Over the past few years, as the magnetic gap of the undulators was reduced from 20 to 10 mm, excessive Bremsstrahlung has been observed downstream from the beamlines. This Bremsstrahlung is created by the collision of the electron beam with the residual gas inside the ID vacuum chambers and becomes a serious problem for small aperture chambers where the small conductance results in high pressure. The Bremsstrahlung is scattered off-axis by the optical component in the beamline (monochromator, mirror,) and can be detected in the air outside the lead hutch. Even though more shielding can be added to the beamline hutches, for a number of beamlines this is not desirable or practical.

During the early times, some freshly-installed 10 mm external chambers showed a high level of Bremsstrahlung and a very slow conditioning with time to the point that the ID chambers had to be removed and replaced with wider aperture chambers to allow safe operation of the beamline. To cope with this situation, two solutions were found: one consisting in using in-vacuum undulators with a large pumping capacity, the other in designing the ID chambers with an antechamber. However, the presence of such an antechamber restricts the space available for the magnets of the associated insertion device and therefore, given the large number of existing undulator magnet arrays, it was deemed inappropriate for the ESRF to adopt such a solution.

A third solution was explored and developed following collaboration with CERN. It consists of depositing a thin layer of a Non Evaporable Getter (NEG) material on the surface of the chamber. The NEG material is an alloy of titanium, zirconium and vanadium deposited by magnetron sputtering. The vacuum in a NEG coated chamber benefits from two effects: a reduced desorption yield as well as direct pumping by the NEG material.

From 1999 to 2003, a number of chambers made from either stainless steel or aluminium were coated at CERN and gave satisfying results. In 2003, a NEG coating facility was commissioned at the ESRF with the ability to coat 5 m long, 8 mm aperture chambers. This facility was operated throughout 2004 and 6 ID chambers made of aluminium (5 m long, 8 mm aperture) were coated. Five of these were installed on the ID6 beamline for vacuum pre-conditioning and Bremsstrahlung measurement vs. integrated dose. When qualified, the chamber is moved to its final beamline destination during the following shutdown. The low desorption of the NEG, its pumping effect as well as the memory of the conditioning made on ID6 are such that, under normal conditions, the beamline receiving such a new chamber restarts user operation immediately after the shutdown with no delays to the user program. As of January 2005, 17 ID vessels installed in the ring are NEG coated.

The initial coating thickness, which was close to 1 micron, has been reduced recently to 0.5 micron. Measurements show that the smaller thickness does not change the pumping and desorption capacity significantly but such chambers could be coated more rapidly with a lower risk of insufficient adhesion of the NEG layer to the chamber walls. Indeed, out of the 6 chambers processed in 2004, 5 perform as expected vacuum-wise while one has shown adhesion problems with NEG strips detaching from the wall. This adhesion problem, which is specific to aluminium chambers (not observed on stainless steel), is believed to be linked to an insufficient surface preparation before the coating is applied and is presently being looked into.

A short sample chamber was successfully coated in 2004 with a magnetic field generated by a permanent magnet array rather than a solenoid. The interest of this procedure is the ability to coat chambers of complex shape than cannot be embedded into a cylindrical solenoid.