Reaction of Germanium Tetrachloride with Oxygen Under MCVD Fiber Preform Fabrication Conditions M

ISSN 0020-1685, Inorganic Materials, 2007, Vol. 43, No. 9, pp. 968Ð971. © Pleiades Publishing, Inc., 2007. Original Russian Text © M.M. Bubnov, A.N. Gur’yanov, M.Yu. Salganskii, V.F. Khopin, 2007, published in Neorganicheskie Materialy, 2007, Vol. 43, No. 9, pp. 1081Ð1085.

Reaction of Germanium Tetrachloride with Oxygen under MCVD Fiber Preform Fabrication Conditions M. M. Bubnova, A. N. Gur’yanovb, M. Yu. Salganskiib, and V. F. Khopinb a Fiber Optics Research Center, Russian Academy of Sciences, ul. Vavilova 38, Moscow, 119333 Russia b Institute of Chemistry of High-Purity Substances, Russian Academy of Sciences, ul. Tropinina 49, Nizhni Novgorod, 603950 Russia e-mail: [email protected] Received December 18, 2006

Abstract—The GeO2 yield in the reaction between GeCl4 and oxygen has been determined as a function of the reaction time under typical MCVD ﬁber preform fabrication conditions. It is shown that the yield increases steadily over time and may attain 100%. In the case of the cooxidation of germanium and silicon tetrachlorides under the same conditions, there is an optimal reaction time corresponding to a maximum in GeO2 yield. The temperature proﬁle along the reaction zone has been optimized in terms of germania yield. DOI: 10.1134/S0020168507090105

INTRODUCTION At the same time, when reactions (1) and (2) are run Chemical vapor deposition on the inside surface of concurrently, as is typical in the MCVD process, the GeO ~0.5 a silica glass substrate tube (MCVD process) is among 2 yield is even at reaction times close to 1 s the most widely used vapor-phase processes for the fab- [3], which seems to be associated with the small equi- rication of single-mode silica-based optical fiber pre- librium constant of reaction (2) [1, 3]. The considerable forms. Typically, the process utilizes high-temperature volume of additional molecular chlorine resulting from reactions of silicon tetrachloride and dopant halide reaction (1) must reduce the GeO2 yield, especially because, in the MCVD process, the silicon tetrachlo- (most frequently, GeCl4) vapors with oxygen. A gas mixture is passed through a rotating silica glass tube, ride concentration in the gas mixture, as a rule, notably and an oxygenÐhydrogen flame slowly traveling along exceeds the germanium tetrachloride concentration. the tube (usually, in the gas flow direction) initiates the High-purity germanium tetrachloride is among the reactions most expensive starting chemicals in the MCVD process. Clearly, a detailed understanding of the processes SiCl4 + O2 SiO2 + 2Cl2, (1) underlying the deposition of fiber preform material is of GeCl + O GeO + 2Cl . (2) practical interest for the fabrication of both conven- 4 2 2 2 tional germanium-doped silica glass fiber preforms and The resulting oxide particles are deposited on the inside germania-glass-core preforms. tube surface under the action of thermophoretic forces and The purpose of this work was to study the effect of are fused by the flame to form a transparent glass layer. reaction time on the germania yield in the reaction of The effectiveness of the reactions in the gas flow can germanium tetrachloride with oxygen under typical be characterized by the reaction time τ and conversion MCVD fiber preform fabrication conditions. β. The latter is the ratio of the number of moles of a component in the reaction product to the initial number of moles of that component: EXPERIMENTAL AND RESULTS Optimization of the reaction temperature. Our N Ð N β = ------1 -2, (3) experiments were conducted in a standard MCVD N1 apparatus for fiber preform fabrication, equipped with a system for forced cooling of the tube in the thermo- where N1 and N2 are the numbers of moles of the start- phoretic deposition zone [4]. ing substance at the inlet and outlet of the reaction zone, First, we optimized the GeCl4 conversion tempera- respectively. ture. To this end, a gas mixture (table) was passed At typical MCVD temperatures, full SiCl4 conver- through the hot zone produced inside the tube by a sion to silica takes on the order of 0.01 s [1], whereas torch (the maximum temperature of the outer tube sur- the GeCl4 conversion is 0.31 at this reaction time [1] face was measured by an IR pyrometer with an accu- and takes 0.5 s to reach unity [2]. racy of ±15°ë).

968

REACTION OF GERMANIUM TETRACHLORIDE 969

In our experiments, germanium tetrachloride was Flow rates of gas-mixture components (normal conditions) oxidized at different temperatures of the outer tube sur- 3 face in the reaction zone, both individually (Fig. 1, Curve no. Flow rate, cm /min curve 1) and in the presence of silicon tetrachloride in Fig. 1 oxygen GeCl SiCl CCl (curve 2) or carbon tetrachloride (curve 3). The germa- 4 4 4 nium tetrachloride concentration in the gas mixture at 1 1325 11.9 Ð Ð the tube outlet end was determined by IR spectroscopy. 2 1325 11.9 39.7 Ð In all of the experiments, the germanium tetrachlo- 3 1325 11.9 Ð 16.9 ride conversion increased with increasing temperature in the reaction zone. The highest conversion was achieved when the temperature of the outer tube surface which reacts with oxygen at a considerably lower tem- in the reaction zone was ~2000–2050°C. At this temperature in comparison with germanium tetrachloride, perature, the GeCl4 conversion in reaction (2) attained must raise the excess chlorine concentration in each 80% (Fig. 1, curve 1). If reactions (1) and (2) were run section of the reaction zone compared to SiCl4, which concurrently, the GeCl4 conversion was 60% (curve 2). seems to be responsible for the lower germanium tetra- The use of carbon tetrachloride instead of silicon tetrachloride conversion. chloride reduced the GeCl4 conversion to ~30% (the If the above reasoning is correct, increasing the initial carbon tetrachloride concentration in the gas reaction time for concurrent reactions (1) and (2) must mixture was lower than the SiCl4 concentration by reduce germanium tetrachloride conversion. more than a factor of 2). Effect of reaction time on GeCl4 conversion. The It follows from the kinetic curves of reactions (1) average reaction time in a flowing gas mixture can be and (2) [5] that the temperature range of reaction (1) is determined as ~100°ë wider than that of reaction (2). The tempera- Sl ture profile created by the traveling torch in the reaction τ = -----, (4) zone (Fig. 2) is asymmetric along the tube axis, with a W flat trailing edge and steep leading edge. Therefore, for a laminar gas-mixture flow and at moderate tempera- where S is the cross-sectional area of the channel, W is tures, the extent of reaction (2) in each section of a sig- the volumetric flow rate of the gas mixture, and l is the nificant part of the reaction zone is a factor of 3Ð effective length of the reaction zone (hot zone). How- 5 greater than that of reaction (1). For this reason, the ever, since only the temperature of the outer tube sur- excess chlorine concentration in the gas mixture in each face in the reaction zone was measured in our experi- section of the reaction zone must be substantially lower ments, the true size of the hot zone inside the tube was than might be expected. Replacing SiCl with CCl , difficult to determine. In connection with this, as a reac- 4 4 tion time criterion we used the flow speed along the tube, W/S. To assess the efficiency of germanium tetrachloride conversion in reaction (2) as a function of reaction time, 0.8 a layer of SiO2 particles was deposited on the inside tube surface by translating the flame in the direction opposite to the flow of a mixture of oxygen and silicon 1 3 0.6 tetrachloride (SiCl4 flow rate, 100 cm /min) (Fig. 3a).

2 0.4 conversion

4 2000 C GeCl ° 1500 0.2 3

1000 0 Temperature, Temperature, 500 Gas-mixture ﬂow Torch travel 12001600 2000 Temperature, °C 0 10 20 30 40 Length, cm Fig. 1. Germanium tetrachloride conversion as a function of the outer-surface temperature in the reaction zone for differ- Fig. 2. Axial temperature proﬁle produced by the traveling ent gas mixtures: (1Ð3) see the table. torch.

INORGANIC MATERIALS Vol. 43 No. 9 2007

970 BUBNOV et al.

Layer of SiO particles 2 Silica tube

(a) SiCl4 O2

Torch Cooling

Fused SiO2/GeO2 layer

(b) GeCl4 O2

Fig. 3. Schematic illustrating the preparation of germanosilicate glass layers via successive oxide deposition from the vapor phase: (a) the flame is translated in the direction opposite to the flow of a mixture of oxygen and silicon tetrachloride; (b) the flame is translated along the flow of a mixture of oxygen and germanium tetrachloride.

The zone behind the flame, where particles were depos- The arrow in Fig. 4 marks the data point obtained at ited, was cooled in order to create conditions similar to the same flow speed as curve 1 in Fig. 1. Since this those produced by the flame traveling in the flow direc- point corresponds to 80% germanium tetrachloride tion. After deposition, the layer of SiO2 particles was conversion, further reduction in flow speed under such fused in a mixture of oxygen and germanium tetrachlo- deposition conditions makes it possible to achieve a ride at ~2050°ë by translating the flame in the flow near 100% GeO2 yield. direction to give a transparent glass layer (Fig. 3b). In In codepositing oxides by reactions (1) and (2), we our experiments, the flow speed of this mixture was used gas mixtures with a constant O2 : SiCl4 : GeCl4 varied by changing the inner tube diameter and oxygen volume ratio of 250 : 10 : 3. The flow speed was varied flow rate, with no changes in GeCl4 flow rate. In this by changing the inner tube diameter or volumetric flow manner, in each experiment we deposited and fused rate in such a manner that the mixture composition several germanosilicate glass layers under identical remained unchanged since the germania yield in oxide conditions. codeposition depends on the oxygen concentration in After collapsing the tube into a rod, we measured the gas mixture [3]. The temperature of the outer tube the difference in refractive index (∆n) between silica surface in the heating zone was maintained at 2050°ë. glass and the deposited material using a Photon Kinet- After deposition, the tube was collapsed into a rod, as ics 2600 preform analyzer. The conversion efficiency above, and conversion was inferred from the refractive was inferred from the variation of ∆n with preform fab- index of the deposited germanosilicate glass. rication conditions. Figure 5 plots ∆n against flow speed for glass deposited at a constant gas mixture composition. As seen, the Figure 4 plots ∆n against the inverse of the flow germanium tetrachloride conversion first rises with speed relative to the reaction zone for silica glass fused increasing flow speed (decreasing reaction time), as into a transparent layer in a flow of germanium tetra- would be expected from the data in Fig. 1. At the same chloride (30 cm3/min) and oxygen. These data indicate time, the germanium tetrachloride conversion as a func- that the GeO2 yield (GeCl4 conversion) increases with tion of flow speed has a maximum (Fig. 5). At lower decreasing flow speed. flow speeds, the germania yield is limited by the equi-

INORGANIC MATERIALS Vol. 43 No. 9 2007 REACTION OF GERMANIUM TETRACHLORIDE 971

∆n point corresponds to 60% germanium tetrachloride conversion, the maximum conversion under these con- 0.026 ditions can be estimated at ~70%.

0.024 CONCLUSIONS The germania yield in the reaction between germanium tetrachloride and oxygen was determined as a 0.022 function of the flow speed of the gas mixture under typical MCVD silica fiber preform fabrication conditions. The germania yield in the reaction between germanium 0.020 tetrachloride and oxygen was shown to significantly depend on the reaction time. The GeO2 yield increases with reaction time and may attain 100%, which means that the germanium tetrachloride conversion under such 0.018 conditions is limited only kinetically. In the case of the cooxidation of germanium and sil- 1020 30 40 50 icon tetrachlorides in the CVD process, there is an opti- Inverse flow speed, s/m mal reaction time corresponding to a maximum in GeO2 yield. Under such conditions, the germanium tet- Fig. 4. Plot of ∆n vs. inverse flow speed for silica glass fused rachloride conversion is limited both by the equilibrium in a flow of oxygen and germanium tetrachloride. of the reaction between GeCl4 and oxygen in the presence of excess chlorine and by the kinetics of this reaction. ∆n Our results on silicaÐgermania codeposition indicate that the optimal temperature profile along the reac- 0.014 tion zone is the one which ensures the maximum germanium tetrachloride conversion and the minimum silicon tetrachloride conversion over most of the reaction zone. SiCl4 conversion must be completed at the down- 0.013 stream end of the heating zone, where the temperature is higher compared to the rest of the zone. The parame- ters of the temperature profile along the reaction zone depend on the inner diameter of the silica glass tube and 0.012 the flow rate of the gas mixture.

REFERENCES 0.011 1. Kleinert, P., Kirchhof, J., and Schmidt, D., Principles of the MCVD-Process (Formation of High Silica Glasses from the Gaseous Phase), Proc. 5th Int. School of Coher- ent Optics, Jena, 1984, pp. 42Ð49. 0.010 2. French, W.G. and Pace, L.J., Chemical Kinetics of the 0.10.2 0.3 Reaction of SiCl4, SiBr4, GeCl4, POCl3, and BCl3 with Flow speed, m/s Oxygen, J. Phys. Chem., 1978, vol. 82, pp. 2191Ð2194. ∆ 3. Wood, D.L., Walker, K.L., MacChesney, J.B., et al., Ger- Fig. 5. Plot of n vs. flow speed for glass deposited by reac- manium Chemistry in the MCVD Process for Optical tions (1) and (2) at a constant gas mixture composition. Fiber Fabrication, J. Lightwave Technol. 1987, vol. LT-5, no. 2, pp. 277Ð285. librium of the reaction between germanium tetrachlo- 4. Khopin, V.F., Umnikov, A.A., Gur’yanov, A.N., et al., ride and oxygen in the presence of excess molecular Doping of Optical Fiber Preforms via Porous Silica Layer Infiltration with Salt Solutions, Neorg. Mater., chlorine, resulting from the concurrent oxidation of sil- 2005, vol. 41, no. 3, pp. 363Ð368 [Inorg. Mater. (Engl. icon tetrachloride. At higher flow speeds, the kinetic Transl.), vol. 41, no. 3, pp. 303Ð307]. limitations of reaction (2) prevail. 5. French, W.G. and Pace, L.J., Chemical Kinetics of the The arrow in Fig. 5 marks the data point obtained at Modified Chemical Vapor Deposition Process, Proc. the same flow speed as curve 2 in Fig. 1. Since this IOOC`77, 1977, paper C1-2, pp. 379Ð382.

INORGANIC MATERIALS Vol. 43 No. 9 2007