The viscous behavior of fluorine-doped synthetic silica is studied using collapsing experiments with different fluorine-doped tubes on a modified chemical vapor deposition (MCVD) lathe. The principles, techniques, and evaluations of this method are the same as the ones demonstrated previously in detail with pure and doped silica. The present investigations provide information about the influence of fluorine doping up to a concentration of about 10 mol% F (3.4 wt% F) in a temperature range between 1600°C and 2000°C. Fluorine doping leads to a systematic decrease in the viscosity, combined with a decrease of the activation energy of the viscous flow and a certain increase of the pre-exponential factor. In summary, this demonstrates the weakening influence of fluorine on the glass network, similar to the incorporation of hydroxyl or chlorine.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
We have recently shown that the viscosity of silica tubes can be determined using collapsing experiments on a MCVD lathe . This method applies to the high temperature region (i.e., for viscosity values between 107 Pa∙s and 104.5 Pa∙s) and has the advantage that is does not require a crucible and, therefore, prevents the contamination of the glass melt, which is a serious problem at the high temperatures of silica glass processing. Therefore, this method can be used as an advantageous addition to classic methods of viscosity measurement [2–6]. For the exact determination of real viscosity values, however, a set of experiments with tubes of different wall thicknesses is necessary. This requirement is due to the fact that a thermal effect leads to a deviation in the measured temperature at the tube surface and the temperature within the tube bulk, which is responsible for the measured viscosity value. It was shown that the reason for this thermal effect is the radiation loss in regions where the quartz glass is semitransparent (i.e., in a wavelength range between 2 μm and 4.5 μm) . It was possible to be shown that the extrapolation to a wall thickness of zero is a well-suited procedure for derivation of correct viscosity-temperature relationships . Even if the demonstrated method seems to be more difficult than classic methods (e.g., rotating crucible and falling ball), it yields process-related data and – beyond that– valuable information regarding the temperature in the tube interior, which cannot be directly measured but is very important for preform and fiber technology.
Here, we use this method in order to measure the temperature-dependent viscosity of a series of synthetic silica tubes with different fluorine concentrations between 0 wt% and 3.4 wt% fluorine. The dependence of the viscosity on the fluorine content is derived from the results, which include the earlier findings for both fluorine-free silica and silica with a high fluorine content (i.e., F300). The underlying theory of this method and the total procedure is completely identical with that described in detail in . Therefore, in the following we provide only a brief outline of the procedure and evaluation and mainly concentrate on the discussion of the results and on a comparison with data from the literature.
For many years, the fluorine doping of pure silica glass has played an important role in optical fiber and waveguide technology. Historically, the interest in fluorine doping emerged because fluorine gives rise to a refractive index depression in silica glass , unlike all other dopants with the exception of boron. Subsequently, it has been shown that fluorine also influences other important glass parameters such as viscosity, density, transformation temperature, atomic defect content, and reactivity [8–13]. Therefore, it is applied in order to tailor optimized fiber and waveguide properties by doping both the fiber core and cladding. Important progress has been achieved through the use of fluorine-doped carrier tubes and glass materials, especially in the field of microstructured fibers [14–16]. It has been demonstrated that the special influence of fluorine on the atomic structure of quartz glass can be used to produce optical fibers for the far UV application at the wavelength around 154 nm .
For the experiments both pure silica and different types of fluorine-doped silica were used. All glass tubes were provided by Heraeus Quarzglas GmbH [17,18]. The synthetic silica tube material F300 was produced via the flame hydrolysis of SiCl4 as a porous body, which was then dehydrated using gaseous chlorine. The content of metallic impurities (a few tens of ppb) and OH groups (below 1 ppm) is very low, but it contains about 0.16 wt% Cl. The other types are fluorine-doped synthetic silica, which also have similarly low metallic impurities and OH groups but an increasing fluorine content up to 3 wt% F. The chlorine contents are distinct and vary between very low (for the F320-14 glass) and a maximum of 0.2 wt% Cl (for the F325 glass). A correlation between the fluorine and chlorine amounts does not exit. The mean concentrations of fluorine and chlorine are displayed in Table 1. For a better compatibility of the activity of fluorine and chlorine, we state their concentrations cF and cCl in mol% at the following ratios (valid for low concentrations)Eqs. (1) and (2). They refer to a molar formula of
Exact radial concentration profiles of fluorine and chlorine were measured on thin, polished, and carbon-coated slices using wavelength-dispersive electron probe microanalysis (WD-EPMA) at intervals of 50 μm with an accuracy of about 10 relative percent . The radial profile’s shape of the tubes used (between inner radius ri and outer radius ro) are displayed in Figs. 1(a) and 1(b). They are not always constant but show in part pronounced gradients and fluctuations.
Collapsing experiments were carried out with a series of tubes of different sizes, an outer diameter between 8 mm and 33 mm, a wall thickness between 0.6 mm and 6 mm, and a uniform length of 500 mm, rotating on a MCVD lathe at 90 rotations per minute. The outer diameter of the tubes before and after the burner pass was automatically recorded with a laser scanner at several axial and angular positions (Laser-Scan-Micrometer LS-3101 (Keyence), a resolution of 0.1 µm, 400 scans/s). The accuracy of the diameter determination is 0.002 mm. The inner diameter was calculated from the outer diameter, the weight of the tube, and the density of the glasses . The oxy-hydrogen burner was operated with gas flows between 10 slm and 50 slm (“standard liter per minute”) of hydrogen and oxygen, respectively, in order to implement different temperatures between 1600°C and 2000°C. Depending on the temperature and tube dimensions, the burner velocity was set up to values of between 0.3 cm·min−1 and 4 cm·min−1 in order to carry out a small collapsing effect of ≤5% of the tube diameter. The temperature of the tube at the outer surface was measured in the wavelength region of 4.8 μm … 5.6 μm with an infrared thermometer, which could be automatically shifted relative to the burner position to record the axial temperature profile. The inner pressure in the tube was set and regulated by a gas system without gas flow through the tube at an accuracy of about 2 Pa.
In order to implement a defined temperature, two different operating regimes of the burner were used. The first regime (“hydrogen excess”) works with a H2:O2 ratio of about 4-5:1 and the second regime (“oxygen excess”) with a ratio of roughly 1. The absolute oxygen flow ranges in the “hydrogen excess” regime between 5 slm and 15 slm and in the “oxygen excess” regime between 30 slm and 50 slm (i.e., much higher).
In each case, two successive experiments were carried out under identical conditions (tube dimensions, burner gas amounts, temperature, burner velocity, etc.) with the exception of a change in the inner tube pressure pi relative to the external atmospheric pressure pa. The value of pi – pa was set to zero in the first experiment, which led to a diameter decrease via collapsing. In the subsequent experiment, it was set to a positive value (pi > pa) in order to create a blow up in the tubes (i.e., an increase in the tube diameter), which results – approximately – in the starting tube dimensions. With the measured change in the tube diameter, the inner tube pressure, the burner velocity and the width of the temperature profile, the viscosity at the temperature maximum of the axial profile η can be determined using the Eqs. (5), (6) and (8) given in .
The evaluation procedure and detailed results will be demonstrated in the following on a typical example that is selected. The summary of the results for all materials is then presented in the next paragraph.
In Fig. 2, the determined viscosity values are displayed depending on the reciprocal temperature as an example of the tube material F520-28 (see Table 1), using three tubes of different dimensions. The error of lnη (i.e the error of the single measurement) was calculated on the basis of the uncertainty of the measured quantities which enter the Eqs. (5), (6) and (8) given in , considering the error propagation. It is primarily determined by uncertainties and small fluctuations in temperature, profile width, and pressure; other parameters such as tube dimensions and burner velocity could be specified with high accuracy.
The Arrhenius plots lnη against 1/T yield straight lines with a nearly identical slope so that the activation energy EA of the viscous flow can be derived. The viscosity values determined in this manner are virtually dependent on the wall thickness of the tubes. Even, if the slope is nearly independent of the dimensions, the pre-exponential factor lnη0* changes with varying wall thickness w. This phenomenon is due to the fact that the radiation loss in regions where the quartz glass is semitransparent (i.e., in a wavelength range between 2 μm and 4.5 μm) leads to a deviation between the measured temperature at the tube surface and the temperature inside the tube, the latter being responsible for the measured viscosity value. The temperature difference depends on the wall thickness and vanishes, if the wall thickness becomes small. The true viscosity-temperature relationship can therefore be simply derived by extrapolation of lnη0*(w) to a wall thickness w of zero, as shown in Fig. 3, with a nearly linear relationship
In the result, the “true” viscosity relationship for our example F520-28 is shown as1].
The experiments with “hydrogen excess” flames and “oxygen excess” flames yield completely agreeable results. In addition to η, a quantity σ* can be derived from the experiments, which is mainly determined by the surface tension of the tube glass but still contains a certain impact of the flame pressure, which is remarkable for “oxygen excess” flames only. This value is shown in Fig. 4 for the same experiments with F520-28. The value σ* shows only a small dependence on the tube dimensions. A remarkable influence of the temperature was not observed. In a similar approach to η, σ* can be linearly extrapolated to a wall thickness of zero, which is a procedure that should provide for a surface tension σ of F520-28 with σ = 0.39 ± 0.02 N∙m−1in this case. For further details we refer to  as well.
The principle behavior of our example F520-28, shown in Figs. 2-4, is typical for all investigated materials. Table 1 provides a compilation of the results, together with the composition of the tube materials.
In the fourth column, the adjustment parameter y for the determination of the pre-exponential factor is indicated. It is similar within certain limits for all materials but shows some variations and a significant decrease towards high fluorine values (noticeable for F520-38 and Fluosil).
The fifth and sixth columns show the “true” viscosity relationships by the pre-exponential factor lnη0 and the activation energy EA. The standard error of the linear regression for the calculation of EA is about ± 14 kJ∙mol−1. The error of lnη0 is ± 0.6. It follows from the “tilting” of the Arrhenius lines in consequence of the error of EA plus the individual error of lnη ( ± 0.1), which however is only a small correction. In reality, all calculated errors have a certain variation in dependence on the material, but the specified values are well representative within ± 15%. Bear in mind that the errors of E/R and lnη0 are interdependent in effect on the viscosity, corresponding with the “tilting” of the Arrhenius lines. This is the result of the limited temperature range of the investigations. A change in the activation energy is largely compensated by a change in lnη0.
On the basis of the determined data in Table 1, Arrhenius plots, the activation energy and pre-exponential factor are shown in Figs. 5, 6 and 7, respectively, depending on the fluorine content of the silica materials.
In addition to the viscosity values, the extrapolated values of the surface tension σ are displayed in the seventh column of Table 1. Within the error limit of about ± 0.02 N∙m−1, the value is the same for all materials.
The parameter y, which describes the wall thickness extrapolation of the pre-exponential factor of the viscosity function, shows an apparent decrease in the high fluorine content of the tubes. This effect has been discussed in  on the occasion of the comparison of F300 and Fluosil. Without going into more detail, we only realize that it is a consequence of the decreasing activation energy. It does not point to a different but rather to a similar thermal behavior in the different doped materials (i.e. the same temperature difference leads to a larger viscosity effect in the case of a higher activation energy). The similar thermal behavior can be understood because the critical wavelength region for the thermal radiation is between 2 µm and 5 μm, where a remarkable effect of fluorine doping is not considered in the absorption and emission spectrum [9,21]. Note that the y value additionally depends to some extent from the temperature of investigation . This explains certain virtually unsystematic fluctuations, particularly noticeable in the case of F325.
The values of σ are fully identical for all materials (i.e. the fluorine seems to be without effect on the surface tension of the glass). In fact, we must treat this conclusion with caution. The thermal treatment of the tubes before and during the collapsing process leads to a fluorine loss at least in a small layer at the outer and inner surface. This layer cannot be clearly resolved in the graphs of Fig. 1 and is too thin to influence the viscosity measurements . For the surface tension however, a very thin outer layer of practically pure silica could lead to the results observed.
Although it has long been known that fluorine doping reduces the viscosity of silica , detailed results in a wide temperature range have only been presented by M. Ohashi, and M. Tateda et al. [23–26], to the best of our knowledge. Additional results have also been presented by M. Kyoto et al. in a low temperature range between 1000°C and 1300°C , which however does not cover the field of our own results. The intention of the authors [23–26] was to achieve a reduction of Rayleigh scattering losses in optical fibers to extremely high transparency by means of a viscosity adjustment of core and cladding glasses. They used the drawing method in the temperature range between 1800°C and 2200°C , as well as the penetration method and preform elongation method at lower temperatures between 1100°C and 1500°C for an estimation of the glass viscosity depending on the fluorine content. They investigated different samples, fabricated by vapor phase axial deposition (VAD) , with fluorine contents between 0 wt% F and 2.7 wt% F (i.e. in our nomenclature between 0 mol% F and 8.6 mol% F). Unfortunately, their reported results differ strongly from ours. The viscosity values for 2000°C, as an example, range between 107.7 Pa∙s … 106.3 Pa∙s with increasing fluorine content . Our results are nearly three orders of magnitude lower! The activation energies reported in  are between 390 kJ∙mol −1 and 320 kJ∙mol −1, in contrast to our results which were between 570 kJ∙mol −1 and 400 kJ∙mol −1. Moreover, the viscosity data, which are provided in  and  for temperatures of about 1400°C, are not very consistent. If we consider additional data from  in the temperature region between 1000°C and 1300°C, we find great discrepancies between [23,24] and . Even if the viscosity of fluorine-doped silica is a new research area, the viscosity of pure silica has been investigated by a variety of authors. It depends on the nature of the silica according to the nomenclature of Brückner [3,31]. Without going into detail, we can state that based on the comprehensive compilation of Nascimento , the values of the silica viscosity at 2000°C, reported by several authors over a few decades, range between 103.5 Pa∙s and 105 Pa∙s, with activation energies between 450 kJ∙mol −1 and 600 kJ∙mol −1. This is in strong contrast to the values, reported in [23,24], but is in agreement on the whole with our results, derived for the undoped silica samples (see Table 1 and Fig. 6). The viscosity of quartz glass is definitely determined by the impurity content (at least in the high temperature region around 2000K) . In principle, the VAD method used for the fabrication in [23,24] yields high purity samples, comparable with our glass materials. In our opinion, the preparation technologies cannot be responsible for different viscosity results. So we cannot explain the deviation of the viscosity data from the literature data and cannot consider the data in [23,24] for an exact comparison with our results.
Our investigations have shown that fluorine doping leads to a systematic decrease in the viscosity of the silica glass, as a consequence of a systematic decrease in the activation energy of the viscous flow (see Table 1 and Figs. 5 and 6). Actually, the pre-exponential factor increases with the fluorine concentration, but the decreasing activation energy overcompensates for it. This general behavior can be understood by the structural effect of fluorine in the silica network. The majority of fluorine atoms replace one of the bridging oxygens around a silicon atom with a non-bridging fluorine atom, forming a SiO3/2F polyhedral. This results in the weakening of the glass network with remarkable influence on transport properties and fictive temperature of the glass [18,21]. Even if a certain part of the fluorine environment is composed of fivefold coordinated silicon of the type SiO4/4F, its amount is very low at low fluorine concentrations; and, it remains below 20% of the total fluorine content at a concentration of about 10 mol% (3 wt%) (see [29,30]). In this respect, the role of fluorine can be compared with the role of hydroxyl and chlorine. Both lead to the interruption of the network by a SiO3/2OH and SiO3/2Cl polyhedral, respectively [2,5].
It is believed that the metal impurities of the glasses and their hydroxyl contents are too low to have a measurable influence on the viscosity in our case, even if the OH were of much greater influence than the fluorine, considering comparable concentrations. According to , the activation energy should decrease by about 10% for an OH concentration of 0.05 mol%. To achieve the same effect, the fluorine concentration must be increased to about 3 mol% of fluorine. Nevertheless, the real OH concentration of our glasses below 1 ppm is far too small to produce a measurable effect on the viscosity.
In the case of chlorine, the situation might be different because the chlorine concentrations cCl in our glasses are in most cases much higher than the OH and metal impurities, although still substantially smaller than the fluorine concentrations cF. In , a relatively strong effect of chlorine on the viscosity is assumed, comparable with the OH influence. In order to estimate the potential influence of chlorine on the viscosity, we have made a polynomial regression analysis of the data in Table 1. It is quadratic in respect of fluorine and linear in respect of chlorine (because of the low concentration of Cl in comparison with F). The best fits and the root mean square deviations (RMSDs) areTable 1.
The RMSD values as mean errors are in reasonable agreement with our experimental maximum error limits (see above). For both EA and lnη0, the fit yields a chlorine effect, which can be roughly compared with the effect of fluorine. Note, however, that the chlorine term lies –in almost all cases– practically within the error limits for EA and lnη0, so we have to consider a great relative error of the determined chlorine coefficients. If we combine Eqs. (6) and (7) in order to determine the viscosity in the region between 1600°C and 2000°C, the situation is slightly changed; we get a chlorine influence which may exceed the error limits. As an example, the change of the viscosity by chlorine Δlnη0 (Cl) is determined for 1800°C as −0.65∙cCl. For chlorine concentrations >0.15 mol%, the chlorine term is slightly outside the experimental error limits of the viscosity determination (see above). This is related to the fact that the relatively large errors of EA and lnη0 are actually interrelated with respect to the viscosity.
In any case, the viscosity data presented here are predominantly determined by the fluorine content of the glasses. Of course, our consideration does not allow the exact calculation of the chlorine influence; however, Eqs. (6) and (7) can be used in order to correct a small influence of minor chlorine amounts (≤0.3 mol%) to the viscosity. The consideration of chlorine provides a certain “smoothing” of virtual fluctuations (i.e. a better agreement between experimental and expected values) for materials with high chlorine content, e.g. in the case of F325. For more precise statements, further investigations are necessary.
In the application of a method demonstrated previously , the viscous behavior of fluorine-doped synthetic silica was investigated using collapsing experiments with different fluorine-doped tubes on a modified chemical vapor deposition (MCVD) lathe. These investigations provide comprehensive information on the influence of fluorine doping up to a concentration of about 10 mol% F (3.4 wt%) in a temperature range between 1600°C and 2000°C. In all cases, the viscosity in this field can be described by Arrhenius expressions in good approximation. Fluorine doping leads to a decrease in the activation energy and an increase in the pre-exponential factor. The interaction of both effects provides a systematic decrease in the viscosity by about one order of magnitude at about 1700°C. This is in general due to the weakening effect of fluorine on the glass network. In principle, the effect can be compared with the role of hydroxyl and chlorine in silica, both of which lead to the weakening of the glass bond and decrease the viscosity. The effect of chlorine on the viscosity is roughly comparable to that of fluorine. An exact comparison of both dopants, however, cannot be made here because too little data is available for chlorine. It remains reserved for later investigations.
References and links
1. J. Kirchhof and S. Unger, “The viscous behavior of synthetic silica glass tubes during collapsing,” Opt. Mater. Express 7(2), 386–400 (2017). [CrossRef]
2. G. Hetherington, K. H. Jack, and J. C. Kennedy, “The viscosity of vitreous silica,” Phys. Chem. Glasses 5(5), 130–136 (1964).
3. R. Brückner, “Properties and structure of vitreous silica,” J. Non-Cryst. Solids 5(2), 175–216 (1970). [CrossRef]
4. B. K. Leko, “Viscosity of vitreous silica,” Glass Physics and Chemistry 5(3), 258– 278 (1978).
5. M. L. F. Nascimento and E. D. Zanotto, “Diffusion processes in vitreous silica revisited,” Phys. Chem. Glasses Eur. J. Glass Sci. Technol. B 48, 201–217 (2007).
6. U. C. Paek, C. M. Schroeder, and C. R. Kurkjian, “Determination of the viscosity of high silica glasses during fiber drawing,” Glass Technol. 29(6), 265–269 (1988).
7. A. Mühlich, K. Rau, F. Simmat, and N. Treber, “A new doped synthetic fused silica as bulk material for low-loss optical fibers,” First European Conference on Optical Fiber Communication, London (1975).
9. M. Kyoto, Y. Ohoga, S. Ishikawa, and Y. Ishiguro, “Characterization of fluorine-doped silica glasses,” J. Mater. Sci. 28(10), 2738–2744 (1993). [CrossRef]
10. T. Kakuta, K. Ara, K. Sanada, N. Shamoto, T. Tsumanuma, and S. Chigira, “Radiation resistance of fluorine doped silica core fibers,” Fujikura Techn. Rev. 24, 17–19 (1995).
12. M. Oto, S. Kikugawa, T. Miura, M. Hirano, and H. Hosono, “Fluorine doped silica glass fiber for deep ultraviolet light,” J. Non-Cryst. Solids 349, 133–138 (2004). [CrossRef]
13. J. Kirchhof and S. Unger, “Thermodynamics of fluorine incorporation into silica glass,” J. Non-Cryst. Solids 354(2-9), 540–545 (2008). [CrossRef]
14. K. Schuster, S. Unger, C. Aichele, F. Lindner, S. Grimm, D. Litzkendorf, J. Kobelke, J. Bierlich, K. Wondraczek, and H. Bartelt, “Material and technology trends in fiber optics,” Adv. Opt. Technol. 3, 447–468 (2014).
15. B. Hooda, A. Pal, V. Rastogi, R. Sen, J. Gandhi, and J. Kobelke, “Segmented cladding fiber fabricated in silica-based glass,” Opt. Eng. 54(7), 075103 (2015). [CrossRef]
16. K. Jens, B. Jörg, W. Katrin, A. Claudia, P. Zhiwen, U. Sonja, S. Kay, and B. Hartmut, “Diffusion and Interface Effects during Preparation of All-Solid Microstructured Fibers,” Materials (Basel) 7(9), 6879–6892 (2014). [CrossRef] [PubMed]
17. Heraeus, “Fused silica tubes for fiber production,” https://www.heraeus.com/en/hqs/products_hqs/optical_fiber/tubes_fiber/Tubes_fiber_production.aspx.
18. U. Haken, O. Humbach, S. Ortner, and H. Fabian, “Refractive index of silica glass: influence of fictive temperature,” J. Non-Cryst. Solids 265(1-2), 9–18 (2000). [CrossRef]
19. S. J. B. Reed, Electron Microprobe Analysis, Second edition (Cambridge University Press 1993).
20. H. Wehr and D. Wiechert, “Refractive index and density of fluorine doped silica prepared by the PCVD process,” Mater. Res. Bull. 21(5), 559–566 (1986). [CrossRef]
21. P. Dumas, J. Corset, Y. Levy, and V. Neumann, “Raman spectral characterization of pure and fluorine-doped vitreous silica material,” J. Raman Spectrosc. 13(2), 134–138 (1982). [CrossRef]
22. J. Kirchhof, S. Unger, K.-F. Klein, and B. Knappe, “Diffusion behaviour of fluorine in silica glass,” J. Non-Cryst. Solids 181(3), 266–273 (1995). [CrossRef]
23. M. Ohashi, M. Tateda, K. Tajima, and K. Shiraki, “Fluorine concentration dependence of viscosity in F-doped silica glass,” Electron. Lett. 28(11), 1008–1010 (1992). [CrossRef]
24. K. Shiraki and M. Ohashi, “Scattering property of fluorine-doped silica glasses,” Electron. Lett. 28(17), 1565–1566 (1992). [CrossRef]
25. K. Shiraki, M. Ohashi, K. Tajima, M. Tateda, and K. Tsujikawa, “Viscosity of F and GeO2 codoped silica glass,” Electron. Lett. 29(14), 1263–1265 (1993). [CrossRef]
26. M. Ohashi, M. Tateda, K. Shiraki, and K. Tajima, “Imperfection loss reduction in viscosity-matched optical fibers,” IEEE Photonics Technol. Lett. 5(7), 812–814 (1993). [CrossRef]
27. U. C. Paek, C. M. Schroeder, and C. R. Kurkjian, “Determination of the viscosity of high silica glasses during fibre drawing,” Glass Technol. 29(6), 263–266 (1988).
28. H. Murata, “Recent developments in vapor phase axial deposition,” J. Lightw. Techn. 4(8), 1026–1033 (1986). [CrossRef]
29. R. E. Youngman and S. Sen, “Letter to the Editor: Structural role of fluorine in amorphous silica,” J. Non-Cryst. Solids 337, 182–186 (2004).
30. R. E. Youngman and S. Sen, “Structural role of fluorine in amorphous silica,” J. Non-Cryst. Solids 349, 10–15 (2004). [CrossRef]