GB2256723A - Annealed fibre optic fusion splices between single mode and erbium-doped fibres - Google Patents

Abstract

Fusion splices at A and B between single-mode-fibre and erbium-doped fibre are repeatedly annealed. <IMAGE>

Description

FIBRE OPTIC SPLICING Erbium-doped fibre amplifiers (EDFAs) are very attractive for optical transmission in the third telecommunication spectral window. Extremely high efficient EDFAs have been achieved by using small-core, high numerical aperture (NA) doped fibre. Both parameters improve the power density in the core of the erbium-doped fibre (EDF) by reducing the signal and pump mode radii. It has been shown that the NA should be as high as possible to achieve a high efficiency fibre. However, a decrease in fibre core diameter and increase in NA lead to a decrease in spot size. This will therefore increase the mismatch between the field distributions in the EDF and standard fibre, leading to large splice losses. This increase in loss degrades the amplifier gain noise figure and amplifier output power.

Hitherto, two methods have been reported to overcome this problem. The first method uses thermal diffusion of the index raising codopant such as germanium to increase the mode field radius of EDF. This will reduce, therefore, the mismatch between the field distributions in the EDF and standard fibre, leading to smaller splice losses. It has been demonstrated that the mode radius of a single-mode fibre (SMF) with relative refractive index difference of 0.004 can be doubled from 4.3yum to 8.7jim by a heat treatment at 1300to for 10 hours. An alternative method does not require the long heat treatment time and uses one or more matching fibres. The splice loss reduction is achieved by splicing one or more (untapered) fibres between an EDF and standard SMF.The spot size of every intermediate fibre is the same multiple of the spot size of the previous fibre. This method can yield results as good as those reported for thermal diffusion tapering. Theoretically, a splice loss of O dB can be achieved with an infinite number of intermediate fibres. In practice only a few intermediate fibres can be used.

This is because each addition of intermediate fibre introduces a residual splice loss due to lateral misalignment, non-circular cores etc., which is typically 0.1 dB between identical fibres. However, a simple method using a commercial arc fusion splicer to reduce the splice loss between EDF and standard SMF has been discovered.

Significant reduction of the splice loss from 1.8 dB to less than 0.1 dB was achieved in tests, which results are much better than reported from the two methods mentioned above.

It has been widely believed that Ge atoms in silica fibres are stable. Experimental results with weakly fused SMFs, however, have shown that the peak refractive index of the core decreases and the fibre core size is actually increased by the fusion process, contrary to previously proposed models. An increase in the silica fibre core size by about 67% was measured during a fusion process of 6 minutes at 1100-C. Generally, the diffusion coefficient of Ge in silica fibres is known to be expressed as D = DOexp(-Q/RT), where T is the temperature (K) and R = 8.31 (J/K/mol). The parameter Q has a value in the range from 1.5 x 105 to 3.0 x 105 for high silica content fibres.The diffusion coefficient strongly depends on the temperature (e.g. the diffusion coefficient at 1400"C is about 20 times that at 1200-C as well as on the Ge concentration in the core. The Ge concentration in the core can be assumed to depend linearly on the relative refractive index difference (i.e. (NA)2 ).

It is therefore possible to achieve a substantial increase in core size at a reduced fusion time by working at a higher temperature with a high NA EDF. The temperature at the arc of an arc fusion splicer is much higher than 1100it and was used in this investigation.

Excess loss due to expansion of the modal field depends on the transition length, L. Calculation based on the beam propagation method shows that the excess loss at L > 700 is small, even if the spot size is expanded to more than twice the original size. At an operating wavelength of 1550 nm, a L of about 1 mm is required to obtain low excess loss due to field expansion. This can be achieved with a commercial arc fusion splicer with a wider electrode gap than the optimum gap of 0.7 mm for 125 Sm outside diameter fibre.

According to the present invention there is provided a method of reducing the coupling power loss in a fusion splice between single-mode optical fibre and erbium-doped optical fibre comprising the steps of repeatedly annealing the splice.

The present invention will now be described, by way of example, with reference to the accompanying drawings, in which : Figure 1 shows an experimental set-up for carrying out the method of the invention and evaluating the results of said method; Figure 2 shows graphically the results of carrying out the method experimentally; Table 1 shows the parameters used during the experiment; and Table 2 shows a comparison between the splices according to the invention and prior art splices.

In Figure 1 the experimental set-up used to measure the splice loss between standard SMF and EDF is shown. Light from a laser 10 operating at either 1300 nm or 1550 nm wavelength was launched into a first arm 12 of a 3 dB coupler 14. The optical power of the laser was monitored through a second arm 16 of the coupler with an optical power meter 18. A third arm 20 was connected to a length of SMF 22 which was fusion spliced at A to 10 m of EDF 24 which was in turn fusion spliced at B by an electrode arrangement such as 30 to another length of SMF 26 which was coupled at C to an optical power meter 28.

The characteristics of the EDF 24 were: a core radius of 1.85sum, an NA of 0.24 and a maximum absorption around 1530 nm of 7.25 dB/m. The splice loss from SMF to EDF (large core radius to a smaller receiving core radius) was obtained by measuring the optical power at A before splicing and that out of a short length of EDF (cut-back from the 10 m of EDF) which was fusion spliced to the SMF. The splice loss from EDF to SMF (small core radius to a larger receiving core radius) was obtained from the powers at B (before splicing) and then at C after joint B was fusion spliced. The splice loss from SMF to EDF is difficult to measure accurately because of the high absorption of the EDF around 1550 nm which affects the result and therefore a 1300 nm source was also used. Splice loss at 1550 nm is in general better than at 1300 nm.This is because as the wavelength of light increases, the width of the fundamental mode field increases and the splice loss decreases for a given offset or tilt. The theoretical splice loss between two fibres with spot sizes ofCA) and ftA is given by 20log(2/(f+1/f)), which is independent of the direction of transmission. However, in practice it was found that, contrary to multimode fibre experience, a greater splice loss occurs when light was launched from a small mode radius into a larger mode radius than when the same splice is measured in the opposite direction. Consequently, the splice loss from EDF to SMF can be considered as the worst case.

Enlargement of the spot size of the EDF was realised using a comercial arc fusion splicer with a wide electrode gap of about 3.0 mm. A splicing operation take place in four steps: withdrawal to form gap, pre-heating, fusion and annealing. During pre-heating, the fibres move together. Fusion begins when the fibres come in contact with each other. During fusion, a certain amount of overlapping of the fibres take place. The joint is then annealed. Table 1 shows three sets of fusion parameters stored as programs &num;1, &num;2 and &num;3 which were used in the investigation. The parameters of programs &num;2 and &num;3 are similar to program &num;1 except the annealing time was lengthened from 2.0 sec to 10.0 sec and 20.0 sec respectively.

The reduction of the splice loss in either direction was achieved by repeatedly applying heat treatment, which includes pre-heating, fusing and annealing at the joint, after the initial splicing. The 10sum of stuffing length (overlap) during the fusing process did not seem to affect the splice loss. An optimum splice was regarded as being achieved when no further reduction in loss was observed after an additional fusion process. Using this method the diffusion of the Ge can be controlled to increase the spot size of the EDF until it matches that of the SMF.

Figure 2 shows the splicing loss against the number of heat treatments using the three splicing programs.

Symbols (o), (A) and (a) represent the measured splice loss from SMF to EDF (large mode radius to smaller mode radius) with a 1559 nm laser using programs &num;1, &num;2 and &num;3 respectively. The minimum splice loss in these measurements includes losses in the small length of the doped fibre. To overcome this, a 1322 nm laser was used because the doped fibre's attenuation around 1300 nm is very low; the measured minimum splice loss (o) in this case (including the loss of 10 m of doped fibre) was less than 0.1 dB. The measured minimum splice loss in the opposite direction, i.e. from EDF to SMF using programs &num;2 and &num;3 was less than 0.1 dB and are represented by the symbols (A) and (i) respectively. The accuracy of the splice loss measurement is estimated to be + 0.05 dB.It was found that the minimum splice loss is insensitive to the initial splicing loss. A minimum splice loss of less than 0.1 dB was achieved even when the fibres were intentionally misaligned to introduce an additional loss of 1 dB. As expected, the total heating time required to reach the minimum splice loss was reduced using a longer annealing time; this is because energy was lost between adjacent heat treatments. The time required to reach 0.1 dB loss is about 1 min using program &num;3 and is about 3 min using program &num;1. The increase in splice loss in curves (6) and (o), following the minimum loss, is due to the field expansion loss. The transition length was fixed at about 1.0 mm by the electrode gap, if the spot size is increased further by the heating process, a longer transition length is needed to reduce the excess loss due to field expansion. Table 2 compares the reported splice loss between standard SMF and EDF obtained by the various methods.

A method for reducing splice loss between SMF and doped fibre using a commrcial arc fusion splicer has been described. An optimum splice only required about 1 min heating time. Minimum splice loss of less than 0.1 dB was consistently achieved with this method. This represents at least 0.5 dB improvement over other reported methods and will therefore improve erbium-doped fibre amplifier performance in terms of noise figure and saturation output power.

Claims (9)

1. A method of reducing the coupling power loss in a fusion splice between single-mode optical fibre and erbium-doped optical fibre comprising the steps of repeatedly annealing the splice.

2. A method as claimed in Claim 1 wherein the annealing time is approximately 10 seconds.

3. A method as claimed in Claim 1 wherein the annealing time is approximately 20 seconds.

4. A method as claimed in Claim 2 wherein the number of annealing steps is between 8 and 12.

5. A method as claimed in Claim 3 wherein the number of annealing steps is between 3 and 6.

6. A method as claimed in Claim 1, substantially as hereinbefore described, with reference to and as illustrated in the accompanying drawings.

7. A fusion splice between a single-mode optical fibre and an erbium-doped optical fibre, the splice having been repeatedly annealed.

8. A splice as claimed in Claim 7 wherein the coupling loss is less than 0.1 dB.

9. A splice as claimed in Claim 7, substantially as hereinbefore described, with reference to and as illustrated in the accompanying drawings.