Thought's on what makes for a High Throughput XAS beamline

One of the cardinal assumptions about NSLS-II is that higher efficiencies of operation at NSLS-II beamlines will make for the more limited capacity in relation to NSLS. On this page, I will jot down some notes on what the efficiency limiting aspects of an XAS beamline are and how those aspects might be improved at a beamline designed and built from scratch.

Detection

Ge and Si-drift detectors

For fluorescence experiments, detection is the main rate limiting step. The solution is highly integrated energy discriminating detection. For energies below about 20 keV, silicon drift detection (SDD) is vastly preferable for ease of use (no LN2!) and flux per element. Above 20 keV, germanium detection is preferable. In either case, subtending as much solid angle as possible is desirable. This means that increasing the number of detector elements is key.

A CLS-style 100 element Ge detector would be great for the higher energy applications. For an NSLS-II 3PW beamline, a highly integrated SDD would be preferable. Currently IIS sells a 4-element detector. Higher integration (7, 9, 16 and so on) would be great. Given the higher flux that the SDD can tolerate, a 16 element SDD would out-perform a 100-element Ge detector while being vastly easier to operate and substantially less expensive.

That said, for fluorescence XAS from a small spot, such as the spot from the focusing mirror, and for certain kinds of samples, there is an advantage to the one-element SDD. If the snout of the SDD can be placed at 45 degrees, elastic and Compton scattering are minimized. If the single element is placed very close to the sample, it will subtend a substantial solid angle. Running at high count rates and using an effective dead time correction (see this recent paper), this geometry is very effective.

Detector configuration

Software tools for setting regions of interest

Other detectors

(multi-channel)

Slew scanning

Step scanning is, for most experiments, a major time sink. The overhead associated with mono motion and settling can be as much as half the time of a scan. All beamlines should operate in slew scanning mode, even those not explicitly designed as QEXAFS/time resolve beamlines. Three five-minute slew scans will probably result in much better data than one 15 minute step scan.

Beamline configuration management

Motor configurations

Depending on optics complexity, there are many actuators that have to be set to particular positions for experiments at particular energies. All of these positions should be pre-measured and stored for speedier changes between edge energies.

Gas handling

Oddly, the simple chore of flushing ion chambers can be quite time consuming. In a continuous flow system, equilibration times can be 10 of minutes. An automated system that pumps and refills the ion chambers could reduce that time from 10s of minutes to less than 10 minutes. With several gas sources (He, N2, Ar, Kr), several output channels (I0, It, Ir, If, user defined), enough safety features to meet DOE requirements, and labor, a fully automated, PLC controlled gas handling system would spec out to over k$100 unless some kind of efficiencies of scale can be found.

Edge energy specification

All of the configurational aspects described above should be keyed to the desired edge. Configuration data can be stored in a data base. The user requests configuration for a particular element. The FOE and hutch undergo a whirlwind of activity and a few minutes later are ready to start an experiment.

Data management

• experimental configuration
• data storage and processing
• extrinsic data referencing
• bar coding of samples, remote and early management of scan parameterization

Sample mounting

Many experiments could be made to work with a standardized sample cartridge. This cartridge could be made from a number of materials for use in different environments (furnace, cryostat, etc). The cartridge would have an obvious space for mounting samples and some way of pseudo-kinematically fixing it to sample holders (i.e. pins and holes, embedded hard magnets, ball and groove, etc). Sample holders can then have well defined mounting geometries all leading to easier creation of high throughput, "combinatorial" style experiment.

If these cartridges are cheap enough, they can be made by the 10s of thousands and sent to users by the hundreds prior to an experiment. Samples can be mounted in cartridges at home and quickly mounted once at the lab.

Of course, many experiments are quirky and will have quirky sample mounting needs. But a surprising number could be fit into a standardized package.