Hydrofluoric acid flow etching of low-loss subwavelength-diameter biconical fiber tapers

An etch method based on surface tension driven flows of hydrofluoric acid microdroplets for the fabrication of low-loss, subwavelength-diameter biconical fiber tapers is presented. Tapers with losses less than 0.1 dB/mm are demonstrated, corresponding to an order of magnitude increase in the optical transmission over previous acid-etch techniques. The etch method produces adiabatic taper transitions with minimal surface corrugations. A biconical fiber taper fabricated using this method is used to demonstrate an erbium doped silica microsphere laser. ©2010 Optical Society of America OCIS codes: (220.4241) Nanostructure fabrication; (220.4610) Optical fabrication; (230.2205) Fiber devices and optical amplifiers. References and links 1. J. C. Knight, G. Cheung, F. Jacques, and T. A. Birks, “Phase-matched excitation of whispering-gallery-mode resonances by a fiber taper,” Opt. Lett. 22(15), 1129–1131 (1997), http://www.opticsinfobase.org/abstract.cfm?URI=ol-22-15-1129. 2. J. P. Laine, B. E. Little, and H. A. Haus, “Etch-eroded fiber coupler for whispering-gallery-mode excitation in high-Q silica microspheres,” IEEE Photon. Technol. Lett. 11(11), 1429–1430 (1999). 3. M. Cai, O. Painter, K. J. Vahala, and P. C. Sercel, “Fiber-coupled microsphere laser,” Opt. Lett. 25(19), 1430– 1432 (2000), http://www.opticsinfobase.org/abstract.cfm?URI=ol-25-19-1430. 4. M. T. Rakher, R. Bose, C. W. Wong, and K. Srinivasan, “Spectroscopy of 1.55 μm PbS quantum dots on Si photonic crystal cavities with a fiber taper waveguide,” Appl. Phys. Lett. 96(16), 161108 (2010). 5. W. Liang, Y. Huang, Y. Xu, R. K. Lee, and A. Yariv, “Highly sensitive fiber Bragg grating refractive index sensors,” Appl. Phys. Lett. 86(15), 151122 (2005). 6. K. Q. Kieu, and M. Mansuripur, “Biconical fiber taper sensors,” IEEE Photon. Technol. Lett. 18(21), 2239–2241 (2006). 7. A. Leung, P. M. Shankar, and R. Mutharasan, “A review of fiber-optic biosensors,” Sens. Actuators B Chem. 125(2), 688–703 (2007). 8. T. A. Birks, W. J. Wadsworth, and P. St. J. Russell, “Supercontinuum generation in tapered fibers,” Opt. Lett. 25(19), 1415–1417 (2000), http://www.opticsinfobase.org/abstract.cfm?uri=ol-25-19-1415. 9. M. A. Foster, A. C. Turner, M. Lipson, and A. L. Gaeta, “Nonlinear optics in photonic nanowires,” Opt. Exp. 16, 1300–1320 (2008), http://www.opticsinfobase.org/abstract.cfm?URI=oe-16-2-1300. 10. G. Brambilla, V. Finazzi, and D. J. Richardson, “Ultra-low-loss optical fiber nanotapers,” Opt. Exp. 12, 2258– 2263 (2004), http://www.opticsinfobase.org/abstract.cfm?URI=OPEX-12-10-2258. 11. L. Tong, R. R. Gattass, J. B. Ashcom, S. He, J. Lou, M. Shen, I. Maxwell, and E. Mazur, “Subwavelengthdiameter silica wires for low-loss optical wave guiding,” Nature 426(6968), 816–819 (2003). 12. G. Brambilla, “Optical fiber nanowires and microwires: a review,” J. Opt. 12(4), 043001 (2010). 13. H. S. Haddock, P. M. Shankar, and R. Mutharasan, “Fabrication of biconical tapered optical fibers using hydrofluoric acid,” Mater. Sci. Eng. B 97(1), 87–93 (2003). 14. A. W. Snyder, “Coupling of modes on a tapered dielectric cylinder,” IEEE Trans. Microw. Theory Tech. 18(7), 383–392 (1970). 15. R. Tadmor, “Marangoni flow revisited,” J. Colloid Interface Sci. 332(2), 451–454 (2009). 16. U. Roth, O. Paulus, and U. Menyes, “Surface activity of amphiphiles in hydrogen fluoride – water solutions,” Colloid Polym. Sci. 273(8), 800–806 (1995). 17. M. T. Lee, “Reaction of high-silica optical fibers with hydrofluoric acid,” J. Am. Ceram. Soc. 67(2), C-21–C-22 (1984). 18. S. H. Behrens, and D. G. Grier, “The charge of glass and silica surfaces,” J. Chem. Phys. 115(14), 6716–6721 (2001). 19. J. A. Voorthuyzen, K. Keskin, and P. Bergveld, “Investigations of the surface conductivity of silicon dioxide and methods to reduce it,” Surf. Sci. 187(1), 201–211 (1987). 20. T. J. Kippenberg, Nonlinear Optics in Ultra-high-Q Whispering-Gallery Optical Microcavities (Doctoral dissertation, California Institute of Technology, 2004), pp. 30. #132883 $15.00 USD Received 5 Aug 2010; revised 1 Oct 2010; accepted 2 Oct 2010; published 8 Oct 2010 (C) 2010 OSA 11 October 2010 / Vol. 18, No. 21 / OPTICS EXPRESS 22593


Introduction
Subwavelength-diameter biconical fiber tapers (SBFTs) have diverse applications ranging from evanescent coupling [1][2][3][4] to optical sensing [5][6][7] and nonlinear optics [8,9]. SBFTs are generally produced from single-mode-fibers (SMFs) and consist of three contiguous regions: a waist region of minimum diameter connected to transition regions on both sides that taper into the nominal fiber diameter ( Fig. 1(d)). Controlled heating and pulling of a SMF is the most widespread method of fabricating low-loss SBFTs, with optical losses of about 0.1 dB/mm for waist diameters around 500 nm [10][11][12]. Alternatively, fiber tapers can also be formed by chemical etching using hydrofluoric acid (HF). However, a main challenge with existing acidetch techniques is the resultant high optical losses. To date, demonstrated chemically etched biconical fiber tapers have exhibited insertion losses in excess of 10 dB even at micron-scale waist diameters [2,13]. The high losses have been attributed to the formation of surface corrugations during the etch [2].
In this article, we present a new acid-etch method to produce SBFTs with losses comparable to heat-pulled tapers. Instead of the acid baths in existing etch techniques, we use a micro-droplet of HF and surface tension driven flows of the acid for the etch. At a wavelength of 1550 nm, the losses through our SBFTs are generally below 0.1 dB/mm for waist diameters between 500 nm to 1 µm, and losses below 0.05 dB/mm are easily achievable for waist diameters above 1 µm. The production of these low-loss tapers requires a combination of smooth fiber surfaces to minimize scattering losses, as well as adiabatic taper transition regions to reduce power coupling between core and cladding modes [14]. Figure 1(a) shows a schematic of our etch setup. To fabricate a taper, we began by mechanically stripping a standard SMF-28 (Corning) fiber and cleansing it with isopropyl alcohol. The fiber was then held by clamps across an HF-resistant Nalgene petri dish. We then injected 100 µL of 49% HF (by weight) onto the dish. Due to the hydrophobicity of the Nalgene dish, an HF droplet formed so that the dish could be raised by a translation stage to immerse the fiber in the droplet to initiate the etch. The droplet shape allowed the length of the waist region to be controlled by the immersion depth; short SBFTs required a shallower immersion compared to long SBFTs. The apparatus was situated on an optical table for mechanical stability, and the experiments were performed at room temperatures of 22°C ± 1°C. Each etch lasted approximately 75 minutes and only required occasional monitoring. We terminated an etch by extracting the HF using a pipette and flushing the fiber with deionized water. Figure 1(b) shows a typical etch progression.

Transition region formation
During the etch, HF flowed from the droplet and traversed along the fiber. This phenomenon can be explained by the Marangoni effect, a surface tension driven flow originating from the droplet meniscus at the droplet-fiber interface. To understand this phenomenon, we note that the surface tension,  , can be expressed as where HF C is the HF concentration, and x is the direction of HF traversal [15]. Due to the higher volatility of HF molecules compared to water, the consumption of HF by the etch reaction, and the larger surface-area to volume ratio at the menisci regions near the dropletfiber interface compared to the bulk droplet, 0 HF dC dx  . Furthermore, since 0 HF d dC   for HF [16], it follows that 0 d dx   in Eq. (1), supporting our observation of a net outflow of acid from the droplet. Figure 1(c) shows an optical micrograph of the outflow along a taper. The Marangoni flow caused the etch to be a traveling process, rather than a static case in an acid bath. This process introduced two benefits: first, as the HF traversed along the transition, its concentration decreased, forming a graded diameter profile and elongating the transition region; second, the continuous motion of the etchant eliminated surface corrugations on the taper transition by removing pockets of HF that aggregated on the fiber surface.

In situ etch monitoring
We measured the waist diameters and transmission losses to monitor the etch progress and taper quality. Figure 2(a) shows a measurement of the waist diameter as a function of etch time for a 5.5 mm long, 920 nm diameter SBFT. The rate of change in the waist diameter at time t is proportional to the etchant concentration and is given by: where () Dt is the waist diameter, and ()

HF
Ct is the HF concentration [13]. Using Eq. (2) and Fig. 2(a), we find that the HF concentration decreased to approximately one-third of its initial value at t = 4000 s. The nonlinear dependence of waist diameter on etch time was in direct contrast to previous theoretical and experimental results [13,17], and was caused by the higher evaporation rate of HF molecules compared to water. This effect is unique to our technique and absent in acid bath etching because of the large surface-area to volume ratio of the HF droplet. The slower etch rates contributed to smooth waist surfaces and greater control over final waist diameters. Previous research demonstrated shorter etch times but resulted in tapers with significantly higher losses [13].  Since glass surfaces become charged in aqueous environments and the charge density increases with decreasing radii [18], if the SBFTs were not treated after the deionized water rinse, their optical transmission rapidly deteriorated in the ambient environment due to a strong accumulation of dust particles on the surface. By exposing the waist surface to hexamethyldisilazane (HMDS) after the rinse [19], we found that the surface could be substantially passivated, and SBFT losses deteriorated by < 1 dB even when left in the ambient for over 36 hours. Figure 3(a) summarizes the optical losses of a series of SBFTs with varying diameters obtained using our setup. The lengths of the waist regions for these tapers were approximately 5.5 mm. An insertion loss of 0.37 dB (0.07 dB/mm) was observed for a waist diameter of 480 nm, and micron-scale diameter tapers had losses slightly above 0.1 dB (0.02 dB/mm). Figure  3(b)-(c) are scanning electron microscope (SEM) images of the gold-sputtered waist region showing the top and cross-section views of the 480 nm diameter SBFT respectively. The waist exhibited excellent surface smoothness. The cross-section of Fig. 3(c) is not at the thinnest portion of the waist, but illustrates the production of cross-sections with good circularity at a diameter of about 1 µm. The slight edge visible at the top of the fiber in Fig. 3(c) was due to the etch asymmetry between the top and bottom of the fiber, and was more pronounced for shallower immersion depths (and therefore shorter waists) which could result in elliptical fiber cross-sections. When viewed under an optical microscope with halogen lamp illumination, the SBFT waist appeared blue-green as in Fig. 3(d). The color was due to interference effects and provided a means to roughly estimate waist diameters [20].

Insertion loss measurements
We etched dozens of tapers using this method and have not observed the losses to exceed 0.15 dB/mm. We emphasize that the per-length, normalized loss is only a rough figure-ofmerit calculated by the ratio of taper loss to waist length. It does not properly account for transition region losses and waist diameter non-uniformities; instead, the insertion loss is a more meaningful metric.

Application: fiber taper coupled erbium microsphere laser
As a demonstration of a potential application of our SBFTs, we fabricated a fiber taper coupled microsphere laser similar to that in Ref [3]. The waist diameter of the taper was approximately 1 µm. The taper coupled 980 nm pump light into an Er 3+ doped silica microsphere and out-coupled the laser light. The Er 3+ microsphere was 30 µm in diameter and was formed using a fusion splicer by arc-melting an HF-etched highly doped Er 3+ fiber taper (Leikki ER110-4/125). The etching removed the cladding from the doped fiber to preserve the Er 3+ concentration during the arc-melting process. Figure 4(a) shows the output laser power as a function of launched pump power for the dominant lasing mode at 1534 nm measured using an optical spectrum analyzer. The laser threshold occurred at a launched pump power of about 2.2 mW. The insets show the microsphere coupled to the fiber waist region and the upconversion fluorescence. Figure 4(b) shows the multi-mode laser spectrum at a launched pump power of 6 mW.

Conclusion
In summary, we have demonstrated the fabrication of low-loss, subwavelength-diameter biconical fiber tapers (SBFTs) using HF etching. Low surface roughness and high taper adiabaticity are general characteristics of the SBFTs produced using this method. Millimeterscale waist length control could be achieved by changes in immersion depth within an HF microdroplet; however, production of centimeter-scale waist lengths requires the deliberate control of droplet geometry. In contrast to previous chemical etching methods, our present method produces SBFTs with optical losses under 0.1dB/mm, which is comparable to heatpulled SBFTs [12]. An added benefit of our technique is that the composition of the core is preserved; thus our approach applies equally well to doped, index-graded, or nonlinear fibers. Finally, by controlling the post-etch surface charge density, our SBFTs can either be passivated or further functionalized with select particles or molecules.