Partner User Proposal Abstract
Development of X-ray Photon Correlation Spectroscopy Techniques to Probe Soft Matter and Complex Fluids (PUP-17)
At the beamline planning stage, it was widely recognized that one of the most exciting scientific opportunities offered by the then-unprecedented brilliance of the APS was the possibility of carrying out x-ray photon correlation spectroscopy experiments. Such experiments promise new insights into dynamical phenomena in condensed matter occuring on shorter shorter length scales than can be reached in light scattering and longer length scales than can be achieved with the neutron-spin-echo technique. Our goal under this PUP is to create at 8-ID a national user facility for XPCS studies.
Unfortunately, XPCS cannot presently be exploited for many of its potential applications, because a poor SNR limits the range of wave-vectors or time scales that can be accessed. While some gains in the XPCS SNR can be expected from gains in the APS brilliance, improvements in optics and detectors promise even more significant benefits. In addition, the current user environment for XPCS is not yet as user-friendly as it needs to be.
To realize the maximum possible SNR, the beam coherence must not be degraded as the beam propagates through the beamline. Ideally, the coherence lengths of the beam increase linearly with distance from the source. However, a less-than-ideal optic tends to act as a new source point, reducing the coherence of the beam. 8-ID was designed from the outset to minimize the number of optics in the beam path. Nevertheless the present endstation beamline uses two berylium windows, a single bounce mirror and a double-bounce germanium monochromator. We intend to systematically re-evaluate the effects of each of our optics on the beam coherence, and where appropriate replace less-than-perfect optics, such as replacing Be windows with thin SiN membranes, or, better yet, differential pumps, re-working our monochromator, and carefully evaluating and reducing possible thermal distortions on the primary mirror.
Beyond upgrading the existing optics, substantial gains can be achieved by introducing vertical focusing. This is because the vertical coherence length at the sample is 100 microns or more, which is too large to properly exploit, because the corresponding speckle size is too small to resolve with current CCD pixel sizes. Instead, we typically use a 20 micron vertical beam dimension. Thus, 5-to-1 vertical focussing at 8-ID-I will increase the usuable coherent flux by a factor of 5. It turns out that this implies we will be able to study 25 times faster processes, for samples with the same scattering strength as now, or 5 times more weakly scattering samples at the same time scales.
The flux gain from these strategies does not come without a potential cost, namely that the x-ray irradiation of the sample is correspondingly increased, with an increased possibility for x-ray damage. This does not present a problem in some cases. But for polymeric and other soft-matter systems it seems certain that the implementation of strategies to ameliorate the effect of x-ray sample damage will be necessary. This issue is relevant to any study of soft matter at the third generation, and we will address it under this PUP.
A major improvement in our XPCS capability has followed from our recent implementation of a new, fast CCD-based x-ray detector, the SMD1M60. In the future, we will further develop this system to make our technology and methodology generally available to 8-ID-I users and others. Currently, we acquire 1000 CCD images into memory. Then, we interrupt data acquisition in order to compress and save the compressed data to disk. For a 17~ms-exposure time, it takes as long or a little longer to compress the images as to acquire them. This is unnecessarily inefficient, since one half of the time the x-ray beam is not being used. Thus, on our immediate agenda is the replacement of the original 2000-vintage Dell Poweredge 6400 server. In 2000, this was leading-edge consumer technology, but nowadays, considerably more capable computers are available at considerably less cost. Therefore, we will replace the old Dell with a new machine, and migrate our data acquisition to this new computer. Then, we then plan to modify our software for on-the-fly data compression. Since the compression factor for XPCS data at 8-ID-I is typically about 50, we will then be able to acquire 50,000 images before needing to save the data to disk.
Eventually, our goal is to be able to display intensity-intensity correlation functions withing minutes after data acquisition. This will permit the data to be assessed during the experiment, and thus allow the experiment to be modified as appropriate. Being able to see the correlation functions in real-time, or at least nearly in real-time, while the data is coming in, will contribute greatly to increasing the accessibility of XPCS to a wider community of users.
The calculation of one million intensity correlation functions, one for each pixel, in real-time, when a new image to be correlated appears every 17 milliseconds, is a formidable computational task. However, a back-of-the-envelope estimate indicates that it will be possible with a small cluster of them, since the correlation function calculation is highly parallel. Towards this goal, therefore, we plan to initiate development of a parallel computing capability and the software to display intensity correlation functions in near-real-time. The benefits to the APS of this PUP are clear: there will exist an XPCS capability that is unique in the world and that will attract both established synchrotron users -- from the SAXS and SANS communities -- as well as new users -- from the light scattering community. Beyond XPCS, major benefits to a wide range of APS users will follow from our detector-system development efforts. In our opinion, for small-angle applications even outside XPCS, our implementation of the SMD1M60 is the most capable time-resolving, two-dimensional x-ray detector currently in existence. Coupled to a powerful (parallel) data analysis system, it will be even more capable.
