At first it was a perfect match. The physical constraints of its site at the Lawrence Berkeley National Laboratory on a hillside above the University of California's Berkeley campus, the research interests of its initial proponents and the fiscal realities of the times all pointed to the same conclusion in the early 1980s: the Advanced Light Source (ALS) should be a third-generation, but low-energy, synchrotron radiation source designed for highest brightness in the soft-X-ray and vacuum-ultraviolet spectral regions.

While the ALS has turned out to be a world leader in providing beams of soft X-rays - indeed, furnishing these beams remains its core mission - there has nonetheless been a steadily growing demand from synchrotron radiation users for harder X-rays with higher photon energies. The clamour has been strongest from protein crystallographers whose seemingly insatiable appetite for solving structures of biological macromolecules could not be satisfied by the number of crystallography beamlines available worldwide.

The question was how to provide these X-rays in a cost-effective way without disrupting the thriving research programmes of existing ALS users. Superconducting bend magnets (superbends) provided the answer for the ALS and a proposal was adopted (a proposal that was originally made in 1993 by Alan Jackson of Berkeley and Werner Joho of Switzerland's Paul Scherrer Institute) to replace some of the normal combined-function (gradient) magnets in the curved arcs of the storage ring with superconducting dipoles that could generate higher magnetic fields and, thus, synchrotron light with a higher critical energy.

A team headed by David Robin, the leader of the ALS Accelerator Physics Group, took on the pioneering task of retrofitting superconducting bend magnets into the magnet lattice of an operating synchrotron light source. In particular, three 5 Tesla superbends were to replace the 1.3 Tesla centre gradient magnets in Sectors 4, 8, and 12 of the 12-fold symmetric ALS triple-bend achromat storage-ring lattice. The long project culminated early last October when, after a six-week shutdown to install and commission the superbends, the ALS reopened for users with a new set of capabilities.

The superbends have extended the spectral range of the ALS to 40 keV for hard-X-ray experiments. They do not degrade the high brightness of the ALS in the soft-X-ray region, for which the ALS was originally designed, nor do they degrade other performance specifications, such as beam stability, lifetime and reliability. They do not require that any straight sections normally occupied by high-brightness undulators be sacrificed to obtain high photon energies by filling them with high-field, multipole wigglers. Superbend magnets are already serving the first of a new set of protein crystallography beamlines. Ultimately, 12 new beamlines for crystallography and other applications, such as microtomography and diamond-anvil-cell high-pressure experiments, will be constructed.

Superbend history

The ALS was originally based on an electron storage ring with a 198 m circumference and a maximum beam energy of 1.9 GeV to provide peak performance in the vacuum-ultraviolet and soft-X-ray spectral regions. One way for the ALS to respond to the demand that arose in later years for higher photon energies would have been to use some of its scarce straight sections for high-field, multipole wigglers. Later, in 1997 the ALS did install one such wiggler - a device that provides the hard X-rays for an extremely productive protein crystallography beamline (Beamline 5.0.2) operated by the Berkeley Center for Structural Biology.

However, the drawback of the wiggler route was immediately obvious: many wigglers would limit the number of high-brightness undulators that give the ALS its state-of-the-art, soft-X-ray performance and that justified its construction in the first place. Moreover, a wiggler cannot readily service more than one beamline capable of the demanding multiwavelength anomalous diffraction experiments that many crystallographers want to perform, whereas a bend magnet can. In the end, the ALS adopted the superbend alternative proposed by Jackson and Joho - a choice that brought along some imposing challenges.

Superconductivity is no stranger to synchrotron light sources, where superconducting bend magnets have been used in small (mini) synchrotrons dedicated to X-ray lithography. In addition superconducting insertion devices in straight sections are, if not common, a venerable technology. Unlike wigglers and undulators in straight sections, however, superbends would be an integral part of the storage-ring lattice in a large multi-user facility and could not simply be turned off in case of failure or malfunction. So, the stakes were very high - the pay-off would be an expanded spectrum of photons to offer users; the risks included the possibility of ruining a perfectly good light source or, at the very least, causing unacceptable downtime.

In 1993, newly hired accelerator physicist Robin was set to work on preliminary modelling studies to see how superbends could fit into the storage ring's magnetic lattice and to determine whether the lattice symmetry would be broken as a result. He concluded that three superbends with fields of 5 Tesla, deflecting the electron beam through 10° each, could be successfully incorporated into the storage ring. Later, beginning in 1995, Clyde Taylor of Berkeley's Accelerator and Fusion Research Division (AFRD) led a laboratory-directed R&D project to design and build a superbend prototype.

By 1998 the collaboration (which included the ALS Accelerator Physics Group, the AFRD Superconducting Magnet Program and Wang NMR Inc) had produced a robust magnet that reached the design current and field without quenching. The basic design has remained unchanged through the production phase. It includes a C-shaped iron yoke with two oval poles protruding into the gap. A mile-long length of superconducting wire made of niobium-titanium alloy in a copper matrix winds more than 2000 times round each pole. The operating temperature is about 4 K.

With the strong support of ALS advisory committees and Berkeley laboratory director Charles Shank, Brian Kincaid - at that time the ALS director - made the decision to proceed with the superbend upgrade, and his successor, Daniel Chemla, made the commitment to follow through. The superbend project team, now including members of Berkeley's engineering division, held a kick-off meeting in September 1998 with Robin as project leader, Jim Krupnick as project manager and Ross Schlueter as lead engineer. Christoph Steier then joined the team a year later as lead physicist.

Subsequently, the success of wiggler Beamline 5.0.2, combined with some pioneering work on normal bend-magnet beamlines by Howard Padmore and members of his ALS Experimental Systems Group, led to the formation of user groups from the University of California, the Howard Hughes Medical Institute and elsewhere that were willing to help finance superbend beamlines, further adding to the momentum of the project.

Superbend team work pays off

For the next three years, the superbend team worked towards making the ALS storage ring the best understood such ring in the world. In every dimension of the project, from beam dynamics to the cryosystem, from the physical layout inside the ring to the timing of the shutdowns, there was very little margin for error. To study the beam dynamics, the accelerator physicists adapted an analytical technique used in astronomy called frequency mapping (see CERN Courier January 2001). This provided a way to "experiment" with the superbends' effect on beam dynamics both theoretically and experimentally before the superbends were installed.

Another technical challenge was to design a reliable, efficient and economical cryosystem capable of maintaining a 1.5 ton cold mass at 4 K with a heat leakage of less than 1 watt. Wang NMR was contracted to construct the superbend systems (three plus one spare). Wang designed a self-sustaining cryogenic system based on a commercial cryocooler, leads made of high-temperature superconductors and a back-up cryogenic reservoir.

Following some preparatory work during previous shutdowns, the installation of the superbends began in August 2001. The initial installation plan was very tight. In one 11-day period, the superbend team removed three normal gradient magnets and a portion of the electron-beam injection line in straight section 1 just upstream of Sector 12; installed the superbends; modified cryogenic systems; and completed extensive control system upgrades. They also installed many other storage-ring items and prepared for start-up with a beam.

After the installation phase, the goal was to commission the ALS with superbends and return the beam to users by 4 October. This schedule allowed the month of September to commission the ring (with the exception of a four-day break for the installation of the front ends for two superbend beamlines) and a three-day period for beamline realignment. However, commissioning proceeded much faster than had been expected and it was less than two weeks after the start of the installation when the machine was ramped up to full strength, and the effects of the superbends on the performance of the storage ring were fully evaluated.

Because so much was at stake, the storage ring had been studied and modelled down to the level of individual bolts and screws to ensure a smooth, problem-free installation into the very confined space within the storage ring. This attention to detail also paid off in the rapid commissioning. To take one example, the superbends were very well aligned, as demonstrated by a stored beam with little orbit distortion and small corrector-magnet strengths.

At the end of the first day, a current of 100 mA and an energy of 1.9 GeV were attained. At the end of the first weekend, the injection rate and beam stability were near normal. By the end of the first week, the full 400 mA beam current was ramped to 1.9 GeV and studies of a new, low-emittance lattice with a non-zero dispersion in the straight sections (designed to retain the high brightness that the storage ring had without superbends) were begun. By the end of the second week, test spectra taken in some beamlines showed no change in quality due to the presence of superbends.

Since reopening for business in October, the ALS has not experienced any significant glitches that might be associated with such a major change. Overall the ALS has made good on its promises to users of installing and commissioning the superbends without disrupting or delaying their research programmes and operating them with no adverse effects on performance in the bread-and-butter soft-X-ray spectral region, as demonstrated by the values of the storage-ring parameters (see table).

Superbend beamlines are already taking data and more are under construction or planned. Three superbend protein-crystallography beamlines are now taking data, and researchers at the first of these to come on line have already solved 15 structures. Three more crystallography beamlines are on the way. Non-crystallography beamlines currently in the works include one for tomography and one for high-pressure research with diamond-anvil cells, two areas for which superbends are even more advantageous than they are for protein crystallography, because they more fully exploit the higher photon energies that superbends can generate. Many other areas, including microfocus diffraction and spectroscopy, would also benefit enormously through the use of the superbend sources.

In summary, a new era at Berkeley's ALS is under way.

Further reading

S Marks et al. 2001 "ALS Superbend magnet performance" Proceedings of the 17th International Conference on Magnet Technology (Geneva)
D Robin et al. 2001 "Superbend Project at the Advanced Light Source" Proceedings of the 2001 Particle Accelerator Conference (Chicago, Illinois).
D Robin, C Steier, J Laskar and L Nadolski 2000 "Global dynamics of the ALS revealed through experimental frequency map analysis" Phys. Rev. Lett. 85 558.
J Zbasnik et al. September 2000 "ALS Superbend magnet system" Proceedings of the 2000 Applied Superconductivity Conference (Norfolk, Virginia).