A silica-based highly acoustically-anti-guiding optical fiber, fabricated using a molten core approach employing spinel (MgAl2O4) for the first time, is presented. To our knowledge, this is the first truly single mode optical fiber fabricated from a precursor crystal. It is shown that MgO increases the acoustic velocity when added to silica (some physical parameters of MgO are identified) and that the Brillouin gain in the core is less than one third that in the cladding in one of the fibers. This results from a massive acoustic waveguide attenuation term that broadens the spectrum to well over 200 MHz. For the first time, to the best of our knowledge, this also enabled the validation of the intrinsic Brillouin line-width (~20 MHz) of pure silica in fiber form via a direct measurement.
©2013 Optical Society of America
Recently, it was demonstrated through the use of novel and nonobvious fabrication methods that optical fibers possessing equally unconventional and extraordinary properties may be realized. First, these fibers may have highly unconventional compositions such as ultra-high alumina-content fibers fabricated by drawing a sapphire (Al2O3) rod inside a silica (SiO2) tube . While alumina is not an unconventional dopant per se, other fiber variants such as YAG-derived fiber  allowed for the characterization of uncommon glass core constituents (i.e., yttria) to be made, hence adding yet another material to the optical fiber designers toolbox, especially within the realm of Brillouin scattering .
Second, the realization of fibers with unconventional, yet still practical, compositions then leads to optical fiber embodiments with unconventional performances. With respect to Brillouin scattering, this has been exemplified by fibers with Stokes’ frequencies that are independent of temperature (i.e., Brillouin athermal, or BAT) , and optical fibers that potentially have zero Brillouin activity, ZEBRA . It stands to reason then that the continued investigation of fibers fabricated using unconventional precursor materials, in particular those that remain ‘uncharacterized’ in terms of important fiber design parameters, can lead to a broadening of the scope of specialty optical fibers to a variety of new applications. For completeness, it is worth mentioning that these unconventional core composition fibers still possess pure silica claddings and so are generally compatible with conventional optical fibers, such as SMF-28TM.
Here, Brillouin measurements on a single-mode spinel-derived optical fiber (SpDF) with acoustic anti-guiding design are presented. These fibers are referred-to as ‘derived’ since the precursor core material is bulk crystalline spinel. The fiber is drawn at a temperature where the core phase is molten (hence the ‘molten core’ method appellation) and dissolves into it some of the cladding glass ultimately yielding a multicomponent all-glass fiber upon cooling. Since one starts with core compositions that are very high in selected species (e.g., pure Al2O3 in the sapphire-derived fiber) the dissolution of some silica into the melt still yields a high alumina content fiber. This is opposed to conventional CVD processes where dopants are added into the silica (e.g., solution-doping). In the molten core approach, the silica cladding essentially is doped into the core phase.
“Spinels” represent a mineralogical class of compounds with general formula A2+B23+O42-, where the A and B cations occupy the octahedral and tetrahedral sites, respectively, in the cubic (isometric; 4/m32/m) lattice. One of the more common compositions, indeed the one employed here and the one for which the general class is named, is the aluminum spinel, MgAl2O4, which possesses a 1-to-1 molar ratio of alumina (Al2O3) and magnesia (MgO) and is well-known for its use in high temperature refractories and transparent armor. In the present work, the aluminum spinel was selected in part in order to validate that MgO increases the acoustic velocity when added to silica, thus joining alumina  and yttria  with this property. More specifically, demonstrated in this paper are 1) the first SpDF and truly single-mode fiber derived from a precursor crystal; 2) that MgO increases the acoustic velocity when added to silica (some physical parameters of MgO are identified); 3) a core Brillouin spectrum that possesses the largest-reported waveguide-loss-induced spectral broadening; and 4) validation of the ~20 MHz intrinsic Brillouin line-width of silica in fiber form via a direct measurement.
2. Optical fiber and experimental details
In order to fabricate the optical fiber via the molten core approach, a rod of transparent ceramic spinel (Technology Assessment & Transfer, Inc., Millersville, MD) was sleeved into a high purity silica tube and drawn at 2175°C on a custom-built fiber draw tower (Clemson University). Because spinel is exceedingly refractory, with a (congruent) melting point of about 2135°C, a thick-walled silica tube (3.5 mm inner diameter, 40 mm outer diameter) was used for the cladding. By employing a thick-walled cladding tube, the effective draw temperature can be raised to above the melting point of the core phase, as required by the molten core method. Several hundred meters of 125 micron diameter silica-clad spinel-derived glass (i.e., now magnesium aluminosilicate) core fiber was drawn and coated with a conventional telecommunications UV-cured polymer coating, and two segments with different core sizes (and hence compositions), denoted hereafter as A and B, were utilized in the experiments. The refractive index profiles (RIPs) were measured (Interfiber Analysis, Livingston, NJ) at a wavelength of 1000 nm, and the compositions were determined by energy dispersive X-ray (EDX) analysis (Clemson University). A representative optical image of the fiber (Fiber A) is shown in Fig. 1(a) .In order to characterize the fibers’ acousto-optic properties, the Brillouin spectrum was measured using a heterodyne method described in detail elsewhere [5,6]. The heterodyne setup consists of a 1,534 nm source, an optical circulator, amplifier, and heterodyne receiver. The test fiber was spliced at the end of a circulator, through which a 1,534 nm source was launched and the backward propagating Stokes’ signal from the sample fiber passed back through the circulator to be filtered, amplified, and measured using the heterodyne receiver with a 1 MHz resolution. As an aside, an intermediate single mode fiber (SMF, Corning SMF-28TM) was used between the circulator and the sample SpDF fiber. The splicing of the SpDF to standard telecom-grade SMF was accomplished utilizing a standard SMF-SMF routine on a Vytran GPX glass processing system. An example of such a splice is shown in Fig. 1(b). The splice losses were estimated to be less than 0.1 and 0.4 dB for fibers A and B, respectively. The higher splice loss to Fiber B was due mainly to a mode mismatch between it and the SMF fiber (the mode diameter in Fiber B was smaller than in the SMF fiber).
3. Experimental results
Figure 2 provides the measured refractive index profile (RIP) and compositional profile of Fiber A (see Table 1 ) as an illustrative example. As expected, the index profile correlates with the MgO + Al2O3 content. More specifically, Fiber A exhibits a peak MgO + Al2O3 concentration of about 4.5 weight % whereas Fiber B (not shown graphically but detailed in Table 1) has about 5.5 weight percent at its peak value. The corresponding change in refractive index is about 0.01 and 0.013 for Fibers A and B, respectively. More detail is provided below on the underlying glass science of these compositions. However, it is appropriate at this stage to note that while these MgO and Al2O3 concentrations seem relatively low, they exceed the miscibility limit of the constituents in silica at conventional silica-based optical fiber processing temperatures (e.g., MCVD, OVD, VAD methods) and so such compositions would likely be difficult to achieve using standard approaches. For completeness, it is also noteworthy that the 1-to-1 MgO-to-Al2O3 (molar) stoichiometry of the precursor spinel is not preserved in the drawn fiber. For example, the peak weight percentages of MgO and Al2O3 in the Fiber A sample exemplified in Fig. 1 are 1.58 and 2.98, respectively, which converts to 2.37 and 1.76, respectively, in mole percent. The exact reason for this remains unclear, and is under further study, since it suggests the preferential removal of Al2O3 from the system despite the fact that MgO is considerably more volatile than Al2O3 at the draw temperature . This finding with regard to the MgO/Al2O3 ratio also contrasts that observed in evaporated films . Accordingly, such unexpected chemistries speak to the potential materials science complexity and versatility possible with the molten core method.
The Brillouin spectra are shown in Figs. 3(a) and 3(b). Both spectra have characteristic peaks corresponding to scattering from an acoustic wave in the pure-silica cladding (lower-frequency peak) and the spinel-derived magnesium aluminosilicate glass core, consistent with calculations and measurements found in . A curve is fit to the rising-edge cladding and falling-edge core spectral components using Lorentzian curves (orange) to estimate the widths of these interactions (Δν). The region between the low-and-high frequency peaks has previously been identified as acousto-optic interactions with higher-order cladding acoustic modes , which is particularly pronounced for Fiber B, and far less significant for Fiber A. As a result of these contributions significantly obscuring the cladding interaction, the low-frequency fit was not included for Fiber B. Additionally, due to the weakness of the Fiber A signal relative to the contribution by the apparatus, the apparatus spectrum was measured and subtracted from the Fiber A spectrum. This was not necessary for the Fiber B spectrum and therefore was retained as a reference.
By way of comparison, Fig. 4 provides a plot of the normalized Brillouin spectrum for the SMF described in Section 2. In order to compare the relative Brillouin gain of the SMF with the fibers of this study, the assumption is made that the integral of the gain spectrum is conserved from fiber-to-fiber , and that a broadening of the spectrum results in a commensurate reduction in the Brillouin gain. Thus, Fiber A has a peak Brillouin gain about 4.9 dB lower than the SMF. Since Fiber B does not possess the large narrow peak seen in Fiber A, with its gain spectrum being more uniformly distributed, its Brillouin gain is similarly estimated to be approximately 9.3 dB lower than the SMF.
The Brillouin spectra from both fibers are considerably broadened due to the acoustically anti-guiding nature of the fiber (more on this below). In fact, the spectral width of the core interaction has broadened to well over 200 MHz, diminishing its contribution relative to the cladding for Fiber A, and hence enabling a largely un-obscured measurement of the spectral width of the pure silica cladding; i.e., the intrinsic Brillouin line-width for silica. A value of 23 MHz is obtained; strikingly similar to that of bulk silica . The remaining measured values are summarized in Table 1. The core diameter, 2a, is defined to be the full-width at the half maximum of the RIP (Δn). The acoustic velocities (V) were determined from the measured Brillouin frequencies (ν), modal indices (nmode), and optical wavelength (λ) using V = λν/2nmode. The acoustic velocity in the cladding is taken to be that of silica in the present glass system. The modal indices and cutoff wavelengths for single mode operation for the fibers were calculated from the RIPs and are also provided in Table 1. The cutoff wavelengths are consistent with purely single mode fibers at the Brillouin test wavelength of 1534 nm. The attenuation coefficient (also in Table 1) at this wavelength was measured utilizing a cutback method, and, while reasonable for the present analysis, its value is likely dominated by impurities in the commercial precursor crystal, as no effort was undertaken to optimize its purity. Future efforts to further purify the precursors prior to use could be employed to reduce losses in the fibers.
In order to characterize the bulk properties of both the magnesia and alumina components of the core after transition to the glassy state, a ternary additive model, described in detail elsewhere  was applied to the data. In short, all constituent materials (including silica) are assumed to obey a law of mixtures. The parameter used to weight the effect of each oxide component is the total volume of glass occupied by that component. This is calculated using molar percentage of the constituents present in the glass, measured by EDX, and the molar volume. Utilizing this model, the bulk values of the acoustic velocity, mass density, refractive index, etc. can be determined as fit parameters for each individual oxide species. The assumption is that these respective species can be treated as ‘separable’ and do not experience effects such as with, for example, the Al-P join in AlPO4, wherein the net refractive index is actually lower than either the Al2O3 or P2O5 components taken individually .
This additive model described was used in conjunction with another boundary-value model  in order to implement the fit-to-data. More specifically, since the compositional profile is not uniform, but rather graded, it is approximated by an eight-layer step-like one. Since each layer possesses a unique composition, each layer also has a unique set of material values, i.e. acoustic velocity, refractive index, etc. The values for the bulk components (magnesia and alumina) are iterated until the calculated values for the optical and acoustic modes best fit those measured for each fiber of this study, and then the results for the two fibers are averaged. The final results are summarized in Table 2 and the uncertainty is half the difference between the best-fit parameters for Fibers A and B.
The spectra measured for the two SpDF fibers are highly acoustically-antiguiding, resulting from the fact that both magnesia and alumina increase the acoustic velocity relative to silica, resulting in very broad Brillouin spectra . Therefore it can be concluded here that in addition to yttria  and alumina , magnesia also increases the acoustic velocity when added to silica. To the best of our knowledge, this is only the third material having been proven to exhibit this behavior when utilized in a silica-based fiber. Additionally, while the model accommodates acoustically anti-guiding fibers by determining a complex propagation constant (which includes the waveguide attenuation term), the waveguide loss term greatly dominates the material damping associated with the addition of alumina and magnesia. Therefore due to uncertainties in the fitting, reliable measurements of the visco-elastic damping loss for magnesia could not be made. In support of this, it is pointed-out that the spectral width of the Fiber A core interaction is likely the broadest-ever reported from a silica-based fiber measured at 1,534 nm, thus having enabled the unobstructed measurement of the 23 MHz intrinsic spectral line-width of pure silica as described above.
Second, and as expected, the deduced densities of the alumina and magnesia are lower than their crystalline counterparts , as are the acoustic velocities . However, the alumina density value is somewhat larger relative to previous observations . One possible explanation is the presence of magnesia and its effect on the glass network. It was shown previously that the mass density of alumina can vary significantly with molecular coordination , with higher coordination numbers favoring higher densities. In the case of normal (as opposed to inverse) spinels, such as MgAl2O4, the divalent cation (Mg) partitions to the tetrahedral site, of which there are 8 cations per unit cell, and the tetravalent cation (Al) partitions to the octahedral site, of which there are 16 cations per unit cell [18,19]. These octahedral sites, on which the Al ion resides, are coordinated by 6 oxygen atoms, which qualitatively support the suggestion above regarding the higher coordination and density.
While there may be some effects brought about by the presence of magnesia, the density and acoustic velocity modeling parameters deduced above suggest that alumina and magnesia can be treated as independent species in the drawn fiber. Alternatively, the coordination may have imprinted from the precursor crystal and rapid processing and quenching of the resultant glass simply made the final glassy phase more closely resemble the precursor crystalline phase. Such possible “structural inheritance” is known in silica where cristobalite, even though it is not the lowest free-energy phase, is the first crystal to form upon devitrification due to structural similarities between the glass and the (non-equilibrium) crystal phase . The ability to mathematically treat the MgO and Al2O3 components as independent in the determination of physical and optical properties may result from the natural partitioning of the respective cations into such different types of crystallographic sites coupled with some degree of structural inheritance based on the rapid cooling from the melt.
Also interesting in the present case is the fact that the deduced refractive index contributions from both the alumina and magnesia are higher than their respective crystalline counterparts [1,8]. This may be due to an effect similar to that of the Al-P join described above although with the effect to increase the refractive index. Such seemingly countervailing properties speak to the complexity of this material system despite what, at first glance, appears to be relatively common species: MgO, Al2O3, and SiO2; arguably 3 of the 5 most abundant compounds in the earth’s surface. In fact, though, the combination of very high melting point and complex crystal structure of spinel, have made accurate determination of its crystal chemistry and phase diagrams a topic of discussion for nearly 100 years . As discussed in , there is marked liquid-liquid immiscibility between Al2O3 and SiO2. There is a similar immiscibility in the MgO – SiO2 system [21–23] with the solubility of MgO in SiO2 being only about 1.5 mole percent at a temperature of 1694°C . Consolute temperatures where phase separation between the immiscible (MgO + SiO2) liquids occurs has been reported as low as 1990°C and as high as 2200°C . Regardless of which consolute temperature is correct, the temperature at which conventional preforms are fabricated and fibers drawn are commensurate with these conditions for phase separation. This fact suggests that the molten core method may be the only practical way to fabricate optical fibers of these compositions and both intriguing and useful properties.
In conclusion, the development of the first known spinel derived single mode fiber has lead to the validation of its highly acoustically-anti-guiding waveguide properties. Measurements of the Brillouin spectra at 1,534 nm have resulted in the observation of the largest-recorded core Brillouin spectral broadening to around 230 MHz. This results in a considerable reduction in the Brillouin gain in the core, however rendering the fiber to be dominated by Brillouin scattering in the cladding. Consequently, due to the large waveguide-loss-induced spectral width of the SpDF, the measurement of the Brillouin spectral width (23 MHz) of pure silica was made possible. In addition, through the use of the additive and the boundary value modeling, the core and cladding acoustic velocities were calculated from the results of the Brillouin measurements, from which some basic properties of magnesia were deduced. This has led to the understanding that magnesia doped into silica will have higher acoustic velocities than pure silica. Magnesia has, therefore, been added as an additional material to the list of those exhibiting this behavior.
The authors wish to acknowledge S. Morris (Clemson University) for the compositional analyses and Lawrence Shaffer (Armorline Corporation) and Larry Fehrenbacher (Technology Assessment & Transfer, Inc.) for the spinel rods. Author J. Guerrier was supported as a Charles Townes Fellow through the joint Clemson University / Furman University Charles Townes Optical Science and Engineering program. This work was supported by the Joint Technology Office through contract W911NF-12-1-0602. The splice machine utilized in this work was originally funded by DURIP award W911NF-07-1-0325.
References and links
1. P. Dragic, T. Hawkins, P. Foy, S. Morris, and J. Ballato, “Sapphire-derived all-glass optical fibres,” Nat. Photonics 6(9), 627–633 (2012). [CrossRef]
4. C. Jen, C. Neron, A. Shang, K. Abe, L. Bonnell, and J. Kushibiki, “Acoustic characterization of silica glasses,” J. Am. Ceram. Soc. 76(3), 712–716 (1993). [CrossRef]
5. P. Dragic, “Estimating the effect of Ge doping on the acoustic damping coefficient via a highly Ge-doped MCVD silica fiber,” J. Opt. Soc. Am. B 26(8), 1614–1620 (2009). [CrossRef]
6. P. D. Dragic, “Brillouin spectroscopy of Nd-Ge co-doped silica fibers,” J. Non-Cryst. Solids 355(7), 403–413 (2009). [CrossRef]
7. V. Lou, T. Mitchell, and A. Heuer, “Graphical displays of the thermodynamics of high-temperature gas-solid reactions and their application to oxidation of metals and evaporation of oxides,” J. Am. Ceram. Soc. 68(2), 49–58 (1985). [CrossRef]
9. Y. Koyamada, S. Sato, S. Nakamura, H. Sotobayashi, and W. Chujo, “Simulating and designing Brillouin gain spectrum in single-mode fibers,” J. Lightwave Technol. 22(2), 631–639 (2004). [CrossRef]
10. B. Ward and J. Spring, “Finite element analysis of Brillouin gain in SBS-suppressing optical fibers with non-uniform acoustic velocity profiles,” Opt. Express 17(18), 15685–15699 (2009). [CrossRef] [PubMed]
11. D. Heiman, D. S. Hamilton, and R. W. Hellwarth, “Brillouin scattering measurements on optical glasses,” Phys. Rev. B 19(12), 6583–6592 (1979). [CrossRef]
12. P. Dragic, “Brillouin gain reduction via B2O3 doping,” J. Lightwave Technol. 29(7), 967–973 (2011). [CrossRef]
13. D. J. DiGiovanni, J. B. MacChesney, and T. Y. Kometani, “Structure and properties of silica containing aluminum and phosphorus near the AlPO4 join,” J. Non-Cryst. Solids 113(1), 58–64 (1989). [CrossRef]
14. P. D. Dragic, “Novel dual-Brillouin-frequency optical fiber for distributed temperature sensing,” Proc. SPIE 7197, 719710, 719710-10 (2009). [CrossRef]
15. S. V. Sinogeikin and J. D. Bass, “Single-crystal elasticity of MgO at high pressure,” Phys. Rev. B 59(22), R14141–R14144 (1999). [CrossRef]
16. P. Dragic, J. Ballato, A. Ballato, S. Morris, T. Hawkins, P.-C. Law, S. Ghosh, and M. C. Paul, “Mass density and the Brillouin spectroscopy of aluminosilicate optical fibers,” Opt. Mater. Express 2(11), 1641–1654 (2012). [CrossRef]
17. G. Gutiérrez and B. Johansson, “Molecular dynamics study of structural properties of amorphous Al2O3,” Phys. Rev. B 65(10), 104202 (2002). [CrossRef]
18. W. Kingery, H. Bowen, and D. Uhlmann, Introduction to Ceramics, 2nd ed. (John Wiley & Sons, 1976).
19. A. von Hippel, Dielectrics and Waves (John Wiley & Sons, 1956).
20. G. Rankin and H. Merwin, “The ternary system MgO – Al2O3 – SiO2,” Am. J. Sci. s4-45(268), 301–325 (1918). [CrossRef]
21. N. Bowen and O. Andersen, “The binary system MgO – SiO2,” Am. J. Sci. s4-37(222), 487–500 (1914). [CrossRef]
22. J. Greig, “Immiscibility in silicate melts: Part I,” Am. J. Sci. s5-13(73), 1–44 (1927). [CrossRef]
23. P. Wu, G. Eriksson, A. Pelton, and M. Blander, “Prediction of the thermodynamics properties and phase diagrams of silicate systems – evaluation of the FeO – MgO – SiO2 system,” ISIJ Int. 33(1), 26–35 (1993). [CrossRef]