Chapter 1. Introduction 1.1 Brief historical review of silica optical fiber The invention of the low-loss silica glass optical fiber using chemical vapor deposition in 1970 [1] has opened new horizon in communication technology, and led to the possibility of ultra-high capacity information flow. Worldwide competitive research and development efforts followed that invention and brought about astounding progress in photonics technology. It must be noted that two inventions made almost simultaneously in 1970, silica-based glass fiber [1] and continuous wave semiconductor laser operating at room temperature[2], are now clearly seen as the seminal events that marked the inception of the photonics era. Owing to excellent transmission properties of low-loss and high bandwidth as predicted by Kao et al.[3], silica optical fiber is the status quo broadband telecommunications media such as transoceanic cables, terrestrial cables, metro-loops, and customer premises loops. Real time video and audio applications for advanced broadband systems prompted optical manufacturers to develop high volume mass production facilities of fibers such as large perform fabrication[4] and high speed drawing [5] and consequently lower the cost of the fiber. As a result of these efforts, fiber fabrication technology has produced various types of fibers and fiber devices that prompted all-fiber platform in telecommunications, sensing, and metrology. All of these fiber optic innovations and contributions to todays information technologies have led to recent recognition of optical fiber such that Dr. Charles Kao have won the Nobel prize in physics in 2009.[6]

Major research efforts began with achieving low-loss fibers to extend the optical link distance. After the demonstration of fiber with loss of 20 dB/km in 1970, the fiber

process advanced very quickly. By 1973, the loss of less than 5 dB/km was reported at 850 nm [7], approaching the intrinsic scattering limit at that wavelength. Researchers started to explore longer wavelengths where the intrinsic scattering loss is lower. In 1976, the first fiber with loss of 0.47 dB/km at 1200 nm was reported [8].Within three years, the fiber attenuation reached 0.2 dB/km at 1550 nm [9,10], close to the theoretical limit of 0.15 dB/km. The evolution of optical loss reduction efforts is summarized in Fig.1.1.1

Fig.1.1.1 Optical loss spectrum of silica glass single-mode fiber [10] In 1970s, the available light sources near ~850nm, has named the corresponding spectral range in the optical fiber as the First Window. With the advancement of both compound semiconductor technology and fiber fabrication process, the second window

near 1300 nm was flourished in 1980s. As the ultimate low loss was achieved near 1550nm region, the communication widow shifted to this third window in order to take fullest advantage of longer link distances. The third window is also called as C-band, conventional band. With further development of bending loss reduction technology the third window expanded into longer wavelength near 1620nm, which is called as L-band, long band. In recent international standards the optical communication bands have been further classified as O, E, S, C, L, and U bands, which will be discussed in more detail in the following section.

Along with the efforts to reduce the optical loss, there have been intensive researches to control the waveguide properties of optical fibers in order to control the modal dispersion and the chromatic dispersion. As a first step, two types of fibers have been developed in parallel; multimode fiber (MMF) and single mode fiber (SMF). Li and Nolan [12] have made a detailed review on optical fiber development history and schematic routes of development for multimode and single mode optical fibers are summarized in Fig.1.1.2, and Fig.1.1.3, respectively. Starting from a step index fiber, multimode fibers (MMFs) quickly evolved to graded index profile MMF to increase the bandwidth for 850nm light emitting diode (LED) light sources. Graded index MMF diameter started from 50m, and then increased to 62.5m to accommodate 1300 nm LED. As high speed narrow linewidth light sources such as vertical cavity surface emitting lasers (VCSELs) and laser diodes (LDs) replaced LEDs, graded index MMF diameter once again reduced to 50m with further more tailored graded refractive index profile near the center to secure high bandwidths. Presently 10Gbit/sec signals can be transmitted over a several hundred

meters long, using graded index 50m core MMFs and further increase in both data rate and link distance is being explored.

Figure 1.1.2 Multimode fiber technology evolution.[12] Here FDDI is fiber distributed data interface, ATM is asynchronous transfer mode, GbE is giga bit ethernet, FC is fiber channel, and SDH is synchronous digital hierarchy.

In the case of single mode optical fibers the evolution routes are more complicated due to introduction of wavelength division multiplexing (WDM) based on optical amplifiers [13] such as erbium doped fiber amplifier (EDFA) [14] and Raman fiber amplifier (RFA)[15] and their combinations. In WDM-SMF links multitudes of channels are carried within a certain wavelength range where the channels are amplified to restore the attenuated intensity. However chromatic dispersion linearly accumulates and the signal pulses spread out in the time domain to result in inter-symbol-interferences and subsequent bit errors in the data streams as shown in Fig. Therefore it is imperative that chromatic dispersion be properly managed or compensated in long haul WDM systems, which opened various avenues of dispersion controlled fibers such as dispersion shifted fiber (DSF), dispersion flattened fiber (DFF), non-zero dispersion shifted fiber (NZDSF), and dispersion compensating fiber (DCF). Dense WDM systems have also introduced new concepts in fiber design to manage

the high optical power and consequent optical nonlinearity, which brought standardization of effective mode areas, and nonlinear coefficients. Nonlinear effects in optical fibers can be significantly suppressed if the modal area is large enough, and large mode area (LMA)-SMFs have been developed with novel refractive index profiles. [16]

Figure 1.1.3 Single mode fiber technology evolution.[12] Here DWDM, and CDWDM are dense and coarse wavelength division multiplexing, respectively. ULH is ultra long haul, OADM is optical add drop multiplexer, FTTH is fiber to the home, SBS is stimulated Brillouin scattering, EDC is electric dispersion compensation, WP is water-peak.

Massive deployment of optical fiber networks into the customer premises such as in fiber to the home (FTTH) networks has raised more stringent optical nonlinearity issue especially for the case of passive optical network, where a high power signals are distributed to several customer premise using optical power splitters. Especially stimulated Brillouin scattering (SBS) has become a critical issue because the incident optical power is transferred to backward propagating Stokes shifts.[17] Spatial distribution of compensating dopants such as Al, and Ge across the core has been proposed to suppress SBS.[18] FTTH applications have also raised another fundamental

issue in optical fiber desing-macro bending loss, an essential factor in fiber installation in FTTH environments where tight bending of optical fiber is unavoidable. Bend insensitive fibers have been recently developed to overcome the macro bending loss.[19] The optical loss due to the overtone of hydroxyl (OH) bonds near 1380nm has been recently removed by developing gas-phase dehydration process [20] and expanded the optical communication bands into S and E bands. These suppressed water peak SMFs are being widely deployed in optical networks that look forward expansion from C, L bands to S bands. The communication bandwidth and channel speed have been continuously increasing with new developments in optical device and system technologies, which subsequently demands modification of SMFs structures for optimal dispersion management within the new available bands. Especially the optical wave guide design and optical amplifier bandwidth are strongly correlated and they are continuously evolving to further expand the optical communication bandwidth.

In this book we will focus on dispersion, birefringence and nonlinear properties in optical fibers and describe various types of optical fiber design, fabrication and related international standards.

Dispersion in optical fibers can be categorized into three different types depending on their origins:1) modal dispersion in MMFs, 2) chromatic dispersion in SMFs, and 3) polarization mode dispersion in SMFs.

Modal dispersion in MMF is schematically illustrated in Fig. 1.1.4. The number of guided modes in MMF is in the order of 103 and each mode has different group velocity. As a result of the distribution in group velocities in step index MMF, the effective modal delay is of a few nano-second, 10-9 sec.[21] The effect of modal dispersion in MMFs are schematically illustrated in Fig. 1.1.4. The magnitude of the modal delay in step index MMFs is comparable to the bit period of a few 100 Mbit/sec, ~a few 10-9 sec, which can result in severe overlap and spread of input pulses. Therefore the maximum data rate along a step index MMF cannot exceed 100Mbit/sec for 1km length.

Fig.1.1.4 Impacts of modal delay in multimode fibers (MMF) over digital pulse spreads in comparison with single mode fiber (SMF). The pulse spread due to modal dispersion in MMFs can be significantly reduced by

changing the refractive index distribution in the core from the step profile to a graded one, which allocates a higher refractive index to lower order modes propagating in the shorter optical paths and a lower refractive index to higher order modes propagating in the longer optical paths to effectively synchronize them in the output. This concept is schematically illustrated in Fig. 1.1.4. The issues in MMF designs are discussed in Chapter 5. Despite the elaborated graded index core design in MMFs, there still exists a finite modal delay and reducing the number of modes drastically was sought after as a fundamental solution to overcome the modal dispersion. Single mode fiber (SMF) has achieved an efficient guidance of only the fundamental mode in the second and third window of communications, 1300~1600nm, and the higher order modes were cut-off in the spectral range over ~1200nm. In the fundamental LP01 mode in SMF, there exits two degenerate polarization modes with very similar effective index and they behaves as a single mode. Even though the modal delay in MMF disappears in SMFs as in Fig.1.1.4, there exists chromatic dispersion that linearly accumulates over the propagation length. The effect of chromatic dispersion in SMF is schematically illustrated in Fig. 1.1.5.

Figure 1.1.5 Impacts of chromatic dispersion (D) in single mode fiber (SMF) over the optical signals at =1550nm. Here Gbps is Giga bit per second.

Optical pulse trains at =1550nm experiences the chromatic dispersion of ~17psec/nm km in typical SMFs to result in pulse broadening in the time domain. The chromatic dispersion linearly accumulates along the fiber length and further broadening of pulses will result in distortion of signal and bit errors, as shown in Fig. 1.1.5. The chromatic dispersions penalties in SMFs have been overcome by the waveguide design with novel refractive index profiles, which significantly modify the waveguide dispersion to result in appropriate dispersion value and slope within the optical bands of interests. We will discuss these waveguide design issues for dispersion managements in Chapter 4, and Chapter 9. Chromatic dispersion spectra of the dispersion controlled fibers are shown in Fig. 1.1.6. Conventional SMFs had the zero dispersion near 1310nm and its value monotonically increases to ~17psec/nmkm at 1550nm with the dispersion slope of 0.58psec psec/nm2km. In order to shift the zero dispersion to the lowest loss spectral position ~1550nm, dispersion shifted fibers (DSF) have been developed for single wavelength signal transmission.[23] In order to utilize wider optical communication window, dispersion flattened fibers (DFFs) developments were followed, where zero dispersion occurred near both 1310nm and 1550nm.[24] In WDM systems, however, the having the zero dispersion wavelengths within the WDM channels was found to be critically detrimental by four wave mixing (FWM) penalties.[25] A small but finite chromatic dispersion was found to be optimal and optical fibers with non-zero chromatic dispersion within the band of interests and so called non-zero dispersion shifted fibers (NZDSFs) with optimal dispersion slopes have been playing major roles in long haul applications.[25] As an alternative way to control the chromatics dispersion of SMF, fibers with a negative dispersion and a negative slope near 1550nm were

developed as dispersion compensating fibers (DCFs).[26] Detailed discussions of these dispersion controlled fibers will be given in Chapter 4.

Fig. 1.1.6. Chromatic dispersion spectra of SMFs with various dispersion optimizations. Here NZ-DSF is non zero dispersion shifted fiber. S, C, and L-bands are short wavelength, conventional, and long wavelength band, respectively and their definitions are summarized in Table 1.2.2.

Nominal chromatic dispersion value of DSFs is a few psec/nmkm in the wavelength region of interests, which corresponds to an order of pico-second (~10-12 sec) group delay for 10km fiber with a 0.1nm linewidth () light source. This delay is negligible in comparison to bit period of 10 Gbit/sec digital signals, which is an order of 10-10 sec, and the chromatic dispersion can be overcome.

However, as the data rate further increases to 40 Gbps (Giga bit per second) and higher

whose bit period is the order of 10-11 sec (or a few tens of pico-second), another type of dispersion in optical fiber becomes the major limitation in optical transmission. By slight random deviation of the optical fiber from the perfect circular symmetry, the fundamental LP01 mode is no longer degenerate and two modes in orthogonal polarizations will have different effective indices to result in a group delay as schematically shown in Fig.1.1.7. This dispersion is called as polarization mode dispersion (PMD).

Figure 1.1.7 Impact of polarization mode dispersion in SMF over the time delay between two orthogonal polarization modes in an elliptic core optical fiber. PSP, PMD, and DGD are acronyms for principal state of polarization, polarization mode dispersion, and differential group delay, respectively.

The group delay between two orthogonal polarization modes could exceed a few picosecond for SMF link of a few tens of km, which is comparable to bit period of high data rate digital signals. In order to reduce PMD various fabrication processes have been developed in both perform and optical fiber drawing. Presently the PMD nominal values are less than 0.2 psec/ km .

The future trends in optical communications are summarized in Fig.1.1.8, where the

horizontal axis is the number of WDM channels per optical fiber and the vertical axis is the data rate for each channel. It is noted that the system capacity upgrade direction should face different challenges in dispersion managements. For example if the number of channels are increased for a fixed relatively low data rate, say OC-48 (2.5Gbit/sec), then transmission fibers should resolve the issues in chromatic dispersion and its slope mismatches within the communication band of interests. If the number of channels are fixed to a relatively small number (