The radiation is produced by a beam of 6.04 GeV electrons circulating in a storage ring. The electrons radiate synchrotron radiation in the bending magnets of the ring lattice and in magnetic insertion devices ­ wigglers and undulators ­ that shake the beam.
However, the new Graal experiment at the ESRF generates radiation in another way to probe nuclear and nucleon structure.

Graal has been realized by a French-Italian-Russian collaboration with primary financial support from INFN in Italy and IN2P3 in France through several Italian laboratories and universities (Laboratori Nazionali di Frascati, Laboratori Nazionali del Sud, and the universities of Catania, Genova, Roma II and Torino), two French Institutes (Institut des Sciences Nucleaires and Institut de Physique Nucleaire) and the Institute for Nuclear Research of the Russian Academy of Science.

Undulators are normally composed of physical magnets, but Graal uses a micro-undulator in the ESRF straight section D7,which is a beam of ultraviolet laser light moving against the electron beam. The electrons scatter the laser photons, transferring energy to them and producing narrowly collimated gamma rays.

At the ESRF, a laser beam in the near-ultraviolet produces gamma rays with a maximum energy of 1.47 GeV, a maximum linear polarization of 98% and an intensity of a few millions of photons per second. In this way, Graal extends the ESRF resolving power to nuclear and that of nucleon structure down to a spatial resolution of 0.2 fm (0.2 x 10-13 cm).
The first advantage of this technique over normal electron bremsstrahlung synchrotron radiation is the almost flat energy spectrum. Polarization is the second advantage: photons scattered in the electron direction maintain their polarization. Therefore, at the higher end of the spectrum, the polarization is very close to that of the laser light. Rotating or changing the polarization of the gamma rays is easily accomplished by rotating or changing the polarization of the laser light.

A gamma beam of a useful intensity for studying photonuclear reactions was first produced at Frascati using the Adone storage ring. After this success, several more such beams were produced.

A beam with a maximum energy of 2.4 GeV is now under construction at Spring8 in Japan.
Graal's main goal is the study of photonuclear reactions in the intermediate energy region, where the nucleons cannot be treated as elementary particles and their internal degrees of freedom cannot be ignored, but away from the asymptotic freedom of quarks and gluons. In this region, many excited baryon states are clearly visible and many others await careful exploration
As highlighted by Feynman's quotation, photons are an interesting probe of hadronic structure because the interaction is given by the product of the electromagnetic vector potential and the hadronic current. The former is well known from quantum electrodynamics, and the relative weakness of electromagnetic coupling makes second-order effects small, thus models of photonuclear reactions are possible. Moreover, the possibility of using linearly and circularly polarized gamma rays makes several single- and double-polarization observables experimentally accessible, providing strong constraints on theoretical models. A linearly polarized photon beam introduces a fixed direction for the electric field, so that the reaction yield is no longer cylindrically symmetric with respect to the beam direction.

In a circularly polarized beam, the photons have well defined helicity and their spins are aligned parallel or antiparallel to their momentum. Parity conservation in photoreactions dictates the overall form of the interaction. The asymmetry in the weak decay of the (strange) lambda provides information on lambda polarization, so it is possible to measure the correlation between the gamma and lambda polarizations in the photoproduction of strange particles.

Ultraviolet photon beam
In Graal, an argon-ion laser provides a beam of ultraviolet photons. A three-lens zoom focuses them at the center of the laser­electron interaction region. Two precisely adjustable mirrors align the laser light with the electron beam 35 m away to within 3 µrad. A retardation plate rotates the plane of linear polarization of the beam. The ultraviolet enters the storage ring vacuum system through a quartz window and is subsequently reflected through 90° by a beryllium mirror coated with aluminium. This mirror lines up the laser beam with the electron beam. (Beryllium minimizes the absorption of back scattered gamma rays that travel in the opposite direction to the ultraviolet.)

Electrons that have transferred part of their energy to a photon, move with the electron bunch along the straight section but, owing to their lower energy, veer away in the next dipole and become separated from the un scattered electrons by a few centimeters. Measuring the distance between a scattered electron and the electron beam is a measure of its energy loss and therefore of the gamma energy.
The detection of the scattered electron and the measurement of its precise position are done with the tagging detector, which comprises plastic scintillators and silicon microstrips. The microstrips give the position of the electron and the scintillators give its precise timing. The electron timing correlates an event in the hadronic detector with the corresponding electron, thus providing the energy of the gamma ray that produced it. It also provides a precise starting signal for measuring the time of flight (TOF) of photoproduced particles. The jitter of the TOF start pulse, provided by the scintillators, is reduced to 120 ps, effectively synchronizing this pulse with the phase of the accelerating radiofrequency of the ring. This is possible because the electrons travel in short bunches separated by 2.8 ns.
The Graal hadronic detector covers the entire solid angle except for two small entry and exit holes along the beam axis. The detector is made of three parts. In the central part, between 25° and 155°, the emerging particles pass through two cylindrical wire chambers and a barrel of 32 thin plastic scintillators, then enter a calorimeter made of 480 BGO crystals, each 24 cm long, and arranged, like an orange, in 32 sectors of 15 crystals each.

Particles emitted at angles of less than 25° go through two plane wire chambers and three plastic scintillator walls. The first two thin walls are used to measure the specific ionization of the particles. Then a thick wall, with alternating layers of plastic scintillator and lead, measures the total charged particle energy and detects neutrons and gamma rays. All three walls provide a measurement of the position and time of flight of the particles.
Particles emitted backwards encounter two plastic scintillator discs separated by lead. Each disc has a small central hole for the passage of the beam and is viewed by 12 photomultipliers to reconstruct the position and timing of a particle. The responses of the two discs allow charged particles and gamma rays to be differentiated.
The main features of the Graal detector are high efficiency, good energy resolution for gamma-ray detection and complete angular coverage. The detector is well suited to events producing several photons, like the photoproduction of neutral pions and etas, and the identification of the various eta decay channels. The first results to emerge are extensive measurements of the beam polarization asymmetries for the photoproduction of positive and neutral pions and etas. The two-photon and three-neutral-pion (giving six photons) decay channels of the eta have been detected simultaneously.
Polarization asymmetries, derived experimentally from the ratio of successive measurements with the same apparatus, are immune to otherwise common systematic experimental errors, such as the knowledge of the solid angle, the efficiency of the apparatus, the measurement of the dose and the size of the target. From a theoretical point of view, polarization asymmetries are given by the interference of different amplitudes and are therefore more sensitive to small, hitherto unobserved, contributions ­ if b is much less than a, then ab is more sensitive to b than is a2 + b2.

Event discrimination
Another advantage of full solid-angle apparatus, with a high overall efficiency for the detection of gamma rays, is its ability to discriminate rare events, where only one or a few photons are produced, from the more frequent events containing many gammas.
One example is Compton scattering, which is about 50 times as rare as neutral pion photoproduction and can be difficult to single out using only kinematics. However, Compton scattering has only one photon in the final state, while neutral pion photoproduction has two. Another example is the rare decay of the eta into a neutral pion, which has four photons in the final state compared with the frequent decay into three neutral pions, which has six.

Graal is now in full operation. It can collect data for more than six months per year ­ a large fraction of the time that the ESRF ring is available to the experimenters. A 10 mK dilution refrigerator and a 16 T magnet are now being delivered for the construction of a polarized target. With polarized targets, double-polarization experiments will be possible in all channels.