In the evaluation a fused biconical taper 1480/1580 nm WDM’s ability to handle high power cascaded Raman laser throughput (>100 W) a significant degradation in performance was observed. A systematic root cause investigation was conducted and it is experimentally confirmed that the WDM degradation was caused by an interaction between the high power 1480 nm line, an out-of-band Stokes line, and the -OH content of the glass optical fiber. Slanted fiber Bragg grating (SFBG) was introduced to filter out the 1390 nm out-of-band Stokes line in an attempt to avoid this interaction. Ultimately a series of tests were conducted and it was confirmed that the addition of a 1390 nm SFBG in between a high power Raman laser and the high power WDM has successfully prevented the degradation which therefore allowed the continued high power operation of the WDM. NAVAIR Public Release SPR 2013-469 Distribution Statement A-“Approved for Public release; distribution is unlimited”.
© 2013 Optical Society of America
High power cascaded Raman fiber lasers are an appealing source for pumping of rare earth doped optical devices such as large mode area (LMA) fiber amplifiers , very large mode area (VLMA) higher order mode (HOM) fiber amplifiers [2,3], distributed signal gain in telecom systems , and frequency doubling in laser guide starts [5,6]. This desirability is due primarily to the Raman laser’s high brightness and low generation of quantum defect. As such, high power cascaded Raman fiber lasers have been under continuous development over the past several years with output powers recently reaching 301 W at 1480 nm . The use of high power wavelength division multiplexers (WDM) are a common means of combining the Raman laser light with an amplifier’s signal light [1–3,8]. The reliability of these high power WDMs are therefore essential to the overall reliability of these high power, high gain fiber amplifier systems.
As described by Supradeepa, et al.  a cascaded Raman resonator consists of a Raman fiber and multiple fiber Bragg grating pairs with center wavelengths located at the specific Raman Stokes lines. For the 1480 nm cascaded Raman fiber lasers pumped by a high power cladding pumped Yb3+-doped fiber laser source near 1.1 µm, the nearest Stokes lines to 1480 nm are 1119 nm, 1175 nm, 1240 nm, 1312 nm, 1390 nm, and 1580 nm. In these systems, multiple Stokes lines coexist, each containing various amounts of power. The optical spectrum for a Raman fiber laser capable of delivering up to 111 W of optical power at 1480 nm (130 W total power) when measured at its maximum output power is shown in Fig. 1. In addition to the main Stokes line centered at 1480 nm, there are several other discrete, so called out-of-band Stokes lines spanning from 1000 nm up to 1600 nm.
At this maximum output power, the measured distribution of the Stokes lines is listed in Table 1. The two strongest Stokes lines after the main 1480 nm, are at 1312 nm and 1390 nm, and contain 5.13% and 4.44% of the total power, respectively.
Stokes lines with wavelengths shorter than 1500 nm are beyond the Er3+ ion’s emission spectrum. However, the Stokes line near 1580 nm (shown in Fig. 1) is within the Er3+ ion’s emission band and contains several hundred milliwatts. If this 1580 nm Stokes line is coupled into an Er3+-doped fiber amplifier it can be amplified, thus depleting the amplifier’s ability to boost the main 1550 nm signal. To eliminate this potential gain, a fiber based fused coupler is used to filter out the parasitic 1580 nm Stokes line while minimizing any introduced insertion loss for the 1480 nm main power line. A 1480/1580 nm WDM was chosen for this purpose due to its ultra-low insertion loss at 1480 nm and high isolation capability at 1580 nm. In the schematic shown in Fig. 2, the input channel on the high power 1480/1580nm WDM is labeled as P1, the 1480 nm output channel is labeled as P2, and the 1580 nm output channel is labeled as P3. Both the input and output channels typically use SMF28 fibers.
The high power handling capability of the 1480/1580 nm WDM was evaluated by splicing P1 to the high power Raman laser output fiber and angle cleaving both P2 and P3 at the WDM output side to avoid back reflections which may result in damage to the high power Raman fiber laser. The optical power from the two angle-cleaved fibers was monitored by separate power meters. The 1480/1580 nm WDM and all splice points were then mounted on a chiller plate that was maintained at 22°C.
Figure 3 shows the monitored optical power at the WDM 1480 nm output channel (P2) over a continual operation burn-in period of 46.5 hours. As shown, we for the first time observed the output power reduced from approximately 111 W to nearly 104.5 W with an accelerating decay rate. This equates to an increase in the insertion loss at 1480 nm between P1 and P2 by 0.26 dB over the 46.5 hour duration. The power degradation seen at P2 was not recoverable.
The same test was conducted on numerous high power 1480/1580nm WDM and 1480/1550nm WDM used for signal and pump combining in Er3+-doped fiber amplifiers with a very similar degradation noted. It was therefore critical to identify the root cause of this degradation to achieve long term stability and reliability of the high power fiber amplifier or laser systems containing these components.
2. Root cause investigation
To better understand the nature of the WDM performance degradation, the optical spectral transfer functions of the 1480/1580 nm WDM were measured both before and after conducting a burn-in test. These optical spectral transfer functions are shown in Fig. 4. Prior to the high power burn-in test, the optical spectral transfer functions from P1 to P2 and from P1 to P3 were centered at 1481.1 nm and 1579.7 nm respectively, with a channel spacing of 98.6 nm (as was expected from the design). Following the high power burn-in test however, both central wavelengths had shifted toward longer wavelengths (from 1481.1 nm to 1496.0 nm, and from 1579.7 nm to 1595.9 nm). In addition, the channel spacing of the WDM before and after the burn-in test changed slightly (from 98.6 nm to 99.9 nm). Based on the fused fiber coupler model [9–11], the channel spacing is dependent on the length of the interaction taper region as well as the cross sectional area of the fused section. As no change in the physical dimension of the fused section was observed following the burn-in test, this rather slight change channel spacing was within expectations. In addition, the total power measured at the P2 and P3 outputs remained constant over the entire period, indicating a negligible change in excess loss change during the burin-in test.
Also shown in Fig. 4(a), is the increase in insertion loss at 1480 nm from P1 to P2 of 0.3 dB following the burn-in test. This value is very close to the insertion loss calculated from the power measurement (0.26 dB). From these measurements it is believed that the observed power degradation at P2 was caused by the change in the WDM’s coupling ratio during the burn-in test. At this time however, the root cause for this change remained unclear.
There are several reasons the refractive index of a fused biconical taper WDM may change including radiation , UV light , and Gamma light . In order to understand the observed coupling shift, the refractive index dependence of the WDM coupling ratio between P1 and P2 was numerically simulated. The physical dimensions of the WDM fused taper section were measured using a high resolution optical microscope and no noticeable dimensional difference between the measurements before and after the high power burn-in test were observed. For this simulation, the refractive index was increased from the nominal refractive index of silica at 1500 nm, 1.444, to 1.474. As plotted in Fig. 5, the minimum coupling ratio wavelength shifts towards the longer wavelength; however, the channel spacing exhibits negligible change with increasing refractive index.
The minimum coupling ratio wavelength was then calculated as a function of refractive index and plotted in Fig. 6(a). At the nominal refractive index of 1.444 for 1500 nm light, the minimum coupling ratio wavelength is 1478 nm between P1 and P2. This minimum value then increases linearly with the refractive index with a slope of 10 nm per change of 0.01 in the refractive index. At 1480 nm, the insertion loss increases with the minimum coupling ratio wavelength. Using the measured shift in the minimum coupling ratio wavelength (from 1481.1 nm to 1496.0 nm), the simulated insertion loss increases by 0.22 dB as shown in Fig. 6(b), notably close to the measured insertion loss increase of 0.26 dB. In addition, assuming the minimum coupling ratio wavelength shift was solely caused by the refractive index increase, the change in the simulated refractive index would be 0.015.
The refractive index profile of the fused biconical taper section of the degraded 1480/1580 nm WDM was measured using spatially resolved Fourier transform spectroscopy . A high resolution microscope image showing interference fringes across the sample is in Fig. 7. There is no gross evidence of anything unusual in the sample such as bubbles or obvious damage sites. However, we were not able to provide quantitative data due to the taper waist not being axis-symmetric as it is a fusion of the two fibers. To obtain quantitative refractive index data it would be necessary to be able to rotate the sample and take data at many different angles .
The continuation of the root cause analysis sought to determine if the WDM high power performance degradation was resultant of the high power 1480 nm light, by the out-of-band Stokes lines, or if this degradation was caused by a combination of these. To investigate this, a test setup was prepared as shown in Fig. 8 with three cascaded 1480/1580 nm WDMs. The input fiber of WDM A was spliced to the high power Raman laser. The input fiber of WDM B was spliced to the 1580 nm output Channel of WDM A, and the input fiber of WDM C was spliced to the 1480 nm output Channel of WDM A. All three WDMs and their splice points were mounted on a chiller plate and maintained at 22°C and the Raman fiber laser was operated slightly lower than its maximum power setting. In this configuration, the majority of the out-of-band Stokes lines are segregated by WDM A and coupled to P3A and into WDM B. P2A then contains only the main Stokes line at 1480 nm. With this arrangement, only the high power 1480 nm line should pass through WDM C. By monitoring the optical power and measuring the optical spectral transfer functions of these three WDMs both before and after the high power test, it would be possible to isolate the root cause.
If no degradation were observed, the monitored optical power would remain stable throughout. However, the optical power measured at P2C, shown in Fig. 9, indicates a clear power drop after a few tens of hours indicating high power WDM degradation from the WDM A and/or the WDM C.
The optical spectral transfer functions from P1 to P2 of WDM A, WDM B, and WDM C were all measured before and after the high power test, as shown in Fig. 10. Here WDM A shows a clear wavelength shift indicating an increase in insertion loss at 1480 nm following the high power test. The optical spectral transfer functions of WDM B and WDM C however show no obvious change.
The insertion losses between P1 and P2 at 1480 nm for WDM A, WDM B, and WDM C before and after the high power burn-in test were also measured. The insertion loss of WDM A increased from 0.10 dB to 0.50 dB, while the insertion losses of WDM B and WDM C remain unchanged.
Based on the test data shown in Fig. 10, it was concluded that the high power 1480 nm line in concert with the out-of-band Stokes lines was causing the degradation of the high power WDM A. The 1480 nm line alone did not result in a degradation of the high power WDM C, and out-of-band Stokes lines by themselves did not result in a degradation of the high power WDM B.
Among the multiple out-of-band Stokes lines, it was suspected that the –OH absorption at 1390 nm was interacting with the high power 1480 nm light resulting in the observed WDM degradation. To confirm this hypothesis, multiple high power 1480/1580 nm WDMs with a range of –OH absorption levels at 1390 nm were fabricated. Replacing the ceramic furnace fabrication process typically used for fabricating standard fused fiber WDMs, so-called “wet” WDMs with higher –OH absorption level were produced using a bare flame process. Figure 11(a) shows the excess losses around the –OH absorption peak of both a standard high power 1480/1580 nm WDM and a “wet” 1480/1580 nm WDM. The excess loss was calculated by measuring the difference between the power input to the WDM (at P1) and the total powers at P2 and P3. The black and the red data sets show the absorption spectra around the –OH absorption peak measured from one of the standard WDMs and one of the “wet” WDMs respectively.
Testing was then carried out on a “wet” 1480/1580 nm WDM using the same configuration described in section 1. The optical power at the output 1480 nm channel P2 was monitored and is plotted in Fig. 11(b). In comparison to the optical power change shown in Fig. 3 as measured from a standard WDM, the optical power change happened immediately following the activation of the Raman laser, without any induction period, dropping from 111 W to 107.5 W in only 20 hours as opposed to the more than 30 hours required for a similar drop in the standard WDM. This measurement indicates that the degree of –OH absorption is directly related to the rate of degradation in the WDM
3. Solution to the degradation of the high power WDM
With the confirmation that higher –OH absorption increases the high power WDM degradation rate a solution was proposed based on filtering out the out-of-band Stokes line near 1390 nm. This solution makes use of a slanted fiber Bragg grating (SFBG) which has been used in the sensing area  to remove this Stokes line from the laser light produced by the Raman pump. Figure 12(a) shows the transmission spectrum of a 1390 nm SFBG that was designed and developed for just this purpose. The transmission spectrum shows a high insertion loss (isolation) of more than 12 dB around the 1390 nm line with a bandwidth of >50 nm.
After splicing the 1390 nm SFBG to the high power Raman laser, the optical spectrum at the 1390 nm SFBG output was measured and compared with the optical spectrum measured directly at the Raman laser output. It is clearly shown in Fig. 12(b) that adding the 1390 nm SFBG successfully removes the 1390 nm Stokes line of the Raman laser with a suppression of more than 12 dB.
With the 1390 nm SFBG spliced between the high power Raman laser and a fresh high power 1480/1580 nm WDM, the optical power at the WDM’s 1480 nm output channel P2 was monitored for more than 300 hours. These data are plotted in Fig. 13(a). During this extended operation, no obvious power degradation occurred. To further confirm that the power degradation had been successfully prevented, the optical spectral transfer functions of P1 to P2 of the high power WDM both before and after the 300 hour burn-in test were measured as shown in Fig. 13(b). The polynomial fitted curves reveal that the minimum coupling ratio wavelength shift was less than 0.1 nm, well within the curve-fitting tolerance. By implementing the 1390 nm SFBG to filter out the 1390 nm Stokes line from the high power Raman laser, the power degradation issue of the high power 1480/1580 nm WDM has been successfully resolved. This analysis and associated solution of implementing 1390 nm SFBG would be applicable to any high power fused coupler, including WDMs and splitters spliced to high power cascaded Raman fiber lasers as studied here.
As one last note, phosphosilicate glass fiber has been reported as a gain medium for Raman laser systems around 1480 nm . In such systems there are only two Stokes shifts, 1060 nm to 1240 nm and 1240 nm to 1480 nm, fully avoiding 1390 nm Stokes line. Therefore no such degradation should occur in these systems containing WDMs and other high power fused coupler devices.
In conclusion, the high power handling capability of the fused biconical taper 1480/1580 nm WDM and observed power degradation due to the coupling ratio shift have been evaluated and investigated. Several tests have been designed and performed to identify the root cause of this degradation. It has been experimentally demonstrated that this WDM degradation is the result of both the interaction between the high power 1480 nm line, a 1390 nm out-of-band Stokes line, and the –OH content of the glass optical fiber.
A 1390 nm SFBG was introduced to remove the 1390 nm content, thus avoiding the detrimental interaction of this light with the –OH ions present in the fiber. In general, this discovery and solution for WDM power degradation will have a significant impact on the improvement of both the stability and the reliability of high power fiber amplifier and laser systems using a high power laser line close to the –OH absorption peak.
Further, the simulated refractive index change of the fused biconical taper section of the degraded 1480/1580 nm WDM described above was 0.015. This permanent refractive index change is even higher than the latest reported peak refractive index change in a hydrogen-loaded optical fiber with extremely high germanium concentration . The mechanisms behind this induced high refractive index change in the fused biconical taper WDMs is worth further investigation and may provide significant insight on further improvements to the refractive index contrast related sensitivity of devices such as fiber Bragg gratings and long period gratings.
Elements of this work were sponsored by Navy Contract N68335-11-C-0020. NAVAIR Public Release SPR 2013-469 Distribution Statement A-“Approved for Public release; distribution is unlimited”.
References and links
1. V. R. Supradeepa, J. W. Nicholson, and K. Feder, “Continuous wave erbium-doped fiber laser with output power of >100 W at 1550 nm in-band core-pumped by a 1480nm Raman fiber laser,” in CLEO: Science and Innovations, OSA Technical Digest (online) (Optical Society of America, 2012), paper CM2N.8.
2. J. W. Nicholson, J. Fini, J. Phillips, A. DeSantolo, K. Feder, X. Liu, P. Westbrook, E. Monberg, F. DiMarcello, C. Headley, and D. DiGiovanni, “Higher order mode erbium-doped fiber amplifiers,” in Optical Fiber Communication Conference, OSA Technical Digest (Optical Society of America, 2012), paper OM3C.5. [CrossRef]
3. J. W. Nicholson, A. M. DeSantolo, S. Ghalmi, J. M. Fini, J. Fleming, E. Monberg, F. DiMarcello, and S. Ramachandran, “Nanosecond pulse amplification in a higher-order-mode erbium-doped fiber amplifier,” in Conference on Lasers and Electro-Optics, OSA Technical Digest (CD) (Optical Society of America, 2010), paper CPDB5. [CrossRef]
4. C. Headley and G. P. Agrawal, Raman amplification in fiber optical communication systems (Elsevier, 2005).
5. D. Georgiev, V. P. Gapontsev, A. G. Dronov, M. Y. Vyatkin, A. B. Rulkov, S. V. Popov, and J. R. Taylor, “Watts-level frequency doubling of a narrow line linearly polarized Raman fiber laser to 589nm,” Opt. Express 13(18), 6772–6776 (2005). [CrossRef] [PubMed]
7. V. R. Supradeepa and J. W. Nichsolson, “Power scaling of high-efficiency 1.5 µm cascaded Raman fiber lasers,” Opt. Lett. 38(14), 2538–2541 (2013). [CrossRef]
8. T. Yilmaz, L. Vaissie, M. Akbulut, D. M. Gaudiosi, L. Collura, T. J. Booth, J. C. Jasapara, M. J. Andrejco, A. D. Yablon, C. E. Headley III, and D. J. DiGiovanni, “Large-mode-area Er-doped fiber chirped-pulse amplification system for high-energy sub-picosecond pulses at 1.55 μm,” Proc. SPIE 6873, 68731I, 68731I-8 (2008). [CrossRef]
10. K. Jedrzejewski, “Biconical fused taper – a universal fibre devices technology,” Opto-Electron. Rev. 8(2), 153–159 (2000).
11. R. C. Gutierrez, G. M. Swift, S. Dubovitsky, R. K. Bartman, C. E. Barnes, and L. Dorsky, “Radiation effects on fused biconical taper wavelength division multiplexers,” IEEE Trans. Nucl. Sci. 41(6), 1950–1957 (1994). [CrossRef]
12. R. J. Orazi, S. D. Russell, T. T. Vu, and P. K. L. Yu, “UV fine tuning of narrow channel fused fibre wavelength division multiplexing couplers,” Electron. Lett. 33(2), 154–155 (1997). [CrossRef]
13. A. F. Fernandez, B. Brichard, and F. Berghmans, “In situ measurement of refractive index changes induced by gamma radiation in germanosilicate fibers,” IEEE Photon. Technol. Lett. 15(10), 1428–1430 (2003). [CrossRef]
14. A. D. Yablon, “Multi-wavelength optical fiber refractive index profiling by spatially resolved Fourier transform spectroscopy,” J. Lightwave Technol. 28(4), 360–364 (2010). [CrossRef]
16. V. I. Karpov, E. M. Dianov, V. M. Paramonov, O. I. Medvedkov, M. M. Bubnov, S. L. Semyonov, S. A. Vasiliev, V. N. Protopopov, O. N. Egorova, V. F. Hopin, A. N. Guryanov, M. P. Bachynski, and W. R. L. Clements, “Laser-diode-pumped phosphosilicate-fiber Raman laser with an output power of 1 W at 1.48 mum,” Opt. Lett. 24(13), 887–889 (1999). [CrossRef] [PubMed]
17. O. I. Medvedkov, S. A. Vasiliev, P. I. Gnusin, and E. M. Dianov, “Photosensitivity of optical fibers with extremely high germanium concentration,” Opt. Mater. Express 2(11), 1478–1489 (2012). [CrossRef]