Glass: The Carrier of Light - A Brief History of Optical Fiber

Abstract

All voice and data communications employ silica glass optical fiber at some point in their nearly instantaneous transmission. This is enabled globally by the annual production of over 180 million kilometers of optical fiber. Since the first low-loss fiber installations, nearly 2 billion kilometers have been manufactured, which is enough to connect the Earth with Jupiter.1 Given such a rare combination of ubiquity and utility, this article reviews the history of glass optical fiber and provides commentary on recent developments, and musings on their future.

History

Light has been illuminating the Universe since its absolute beginnings. The first documented efforts at describing light can be attributed to Euclid (circa 300 BC),2 followed by Abū ‘Alī al-Ḥasan ibn al-Haytham (circa 10151),3 Robert Hooke (1665),4 Isaac Newton (1721),5 and James Clerk Maxwell (1864).6 Indeed, it was Maxwell who is credited with unifying the fields of electricity and magnetism into electromagnetism and providing the mathematical foundation for light as it is described today.

As specifically relates to optical fiber, the guiding principle for light confinement and propagation is total internal reflection. Total internal reflection was used to explain the appearance of rainbows as early as 1300 AD, independently by Al-Farisi7 and Theodoric of Freiberg,8 both of whom based their experiments on the work by Ptolemy.9 However, it was not until the 1600s when total internal reflection was first systematically studied by Johannes Kepler in 1611,10 mathematically defined by Snellius (from whom the “Snell” of Snell's Law is derived) in 1621, but unpublished until mentioned by Huygens in his Treatise on Light in 1690.11 Colladon and Babinet demonstrated a “lighted laminar water fountain” in 1841 and published the results in 1842.12, 13 However, it was John Tyndall who popularized the demonstration in a series of public lectures in the 1850s and history has largely associated his name with the discovery.

Although the physics of total internal reflection was well understood, the potential application of optical fiber beyond illumination14 and imaging15 would not be more fully appreciated for nearly another century; not until, that is, the advent of the laser as a coherent and collimated light source.16 Shortly after the first crystalline17 and gas lasers18 were realized, Snitzer2 developed the first glass laser19 and, subsequently, the first optical fiber amplifier.20 This latter advancement was especially critical to the future development of optical fiber communication and laser systems particularly in light of the invention a few years earlier of the semiconductor laser.21 By 1964, all of the necessary building blocks had been realized, all-be-them independently.

In 1966, Kao2 and Hockham published their now famous work, which evaluated material and loss mechanisms in waveguides and determined that a dielectric fiber “represents a possible medium for the guided transmission of energy at optical frequencies.22” Most memorably, they conclude that “Certainly, the required loss figure of around 20 dB/km is much higher than the lower limit of loss figure imposed by fundamental mechanisms.” This work earned Kao the Nobel Prize in Physics in 2009 for “groundbreaking achievements concerning the transmission of light in fibers for optical communication.23” It is worth noting that this year (2016) represents the 50^th anniversary of this publication.22

The global challenge to make such low-loss optical fibers was accepted by many, though the most successful were those by Corning, Corning, NY (Keck, Maurer,2 and Schultz), Bell Labs, Murray Hill, NJ (MacChesney2), and ITT, Tokyo, Japan (Izawa).24-29 These teams developed the fabrication processes employed to make the bulk preforms from which optical fibers subsequently are drawn. Each of the main processes can be categorized as being chemical vapor deposition (CVD) methods with Corning developing the outside vapor deposition, OVD, method, Bell Labs developing the modified chemical vapor deposition, MCVD, method, and ITT developing the vapor axial deposition, VAD, method. The same basic chemistry is used in each case, which involves the thermochemical oxidation of a volatile halide: SiCl₄ + O₂ → SiO₂ + 2Cl₂. Because of the remarkably large vapor pressure differences (nearly 12 orders of magnitude) between the SiCl₄ precursor and impurities, such as Fe₂Cl₆, the resultant glass is exceedingly pure and this purity enables the attainment of intrinsic optical attenuation from the fibers. Other precursors, also with reasonably large vapor pressure differences, such as GeCl₄ and POCl₃ (precursors to GeO₂ and P₂O₅, respectively), are added into the SiCl₄ + O₂ vapor stream to yield compositional modification to the SiO₂ necessary to create the core/clad waveguiding structure of the optical fiber.

Given the significant potential for communication systems employing low-loss optical fibers, work began almost immediately on fabrication methods following the Kao and Hockham paper. Only a few years later, the Corning team broke the 20 dB/km barrier.30 Within several years, other approaches ensued, each with their own advantages and disadvantages. Within a few additional years, the first nonexperimental optical fiber communication links were being used in the United Kingdom and the United States.

The combination of low loss over long distances with small core sizes made optical fiber a new practical tool for the study of otherwise very weak nonlinear optical effects such as those involving frequency mixing processes (e.g., sum frequency generation [SFG], difference frequency generation [DFG], optical parametric amplification [OPA], and third harmonic generation [THG]) or other nonlinear processes (e.g., optical Kerr effect, self-focusing, self-phase modulation [SPM], cross-phase modulation [XPM], four-wave mixing [FWM], multiphoton absorption, optical solitons, stimulated Brillouin scattering [SBS], and stimulated Raman scattering [SRS]). As Stolen notes in his review “Fiber nonlinear optics has grown from a novel medium for the study of nonlinear optical effects, through a period where these effects appeared as system impairments, to the present day where optical nonlinearities are an integral part of high capacity optical systems.31”

Today, optical fibers specifically, and lasers more generally, are used in countless applications including communications, medicine, energy, manufacturing, sensing, transportation, entertainment, and as tools for scientific inquiry.32 The history provided above was intended only as a brief summary. An exhaustive history of light, light guidance, and the development and application of optical fiber has been the source of many fine articles and books.33, 34 As the purpose of this Review is to provide highlights from the past while provoking the future, Table 1 provides a “modern” history of optical fiber and related discoveries and inventions, with references for further study as the Reader wishes.

Table 1. A Brief History of Optical Fiber

Year	Event
1842	Colladon and Babinet invent the “lighted laminar water fountain” based on total internal reflection12, 13
1854	John Tyndall formulates a ray-tracing approach to optical wave guiding67 James Clerk Maxwell unifies electricity and magnetism into electromagnetism, thereby defining light6
1880	William Wheeling patents “light piping” for illumination14
1888	Austrian medical doctors, Roth and Reuss, employ bent glass rods to illuminate body cavities
1930	Heinrich Lamm first demonstrates imaging using a glass rod bundle15
1933	First television signals optically transmitted (very poorly) over glass rods
1953	First maser was built by Charles Townes, James Gordon, and Herbert Zeiger at Columbia University (United States). Work is published the following year16
1954	Fiber cladding invented by Abraham Van Heel68, 69
1960	Demonstration of the first laser (optical frequencies)17
1961	Elias Snitzer2 reports on the waveguide modes of cylindrical glass waveguides70
	Snitzer2 fabricates a single mode fiber and demonstrates the first bulk glass laser19
1962	Robert Hall demonstrates the semiconductor laser21
1964	First demonstration of fiber-based laser/amplifier20
1966	Kao2 and Hockham identify glass impurities as the leading cause of optical loss and theorize glass optical fibers capable of transmission losses below 20 dB/km22
1970	Corning2 breaks the “20 dB/km” loss mark30, 71
1972	Development of the Raman fiber amplifier72
1973	First diode end-pumped fiber laser73
1974	Wave mixing in an optical fiber74, 75
1975	Discovery of heavy metal fluoride glasses2, 76
	First commercial fiber-optic link installed by the Dorset (U.K.) police
1977	First telephone signals using optical fiber occurs in Long Beach, CA (United States)
1978	Photosensitivity and nonlinear effects in optical fiber77, 78
	Single polarization optical fibers79
	Fiber Bragg gratings77, 80
1979	Dispersion shifted fibers81
	Dispersion flattened fibers82, 83
1986	The erbium-doped fiber amplifier (EDFA) is pioneered by David Payne (Southampton, United Kingdom)84 and Emmanuel Desurvire (Bell Labs)
1988	First transatlantic telephone cable goes into operation
	Experimental realization of a double clad fiber laser85; though suggestions of optimized fiber designs for higher power operation existed as far back as 196186
1995	First photonic bandgap optical fiber87, 88
	Output power from optical fiber lasers hits 10 W89
	Beginning of the “dot-com” boom
1996	First photonic crystal fiber90
1999	Output power from an optical fiber laser exceeds 100 W91
2000	First photonic crystal optical fiber-based supercontinuum generation92
	Beginning of the “telecom bust”
2006	Development of semiconductor core optical fiber that marry silicon photonics with optical fiber93-95
2009	Kao2 awarded Nobel Prize in physics23
	Output power from an optical fiber laser exceeds 10 KW96
2014	Output power from an optical fiber laser exceeds 30 KW97

A Bright Future

Just as Kao and Hockham could never have envisioned what their efforts would have enabled over the intervening 50 years, it is nearly impossible to predict the future opportunities in store for optical fiber. However, there are several trends that have emerged and the section that follows discusses these in general and proposes an unorthodox approach to their handling.

Trends in the Modern Era and State of the Art

Be it more Internet traffic or requisite power levels needed for manufacturing or directed energy applications, the optical power propagating through optical fibers today has grown markedly over the past few decades. As noted in Table 1, it took only 20 years for the output power from a fiber laser to reach 30 kW from 10 W.35, 36 The small mode diameter of conventional fibers (providing a potential for high intensity light confinement) and long propagation lengths lead to the excitation of a bouquet of parasitic nonlinear phenomena. These nonlinearities have caused the growth over time in output powers to plateau and, today, represent the main limitation in fiber system performance. Figure 1 provides a list of the main optical nonlinearities that presently limit the continued scaling to higher powers in fiber-optic systems.

Figure 1

Optical nonlinearities that plague scaling to higher output powers in modern optical fiber-based lasers: SBS is stimulated Brillouin scattering, SRS is stimulated Raman scattering, HOMI (TMI) is higher order mode instability (transverse mode instability), n₂ is the nonlinear refractive index, SPM is self-phase modulation, FWM is four-wave mixing, and MoR is modulus of rupture. Provided is a description of each effect, why it is problematic, and the author's proposed solution as opposed to conventional methods of microstructural engineering of the fiber.

In response to these parasitic effects, the optical fiber community began to develop new optical fibers whose cross-sections possess periodic regions of high and low index glasses (or, possibly air/voids). The periodic structure leads to the formation of optical bandgaps that permit or restrict certain energies or polarizations of light to propagate, thus providing a means of control.37-39 These microstructured or photonic crystal optical fibers can be very complex in structure and have promoted incremental improvements in reducing the aforementioned parasitic nonlinearities via slightly enhancing the mode size. These structured fibers have significant barriers to commercial entry due to their complexity and, therefore, high cost and low yield. Figure 2 provides representative images of selected photonic crystal fibers.

Figure 2

Representative examples of photonic crystal fibers.38

The development of microstructured and photonic crystal optical fibers has led to the realization of a great many new types of fibers from a wealth of optical materials including novel “multimaterial” fibers.40, 41 However, it is the author's supposition that developing ever-more complicated optical fibers as a means of reducing the detrimental effects of optical nonlinearities is equivalent to a medical doctor treating the symptoms of an illness through medication and not attacking the disease itself. At the most fundamental level, it is the interaction of the light with the material from which the fiber is made that is the origin of these parasitic nonlinearities. In order to truly mitigate these effects, one must start with the material.42, 43 Each phenomenon has a materials origin and it is through the appropriate materials coefficients that reductions can be gained, if not fully negated. As previously noted, Fig. 1 provides a compilation of the five main phenomena that presently limit the scaling of fiber lasers to higher output powers.35, 36, 44, 45 Included is a brief description of each phenomenon, the issue it presents, and the material properties that can influence its magnitude.

A canonical case-in-point is SBS.46 SBS is an interaction between hypersonic (thermally excited) acoustic waves and the optical signal. The acoustic wave produces a periodic longitudinal pressure, hence density variation. The spatially modulated density corresponds to a spatially modulated refractive index. Via electrostriction, the interference between the forward-propagating optical signal and back-scattered light (from the original thermally generated periodic refractive index variation) feeds the acoustic wave. This feedback process increases in efficiency with increasing optical power until a threshold is reached where the acoustic wave becomes a highly efficient reflector to the optical signal. SBS generally limits the amount of light per unit bandwidth that can be generated in or transmitted down an optical fiber.47, 48 As such, it typically has the lowest threshold of all the nonlinear processes in narrow line-width systems and is a major limitation in the scaling to higher powers in high energy (fiber) laser systems. Methods presently employed to suppress SBS have focused on fiber engineering and typically involve broadening of the Brillouin gain spectrum by tailoring of the fiber acoustic properties.

However, as has been noted,42 the Brillouin gain coefficient, g_B, at a given wavelength, λ_o, is proportional to the glass' p₁₂ photoelastic coefficient and refractive index, n, and inversely proportional to its density, ρ, acoustic velocity, V_a, and Brillouin linewidth, Δν_B; more specifically, .49 In other words, rather than designing and fabricating a complex, large mode area fiber whose microstructure seeks to control the optical and acoustic properties in order to suppress SBS, one conceivably can make a simple core/clad fiber using a core glass with increased mass density, acoustic velocity, and Brillouin line-width and/or lower photoelastic constant (p₁₂) and refractive index.

When Negative Becomes a Positive

Each parasitic nonlinearity depends on a (differing) set of material properties.42, 43 Modifying these properties, through tailoring of the glass composition to the extent allowable by glass formation and stability, permits reductions to the magnitude of the effects.

However, in certain cases, the compounds comprising the glass can take on property values of opposite sign such that a composition exists (again, assuming glass formation) where the undesired effect is zero.

Here, it is illustrative to return again to the example of SBS, whose gain is proportional to the p₁₂ photoelastic constant. Whereas silica (SiO₂) is known to have a positive p₁₂ photoelastic constant, alumina (Al₂O₃),50-52 baria (BaO),53 strontia (SrO),54 and rare-earth oxides55, 56 have been deduced to have negative p₁₂ values. Accordingly, a binary silicate with any of these compounds has a composition where p₁₂, hence Brillouin gain coefficient, is zero.42 For completeness, it is worth noting that “deduced” means that the p₁₂ values are determined by measurements on the drawn fiber. In comparison to a conventionally prepared and annealed bulk glass, the core glass in the optical fiber might be under additional stress57 and have a different bonding structure due to the rapid fiber quench rates, both of which can influence the resultant p₁₂ value. In practice, one does not need a p₁₂ = 0 glass since reductions in Brillouin gain of even 3 dB (50%) are useful in many applications. Table 2 provides a compilation of performance specifications of interest to optical fiber end-users along with the properties of selected silica-clad, molten core-derived optical fibers.42, 58, 59 Figure 3 shows the deduced Brillouin gain coefficient for the various silica-clad molten core-derived optical fibers referenced in Table 2 in comparison to a conventional telecommunications fiber (left). Figure 3 also provides an image (right) of a strontium aluminosilicate oxyfluoride molten core fiber (Table 2) whose core glass composition is designed to reduce Brillouin and Raman gain as well as transverse mode instability (TMI).

Table 2. Comparison of Selected Properties Critical to Modern Fiber Amplifier and laser Performance and the Values Obtained from the Noted Molten Core–Derived Optical Fibers

Performance specification	Goal	Precursor to molten core–derived optical fiber
		YAG	Al2O3	MgAl2O4	BaO	Li2O-Al2O3	Oxyfluoride
Doping into SiO2	High content	22	54	5.5	18	13	~20
EM modality	Single mode (SM)	Near SM	MM	✓	MM	✓	✓
Brillouin gain	5–10 dB suppression	−10	−19	−9.4	−12	−10	−7
ZeBrA composition	Realistic doping	–	88	84	33
Raman gain	>3 dB suppression	−3	−2.5		−2.0		− 3
Thermo-optic	>3 dB reduction						− 3
Brillouin athermal	Athermal	–	✓	Higher MgAl2O4 levels	Close	✓
Brillouin atensic	Atensic	–	–			Close
Active fiber	Rare-earth doped	✓	✓	✓	✓	Not tried but possible
Attenuation	<100 dB/km	✓	200	200	600	1000	400
Strength	Equivalent to SMF-28^™	✓	Not measured but fine in practice
Fusion splice	Splice to SMF-28^™	✓	Close	✓	✓	✓	✓
Reference		98-100	50-52	101	53	63	–

• ✓ denotes fibers that meet the stated goal.

Figure 3

Left: Comparison of computed and measured Brillouin gain from a series of intrinsically low Brillouin gain optical fibers to a conventional telecommunications optical fiber (Corning's SMF-28^™); after.42 Right: A representative micrograph of a silica-based intrinsically low Brillouin, Raman, and thermo-optic optical fiber for comparison to the fibers shown in Fig. 2.

A second example of this balancing of positive and negative material coefficients relates to TMI, which is also known as higher order mode instability (HOMI). Thermally induced longitudinal refractive index modulations associated with stimulated Rayleigh scattering result in instabilities in the electromagnetic modes (mode coupling) in “effectively single mode” fiber lasers.60 These higher order transverse mode instabilities (TMI) limit power scaling in high-energy laser systems by dynamically randomizing the beam modal distribution.

The causation between stimulated Rayleigh scattering and TMI yields a materials solution. The TMI threshold power is proportional to the mass density and specific heat of the glass and inversely proportional to its thermo-optic coefficient (dn/dT).45, 60, 61 Here, in an analogous manner to SBS, a material with dn/dT = 0 would obviate TMI. Materials such as SiO₂, GeO₂ (dn/dT larger than silica), and Al₂O₃ (dn/dT similar to silica) have positive dn/dT, whereas this value is negative for several materials such as P₂O₅ and B₂O₃, among many others, so that selected binaries can exhibit zero dn/dT values.

As one last example of how this balancing of material properties can be a very powerful approach to either mitigating parasitic optical nonlinearities or creating new opportunities, consider the coupling between applied strain and/or temperature (cause) and acoustic and/or optical properties (effect). The appropriate material parameters in these cases are the strain optic coefficient (SOC = dn/dε), strain acoustic coefficient (SAC = dV_a/dε), thermoptic coefficient (TOC = dn/dT), and thermacoustic coefficient (TAC = dV_a/dT). These properties also can take on positive and negative values, depending on the glass and so can, in some cases, yield a value of zero for the resultant glass. Simple binary glasses that exhibit such atensic (Brillouin frequency is immune to strain53) and athermal (Brillouin frequency is immune to temperature) include aluminosilicates,50 borosilicates,53 strontium aluminosilicates,54 and lutetium aluminosilicates.62 Residual stresses associated with the differential thermal expansion between the fiber's core and clad also can create the conditions for a Brillouin athermal optical fiber.63 Such Brillouin atensic and athermal fibers are of great opportunity to distributed sensor systems whereby the (Brillouin) frequency of the scattered light is shifted by an amount that depends on the optical fiber's temperature and strain environment. However, distinguishing between strain and temperature in a measurement requires the determination of at least two Brillouin frequency components, which usually is achieved through the form of a pair of fibers exhibiting specific property differences. An alternate, and simpler, approach would be to employ an optical fiber, such as these, that is immune to either temperature (athermal) or strain (atensic), therein reducing system complexity.

For completeness, some phenomena, such as SRS, FWM, and SPM, do not appear to have way to be fully eradicated as in the case of p₁₂ = 0 for SBS and dn/dT = 0 for TMI. FWM and SPM originate from the nonlinear refractive index, n₂, which is associated with the instantaneous electronic response of the third-order susceptibility, Re[χ⁽³⁾]. Raman gain, g_R, results from the noninstantaneous (delayed) part of χ⁽³⁾. Thus, from a materials perspective, intrinsically low n₂ glasses would be effective options to lessen the impact of SPM, FWM, and SRS in high-energy laser systems. Relative to conventional germanosilicates and phosphosilicates employed in optical fiber laser applications, one based on fluorosilicates or oxyfluorides should exhibit reduced n₂ values.64, 65 SRS can also be reduced by producing a more highly disordered glass, naturally associated with the molten core method, which broadens the Raman spectrum and reduces the peak value of g_R, replacing high Raman gain materials with low gain ones (e.g., YAG replacing SiO₂), and utilizing materials whose Raman spectral features have minimal overlap.66

Conclusions

Provided herein is a brief history of optical fiber, including major achievements and more recent developments, in commemoration of the 50^th anniversary of the Kao and Hockham publication that originated modern communications.22 Also provided are the authors' musings on present and future trends and steps forward. Of particular importance is the trend of increased optical power propagating through the fibers and the performance limitations created by a host of parasitic nonlinearities that result. Whereas the majority of the optical fiber community has approached the mitigation of such phenomena by designing ever-more-complicated microstructured fibers, the authors propose a route to very simple fibers that attacks the nonlinearities at their fundamental source: the glass from which the optical fiber is made.

Acknowledgments

The authors are especially grateful to Wade Hawkins, Maxime Cavillon, Courtney Kucera, and Dr. Roger Stolen (Clemson University). Thoughtful and insightful communications with Profs. Josef Zwanziger (Dalhousie University) and Anna Peacock (University of Southampton) also are appreciated. The authors also thankfully acknowledge financial support from the U.S. Department of Defense Joint Technology Office through contracts W911NF-05-1-0517, FA9550-07-1-0566, W911NF-12-1-0602, FA9451-15-D-0009/0001, and FA9451-15-D-0009/0002.

Notes

1 The millennial anniversary of Ibn al-Haytham's opus was one of several reasons for 2015 being celebrated globally as UNESCO's International Year of Light (IYL). Other major light-related anniversaries celebrated during 2015 IYL were as follows: the proposition of light as a wave (Fresnel, 1815), the electromagnetic theory of light (Maxwell, 1865), theory of the photoelectric effect (Einstein, 1905), role of light in cosmology through general relativity (Einstein, 1915), discovery of the cosmic microwave background (Penzias and Wilson, 1965), and, as noted herein, Charles Kao's achievements regarding the transmission of light in optical fibers for communication (1965; published in 1966).

2 It is worth noting that the American Ceramic Society, which publishes this Journal, was central to the earliest days of glass and optical fiber development. For example, the George Morey Award, which is the premiere award for the Society's Glass and Optical Materials Division, was awarded to many of the seminal contributors listed in this Review: the glass laser (Snitzer, 1971), low-loss optical fiber (Kao, MacChesney, and Maurer, 1976), and fluoride glasses (Lucas and Poulain, 1989).