Erbium-doped fiber amplifiers

FIBER OPTICS

Abridged

        

Fundamental research in glass science, optics and quantum mechanics has

matured into a technology that is now driving a communications revolution.

                                                                                                                 Alastair M. Glas

 

1 On looking back to the beginnings of Physical Review it is interesting to find that the very first paper¹ in volume 1, number 1, 1893, addressed the "transmission spectra of certain substances in the infrared." The article, authored by Ernest Nichols, included quartz and glass as materials of interest.

2 The study of fiber optics owes much to the vision of Charles Kao (then at Standard Telecommunications Labs in England), who recognized 27 years ago that silica-based waveguides, consisting of a core of high-refractive-index glass surrounded by a cladding with a lower re­fractive index, offered a practical way to transmit light by total internal reflection. Today the field has grown far beyond that vision of a passive communications chan­nel. Research on fibers for active devices such as ampli­fiers and lasers is now leading to a new class of optical devices.  

3 All dielectrics, whether crystalline or amorphous, have an "optic window" of relative transparency to elec­tromagnetic radiation. This window lies between the phonon (and multiphonon) absorption bands at lower energies and the electron (and exciton) absorption char­acteristics at higher energies. In glasses the losses within this optic window are dominated by scattering from static fluctuations of the refractive index and by absorption by impurities and structural defects. In 1970 came the important milestone at the Corning Glass Works⁴ in which researchers fabricated silica fiber with a loss as low as 20 dB/km (that is, 1% transmission through 1 km of fiber). At that time it was impossible to imagine

how rapidly silica fiber technology would evolve. In 1978 AT&T demonstrated the first fiber communications sys­tem, and since that time several million miles of fiber have been installed around the world, both on land and undersea. Low-cost fiber processing techniques have re­duced losses almost to the theoretical limit of 0.15 dB/km (for wavelengths near 1.55 µm), which is set by scattering from density fluctuations in the fiber core. Residual losses are attributed to defects in the silica glass network, the germanium dopant in the fiber core (used to raise the refractive index) and H₂ or OH contaminants.

4 We are indeed fortunate that a material as simple and abundant as silica combines all the features of low loss, high mechanical strength and chemical stability. (Are the parallel revolutions in silica and silicon technol­ogy coincidental?) With outside diameters of about 120 lira, the tiny strands of glass are surprisingly flexible and strong. Glass is vulnerable to damage and stress-accel­erated corrosion, but application of a protective polymer coating provides long-term mechanical and chemical sta­bility. Polymer science has played an essential role in the fiber success story.

5 Early developments of fiber optic communication re­quired a number of advances in material research and semiconductor laser development. The net achievement of this first generation of research and development is a worldwide network of optical fiber that provides low-cost, wide-band­width communications. Optical fiber now carries most long-distance telecommunication, and the bandwidth-distance product is doubling annually. (See figure 2.) Optical fibers are finding new applications in medicine, environmental sensing, and the aerospace and automotive industries, to name a few.

6 For long-distance communication it is necessary to compensate for the residual loss in the fiber by regener­ating the signal typically every 30-100 km (depending on the data rate). This is currently done by detecting the optical signal (that is, converting it to an electrical signal), followed by amplification and pulse shaping, and finally driving a laser to retransmit the regenerated signal over the next leg of fiber. The electronics needed for this procedure is the bottleneck of the fiber network. The bandwidth of such repeaters is typically much less than 10 gigabits (10¹⁰ bits) per second, while the full bandwidth of the optical fiber is in the multiterabit (10¹² bit) per second range. One could increase the communication bandwidth of electronically repeated systems by sending several wavelengths along the same fiber, but each wavelength would require its own set of repeaters That would be an expensive solution!

Erbium-doped fiber amplifiers

7 The emergence of the erbium-doped fiber amplifier in 1987 greatly changed this picture. It took only a few months for the importance of the erbium amplifier to be recognized. This invention is now revolutionizing telecommunication network design and marks the beginning of an exciting new phase of optical fiber research and development.

8 The fiber amplifier is a very simple device. (See figure 3.) The core of the fiber is doped with erbium ion (less than 0.1%) during fiber fabrication. The amplifier is pumped at either 1.48 µm or 0.98 µm with a commercially available semiconductor diode laser coupled into the amplifier with a wavelength multiplexer, which is a fiber device that sends light of two different wavelengths into a single fiber. Signal light near 1.55 µm is amplified by stimulated emission of the excited erbium ions as it passes through the fiber. (See figure 4.) A gain of over a thousandfold is readily achieved with pump powers of about 50 mW.

9   Erbium fiber amplifiers offer amplification independent of polarization in a wavelength range (1.53-1.5 µm) that lies in the region of lowest loss in optical fibers One can splice these amplifiers directly into the transmission fiber. The noncrystalline environment and the long lifetimes of the excited states of the erbium ion cause linewidth broadening, allowing many wavelength channels to be simultaneously amplified without cross talk. The optical amplifier removes the electronic bottleneck and makes for a transmission line that is data-rate transparent, broadband and lossless over a 4-THz bandwidth! Multi­channel operation at several different wavelengths is now commercially feasible, and this has increased the need for tunable semiconductor laser sources (which are used to create the optical signals in the first place).

10 Erbium-doped fiber design and performance continue to improve. The first transatlantic installation of an all-optical amplified system, which will be almost 6000 km long, is scheduled for operation in 1995, followed by a transpacific system 9000 km long in 1996.

11 While it might seem from this discussion that fiber amplifiers offer a potentially unlimited upgrade to already installed optical systems, new limits are evident. First, amplified spontaneous emission adds noise to the system. To minimize amplified spontaneous emission it is important to continue to reduce the loss of the transmission fiber and so minimize the number of amplifiers required. Second, much of the installed terrestrial fiber was designed for operation near 1.3 µm, because of the availability of 1.3-µm transmitters at the time of initial installation. This fiber has significant chromatic dispersion at 1.55 µm, causing pulse broadening, which limits the rate at which one can transmit data over long distances.

12 An alternate solution to upgrading the installed 1.3-µm-based network is to equalize the dispersion at each optical amplifier with special dispersion-compensating fiber having large negative dispersion. A number of such fiber designs have been fabricated, exhibiting negative dispersions up to 30 times the (positive) dispersion of conventional transmission fiber and equalizing the dis­persion over the entire wavelength range of interest. Designs with about -5 times the conventional dispersion are commercially available.

Nonlinearity

 

13 An important new limit to long-distance transmission is caused by optical nonlinearity of the silica fiber.

Although nonlinearities of silica are extremely small, and one never had to consider them in the past because electronic repeaters reshaped the pulses, long optical interaction lengths without regeneration and higher-average optical powers make even these small non-linearities significant. Nonlinearities result in pulse broadening and cross talk between channels. A num­ber of systems