Diffusion of Ge and F was studied during drawing of silica optical fibres. Preforms were drawn using various draw conditions and fibres analysed using the etching and Atomic Force Microscope (AFM) technique. The results were confirmed by comparison with fibre Refractive Index Profiles (RIP). Both Ge and F were found to diffuse at high temperature, 2100°C, and low draw speed, 10m/min. Diffusion simulations showed that most diffusion occurred in the neck-down region. The draw temperature and preform feed rate had a comparable effect on diffusion, whereas preform diameter did not significantly affect the diffusion.
©2004 Optical Society of America
Dopant diffusion in optical fibres is important as the dopant distribution in the core and inner cladding of an optical fibre determines the transmission properties of the fibre. Specialty fibres typically have more complex dopant profiles than long-haul telecommunication fibres. Designs with higher dopant concentrations are common and multidoping is frequently used making the precise control of the fabrication process particularly important. If the fibre is exposed to high temperatures for a period of time, the dopant distribution in these regions can change due to dopant diffusion. Ge is the dopant that has been most frequently reported in diffusion studies in optical fibres. Diffusion of Ge in silica optical fibres has been observed, for example, during splicing, manufacture of fused fibre couplers and beam expanding fibres [1–3]. The reported diffusion data however has significant scatter and it is questionable whether the published values can be used to estimate diffusion during the fibre drawing process. The published data for F diffusivities in silica is relatively consistent and reliable values can be found, e.g., in Ref. . Recent studies of F diffusion in silica fibres have also been conducted in [5,6].
There are very few publications dealing with the effect of the fibre drawing process on dopant diffusion. Due to the high processing temperature it is likely that some dopant diffusion occurs specifically at the low draw speeds characteristic of specialty fibre fabrication. Ge diffusion during drawing was observed by Hersener et al.  who looked at the drawing induced changes of the dopant profile in the neck-down of the drawn preform. No quantitative analysis was made but diffusion was observed through the smoothening of the so-called ‘ripples’ characteristic of Ge-doped MCVD layers. Ge diffusion during drawing was also studied by Pugh et al. , but no diffusion was observed. In the present study we demonstrate the diffusion of both Ge and F during drawing of silica optical fibres. A diffusion coefficient is determined for Ge and results are used to compute the effect of drawing parameters such as furnace temperature and draw speed on dopant diffusion.
2.1 Fabrication and measurements
The fibre designs used in the study are shown in Fig. 1. Ge diffusion was studied using a 3-ring structure with different germania concentration in each ring (17, 8 and 4mol%). The doped rings were separated by pure silica regions. The preform was fabricated using the MCVD technique. The F diffusion study was carried out on a pure silica core fibre with F doped cladding (Fluosil® SWU1.1, 4w%F). The preform was manufactured by plasma outside deposition. Both the preforms were drawn using a standard drawing technique. The draw conditions are listed in Table 1.
Chemical analysis for a section of each preform was carried out by energy dispersive X-ray spectrometry (EDS) in a scanning electron microscope (EDAX + Philips XL30 SEM) for Ge-doped samples and wavelength dispersive X-ray spectrometry (WDS) in an electron probe microanalyser (Cameca SX50 EPMA) for F-doped samples. The obtained dopant concentration profiles provided the required molar concentrations for the diffusion calculations as well as the initial concentration profile for comparisons with fibre samples. The preform radial and longitudinal homogeneity was determined by a tomographic (RIP) method  (Photon Kinetics 2600). The measurement of the fibre dopant profiles proved to be the most challenging task. The AFM and etching technique  was found to provide data with sufficient accuracy. Cleaved fibre ends were etched in 5vol% HF for 2min (Sample 3R) and 4min (Sample F) and the resulting topography profiled using AFM (Digital Instruments Dimension 3100, contact mode). The AFM profiles for the Ge-doped fibre were converted to molar concentration using etching reaction order of 0.6 . Results were confirmed using RI profiling (York S14 and EXFO NR-9200HR).
2.2 Diffusion computation
In order to estimate dopant diffusion during drawing, diffusion was assumed to occur according to Fick’s law and the diffusion coefficient was assumed to be independent of concentration but dependent on temperature according to Arrhenius equation. The fibre profiles studied were circularly symmetric and diffusion was assumed to occur exclusively in the radial direction. The variation in temperature along the preform, neck-down and fibre for various drawing conditions was obtained from heat and mass transfer simulations (for details see Ref. ). The one-dimension, cylindrical, diffusion equation is presented in Eq. (1) and Arrhenius equation in Eq. (2).
where D is the diffusion coefficient [m2/s], c is dopant concentration [mol-fraction], r is radius [m], t is time [s], T is temperature [K], D0 is the pre-exponential term [m2/s], E is activation energy [J/mol] and R is the gas constant, 8.314J/(Kmol).
3. Results and discussion
3.1 Germanium diffusion
The AFM profiles for the Ge-doped fibres are shown in Fig. 2(a). The difference of the profiles is evident for the innermost ring with highest Ge concentration. The fibre that was drawn at the hotter temperature has etched less indicating a lower Ge concentration. The difference in the two outer rings is not so clear. The innermost peak shows broadening of the profile, characteristic of dopant diffusion.
The corresponding fibre RIPs are shown in Fig. 2(b). Again the greatest difference between the fibre samples is seen in the innermost ring, which is consistent with the AFM profiles. The peak refractive index is higher for the samples drawn at a lower temperature. Note that the innermost ring has a FWHM of only 1.5µm making accurate RI profiling difficult. No significant change is seen in the width of the peak within the measurement accuracy. It must be noted that the focus for the fibre RIP was difficult to adjust and could result in some error. Note that the cladding RI is slightly reduced for the fibre drawn at lower temperature. This may be due to draw-induced residual stresses complicating the comparison with the AFM profile.
It must be noted that glass fictive temperature affects both the RI and etching rate . However, the cooling rates of fibres during drawing were calculated to be moderate and of similar magnitude for the samples. The maximum draw temperature range was 300°C, however at moderate cooling rates the induced density change is insignificant. More importantly the HF etching process has been shown to be dominated by the chemical composition in doped silica .
To determine the diffusion coefficient from the measurements, the innermost ring was chosen for computations. The initial preform GeO2 concentration profile for the simulations was taken from preform microanalysis measurements. Figure 2(c) shows the measured fibre Ge molar concentration profile against the initial (scaled from preform measurements) and simulated profile for a diffused fibre. Best fit was found using coefficient D0=2.4±0.1×10-6 m2/s, Eq. (2), which agrees with the coefficient determined in Ref. . E=310kJ/mol was taken from Ref. .
When diffusion simulations were compared with the AFM data for the outermost ring of the fibre, an interesting point was noticed. In the high temperature sample the measured profile showed the preform layer structure that should have been more diffused according to the simulations. There are two likely explanations; (i) there is a high enough temperature gradient radially in the preform and fibre to cause different diffusion rates and/or (ii) the diffusion coefficient depends on Ge concentration. Since the outer layer is less diffused, this means that the temperature should be lower in the outside part of the fibre. This can only be possible during the cooling stage of the draw. Diffusion simulations showed that a 150°C temperature difference would be necessary to cause the reduced diffusion or that the inner layer remains hotter for much longer times. Heat transfer simulations showed that at 10m/min speed, the difference between surface and core temperature was only few degrees. It therefore seems unlikely that a radial temperature gradient is the cause of reduced diffusion. It is possible that the diffusivity of Ge is concentration dependent. Concentration dependent diffusion has been observed before for silica network modifiers in Ref. . The diffusion coefficient increased with concentration which is in agreement with the current study.
3.2 Fluorine diffusion
The AFM profile for the F-doped fibre drawn at 1900°C and at 10m/min is shown together with the F-concentration profile of the preform in Fig. 3(a). The preform shows a step change from the core to the cladding whereas the fibre profile shows reduced slope of the interface indicating some diffusion of F towards the core. The F-depleted dip on the outer edge of the fibre is of similar magnitude. No quantitative comparison can be made as the etching reaction order is not know for these samples, however comparison of the various fibre samples can be performed. The etched fibre AFM profiles for fibres drawn at 10m/min and at 2100°C and 1900°C are shown in Fig. 3(b). At the higher temperature the F-depleted region in the outer edge of the fibre has etched less, indicating lower F concentration. At the core-cladding interface the slope of the interface is reduced for the fibre drawn at hotter temperature. The changes indicate mobility of F both towards the core and out of the fibre.
To compare all the draw conditions, two parameters were chosen to represent the amount of diffusion: (i) etching depth of the F-depleted region on the outer edge, Fig. 4(a) and (ii) the slope of the core-cladding interface, Fig. 4(b). At both temperatures as the draw speed is reduced, similar changes are seen, i.e., reduced etching depth of the dip and reduced core-cladding slope, indicating increased diffusion. To confirm the changes in concentration profiles the fibre RIPs were also measured. The profiler did not resolve the F-depleted region in the outer edge of the fibre but showed the change in core-cladding interface slope. The trends were similar to those shown with AFM profiles, and are plotted in Fig. 4(b).
It is possible that the mobility of F is different in the outer edge of the fibre to that inside the fibre. This is possible as the removal mechanism of F from the fibre edge can affect the diffusion profile. If this were the case, the furnace atmosphere would have an effect. Factors such as selection of furnace gas, the amount of oxygen and humidity in the gas as well as the furnace temperature would affect the surface chemistry on the preform and fibre. The diffusion towards the core is unaffected by these factors but would be affected by the quality of the silica, i.e. OH and impurity content .
3.3 Diffusion simulations
A high NA (35mol% GeO2) fibre design with core diameter of 3µm, was chosen for the simulations. In order to determine the relative importance of the various stages in the preform necking down, diffusion was calculated along the neck-down. Although the preform is exposed to the highest temperature longer than other parts of the neck-down, the diffusion was found insignificant at the chosen preform diameter of 17mm. Significant change to the profile was observed in the neck-down at diameters ranging from about 8 to 2mm. Due to short exposure time only slight broadening of the profile was seen in the fibre.
The draw parameter range was chosen as typical for specialty fibre drawing presenting the maximum change expected due to diffusion. The simulated results supported the experimental results. As expected, higher furnace temperature and slower speed increased diffusion (see Fig. 5) and the relative effects of change in draw speed and temperature were comparable in magnitude. To minimise diffusion the fibres should be drawn at lower temperatures and higher speeds. Typically fibres are drawn at a specific tension, which is kept constant from draw to draw. This is good practice since it gives more accurate control on reproducibility than the furnace element temperature. It must be noted however that diffusion is affected also by the draw speed and the neck-down shape and does not solely depend on tension. Diffusion simulations showed that draw conditions exist where fibre drawn at higher tension has a more diffused profile.
No significant differences in diffusion were found for fibres drawn from different diameter preforms (12mm and 24mm) at the same draw speed and maximum glass temperature. It was found that the temperature profile in the neck-down region (where most diffusion occurs) was similar for the two cases. Furnace design had a significant effect on the temperature profiles of the preform and the fibre. Fibre drawn in a furnace with a longer hot-zone and extension tube experienced more diffusion.
It has been shown that dopant profile change due to diffusion can be induced during drawing in both Ge and F-doped silica fibres. Although for most of applications changes of this magnitudes would not cause significant effects, in some fibre designs optical properties such as dispersion can be altered. It is also possible to use diffusion during drawing to advantage e.g. when modified profiles are required. This however would require detailed knowledge of the effects of co-doping on diffusion for more complicated fibre designs.
K. Lyytikäinen acknowledges the OFTC fibre fabrication staff for their help in fabrication and measurements and K.-F. Klein (Giessen-Friedberg University) for useful discussion on F diffusion. K. Lyytikäinen also acknowledges funding from Australian Government DEST and CSC Scientific Computing Ltd. for computing resources. S.T. Huntington acknowledges the support of the Australian Research Council and DEST.
References and Links
1. J.T. Krause, W.A. Reed, and K.L. Walker, “Splice loss of single-mode fiber as related to fusion time, temperature, and index profile alteration,” J. Lightwave Technol. LT-4, 837–840 (1986). [CrossRef]
2. K. Shiraishi, Y. Aizawa, and S. Kawakami, “Beam expanding fiber using thermal diffusion of the dopant,” J. Lightwave. Technol. 8, 1151–1161 (1990). [CrossRef]
3. H. Yamada and H. Hanafusa, “Mode shape convertor produced by the thermal diffusion of different dopants,” IEEE Photonic Tech. L. 6, 531–533 (1994). [CrossRef]
4. J. Kirchhof, S. Unger, K.-F. Klein, and B. Knappe, “Diffusion behaviour of fluorine in silica glass,” J. Non-Cryst. Solids 181, 226–273 (1995). [CrossRef]
5. M. Fokine, “Thermal stability of chemical compositions gratings in fluorine-germanium-doped silica fibers,” Opt. Lett. 27, 1016–1018 (2002). [CrossRef]
6. B.C. Gibson, S.T. Huntington, J.D. Love, T.G. Ryan, L.W. Cahill, and D.M. Elton, “Controlled modification and direct characterization of multimode-fiber refractive-index profiles,” Appl. Opt. 42, 627–633 (2003). [CrossRef] [PubMed]
7. J. Hersener, G.P. Huber, and J. von Wienskowski, “Semiquantitative X-ray microanalysis on preforms for optical fibres,” Electron. Lett. 20, 208–209 (1984). [CrossRef]
8. A.C. Pugh, R.P. Stratton, and D.B. Lewis, “Investigation of elemental diffusion during the drawing and heat treatment of glass optical fibres,” J. Mater. Sci. 29, 1036–1040 (1994). [CrossRef]
9. Y. Zhao, S. Fleming, K. Lyytikainen, and L. Poladian, “Nondestructive measurement for arbitrary RIP distribution of optical fiber preforms,” J. Lightwave Technol. (to be published).
10. P. Pace, S.T. Huntington, K. Lyytikäinen, A. Roberts, and J.D. Love, “Refractive index profiles of Ge-doped optical fibers with nanometer spatial resolution using atomic force microscopy,” Submitted to Opt. Express Jan 2004.
11. K. Lyytikäinen, P. Råback, and J. Ruokolainen, “Numerical simulation of a specialty optical fibre drawing process,” in Proceedings of the 2002 ASME Pressure Vessels and Piping Conference, S. Kawano and V.V. Kudriavtsev, (ASME, New York, 2002), PVP448-2, pp. 267–276.
12. A. Agarwal and M. Tomozawa, “Correlation of silica glass properties with the infrared spectra,” J. Non-Cryst. Solids 209, 166–174 (1997). [CrossRef]
13. Q. Zhong and D. Inniss, “Characterisation of lightguiding structure of optical fibers by atomic force microscopy,” J. Lightwave. Tech. 12, 1517–1523 (1994). [CrossRef]
14. J. Kirchhof, S. Unger, and B. Knappe, “Diffusion processes in lightguide materials,” in Proceedings of Optical Fiber Communications Conference, (2000), 2, pp. 212–214.