Résumé

Un brillant avenir pour les lasers à électrons libres

Les lasers à électrons libres constituent une application importante des accélérateurs de particules, les faisceaux de particules étant utilisés pour produire des impulsions de lumière intenses et cohérentes. à présent, les sources de lumière cohérentes linéaires du SLAC sont entrées dans la région spectrale des rayons X durs pour la première fois, ouvrant de nouvelles perspectives sur les processus dynamiques au niveau atomique et moléculaire. Leur développement et leurs perspectives sont étroitements liés à la physique des faisceaux de particules. La recherche en cours pourrait conduire à des réductions de dimension, non seulement pour les lasers à électrons libres, mais aussi pour les accélérateurs.

The successful lasing at the Linac Coherent Light Source (LCLS), which makes use of SLAC's famous linac, is a landmark in the development of coherent electromagnetic radiation sources. It is the first hard X-ray free-electron laser (FEL), operating in the wavelength range 0.12–1.5 nm, with a respective pulse duration of 50 to several femtoseconds and the number of coherent photons per pulse ranging from 1013 to 1011 (Emma et al. 2010). Until recently the best hard X-ray source has been undulator radiation from an electron beam of a few giga-electron-volts in a specially designed storage ring. Now, the LCLS has set a new standard. Its peak X-ray brightness is higher than that of the best synchrotron radiation source by 10 orders of magnitude, as figure 1 shows. Its X-ray beam gives us, for the first time, the capability to explore matter with wavelengths and pulse durations as short as 0.1 nm and 1 fs – the atomic scales of length and time.

With these breakthrough characteristics the X-ray FEL provides a new window on dynamical processes at the atomic and molecular level. Imaging of non-crystalline matter at nanometre and subnanometre scales is also made possible by the coherence properties of the radiation. Taking an atomic-scale motion picture of a chemical process in a time of a few femtoseconds or less, and unravelling the structure and dynamics of complex molecular systems, such as proteins, are among the exciting experiments made possible by this novel source of X-rays. The LCLS – and the other hard X-ray FELs now being built at DESY in Germany and at Spring-8 in Japan – will open a new chapter in the biological and physical sciences, together with other hard X-ray FEL projects being developed in China, Korea and Switzerland.

The soft X-ray region, at wavelengths from a few nanometres to about 50 nm, has also seen a dramatic increase in performance with the equally successful operation of FLASH at DESY (FLASH: the king of VUV and soft X-rays). Another soft X-ray FEL, Fermi@Elettra in Trieste, will be completed by the beginning of 2011 and new soft X-ray FELs are being designed at the Lawrence Berkeley National Laboratory and at SLAC, as part of LCLS II, an addition to the existing LCLS.

Another important aspect of FELs has been the successful construction and operation at the Jefferson Laboratory of an FEL with high average power, up to 9–15 kW, at wavelengths ranging from about 1.6 to 6 μm (CERN Courier June 2005 p27). A distinctive characteristic of this FEL, critical for high average-power operation, is the use of a recirculating superconducting linac to recover the electron energy after the beam has amplified the FEL radiation (figure 2). A similar approach to reach even higher average power is being used in an extension of the programme sponsored by the US Navy.

Soft and hard X-ray FELs represent the latest success in the development of FELs, which began a few decades ago. Many FEL oscillators have been in operation since the 1980s, at laboratories around the world, but mainly in the near- to far-infrared region. Scientists have used their unique properties of full tunability from the visible to the far-infrared and high peak power of tens to hundreds of megawatts to explore many areas of physical, chemical and biological sciences.

FEL oscillators operating in the infrared, visible or near ultraviolet part of the electromagnetic spectrum can take advantage of the existence of high-reflectivity mirrors to use an optical cavity and so operate at small gain, as John Madey first proposed (Madey 1971). In this mode of operation the undulator magnet, where the electron beam emits the radiation, can be rather short, although many electron bunches are needed to amplify the radiation in small steps up to the maximum saturation value. For an oscillator system the radiation pulse is nearly Fourier-transform and diffraction limited.

In the X-ray spectral region, research is being done to develop the mirrors and other components, in particular a very high-frequency electron gun, needed for an X-ray oscillator (Kim et al. 2008). Mirrors in general have small reflectivity and an optical cavity is complicated and, until now, impractical. The main approach – used for the LCLS, FLASH and most other new X-ray FELs – is not to use an optical cavity, but to operate in the high-gain regime, reaching saturation in a single pass of an electron bunch through a long undulator. The system can start either by amplifying the spontaneous radiation in the undulator within the bandwidth of the FEL gain (a self-amplified spontaneous emission, or SASE, FEL) or by amplifying an external laser signal (a seeded FEL) (Bonifacio et al. 1984). A SASE FEL is nearly diffraction limited but is not transform limited, except when a very short electron bunch is used to generate and amplify the radiation (Reiche et al. 2008). Seeded FELs can be nearly diffraction and transform limited.

The physics of the high-gain regime is that of a collective instability. Under the effect of an input field, external or self-generated from noise, the longitudinal distribution of the electron beam evolves from a random initial state to one in which the electrons are captured in microbunches separated by one radiation wavelength. In a SASE FEL, such as the LCLS, the collective instability leads to self-organization of the electrons in the equivalent of a 1D relativistic electron crystal propagating through the undulator at a speed near that of light. Because the crystal planes are separated by one radiation wavelength, the electrons emit in phase. As the radiation power grows exponentially, the electron longitudinal distribution changes from disorder to order. The result is that, while in the case of spontaneous radiation the total intensity is proportional to the number of electrons, Ne, in the high-gain case the total intensity is proportional to a power of Ne between 4/3 and 2. The number of electrons in a bunch is typically of the order of 109–1010, so the change in intensity can be quite large. The number of coherent photons emitted spontaneously by one electron going through an undulator is approximately given by the fine structure constant, or about 10–2. When a high-gain FEL reaches saturation the number can be as large as 103–104.

New levels of sophistication

The electron beams for X-ray FELs require a new level of sophistication in their generation, handling and diagnostics. An FEL operating as an oscillator or a single-pass amplifier requires an electron beam with a large six-dimensional phase-space density, becoming larger as the wavelength decreases. However, the scaling is much more favourable than for other kinds of laser, a key reason for the success of FELs in the X-ray spectral region. In the high-gain regime, the six-dimensional density required for the electron beam scales only as the inverse of the square root of the radiation wavelength.

The electron beam for the LCLS is generated by a radio-frequency electron gun before passing down the last kilometre of the 3 km-long SLAC linac (figure 3). This beam has the highest six-dimensional phase-space density ever generated. The density is preserved through the process of beam acceleration and longitudinal compression, minimizing the effect of all of the collective instabilities that can dilute the density of a high-intensity beam. In turn, the high phase-space density is used to allow the self-organization of the beam – the FEL collective instability that leads to lasing. It is interesting to note that much of the beam physics needed to control the instabilities during acceleration and compression has evolved in the study of electron–positron linear colliders.

The main difference between the LCLS at SLAC, the European XFEL and the Spring-8 Compact SASE Source (SCSS), Japan, is in the choice of the linear accelerator to produce the electron beam – an existing S-band room-temperature linac for the LCLS, a superconducting linac for the XFEL and a C-band linac for the SCSS. While the peak intensity and power will be similar for the room-temperature and superconducting systems, a superconducting linac can produce more electrons and more X-rays per second. The wakefields in the two types of linac are also different but in both cases will produce a beam with the required characteristics.

FELs, notwithstanding their larger size, cost and complexity, offer attractive advantages over atomic/molecular lasers: complete and continuous tunability, capability of high average power and a more favourable scaling for gain at short wavelengths. These characteristics, in particular the capability of lasing in the soft and hard X-ray regions, with control of the pulse length from a few to hundreds of femtoseconds, gigawatt peak power and full tunability, are making FELs attractive to an ever wider number of scientists. On the other hand, FELs are large and expensive, justifying their use only when their characteristics are fully utilized and when atomic and molecular lasers are not available to produce radiation with similar characteristics.

With the development of high repetition-rate electron guns and by making use of continuous-wave superconducting linacs, FELs can reach repetition rates of megahertz. Using the extraordinary brightness of electron beams produced by radio-frequency or other novel types of electron sources, together with high-frequency, high--gradient room-temperature linacs, it is possible to reduce the size and cost of the accelerators driving the FELs – a particularly important factor for hard and soft X-ray devices. Ongoing research into short-period undulators, as well as the novel laser/plasma accelerators being developed in many laboratories, might lead in the future to compact, table-top FELs, at a cost and size compatible with a university-scale laboratory.

Such developments in scale hold promise not only for the FELs. The six-dimensional density of the LCLS electron beam is so large that if the beam is further compressed in the transverse direction, it generates on its surface electric fields as high as many tera--electron-volts per metre (Rosenzweig et al. 2010). Because of these properties, the electron beam itself could be used for the exploration of atomic physics or for exciting plasma wakefields exceeding 1 TV/m, thus opening the way to a table-top tera-electron-volt accelerator for use in frontier high-energy physics. While the success of short-wavelength FELs owes much to the advances in the physics of particle beams stimulated by work on electron–positron linear colliders, it is now possible to look to a future in which the development of FELs will help to continue the exploration of matter at the subatomic level.

In summary, FELs are being developed to operate in an ever larger spectral region, from hard X-ray to terahertz frequencies, generating pulses with femtosecond to attosecond duration, or pulses with extremely small line width. Longitudinal coherence can be pushed to near the Fourier-transform limit using seeding and other techniques. Such high-power, diffraction- and transform-limited X-ray pulses, with a duration that can be controlled between attoseconds and hundreds of femtoseconds, will lead to new discoveries and new knowledge in many areas of science.