by Sergei BABIN, RAS Corresponding Member, Head of Fiber-Optic Laboratory, Deputy Director in scientific work, Institute of Automation and Electrometry, RAS Siberian Branch (Novosibirsk)

The high-speed means of communication play an increasingly noticeable role in the present-day dynamic world. Today up to 75 percent of communicated information, from Internet to mobile telephones and television, are conveyed through trunk fiber-optic communication lines, and this figure increases every year. Therefore, developers are facing the problem of elaborating a new generation of communication systems, in particular those capable to deliver a signal with no quality lost for long distances with minimum costs. Laser technologies assist in solving these problems.

Some principles of laser operation is a subject we would like to begin with. It is well known that laser generation requires active medium amplifying the light (for example, ruby crystal) and positive feedback modifying optical amplifier into coherent radiation oscillator. To develop such communication, the active medium is placed in a resonator, which consists usually of two mirrors adjusted in parallel to each other. The mirrors cast the light backward to a gain medium, and if an amplification factor exceeds losses at two-way passage between mirrors, oscillation threshold is reached, and radiation power rises significantly, however not to infinity but stabilizes on the level determined by the saturation effect--amplification in active medium decreases with power rise and becomes equal to losses in the resonator in steady conditions.

The cross section of a laser beam is limited by sizes of the active medium or resonator mirrors, which cannot be very long, as a beam expands in a free space due to diffraction. The less the beam size, the more its diver-

gence and consequently the losses on the mirrors. The closed version of the resonator is completely free of these disadvantages, when in a space between mirrors the beam propagates along a waveguide, for example, a fiber light guide*. The latter is important for further description, therefore we shall touch upon its design briefly.

The fiber light guide core of about 10 micron is made of doped quartz glass and has an increased refraction index. The external sheath of about 100 micron diameter is also made of glass and covered with plastic at the top. Light travels in the core at the expense of total internal reflection practically without losses. Quite negligible losses are determined by the Rayleigh scattering** on refractive irregularities of the submicron scale typical of glass structure. In passive fiber light guides used in telecommunications the loss coefficient decreases with wavelength growth and reaches minimum (~0.2 dB/km) close by ~ 1.55 mkm, i.e. the signal weakens hundredfold after passing 100 km, which determines the maximum length of a section of a fiber communication line between the signal amplifiers.

Such light guides have one more useful property, namely, photosensibility: if any section of a fiber core is exposed to ultraviolet radiation, it is possible to change the refraction index within this section. More simply, by using this method it is feasible to create similarity of filters or, if anything, shutters on the way of light which would also reflect it, i.e. act as mirrors. A periodic structure of the refraction index, the so-called Bragg's fiber

* See: A. Prokhorov, Ye. Dianov, "Fiber Optics. Problems and Perspectives", Science in the USSR, No. 3, 1987.--Ed.

** Rayleigh scattering is light scattering on objects, whose sizes are less than its wave. It is named after British physicist Lord Rayleigh who discovered this phenomenon in 1871.--Ed.

grating* with a reflection index exceeding 99 percent, is formed in the light guide core to reflect a specific light wavelength. These intrafiber laser mirrors withstand high power and operate for many years. Due to them it is possible to modify a light guide into a fiber laser (this idea was suggested by Elias Snitzer, a staff member of the American Optical company already in 1961, i.e. only one year after the start-up of the first ruby laser by the American physicist Theodore Maiman).

Already at that time the advantages of a fiber active element in comparison with crystal were evident: effective heat removal is achieved at the expense of a large specific area of the fiber guide surface, and its waveguide properties provide a high-quality output beam, which is heat-resistant to the active element. But the level of technologies at that time (low-quality optical fiber, complex pumping circuits and necessity for interfacing of active light guides with a bulky optical system, i.e. mirrors and other elements) put back practical application of this laser type to dozens of years.

Advance in their development became practical mainly due to the appearance of optical fiber communication and telecommunication technologies, first of all, creation of quartz glass low-loss light guides due to the reduction of admixure concentration at the end of the 1960s. Chinese (and also British and American) physicist-engineer Charles Kao was awarded the Nobel Prize in 2009 for research in this field. Practical application of optical fiber lines of communication, especially rapid in the 1990s after creation of Internet, resulted in the development of a

* Bragg's grating is named after British physicists William Henry Bragg and William Lawrence Bragg (father and son), founders of the X-ray diffraction analysis, the 1915 Nobel Prize winners.--Ed.

The principle of operation of a laser with random distributed feedback.

When pumping radiation (1,455 nm) is introduced into a light guide in two directions, distributed amplification is created due to the effect of stimulated Raman scattering both for direct and scattered photons, and if the integral amplification index exceeds losses for a full passage, there originates generation of 1,560 nm.

radically new element base of fiber optics: effective sources of optical pumping, namely, semiconductor lasers* with radiation coupling to optical fiber, Bragg's fiber gratings and other elements, which served as a basis for development of effective lasers in a fully fiber design.

The simplest version of such laser unit represents a section of active fiber light guide with a core doped with ions of rare-earth elements (ytterbium, erbium, etc.), on whose ends Bragg's fiber gratings are formed which act as a light beam reflector. Noncoherent radiation of a multi-mode pumping laser diode enters a light guide through a brancher and transforms active ions to an excited state, thereby creating a gain medium. Thus, Bragg's gratings, which reflect light on a resonance frequency, form a laser resonator directly in a fiber light guide.

Realization of a fully fiber circuit has brought about a revolution in laser technology, as it does not require mirror alignment and provides a highly effective and stable generation with a high-quality beam. As it turned out, for the purpose of amplification it is not necessarily to use optical pumping of laser levels of rare-earth element ions. It can be created also in a passive telecommunication fiber at the expense of the stimulated Raman scattering (SRS) on optical phonons (oscillations) in glass. Occurrence of this effect is attributed to concentration of intense radiation in the core of an extended light guide.

In 2006, in the course of experiments with SRS-lasers carried out together with a group of British colleagues from the Aston University (Birmingham) headed by our national professor Sergei Turitsyn we set forth two questions: To what extent can the length of a fiber laser be

* Russian scientist Academician Zhores Alferov was awarded Nobel Prize in the field of physics in 2000 for creation of room-temperature heterostructure lasers.--Ed.

increased? Can it exceed the extension of the passive section of an optical fiber communication line (~100 km)? The work started with realization of SRS-lasers with a resonator length (distance between Bragg's fiber gratings acting as mirrors) of 10-20 km, then this parameter reached ~100 km. In 2009, we succeeded in finding of the required limit, it was equal to 270 km. It has been found that the structure of longitudinal modes* of the linear resonator (with intermode distance ~400 Hz) is observed up to this boundary. It means that between "mirrors" spaced apart to 270 km, there forms a standing electromagnetic wave** (it appears by superposition of the incident and reflected waves of equal intensity moving towards each other), which by itself is amazing. It is even more amazing that with further extension of the length (to 300 km and more) laser also operates but already in "modeless" conditions.

The following assumption was made. In our case, generation is caused by the Rayleigh scattering on submi-cron heterogeneities of the refraction index. It is a scattering, which is responsible for a blue color of the sky overhead and the minimum level of losses in telecommunication fiber light guides. Though scattering in a light guide moves around, a part of radiation going back gets again into the light guide and propagates in the opposite direction. Integrally this effect is very small (at a level of 0.1 percent), but if distributed amplification

* Mode is a type of oscillations. They differ from each other in a propagation speed, intensity distribution throughout the section of a light guide, and direction of the electric field vector. According to the transverse structure, light guides are broken down into the single-mode and multi-mode structures.--Ed.

** Standing waves (in a general state) are oscillations in distributed oscillatory systems with specific distribution of amplitude peaks and lows. They occur in reflections of obstacles and non-uniformities as a result of superposition of the reflected and incident waves; examples -- oscillations of a string or oscillations of air in organ pipes.--Ed.

Generation power on one end of fiber depending on the full pumping power (A).

Generation spectrum p(λ) depending on the full pumping power (B): 2 W--1,557 nm line is generated; 2.4 W-1,557 and 1,567 nm lines are generated simultaneously, which correspond to two local maxima in a spectrum of SRS-amplification; 2.7 W-1,567 nm.

is created in fiber (for example, at the expense of the stimulated Raman scattering), such scattered radiation can be sufficient for overcoming the generation threshold even in case of absence of ordinary point reflectors.

The experiment was carried out for validation of this hypothesis. Distributed amplification with simultaneous elimination of stray reflections (from fiber faces and connections) was created in a fiber of ~100 km long. It turned out that narrow-band laser generation with spectrum localization close by maximum of SRS amplification was observed on both ends of the fiber on exceeding some pumping power threshold (~1.5 W) even without Bragg's mirrors. As the SRS line in quartz glass has two local maxima, two lines (1,557 and 1,567 nm) or one of them are observed in generation depending on pumping power. Besides, it has been proved that generation occurs due to the so-called accidental distributed feedback (ADF) owing to the Rayleigh scattering.

In terms of quality ADF-laser is similar to thoroughly studied fiber lasers with regular distributed feedback (the so-called DF-lasers, in which Bragg's grating is formed along the whole active fiber light guide), but due to the difference in grating intensity, the linear scales of such lasers differ by 7 orders (hundreds of kilometers and centimeters respectively). The random nature of distributed feedback also plays a fundamental role. The regular structure of longitudinal modes lacks in ADF-laser, and a continuum of random spectral components close by two maxima of signal amplification of 1,557 and 1,567 nm takes part in generation. There is another fundamental distinction of ADF-laser. Its feedback depends on distributed amplification determined by optical pumping, and the latter in its turn is exhausted with the rise of laser power generated in the continuum of interacting spectral components. This principally changes the mechanism of generation. Without going into details important for specialists, it should be noted that for its understanding we have to answer questions of the fundamental nature. Such attempts were made also by other specialists--after publication of our results in the Nature Photonics journal in 2010, several groups of scientists got interested in this problem. Today we already can say that the scientific community regards the scheme suggested by us as a new type of laser generation.

Conceptually this problem is close to the rapidly developing lately concept of "random" lasers, i.e. generation in disordered (randomly heterogeneous) gain media, such as powders of laser crystals or semiconductors, dye suspensions with scattering nanoparticles, etc. In contradistinction to regular lasers, in which radiation properties (spectrum and form of an output beam) are determined by resonator modes, the "random" lasers have no optical resonator in the usual sense, and their characteristics are determined by processes of multiple scattering in a disordered gain medium. It should be noted that this concept was first laid down by the pioneer of laser cooling of atoms Professor Vladilen Letokhov back in 1967.

"Random" lasers have a very simple design especially as compared with semiconductor heterostructure and crystal microlasers, which require a precision resonator. Of course, their output characteristics are to be improved because these new systems usually radiate in a pulsed mode and have a complex random spectrum of generation and an oversophisticated beam direction diagram.

Readjustment of a generation spectrum of RDF-laser by means of a special filter in the range of 1,535-1,570 nm.

Change-over to smaller dimensions is one of the updating methods of such systems. Previously it has been proved that one-dimensional random media (a set of plates of random thickness or a dye suspension with nanoparticles in a hollow light guide) provide formation of a directed beam exactly as in regular lasers, however, the time and spectral characteristics of random lasers are inferior to them as yet.

In this respect the fiber laser with a random distributed feedback developed by our specialists can be considered a one-dimensional "random" laser because the light is amplified and dissipated only in one direction, i.e. forward or back along a light guide. However, in regular "random" lasers representing a small "ball" composed of dissipating and/or amplifying particles, the light propagates in all directions. In this case, a "random" fiber laser differs from volumetric lasers by a narrow spectrum, high stability and quality of beam determined by waveguide properties of optical fiber as in case of regular fiber lasers. But as opposed to the latter having a resonator in the form of mirrors (regular point or distributed reflectors), the Rayleigh ADF-lasers have no principal length restrictions, can perform rather simple frequency readjustments and generate on many lines in different spectral ranges. In particular, installation of an acousto-optical filter with a fiber input and output in the circuit center (between pumping inputs) provided smooth rearrangement of the RDF-laser in a wide wavelength range of 1,535-1,570 nm with power variations <3 percent (~0.1 dB), which is by an order of magnitude better than the readjusted SRS-lasers with a linear or ring resonator.

The unique properties of the Rayleigh fiber RDF-lasers are applicable in basic and applied research as well as practical work especially in ultralong optical communication and distributed sensor systems, which is a subject for further studies. In particular, in case of ultralong fiber laser (with a combined resonator composed of point reflectors and a distributed Rayleigh "mirror") realized directly in a fiber optic communication line, the generated radiation can be sufficiently homogeneous lengthwise and used as homogeneous secondary pumping of a distributed SRS-amplifier of an optical signal. If induced amplification and its losses are equal, the communication signal can be transmitted successfully to great distances within a wide frequency band. This is a real base for creation of brand new systems of highspeed information transfer even without intermediate amplifiers. Such work has been already started at our institute under a contract with OAO Rostelecom. In late 2011, the efficient transfer of information over a 250 km long line without intermediate amplifiers was carried out in laboratory conditions. In the near future we plan to test the new technology on the existing main communication line.