High power handling shape memoryalloy optical fiber connector

Dominic Faucher,1,* Alex Fraser,1,2 Patrick Zivojinovic,2 Xavier Pruneau Godmaire,1,2

Éric Weynant,2 Martin Bernier,1 and Réal Vallée1

1Center for Optics, Photonics and Lasers (COPL), Université Laval, Québec, Canada G1K 7P42Phasoptx, Inc., 2740 rue Einstein, Québec, Canada G1P 4S4

*Corresponding author: dominic.faucher.1@ulaval.ca

Received 30 June 2009; revised 11 September 2009; accepted 25 September 2009;posted 25 September 2009 (Doc. ID 113240); published 12 October 2009

We present a novel shape memory alloy-based optical fiber splicing device that can provide robust, lowloss, and high power handling splices between single-mode fibers of identical or entirely different glasscompositions. The achieved splice loss was as low as 0:12dB between two SMF-28 fibers with an averagevalue of 0:23dB. To the best of our knowledge, this is the first demonstration of a purely mechanicalsplicing device that can withstand optical powers in excess of 10W with various combinations of silicaand fluoride fibers. The device can be used in moderate to high power all-fiber components, especiallythose involving junctions unsuitable to fusion splicing, such as fiber lasers and amplifiers based on fluo-ride, chalcogenide, or microstructured fibers. © 2009 Optical Society of America

OCIS codes: 060.2340, 140.3510.

1. Introduction

The development of fusion splicing technology hasgreatly benefitted the optical telecommunicationsand fiber laser industries due to its low loss, high re-peatability, and ease of use. However, this techniquecannot readily be applied to various specialty fiberssuch as plastic optical fibers, microstructured opticalfibers (MOFs), and chalcogenide or fluoride glass fi-bers (FGFs). MOFs are usedmainly for their unusualdispersive and nonlinear properties but are difficultto splice to conventional silica fibers [1]. In the case ofFGFs, their low phonon energy makes them ideal fortransmitting and amplifying mid-IR and visiblelight. Recent reports of highly reflective fiber Bragggratings (FBGs) being written in FGFs [2], all-fiberlasers based on these FBGs [3], and >9W of outputpower from a mid-IR fiber laser [4] have motivatedthe development of an efficient way to splice FGFstogether as well as to silica fibers. Fusion splicingof FGFs has actually been demonstrated with a

low-current arc fusion splicer [5], a filament fusionsplicer [6], or a CO2 laser [7]. However, these meth-ods cannot be readily applied for splicing silica andFGFs together due to the significantly different melt-ing temperatures of the two glasses (300–400 °C forFGFs, compared to about 1600 °C for fused silica).Purely mechanical junctions have been realized byinserting the fibers in a glass capillary or a V groove,or by using a UV-cured adhesive [8]. However, thesemethods involve relatively complex and time-consuming procedures that typically result in lowpower handling junctions. Recently, shape memoryalloys have proven to be excellent candidates for me-chanical splicing of single-mode and multimode opti-cal fibers [9,10]. In fact, their intrinsic hyperelasticbehavior is well suited for high power handling ap-plications because it allows the fibers to expand al-most freely in response to the heat generated by thesplice losses.

In this paper, we present an efficient, fast, and reu-sable method of joining two entirely different single-mode optical fibers of slightly different cladding dia-meters. More importantly, we show the excellentpower handling of the splices in various situations,

0003-6935/09/305664-04$15.00/0© 2009 Optical Society of America

5664 APPLIED OPTICS / Vol. 48, No. 30 / 20 October 2009

including two identical undoped FGFs, an undopedFGF with a 2.5 wt.% thulium-doped FGF, and finallybetween a silica and a FGF with closely matchedmode field diameters (MFDs).

2. Description of the Shape Memory Alloy Connector

A picture of the device is shown in Fig. 1. The CuAlBeshape memory alloy (SMA) used as the basic materi-al for the connector was chosen for its hyperelasticbehavior, which originates from a phase transforma-tion from the martensitic phase to the austeniticphase. It can support strain as high as 10% whensubmitted to a stress or a temperature change. Forthe present device, the phase transformation is acti-vated through the application of a mechanical stressvia an opening tool. The CuAlBe alloy was also de-signed to achieve a phase transformation tempera-ture between −80 °C and −100 °C, which preventsany undesirable thermally induced phase transfor-mation from occurring in the device’s range of oper-ating temperatures.The connector is fabricated by drilling a hole in the

2mm long, 1:4mmwide cylindrical SMA, after whicha notch is carved along its length down to the centralhole, whose diameter is 122 μm across the centralportion. The hole is funneled at both ends up to a dia-meter of 250 μm to ensure easy fiber insertion. Thecleaved or polished unclad fibers are aligned tothe device axis using V-grooves. The notch allowsthe hole to be slightly enlarged as the opening tool,a wedge-shaped insert, is pressed into it. This forcecreates a high stress at the bottom of the notch thatleads to a phase transformation of the SMA, accom-panied by a large elastic deformation. The resultingenlargement of the central hole allows an easy inser-tion of the 125 μm diameter fibers on each side. Oncethe fibers are butted against each other inside theconnector, the wedge is removed from the notch. Thatcompletes the alignment procedure, which takes onlya few seconds and can easily be fully automated. The

almost radially symmetric compressive stress thatthe SMA exerts on each fiber keeps them solidly inplace while virtually eliminating any cladding-to-sleeve eccentricity, even when fibers of slightly differ-ent diameters (perhaps up to a few micrometers) areused. This eliminates one of the dominant contribu-tions to splice loss in sleeve-based fiber splices, aslong as right angle cleaved fibers with similar MFDsand small core-to-cladding eccentricities are used.The hyperelastic behavior of the SMA limits radialstresses on both fibers when they expand thermally,even if their thermal expansion coefficients are dif-ferent. Thus the alignment between the fibers is pre-served due to the symmetric design of the splice.Slightly pulling the fibers demonstrates that the re-sulting connection is sufficiently robust for freely ma-nipulating the splice with some care, so that it caneasily be further reinforced by a suitable packagingprocess for increased robustness. An index-matchinggel can be used to further reduce splice loss, at theexpense of a potentially lower power handling. How-ever, we used no such gel in our experiments.

3. Splice Loss Performance

We first investigated the performance of the devicewith respect to maximum splice loss and repeatabil-ity using SMF-28 fibers. The fibers were cleaved witha high-quality Fitel S325A diamond blade cleaverthat provided cleave angles under 1° with an averageangle of 0:5°. The splice loss was measured bylaunching the light of a highly stable 1550nm laserdiode in a fiber segment and measuring the outputpower with an optical powermeter. The fiber wasthen cut in its middle portion, cleaved, and spliced.The splice loss is calculated by comparing the trans-mitted power before and after the splice. Figure 2shows the statistical distribution of the splice lossof 20 such junctions realized with the same connectorbetween two SMF-28 fiber segments. The minimum,maximum, and average splice losses achieved were

Fig. 1. (Color online) Picture showing the shape memory alloy(SMA) connector used in this study with two SMF-28 fibers splicedtogether.

Fig. 2. Splice loss statistical distribution using two SMF-28fibers.

20 October 2009 / Vol. 48, No. 30 / APPLIED OPTICS 5665

0.12, 0.33, and 0:23dB, respectively, with a measure-ment accuracy of �0:02dB. Monitoring one suchjunction over an observation period of 75 days re-vealed no sign of deterioration. The dominant contri-bution to the loss is attributed to the air gap betweenthe fibers, whose thickness depends on the cleave an-gle magnitude and relative orientation of each fiberend. In fact, it is worth mentioning that rotating thefibers along their axes can reduce the thickness ofthis air gap as well as the core-to-core lateral offsetarising from significant core–cladding eccentricity.Although this can lead to a reduction in the overallsplice loss, no such attempt was made in all the ex-periments described in this paper.

4. Power Handling

We also investigated the power handling capabilityof the splices with various fiber combinations. Table 1states the properties of the four fibers that wereused. The silica fibers were cleaved, while the FGFswere carefully polished to a clean, right angled fin-ish. The loss of each junction was first measuredright after the alignment procedure at 1550nm. Wethen tested the power handling of each splice bylaunching a CW Yb3þ fiber laser at 1064nm throughthem. All the fibers were slightly multimode at thiswavelength. The splice was held in contact with ametallic mount, but no active cooling was used. Ourfirst experiment was conducted using two SMF-28 fi-ber segments (fiber A) with a splice loss of 0:3dB at1550nm. The launch efficiency at 1064nm was mea-sured at 76% with a separate fiber segment. Figure 3shows the linear relationship between launched andoutput power up to the maximum available launchedpower of 12:2W. Each time the power was increased,a 2min pause wasmade to ensure a steady-state con-dition. Themaximum power was thenmaintained foran observation period of 3 h without any sign of de-terioration. The slope of the curve implies a spliceloss of 0:4dB at 1064nm.Then, we repeated this experiment using two iden-

tical FGF segments (fiber B). This time, the spliceloss measured at 1550nm was 0:4dB. Again, we ob-served a linear relationship between launched andoutput power up to the maximum available power.With a launch efficiency of 76%, the implied spliceloss at 1064nm was 0:6dB.In order to more thoroughly test the power dissipa-

tion capability of the SMA connector, we also splicedthe same FGF (fiber B) to a mode-matched, 10 cmlong, 2:5mol% thulium-doped FGF (fiber C). Thesplice loss of 0:4dB at 1550nm was measured by

launching through the doped fiber. The 1064nm lightwas then launched through the undoped fiber, andthe result of the experiment is shown in Fig. 3, wherethe output power is magnified ten times for clarity.The high pump absorption at low power graduallydecreases at higher power owing to strong groundstate bleaching. Upon reaching a launched powerof 5W, a strong scattering center located 2 cm afterthe splice appeared in the doped fiber. Examinationof the fiber revealed a yellow coloration of the jacket,indicating that the fiber temperature must have in-creased significantly. We believe this demonstrationis an indication of the high power handling capabilityof the device, which we attribute to its inherent hy-perelastic behavior as well as to its good thermal con-tact with the fiber. In fact, the SMA allows thermalexpansion of the fibers to occur, which limits the ac-cumulation of a thermally induced stress that wouldotherwise lead to cracking. Also, the low thermal re-sistance between the device and the fiber as well asits small footprint and high heat conductivity ensurean efficient heat removal rate.

Finally, we spliced a silica fiber and a FGF together(fibers B and D) to investigate the performance of thedevice with heterogeneous fiber junctions. This time,the significantly different MFDs of each fiber led to aslightly higher loss of 0:8dB at 1550nm. For thepower handling test, the Yb3þ laser was launched

Table 1. Properties of the Fibers Used: Core Diameter (a), Cladding Diameter (b), Numerical Aperture (NA), LP11 Cutoff Wavelength (λc ), andCore–Cladding Eccentricity (ε)

Fiber Glass a ðμmÞ b ðμmÞ NA λc ðμmÞ ε ðμmÞFiber A SiO2 8.2 125:0� 0:7 0.14 1.26 ≤ 0:5Fiber B ZBLAN 6.8 124:1� 0:8 0.17 1.48 ≤ 0:8Fiber C Tm:ZBLAN 6.8 123:7� 0:8 0.17 1.48 ≤ 0:7Fiber D SiO2 6.2 124:7� 1 0.20 1.47 ≤ 0:6

Fig. 3. Power transmitted by each splice as a function of launchedpower at 1064nm using two SMF-28 fibers (▪, fiber A), two iden-tical fluoride fibers (▴, fiber B), a silica and a fluoride fiber (▿, fi-bers B and D) as well as a doped and an undoped fluoride fiber (□,fibers B and C).

5666 APPLIED OPTICS / Vol. 48, No. 30 / 20 October 2009

through the silica fiber with a launch efficiency of82%. As shown in Fig. 3, a linear fit of the resultsindicates a constant splice loss of about 1:5dB at1064nm throughout the whole power range. Despitethis relatively high loss, no sign of splice failure wasobserved up to the maximum available power. Sincethe achieved splice loss was relatively high in most ofour power handling experiments, mainly due to sig-nificant core-cladding eccentricity and MFD mis-match, we expect the device to be able to sustainmuch higher powers. Thus we believe this device canbe used for moderate to high power optical fiber com-ponents involving junctions unsuitable to fusionsplicing. These include both homogeneous and het-erogeneous junctions between fibers of different geo-metries and compositions, including plastic opticalfibers, FGFs, and tellurite or chalcogenide glass fi-bers as well as MOFs.

5. Summary

We presented a new SMA-based fiber splicing devicethat can provide robust, reusable, low-loss splices ina compact package. We believe this is the first de-monstration of a mechanical splice that can with-stand optical powers in excess of 10W, even withentirely different fibers. Thanks to an adhesive-freedesign, the simple splicing procedure takes only afew seconds and can be fully automated. The varioussplices realized with different fiber combinationscould withstand optical powers of at least 13Wwithout any sign of deterioration. The high powerhandling capability of the device wasmore clearly de-monstrated with the 2:5mol:% thulium-doped FGFexperiment, during which the fiber itself was da-maged before we could observe any splice failure.Our results suggest that much higher powers shouldbe attainable. This device could be a key componentin high power fiber lasers or supercontinuum sourcesinvolving fibers that are a challenge to splice, such asFGFs or MOFs. It could allow the robust integrationof the pump source and specialty fiber, which re-mains a challenge at high launched powers. Moretesting involving the long-term stability of the de-

vice, higher power handling, and suitability for othertypes of specialty fibers is currently under way.

The authors acknowledge the financial support ofthe Canadian Institute for Photonics Innovations(CIPI), the National Sciences and Engineering Re-search Council of Canada (NSERC) and the Centerfor Optics, Photonics and Lasers (COPL).

References

1. B. Bourliaguet, C. Paré, F. Émond, A. Croteau, A. Proulx, andR. Vallée, “Microstructured fiber splicing,” Opt. Express 11,3412–3417 (2003).

2. M. Bernier, D. Faucher, R. Vallée, A. Saliminia, G. Androz,Y. Sheng, and S. L. Chin, “Bragg gratings photoinduced inZBLAN fibers by femtosecond pulses at 800nm,” Opt. Lett.32, 454–456 (2007).

3. G. Androz, D. Faucher, M. Bernier, and R. Vallée, “Monolithicfluoride fiber laser at 1480nm using fiber Bragg gratings,”Opt. Lett. 32, 1302–1304 (2007).

4. X. Zhu and R. Jain, “10W-level diode-pumped compact2:78 μm ZBLAN fiber laser,” Opt. Lett. 32, 26–28 (2007).

5. B. B. Harbison, W. I. Roberts, and I. D. Aggarwal, “Fusionsplicing of heavy metal fluoride glass optical fibers,” Electron.Lett. 25, 1214–1215 (1989).

6. B. Srinivasan, M. Erlandsson, G. S. Feller, E. W. Mies, andR. K. Jain, “Reproducible fusion splicing of low melting point(fluoride) optical fibers with the use of a stable heat source,” inOptical Fiber Conference, OSA Technical Digest Series (OSA,1997), paper TuB1.

7. L. Rivoallan and J. Y. Guilloux, “Fusion splicing of fluo-ride glass optical fibre with CO2 laser,” Electron. Lett. 24,756–757 (1988).

8. M. M. Kozak, W. Kowalsky, and R. Caspary, “Low-loss gluesplicing method to join silica and fluoride fibers,” Electron.Lett. 41, 21–22 (2005).

9. G. Trouillard, P. Zivojinovic, M. Bergeron, A. Fraser, andE.Weynant, “New connectivity solution for optical fibers usingPhasOptx shape memory alloy Optimend,” in Proceedings ofthe Avionics, Fiber-Optics and Photonics Conference (IEEE,2008), pp. 89–90.

10. E. Weynant, P. Zivojinovic, E. Menu, A. Fraser, andM. Berger-on, “Connector for multiple optical fibers and installation ap-paratus,” International patent application WO 2008/151445A1 filed 16 June 2008.

20 October 2009 / Vol. 48, No. 30 / APPLIED OPTICS 5667