical limits as
far as the maximum information density which can be transmitted is concerned. Much higher information densities
should in principle be achievable when optical communication media are used. The prerequisites for an effective transfer of information with the aid of light are:
1. Powerful transmitters.
2. Materials which are capable of fast modulation of the
light waves.
3. Transfer media which permit the transmission of the
modulated light waves across long distances with as low
a degree of attenuation as possible.
4. Sensitive and fast light detectors.
Scientific progress in solid-state chemistry and physics
during the last 20years has played a decisive role in the
development, manufacture and in particular in the undersfanding of the materials referred to under 1, 2 and 4. However, the transfer medium, the optical waveguide, has been
brought to an exemplary state of technical perfection in a
manner which is basically empirical (though by no means
less effective!); the term “empirical” refers not to the wellknown ph,vsical basis of this technology but to the extremely
complex chemical reactions occurring during the manufacturing processes. It is however by no means unknown in
history for technologies to undergo purely empirical development. The best-known example is the history of metallurgy, in the beginnings of which our ancestors were able to
obtain metals from oxide minerals and coal without knowing
anything of the chemistry involved (to say nothing of the
Boudouard equilibrium!). The heretical postulate which
could be put forward at this juncture is: research is not neces[*] Prof. Dr. M. Binnewies, Dipl.-Chem. M. Jerzembeck,
Dipl.-Chem. A. Kornick
Institut fur Anorganische Chemie und Analytische Chemie der Universitit
Albertstrasse 21. W-7800 Freiburg (FRG)
Angew. Chein. In(. Ed. EngL 30 ( I 991) 74s-753
sary. In the case of many technically manufactured materials
the scientist is justifiably still interested even today in the
material itself, while the processes by which it is formed are
of only minor interest. The relationships between the properties and the genesis of a material are often studied only
empirically in the absence of parallel studies which can lead
to an understanding of the chemical mechanisms involved in
their formation.
1.1. The Physical Principle of the Optical Waveguide
The purpose of the present article is not to present the
physical background of light propagation in optical waveguides in detail but to investigate the chemical processes
involved in their manufacture. However, a brief look at the
physical principle[’] appears imperative in order that we can
understand why “chemistry” is in fact necessary here at all.
When a ray of light passing through a medium of refractive index n , is incident on a medium of lower refractive
index n, ( n , > n,) at an angle of incidence below a certain
limiting value it will undergo total reflection, i.e. it cannot
leave the medium. Thus if it is possible to produce a transparent rod consisting of a core of refractive index n , and a
coating of refractive index n z , light entering the core of the
rod at one end will exit at the other end without being able
to leave the rod sideways at any stage; this will be the case
even when the rod is bent. Such a rod can be easily prepared
by fusing a glass tube (n,) on to a glass rod (n,) and readily
be drawn out to give a thin filament without any appreciable
change in the refractive index profile, so that the optical
properties remain unchanged. A so-called step-index fiber in
which the refractive index changes in a stepwise manner over
the fiber cross-section (Fig. 1 a) does however have one principle disadvantage which makes a high transmission density
impossible; this is the dispersion (for details see Ref. [2]).
One of the important points here is that light pulses which
enter an optical waveguide can have differing transit times:
a pulse which enters exactly parallel to the fiber axis will in
an ideal case be able to exit the fiber without undergoing
0 VCH VerlagsgesellschujimhH, W-6940 Weinhelm, 1991
OS70-0833/9ij0707-0745S 3.N+ .2.5/0
reflection (assuming the fiber to be linear). However, a pulse
which is subject to repeated reflection will traverse a longer
path and will thus exit the fiber with a certain small time
delay. This leads to a broadening of the output signal with
respect to the input signal and thus a priori to a lower transmission density. These differences in the transit time can be
considerably reduced if the refractive index does not vary in
a stepwise manner across the fiber diameter but decreases
from the fiber center to its boundary according to a certain
predetermined parabolic function (Fig. 1b). In this case the
The much more successful processes used today involve
the deposition from the gas phase of glass-forming oxides of
higher refractive index.
The starting material is a tube made of high-purity quartz
(external diameter ca. 3 cm, length ca. 2 m). A stream of
oxygen containing SiC1, vapor is passed through the rotating tube. Dopant materials such as GeCl,, BCl, and P(O)CI,
are added to this gas stream at concentrations which can be
metered and varied exactly according to a predetermined
program.r31When the necessary energy of activation is applied, the reaction of the oxygen with the volatile halides or
oxide halides of the elements silicon, germanium, boron and
phosphorus leads to the formation of the oxides of these
elements; these are deposited as solids on the inner surface of
the quartz tube and fuse with it at high temperatures. Figure 2 shows the nature of the industrial process. The gas
Fig. 1. Schematic representation of the variation of the refractive index across
the fiber diameter in step index fibers (a) and gradient fibers (b).
light pulse which passes through the center of the fiber without undergoing reflection is slowed down by the higher refractive index (i.e. by the reduced light velocity) so that differences in transit time can be eliminated to a large extent.
Such a fiber, known as a gradient fiber, can of course not be
obtained simply by fusing a rod and a tube together.
2. The Internal Coating Process for the
Manufacture of Gradient Fibers
Initial attempts to obtain gradient fibers involved the application of glass-forming powder layers of continually increasing refractive index to the inner surface of a quartz
tube; the powder layers were subsequently fused to the
quartz tube, and the blank thus prepared was then allowed
to melt under controlled conditions to give a rod which could
be pulled out to give a fiber. However, this process could not
be developed to a stage suitable for commercial manufacture.
Fig. 2. Manufacture of the blanks for optical waveguides. The photograph
shows a rotating glass tube through which the reaction mixture is passed. The
tube is heated by oxyhydrogen burners which are slowly moved parallel to the
axis of rotation of the tube during the process, which lasts several hours; an
even deposition is thus achieved along the whole length of the tube.
mixture enters the quartz tube from the left and is heated by
a battery of oxyhydrogen burners. The deposition of the
solids occurs within and immediately subsequent to the zone
of heating.
Variation of the concentrations of the added dopant substances permits the predetermination and continual variation of the refractive index of the deposited layers. Such a
CVD (Chemical Vapor Deposition) process makes it possible to obtain the required refractive index profile. When
this goal has been achieved, the tube is fused at high temperatures to give a rod, from which a fiber can be drawn which
has the originally constructed refractive index profile. Such
a blank affords ca. 50 km of optical waveguide material.
Michael Binnewies, who was born in 1947 in Hagenl Westfalen (FRG), studied chemistry at the
University of Miinster, where he gained his Diploma and thereafter his doctorate in 1973 under
the supervision of Harald Schafer. He then spent a few years doing research into the development
of mass spectrometric measuring techniquesfor the invesligation of chemical transport reactions.
His current research interests focus on high-temperature reactions, plasma chemistry, and halogen compounds of semi- and non-metals. After continuing his researches at the Max-Planck-Institut fur Festkorperforschung, Stuttgart (1979), he returned to Miinster, where he habilitated
in 1985. In 1988 he accepted the ofler of a professorial chair at the University of Freiburg.
Angew. Chem. Int. Ed. Engl. 30 (1991) 745-753
Figure 3 shows a raster electron microscope picture of the
end face of such a fiber. The deposited layers can be seen
clearly within the higher-refracting core of the fiber. The
activation of the required chemical reaction can be carrkd
intermediates formed and what are their nature? What are
the thermodynamic properties of the system? How does nucleation occur starting from such a gas phase? When can one
already speak of the existence of a solid and when must one
still speak of a molecule? Questions of this and similar types
are presently being addressed and discussed by many groups
working with "simpler" systems such as metal vapors (cluster formation in molecular beams).
In the following discussion we shall deal with the above
reactions of the non-metal halides and oxide halides of silicon, germanium, and phosphorus, and attempt to find answers to some of the above questions.
3.1. The Reactions of SiCI, and GeCI, with 0,
Fig. 3. Raster electron microscope picture of the end face of an optical waveguide. The external diameter of the fiber is ca. 120 prn.
of the appli-
( T , 1300 K) or by
cation of a non-isothermal plasma.
3. Chemical Reactions
The chemical reactions occurring in the course of the process described Seem at a first glance to be simple [Eq. (a)(d)]. All these reactions are exothermic, the equilibrium constants are larger than unity even at higher temperatures (except for (b)), AGO is negative (except for (b) above 1200 K).
Some numerical values are collected in Table 1. The reactions (a)-(d), like all reactions involved in CVD processes,
+ 0,
G SiO,
+ 2 CI,
2BCI,+ 1 SO,
+3 0 ,
Table 1. Selected thermodynamic data of the reactions ( a H d )
-242.7 [4]
- 79.9 [4, 51
-456.9 [4, 51
-770.7 [41
[12, 131
[12, 131
[12, 131
[12, 131
112, 131
[12, 131
[12, 131
[12, 131
[12, 131
[6], [9-131
[lo, 121
[ l l , 121
[12, 131
[12, 131
describe processes in which solids separate from a gaseous
phase; thus molecular precursors undergo conversion into
infinite three-dimensional lattices (the fact that the oxides
are deposited as a glass rather than in a crystalline form is of
no importance for the following discussion). Such processes
cannot in fact be so simple as the equations (a)-(d) would
lead one to expect. A number of very basic questions must be
addressed when we consider how a solid is formed starting
from a gas phase (or from a solution or melt): are detectable
The first studies of these apparently simple reactions were
carried out long before their future technical importance
could be imagined. In 1868 Friedel and Ladenburg reported
the formation of Si,OCI, during the reaction of SiCI, with
0, in a porcelain tube heated by burning coke.[61Troost and
Hautefeuille obtained mixtures of higher molecular weight
chlorosiloxanes in a similar manner a few years later.[7,81 A
series of further studies of the reaction followed in which the
lol A
original results were confirmed and s~pplemented.[~*
large number of related chlorosiloxanes have been described" - 13] and in part structurally characterized" l l in
recent years, The common feature of all these reports is that
the compounds described were prepared in the reaction between SiCI, and 0, at high temperatures (around 1300 K) or
in a gas discharge (plasma), i.e' in the Same reaction and
under similar conditions as those which are so important for
the technology of the manufacture of optical waveguides.
The chlorosiloxanes presently known are collected in
Table 2.
does not by any
2.3 x 10'
1 . 9 10'
- 40.1 [4]
3.8 x 10"
- 67.8 [4, 51 4.3
- 165.1 [4, 51
- 349.9 141
1 . 7 lOI5
9.6 x loz1
Angeil.. Chem. In[. Ed. Engl. 30 (1991) 745-753
[kJ mol- '1
[7, 81
[11, 131
[I 31
means occur as formulated above via the formation and
separation of solid SO,, but in a much more complex manner. How can we understand the formation of such a large
number of compounds? Let us start by considering their
structures. The plethora of siloxanes listed in Table 2 can be
divided into four groups:
Group 1 :
Group 2:
Group 3:
Group 4:
(n = 3-5, m = 2, 3)
Si~On-lCl,n+ (n = 2-7)
Si,O,CI,, (n = 3-7)
Si,O, +mCIZn-Zm
(n = 4-8, m = 1-3)
The siloxanes thus classified differ in their oxygen content,
which increases on going from Group 1 to Group 4. This has
consequences for the structure of the compounds:
Group 1 . The composition indicates that Si-Si bonds
must be present." 31 These siloxanes have only been observed
in plasma-activated reactions, and they are of no importance
for the following discussion.
Group 2. The compounds in this large group are built up
from chains in which the Si atoms are bridged by oxygen
Group 3. In most cases these molecules have closed monocyclic structures, although monocyclic arrangements with an
exocyclic SiCI, group bonded via an oxygen bridge are
known." The molecular structure of Si,0,C18 was determined by X-ray crystallography['61 (Fig. 4).
Fig. 4. Molecular structure of Si,O,CI,. Open circles are Si atoms, dotted are
0 atoms, and shaded are CI atoms.
Group 4. Chlorosiloxanes which contain more oxygen
than silicon atoms must be oligocyclic in nature. Bicyclic
structures result for m = 1, tricyclic for m = 2,("] etc. The
limiting case in this series (n = m) is SiO,. An initial relationship between the chlorosiloxanes and solid SiO, thus becomes clear.
An answer to the question as to the stability of the
chlorosiloxanes is given by mass spectrometric studies of the
reaction of SiCI, with 0,r121
and the thermolysis of
Si,0CI,.['31 These have shown that the proportion of the
siloxanes in the various groups changes greatly as a function
of the reaction temperature (Fig. 5): the amount of chainlike siloxanes decreases very rapidly with increasing temperature, while the concentrations of the monocyclic and in
particular of the oligocyclic siloxanes increase. Thus, an in-
crease in the temperature accelerates reactions in which relatively oxygen-poor open chain siloxanes such as Si,OCI, are
converted into oxygen-richer compounds. This is shown by
the representative reactions described by Equations (e) to
4Si,0C16 ~Si,0,CI,+4SiCI4
8Si,0C16 =Si,O,CI,+
11 Si,0C16
+ 14SiC1,
Si,O, ,C1,,
The limiting case of such reactions is the decomposition of
a siloxane to give the binary compounds SiCI, and SiO,
[Eq. @)I:
2 Si,OCI,
+ 3 SiCI,
What is the enthalpy and entropy balance of these reactions? Since the total number of Si-0 and Si-CI bonds remains unchanged, the reaction enthalpies are expected to be
close to zero,[12"31as the bonding situation in the starting
materials and products is very similar (see Section 3.2).
However, increasing oxygen content and decreasing chlorine
content in a siloxane corresponds to increasing stabilization.
This can be demonstrated on the basis of the bond energies
of the Si-0 and Si-CI bonds: Si-0 bond 464.8 kJ/bond (in
SO,, data from Ref. [4]), Si-C1 bond 399.5 kJ/bond (in
SiCI,, data from Ref. [4]). The reactions (e)-(h) thus involve
the formation of a product with a lower enthalpy content
(oxygen-richer siloxane and SiO, respectively) and a product
with higher enthalpy content (SiCl,), so that the energy balance will be around 0 kJ.
However, all four reactions involve an increase in the
number of moles present, so that there is a gain in entropy.
This entropy effect favors the formation of the oxygen-richer
reaction products, and in particular that of the oxygenrichest compound SiO,. It can thus be concluded that all the
chlorosiloxanes must be metastable with respect to decomposition to give SiO, and SiCI,. However, the formation of
detectable amounts of S O , is only observed above ca.
1500 K.["] The origin of this reaction inhibition presumably
lies in the fact that the formation of the SiO, structure must
be a kinetically extremely complex process which requires a
high energy of activation.
The increasing degree of formation of oxygen-rich siloxanes with increasing temperature signifies in several respects
that these compounds are continually becoming more
"Si0,-like" :
1. More and more Si-0 bonds are formed.
2. The molecular structures become increasingly more complex (one-dimensional Si-0 skeletons in the chainlike, two-dimensional in the monocyclic, and three-dimensional in the oligocyclic siloxanes).
3. The enthalpy content steadily approaches that of SiO,.
Fig. 5. Temperature dependence of the distribution of the chlorosiloxanes observed during the pyrolysis of Si,OCI,.
Let us return to the technology of the optical waveguide
manufacture (thermal activation of the reaction) and attempt to develop a picture of the reaction scheme on the
basis of the above discussion: the gas mixture SiCl,/O,
(+ dopant materials) enters the glass tube at room temperature and i s heated steadily as it approaches the heating zone.
The formation of halosiloxanes, mainly Si,OCI,, starts at
Angen. Chem. Int. Ed. Engl. 30 (1991) 745-753
around 1300 K. However, this cannot be the first step in the
long reaction sequence which leads to solid SiO,, as it would
require a thermolecular reaction step in which two molecules
of SiCI, and one molecule of oxygen combine with elimination of C1,. The primary formation of a reactive molecule
appears much more likely, the most probable candidate being Si(O)Cl, . Nothing is presently known with respect to the
existence of this molecule in the gas phase; however, it has
been detected in a helium-cooled argon matrix,["1 so that its
intermediate formation in the gas phase also appears likely.
The formation of cyclic siloxanes Si,O,CI,, in higher concentrations thus points to the formation of Si(O)Cl,, since
the former are oligomers of this reactive molecule and could
be formed from it via oligomerization reactions. The formation of open-chain siloxanes such as Si,OCI, from a reaction
between Si(O)CI, and excess SiCl, is also plausible. The formation of the oligocyclic siloxanes is most probably due to
the entropically favored decomposition (see above) of the
monocyclic compounds into oxygen-richer oligocycles and
oxygen-poorer chain compounds or SiCI, (see above); other
reaction pathways are of course possible. There are in principle two types of reaction which could in the final analysis be
responsible for the deposition of SiO,: either the decomposition of oligocyclic chlorosiloxanes which are presumably
oxygen-rich and already contain several Si-0 bonds to give
the binary components SiO, and SiCl, (or oxygen-poorer
siloxanes) or an exothermic reaction between chlorosiloxanes and oxygen via reactive intermediates which eventually
leads to the formation of SiO, and CI, .
Figure 6 shows one of many possible routes from molecular SiCI, to solid SiO,; this includes a number of the compounds listed in Table2. Without any doubt the higher
molecular weight siloxanes such as Si,Ol,CIlo are of partic-
i l
l i
+ chains
Fig. 6. Reaction scheme for a possible course of the reaction between SiCI, and
0, to yield S O , .
Angew. Chent. Int. Ed. Engl. 30 (1991) 745-753
ular interest in this connection; one of its many possible
isomers is shown in Figure 6. A small detail of the SiO,
structure, two SiO, tetrahedra joined via a common apex, is
already preformed. The formation of solid SiO, from such
molecules appears plausible and should be kinetically readily
achievable. The formation of solid SiO, from Si80,,CI,,
requires the formation of oxygen-poorer siloxanes (or SiCl,)
which can then re-enter the reaction cycle. The chlorosiloxanes formed in a stepwise manner can thus be regarded as
intermediates in a chemical reaction between simple molecules (SiCl, and 0,) which eventually lead to the formation
of a three-dimensional solid structure (including glasses).
The formation of the Si-0 bonds occurs in a stepwise manner: one or two Si-0 bonds per Si atom in the chain
molecules, two Si-0 bonds in the monocycles and two, three
(for the bridgehead Si atoms) or even four Si-0 bonds
(Si,O, ,CI,,,) in the oligocyclic siloxanes.
Thus the system Si/Cl/O offers fascinating possibilities for
studying the stepwise transition from molecular compounds
to solids and thus for increasing our understanding of reactions of this type in which solids are deposited from the gas
phase. The formation of this large variety of chlorosiloxanes
(which are thermodynamically unstable with respect to SiO,
and SiCl,) is certainly not fortuitous; their stepwise formation is possibly a necessary prerequisite for the construction
of an SiO, structure.
In contrast to the reaction pathways suggested here, other
intermediates and mechanisms have been discussed in the
literature. Thus in connection with optical waveguide technology McAfee, Laudise and Hozack have published a detailed paper dealing with the thermodynamics of gaseous
species in the system Si/Ge/O/Cl ; their discussion mainly
involves the existence of lower (oxidation state + 2) halides
and oxides of Si and Ge (i.e. molecules of very low molecular
weight) in the temperature range 1400-4000 K.['*] However, the evidence presented above shows clearly that the
deposition of SiO, is almost complete at temperatures of ca.
1500 K, so that a discussion of the composition of the gas
phase at temperatures far above 1500 K does not appear
particularly useful. In addition, thermodynamically unstable
compounds such as the siloxanes discussed above are naturally not amenable to thermodynamic calculations.
In contrast to the reaction between silicon tetrachloride
and oxygen, which has been studied in great detail, the literature contains very little information on the analogous reaction of germanium tetrachloride. Ge,OCI, has been described as a reaction product.['g* Our own studies of this
reaction under the conditions of low-pressure gas discharge
have also demonstrated the formation of the additional
chlorogermoxanes Ge,O,CI,,
Ge,O,CI,, , Ge,O,CI,,
Ge,O,Cl, ,Ge,O,Cl,, and Ge,O,Cl, .lZ11 This indicates that
the system Ge/O/Cl is very similar to the system Si/O/Cl
described above.
3.2. The Reaction of BCl, with 0,
Studies of the equilibrium between liquid B,O, and BCI,
vapor at temperatures above 500 K have demonstrated the
existence of the cyclic chloroboroxines B,O,Cl, and
B,O,Cl, as well as of BOCl, which is formed at measurable
concentrations at higher temperatures (> 1200 K).[22-241
The thermodynamic data determined indicate that at temperatures above 1000 K these compounds should be formed
in a direct reaction between BCI, and 0,, i.e. during reactions of SiCI,/GeC1,/BC13/P(0)CI,/0,-mixturesoccurring
during the manufacture of glass fibers. In contrast to the
behavior of the siloxanes, the decomposition of (for example) B,O,CI, into the binary compounds B,O, (liquid) and
BCI, (gas) is not expected. In agreement with this hypothesis
the formation of B,03CI, has been detected spectroscopically in the course of the reaction of BCI, with 0,
of this reaction under plasmachemical conditions show that
apart from the cyclic boroxines the linear molecules B,OCI,
and B,O,CI, are also formed.[z61Oligocyclic analogues of
the siloxanes are unknown in the system BjOjCl. If the decomposition of the chloroboroxines cannot lead to the formation of B,O,, the deposition of which certainly occurs,
the formation of the boroxide must occur via another reaction pathway. This can only involve the reaction of the initially formed boroxines with oxygen, which is present in
excess. The thermodynamic datac4]for such reactions do
indeed lead us to expect that the deposition of B,O,
will occur. Thus the reaction of B,O,CI, with 0, as
shown in Equation (i) is strongly exothermic (AH'&8 =
- 551.7 kJmol-',[41 though it is accompanied by an entropy
loss (AS:98 = - 295.5 J K - ' ~ o I - ' ) . [ ~ ]
2 B,O,CI, (gaseous)
+ 1.5 0,+2
B,O, (liquid) + 3 CI,
These values lead to equilibrium constants much greater
than unity even at high reaction temperatures.
Why is the thermal behavior of the gaseous chlorosiloxanes and chloroboroxines predicted on the basis of thermodynamics so different? To answer this question we shall take
as an example the monocyclic compounds Si,O,Cl, and
B,O,CI, and their decomposition according to Equations (j)
and (k) to give the binary compounds ( T > 1500 K).
si,o,CI, (gaseous) + 1.5 SiO, (solid) + 1.5 SiCl,(gaseous)
B,O,CI, (gaseous) + B,O, (solid) + BCI, (gaseous)
Both reactions are expected to have low reaction enthalpies, since (with similar bonding conditions in starting
materials and products) the number of element-oxygen and
elementxhlorine bonds remains unchanged (AH&,8 is unknown, AH:98 = - 42.9 kJmol-'[41). The reason must thus
lie in the entropy balance. Reaction (j) is expected to have a
reaction entropy considerably larger than zero (see above),
while for reaction (k), in which the number of moles is constant, there should be no great entropy change (experimental
value: AS'&* = - 56.3 JK-'mol-'[41). Thus, the different
thermal behavior depends on the stoichiometry of the decomposition reactions of B,O,CI, and Si,O,CI, and the
resulting difference in the entropy balance.
Attempts to obtain B,O,CI, in the condensed phase have
been unsuccessful; the main reason for this is also the change
in the entropy balance of the decomposition reaction of hypothetical B,O,CI, (solid, liquid) in favor of the decomposition products.
Since the thermodynamic data of several of the chloroboroxines formed in the system B/O/CI are known, it is pos750
sible to discuss the relation between their stabilities and those
of the starting material BCI, and the reaction product B,O,
[in the sense of Equation (c)]. The standard enthalpies of
f ~ r m a t i o n ''*~ ' 261 of the compounds of interest here are
plotted in Figure 7 against the quotients of the number of
moles of B and 0 (as a measure of the composition of the
compound). It is clear that the relationship between composition and energy content is almost linear. This shows that
the bonding energies of the B-C1 bond in BCI, and of the
B-0 bond in B,O, are almost the same as the corresponding
bonding energies in the chloroboroxines. The replacement of
-700 I ,
0 6 0 8 10
Fig. 7. Graphical representation of the enthalpy content of BCI,, B,O,,
B,OCI,. B,O,CI, and B,O,CI,
each R C I bond by a B-0 bond corresponds to a stepwise
approach to the thermodynamically stable final product
B,O, without any marked change in the bonding situation.
Thus, from the thermodynamic point of view we can consider the formation of the chloroboroxines as initial steps towards the formation of B,O, . In this system the formation
of a highly reactive molecule, here O=B-Cl, must also be
considered as a possible first step; the oligomerization of this
intermediate can lead to the formation of the cyclic boroxi11es,[~~]
while its reaction with excess BCI, makes possible the
formation of B,OCI, and longer chain analogues. Structural
relationships between the molecular chloroboroxines and
B,O, are also evident, for example the tendency for the formation of chains and rings containing alternating 0 and B
atoms which is also a feature of the structure of solid B,O,.
3.3. The Reaction between POCl, and 0,
Our own investigations into the plasma-activated reaction
of PCI, with O2f2']have demonstrated the formation of a
Table 3 . Phosphorus oxychlorides observed in plasmachemically and thermally activated reactions.
Angew. Chem. Int. Ed. Engl. 30 (1991) 745-753
large variety of phosphorus oxide chlorides.[281The compounds detected in this reaction can also be obtained in the
condensed phase from the reaction between P,O,, and
POCI, as well as from other reactions. They have been characterized by various spectroscopic (NMR, IR, Raman) and
chromatographic techniques and in some cases structurally
characterized. All the molecular oxide chlorides of phosphorus presently known which contain more than one P atom in
the oxidation state + 5 are collected in Table 3. In analogy
to the chlorosiloxanes these can be divided into several homologous series:
series of reactions leading to the deposition of the solid may
well be the formation of PO,CI, which has been detected and
characterized both in the gas phase by mass spectrometryr5*’
and by means of matrix s p e c t r o ~ c o p y and
[ ~ ~ photoelectron
The formation of oligomers of this molecule (P,O,,CI,, n = 3-7) can be regarded as evidence for its
primary formation.
Group 1 : P,O,,_,CI, + (n = 2-7)
Group 2 : P,O,,CI, (n = 3-7)
Group 3: P,O,, + lCln-z(n = 4-8)
In Sections 3.1 -3.4 we have dealt separately with the
chemical processes involved in the reactions between oxygen
and the chlorides or oxide chlorides of silicon, germanium,
boron and phosphorus. However, in the course of the technological application of these reactions in the manufacture
of optical waveguides they will not occur in a spatially or
temporally separated manner but simultaneously. The question thus arises as to the manner in which the dopants are
incorporated into the SiO, glass occurs during the deposition of the solids are from the extremely complex gas phases
present. Two basically different processes must be considered :
1 . The oxides of the above elements are deposited in separate domains; the homogeneous distribution of the dopant substances occurs via diffusion processes.
2. The molecular “precursors” of the solids described above
already contain the dopants in the form of “alien atoms”.
It is of course not known whether these two possibilities
are of any practical importance. However, since this is a
question of scientific interest, we have carried out a series of
experiments in which the plasma-activated reactions of oxygen with the mixtures SiCI,/GeCI, and SiCI,/BCI, were
studied by mass spectrometry in order to clarify the situation
regarding the possibility of the incorporation of “alien
atoms” into the oxide halides of silicon. Since these gas
phases naturally have enormously complex compositions,
the analysis of the mass spectra obtained is difficult. It was
however possible to identify the molecules BSiOCI,,
BSi,O,Cl,, SiGeOCI,, Si,GeO,CI,, SiGe,O,CI, and
Si,GeO,Cl, unequivocally.121’This makes it clear that the
incorporation of the alien atoms can in principle occur in the
precursors of the solids. It can be assumed that both the
above-mentioned possibilities for dopant incorporation will
play a part in practice.
In contrast to the situation in the siloxane series, the composition of the compounds does not permit us to draw unequivocal conclusions as to the structure of the compounds.
Thus the members of Group 1 could be chain molecules, but
there are also other possibilities: thus P,O,CI, could also be
an oligocyclic compound derived from P,O,, by replacement of three terminal oxygen atoms by six chlorine atoms.
Here we must refer to the experimental studies cited in
Table 3. In analogy to the structural chemistry of the
metaphosphates (the compounds discussed here correspond
formally to the chlorides of the corresponding metaphosphates) we can expect the compounds in Group2 to be
monocyclic. However, oligocyclic compounds are also possible, at least in some cases. Thus while P,O,CI, could be an
eight-membered ring, a structure derived from P,O,, by replacement of two terminal oxygen atoms by four chlorine
atoms would also be in agreement with the structural chemistry of the phosphates. Group 3 must contain only oligocyclic compounds if we exclude structures in which a phosphorus atom is doubly bonded to two oxygen atoms. The
literature contains no information regarding the thermodynamic stability of any of the compounds listed in Table 3.
The most likely decomposition reaction involves the formation of P,O,, and POCI,, here demonstrated for the case of
P,O,,CI, [Equation (I)]:
3 P , O 1 , C I , ~ 2 . 5 P,O1,
+ 5 POCI,
Decomposition to give the binary components P,O and
PCI, (or PCI, and CI,) is not expected because of the extreme
thermodynamic stability of POCI, . The reaction enthalpy
for the process described by (I) is unknown and cannot readily be estimated. The reaction entropy should however be
clearly positive, so that we can assume that the decomposition products are thermodynamically favored. The compounds listed in Table 3 should thus be metastable. This is
441 in which the thermal instability
supported by reportstz8*
of these oxide chlorides of phosphorus is described.
As in the system Si/O/Cl, metastable compounds are thus
formed which can be regarded as intermediates in the reaction between POCI, and 0, (the final product of this reaction in the manufacture of optical waveguides is certainly not
molecular P,O,, but compounds in which silicon atoms in
SiO, are replaced by phosphorus, the excess positive charge
being compensated by the incorporation of boron atoms at
other positions in the lattice). The initial step in the long
A n ~ e w Chem.
Inr. Ed. Engl. 30 (1991) 745-753
3.4. Comments on the Incorporation of the Dopant
Materials into the SiO, Matrix
4. Summary and Future Prospects
The available information summarized above on the apparently very simple chemical reactions of elementary oxygen with the chlorides of silicon, germanium, boron and
phosphorus which occur in the course of the manufacture of
optical waveguides show clearly that the deposition of the
solid oxides is preceded by the formation of a large number
of oxide halides which can in several respects be regarded as
precursor molecules of the oxides:
1 . Their chemical compositions lie between those of the
chlorides and the oxides, and are (in the case of the system
SiCI,/O, studied) strongly temperature-dependent in the
sense that the compounds become steadily oxygen-richer
as the reaction proceeds (i.e. as the temperature increases).
2. The structural principles found in the oxides are already
preformed in the oxide chloride molecules.
3. The thermodynamic stability of the oxide chlorides lies
between those of the chlorides and the oxides.
4. Many of the oxide halides are formed even though they
are thermodynamically unstable with respect to decomposition into the chloride and oxide. This can be explained on the basis that their intermediate formation is
a necessary prerequisite for the construction of the solid
5. The primary formation of reactive, coordinatively unsaturated molecules (SiOCl,, GeOCI,, BOC1, P0,CI) is
probable in all the cases discussed.
On the basis of the still very limited information presently
available on the structures of the complex oxide chlorides it
is possible to develop mechanisms which describe the transition from molecules to solids. These, although unconfirmed,
appear reasonable in view of the establishment of the identities (and in many cases of the structures and stabilities) of the
intermediately formed oxide chlorides. However, the enormously complex nature of the reactions involved makes it
unlikely that it will become possible to obtain unequivocal
evidence for the mechanisms discussed.
All the reactions described here lead to the deposition of
oxides which are typical precursors of glasses. It should be
noted that the information presently available indicates that
the formation of polynuclear molecular oxide halides is limited to exactly these elements. This may be due to the fact
that the latter have been the subject of particularly detailed
studies, but it may also indicate that the prerequisites for the
formation of such compounds are similar to those occurring
during the formation of glasses (low coordination numbers,
apical linking of the polyhedra).
The structural relationships between molecular compounds and “related” solids is by no means limited to the
class of compounds described here. Thus for example we find
details of the corresponding solid structures in many molecular cluster compounds.r611A further impressive example is
provided by the pyrolytic formation from Si(CH,), of a variety of carbosilanes, in which details of the lattice of the
silicon carbide generated at high temperature are already
Detectable unstable intermediates in which the final product of the reaction is already recognizable are also formed in
other completely different reactions. Thus, Goubeau and
Keller have demonstrated that the careful hydrolysis of BC1,
leads to the formation of B,0,CI,,t631 i.e. exactly the same
compound which is formed during the reaction between
BCI, and 0, in the gas phase at high temperatures. The
formation of the siloxanes Si,OCI,, Si,O,CI,, Si,O,CI,,,
Si,O,Cl,, and Si,O,Cl
was observed under analogous
conditions during the hydrolysis of SiCI, and Si,C16 respectively.r64-661It is thus clear that reactions can occur in a very
similar manner even when they are carried out under such
extremely different conditions. From this point of view it
appears desirable to study the processes occurring during the
hydrolysis of halides under the conditions described here as
regards precursors (intermediates).
Numerous technical processes based on high-temperature
reactions or on hydrolyses are carried out (successfully!)
even though nothing is known about the course of the reactions involved. Monographs on industrial chemistry thus
offer a rich source of impulses for scientific studies on the
manner in which such reactions take place.
We thank the Deutsche Forschungsgemeinschaft and the
Fonds der Chemischen Industrie for their support of our investigations. We also thank KabelmetaN Elektro, Stuttgart, for
making available the photographs showing industrialprocesses
and Dr. M . Wittmann of this company for valuable discussions
and suggestions.
Received: December 3, 1990 [A 821 IE]
German version: Angew. Chem. 103 (1991) 762
Translated by Professor Z N. Mitchell, Dortmund (FRG)
[l] Glass fibers used for the long distance optical data transmission are referred to as optical waveguides, while those used for lighting purposes are
known as light-conducting fibers.
[2] P. Geittner, Phys. Unserer Zeit 19 (1988) 37.
[3] BCI, has been replaced increasingly by Freon in recent years.
[4] M. W Chase, C. A. Davies, J. R. Downey, D. J. Frurip, R. A. McDonald,
A. N. Syverud, (JANAF Thermochemical Tables) Phys. Chem. ReJ Data
14 (1985).
[5] I. Barin: Thermochemical Data of Pure Substances, VCH, Weinheim
[6] C . Friedel, A. Ladenburg, Ann. Chem. Pharm. 147 (1868) 355.
[7] L. Troost, P. Hautefeuille, Bull. Soc. Chim. Fr. 16 (1871) 240.
(81 L. Troost, P. Hautefeuille, Ann. Chim. Phys. 7 (1887) 452.
[9] H. Reinboldt, W. Wisfeld, Justus Liebigs Ann. Chem. 517 (1935) 198.
[lo] W. C. Schumb, D. F. Holloway, J. Am. Chem. Soc. 63 (1941) 2753.
[ l l ] H. C . Marsmann, E. Meyer, Z . Anorg. Allg. Chem. 548 (1987) 193.
[12] A. Kornick, M. Binnewies, Z . Anorg. Allg. Chem. 587 (1990) 157.
(131 A. Kornick, M. Binnewies, Z . Anorg. Allg. Chem. 587 (1990) 167.
[14] The relationships between the composition and the structure of the
chlorosiloxanes described here are valid only if no Si-0 double bonds are
[15] W. Arey, C. Glidewell, A. G. Robiette, G. M. Sheldrick, J. Mol. Struct. 8
(1971) 413.
1161 D. Fenske, M. Jerzembeck, M. Binnewies, crystal structure analysis of
Si,O,CI,, unpublished results.
H. Schnockel, Z . Anorg. Allg. Chem. 460 (1980) 37.
K. B. McAfee, R. A. Laudise, R. S. Hozack, J. Lightwave Techno/.4 (1983)
R. Schwarz, P. W. Schenk, H. Giese, Chem. Ber. 64 (1931) 362.
P. Kleinert, D. Schmidt, H. J. Laukner, Z . Anorg. Allg. Chem. 495 (1982)
A. Kornick, M. Binnewies, unpublished results.
J. Blauer, M. Farber, 1 Chem. Phys. 39 (1963) 158.
R. F. Porter, S. K. Gupta, J. Phys. Chem. 68 (1964) 280.
M. Farber, Trans. Faraday Soc. 60 (1964) 301.
D. J. Knowles, A. S. Buchanan, Inorg. Chem. 4 (1965) 1799.
M. Binnewies, 2. Anorg. Allg. Chem. 571 (1989) 7.
The reaction between PCI, and 0, results in the formation of high concentrations of P(O)CI, (281. The products obtained from the reaction mixture
PCI,, P(O)CI,, 0, should thus correspond to those of the reaction between
P(O)CI, and 0,.
H. Bange, M. Binnewies, unpublished results.
G.Miiller-Schiedmayer, H. Harnisch, Z. Anorg. Allg. Chem. 333 (1964)
R. Klement, K. H. Wolf, Z. Anorg. Allg. Chem. 282 (1955) 149.
A. Geuther, A. Michaelis, Ber. Dtsch. Chem. Ces. 4 (1871) 766.
R. Klement, 0. Koch, K. H. Wolf, Naturwissenschaften 41 (1954) 139.
R. Klement, E. Rother, Naturwissenschaften 45 (1958) 489.
C. Oddo, Gazz. Chim. Ital. 29 (1899) 330.
P. C. Crofts, I. M. Downie, R. B. Heslop, J. Chem. Soc. (1960) 3673.
H. Grunze, Z . Anorg. Allg. Chem. 296 (1958) 63.
G. N. Huntly, J. Chem. SOC.59 (1891) 202.
L. C . D. Groenweghe, J. H. Payne, J. R. Van Wazer, J. Am. Chem. Sot. 82
(1960) 5305.
W. E. Morgan, T. Glonek, J. R. Van Wazer, Inorg. Chem. 13 (1974) 1832.
H. Grunze, Z. Anorg. Allg. Chem. 298 (1959) 152.
H. Grunze. Z . Anorg. Allg. Chem. 324 (1963) 1.
Angew. Chem. Int. Ed. Engl. 30 (1991) 745-753
[42] H. Grunze. Z. Chem. 2 (1962) 313.
[43] M. Becke-Goehring, J. Sambeth, Angew. Chem. 69 (1957) 640.
[44] M. Viscontini, K. Ehrhardt in: Silicium, Schwefel, Phosphate, (Colloq.
Sekt. Anorg. Chem. Int. Union Reine Angew. Chemie, Miinster 1954)
Verlag Chemie, Weinheim 1955, p. 232.
[45] E. Fluck, Angew. Chem. 72 (1960) 752.
(461 D. F. Toy, J. E. Blanche, US 3034826 (1962).
[47] R. Appel, G. Eisenhauer, Chem. Ber. 95 (1962) 1756.
[48] K. H. Wolf, H. Grunze, Z. Chem. 22 (1982) 211.
[49] A. Besson. C . R. Hebd. Seances Acad. Sci. 124 (1897) 763.
[50] M. Baudler, R. Klement, E. Rother, Chem. Ber. 93 (1960) 149.
[51] E. A. Robinson, Can. J. Chem. 40 (1962) 1725.
[52] M. Becke-Goehring, A. Debo, E. Fluck, Chem. Ber. 94 (1961) 1383.
[53] E. Fluck, Chem. Ber. 94 (1961) 1388.
[54] R. A. Y. Jones, A. R. Katritzky, Angew. Chem. 74 (1962) 60; Angew.
Chem. Int. Ed. Engl. l(1962) 32.
[55] G. S. Reddy, C. D. Weis, .
Org. Chem. 28 (1963) 1822.
Angew. Chem. Int. Ed. Engl. 30 (1991) 745-753
[56] H. Gerding, H. Gijben, B. Nieuwenhuijse, J. G. Van Raaphorst, Recl. Trav.
Chim. Pays-Bas 79 (1960) 41.
[57] E. A. Kravcenko, V. G. Morgunov, B. N. Kulikovsij, T. L. Novoderezkina, M. Meisel, 2. Chem. 23 (1983) 143.
[58] M. Binnewies, 2. Anorg. Allg. Chem. 507 (1983) 77.
[59] R. Ahlrichs, C. Ehrhardt, M. Lakenbrink, S. Schunck, H. Schnockel,
J. Am. Chem. SOC.108 (1986) 3596.
[60]M. Meisel, H. Bock, B. Solouki, M. Kremer, Angew. Chem. 101 (1989)
1378; Angew. Chem. Int. Ed. Engl. 28 (1989) 1381.
[61] S. C. Lee, R. H. Holm, Angew. Chem. 102 (1990) 868; Angew. Chem. Int.
Ed. Engl. 29 (1990) 885.
(621 G. Fritz, Angew. Chem. 99 (1987) 1150; Angew. Chem. Int. Ed. Engl. 26
(1987) 1111.
[63] J. Goubeau, H. Keller, Z. Anorg. Allg. Chem. 265 (1951) 73.
[64] J. Goubeau, R. Warncke, Z. Anorg. Allg. Chem. 259 (1949) 109.
[65] W. C. Schumb, A. J. Stevens, 1 Am. Chem. SOC.69 (1947) 726.
(661 W. C. Schumb, R. A. Lefever, 1 Am. Chem. SOC.76 (1954) 2091.
Без категории
Размер файла
1 384 Кб
data, chemical, reaction, optical, transmission, long, waveguides, manufacture, distance
Пожаловаться на содержимое документа