| monochromatic | polychromatic | variable monochromatic |
Thanks to the unique advantages of the synchrotron radiation (very high intensities, variable wavelengths, and small focal spots) major improvements in data quality and accessible pressures, compared to laboratory SXD experiments, are possible even for the standard mSXD approach. Unfortunately, at present no beamlines in the United States is optimized for mSXD experiments on micron samples or at extremely high pressures. Furthermore, standard mSXD does not address two requirements defined for optimal ultrahigh-pressure SXD. These are that mSXD requires extensive sample rotation during data collection, and full intensity data collection by is a relatively slow process, on the scale of hours. mSXD methods to 10-to-20 GPa are, however, mature and well tested on laboratory diffractometers and have been successfully used at synchrotrons outside the US. They have benefited from recent advances in CCD detectors that increase resolution and sensitivity and decrease data collection times. Although CCD’s are optimized for rapidly collecting intensity data, they are less well-suited for obtaining high-resolution peak profiles and high-precision cell parameters than traditional point detectors. The background in CCD data also is higher because the diffracted beam cannot be collimated. High-pressure experiments with DAC’s are possible using standard three-circle or platform-type goniometers, and commercial software can be used to process these data when combined with relatively simple custom high-pressure software tools. Although X-rays intensities at synchrotrons would allow extending the pressure range beyond 30 GPa by decreasing sample sizes, the need to rotate the sample during data collection introduces mechanical challenges when the beam is nearly the same size or smaller than the sample. While the hardware (three-circle goniometers) needed for mSXD is still unavailable at micro-focus high-pressure beamlines in the US, this technique will be easy to implement and will rapidly offer a first improvement of the accessible pressures and quality of SXD data. However, the intrinsic limitations of the method (necessary rotation high background, and limited angular access) may never allow mSXD to work with the samples smaller than 10-to-20 micrometers, such as those necessary for experiments at >100 GPa. During the first year, we plan to begin to implement and to optimize mSXD technique for studying crystals down to 20 microns in size (possibly to 50 GPa) at GSECARS 13-BM-C and ALS/12.2.2. This is our first approach to diffraction on the microscale. Use of CCD and imaging plate detectors generally reduces data collection time for mSXD, especially for crystals with large unit cells, for which tens of thousands of reflections must be measured. The most common strategy used with area detectors, “blind” data collection without previously determining the crystal orientation, normally saves considerable time compared to point-detector experiments in which peak-searches must first be performed. This approach, however, loses its time-saving advantage for mineral crystals with small unit cells. In these cases, it is the most efficient to use a combined approach which uses an area detector to obtain limited data, with which the unit cell can rapidly be determined, and then use a point detector to measure peak intensities. We also plan to implement this second approach at GSECARS 13-BM-C. While sample rotation cannot be avoided with fixed-wavelength incident radiation, very intense X-ray sources, fast detectors and very good rotation stages may make possible time-resolved experiments which perform full data collection in one oscillation exposure. This approach will, however, require developing new and more efficient algorithms for determining unit cell parameters, since only lengths of the reciprocal vectors will be determined and their directions will be constrained to arcs with widths equal to the range of the rotation. A major benefit of such mSXD experiment is ability to avoid multiple readout times, the major part of the data collection process at synchrotrons. To explore possibilities for time-resolved mSXD experiments, we will test three new methods for ultrafast detection of mSXD signals at the participating beamlines as the third approach to diffraction with micro-samples: (i) We will participate in the tests of a novel, ultrafast GE medical detector that APS will be leasing during 2005. This detector can be read out at 30 frames/sec and may revolutionize the speed of collecting mSXD data. We will use this detector for high-pressure mSXD experiments. (ii) An interesting alternative to CCD-based detectors is CMOS (Complementary Metal-Oxide Semiconductor) technology (Rossi et al., 1999; Strueder et al., 1998; Yagi et al., 2004). Although this approach sacrifices dynamic range (10000:1 for CMOS vs. 60000: for CCD) and has higher dark current levels, CMOS detectors offer much faster readout times (standard 3 frames/second), are large (standard 100 cm2 without a taper), offer small pixel sizes (~50 m), and are an order of magnitude less expensive than CCD’s. We will purchase and test a CMOS detector built by Rad-icon Imaging Corp. and implement its controls in the EDX4DAC software. This detector will be used mostly for the scanning monochromaor pSXD approach and should reduce scan times from ~4 h to 10 min. We also will test next generation of hybrid detectors utilizing new readout architectures, originally developed by Mark Wadsworth at JPL, now at Tangent Technologies, and Gene Atlas at Imager Laboratories, which combine CMOS analog and digital electronics to provide multiple (up to 512) channels of readout, all operating in parallel, with a high-quality CCD array divided into up to 512 individual CCD sub-arrays using larger (up to 50 micron) pixels (Janesick and Putnam, 2003). With this Hybrid Imaging Technology, or HIT, approach, an array of approximately 8 cm x 8 cm can be read out in 2 msec. Another alternative to standard CCD technology utilizes a new generation of the charge injection devices (CID’s) produced by CID Technologies. These CID arrays allow random pixel addressing with the ability to read a pixel either destructively or non-destructively (Bilhorn et al., 1987; Hanley et al., 1995, Sweedler et al., 1988; Sweedler et al., 1994). They are fabricated using processes that make them significantly more resistant to radiation degradation than conventional CCD’s. These unique addressing and readout modes allow rapid jumping from one pixel to another or reading of small arrays and jumping to a second (third, fourth…) array. Data from two pixels could be acquired in as few as 40 nanoseconds (5 MHz pixel rate). As with the HIT devices, they can be easily fabricated with multiple output ports, greatly speeding the total frame readout. Control of all of these types of detectors will be implemented into the EDX4DAC program. With msec time resolution, time-resolved studies at high-pressure will become accessible. These new detection methods should be especially suitable for future application at Free Electron Laser facilities, where highly coherent and extremely intense monochromatic beams are going to be available.
mSXD with area detector
mSXD with point detector
EDX-SXD
The conventional mSXD approach to structure determination in DAC’s has been optimal for laboratory experiments with conventional X-ray sources at single characteristic wavelength. However, many DAC-experiment specific experimental difficulties could be solved by applying a polychromatic SXD (pSXD) approach. Laue technique, in which a white-beam diffraction pattern from a still crystal is captured with area detector, has been used quite extensively in the protein crystallography field (Ren et al., 1999). Laue diffraction has been demonstrated to be successful also for obtaining structural information about miner and small-molecule crystals (Euler et al., 1994; Kariuki and Harding, 1995; Ravelli et al., 1999). The very short, subsecond exposure time in Laue imaging permits utilization of this technique for studying dynamic processes, such as photochemical reactions (Ren and Moffat, 1994). Unfortunately, the limitations of this technique with respect to unknown structures have prevented it from becoming as widespread as monochromatic diffraction. On the other hand, energy-dispersive methods (EDX) with solid state detectors have been applied to high-pressure pSXD experiments, (Mao et al., 1994; Mao and Hemley, 1996; Shu et al., 1998), which offer very convenient way of precisely determining lattice parameters, but have never been used for structure solution or refinement. Since there has never been a coordinated, focused effort to optimize the methodology and to develop software for this kind of experiment, especially one that would permit full structure determination, EDX also remains restricted to a small group of experts. In general, pSXD techniques all require new instrument control and data processing software, but offer in return great flexibility, the advantages of very rapid data collection, and avoiding sample rotation. While the mSXD technique has been developed and tested in laboratories, the need for high-intensity polychromatic X-ray beams constrains use and development of polychromatic approaches to synchrotrons. Three main problems limit the application of white-beam Laue imaging technique to structure solution: (i) the unknown energies of the diffracted beams, making impossible precise unit cell determination, (ii) harmonic overlaps make it hard to deconvolute the intensities of constituent peaks, (iii) energy-dependent corrections to the measured intensities are complicated. While protein Laue crystallographers have developed satisfactory computational solutions to problems with harmonic overlaps (Hao et al., 1995, Ren and Moffat, 1995) and intensity corrections (Ren and Moffat, 1995b), the issue of determination of unknown unit cell parameters is best solved experimentally. This is a critical problem for high-pressure experiments, since knowledge about the pressure-dependence of unit cell parameters is one of the fundamental quantities determined during the experiment, and non-quenchable high-pressure phases have unknown unit cells. There are three experimental approaches to determining peak energies in Laue technique: Combined Laue&EDX. One solution to the peak energy problem combines Laue imaging with energy-dispersive point detection. An area detector is used first to capture a Laue pattern, possible with a sub-second exposure, and reciprocal vector directions are calculated from the peak positions on the area detector. In the second stage of the experiment, the crystal is reoriented to point each observed diffraction peak towards the energy-dispersive detector; and the energy spectrum of each diffracted beam is measured. An alternative to avoid rotating the sample involves moving the energy dispersive detector to each diffracted beam (not possible in mSXD). If the position and orientation of the area detector are well calibrated, it is unnecessary to center the peaks with the solid-state detector, and the duration of this stage of experiment can be reduced to below 1 minute (~5 seconds per peak). Since information about both directions and lengths of reciprocal vectors is then available, the Laue image can be fully and easily interpreted. Alternatively, full data collection can be done with a very well collimated solid state detector; this helps in deconvoluting harmonic overlaps. During the last year, Dera and Downs started developing instrument control and data processing software for the EDX detector, and performed pilot experiment for such an approach at APS (HPCAT) beamline 16BM-B (Dera et al., 2005a). This work will continue during the first year of the proposed project. The combined Laue&EDX approach offers very simple and reliable way to determine peak energies. However, the relatively long data collection times precludes its application to studying ultrafast processes.
OLA-SXD
An alternative solution to approximately determining energies of peaks draws on Hanley and Denton’s concepts (Hanley et al., 1997; Hanley et al., 1996) for a “foil mask X-ray spectrometer.” The energies can be determined from series of Laue images taken in separate subsecond exposures through foil masks with different absorption profiles and solving simultaneous equations that yield the energy content of each spot. A more sophisticated version of this approach, that has not been tested, involves stacking a series of CMOS with filter media between them. This is conceptually analogous with the technique of stacking numerous photographic emulsion sheets on top each other, exposing them, and then extracting energy data by determining how many sheets the X-ray photons penetrate (blacken). Photon energies are then determined by deconvoluting the filter transmissivities in a manner similar to using the foil mask spectrometer. This approach should be able to resolve energies with 30-to-50 eV precision. Although the foil-mask spectrometer approach does not match the energy-resolutions of the other two pSXD approaches, it combines the convenience of easy peak indexing and unit cell parameter determination with very short readout times, offering possibilities for time-resolved studies.
This approach merges mSXD and pSXD techniques in one experiment to extract energies of peaks from a static sample. The first stage of the experiment involves collecting a Laue image and determining the positions of the observed peaks. The flexibility and accuracy of modern monochromators at synchrotrons can then be exploited so that, instead of moving the sample to bring many diffraction spots to the diffraction condition, one tunes the monochromator in small steps and collects series of monochromatic diffraction images for each small range of energy. For each peak in the Laue pattern, an energy-profile will appear in several consecutive monochromatic images of the energy-scan. This approach both allows keeping the sample static, an advantage for, e.g. working with a laser-heated DAC, and also enables one to apply a large set of well developed techniques from mSXD. Absorption can be separately applied for each small energy step; harmonic overlap does not occur because the beam is monochromatic. An energy range between 10 keV and 35 keV covers a region of reciprocal space that exceeds the range covered by classical mSXD. This scanning monochromator approach is quite new and has been applied only in microdiffraction studies (Tamura et al., 2003) of strain/stress analysis rather than structure determination. Dr. Kunz performed the first tests at the Swiss-Norwegian beamline on a sample in air and obtained sufficient data to index the peaks. In further experiments involving samples in DAC at pressure as high as 50 GPa, co-PI Dera H-K. Mao, N.Tamura, W. Liu and W. Yang successfully indexed peaks and proved that the method can easily be applied to high-pressure experiments. Further analysis of the experimental data, and especially extraction of the structure factor amplitude information, requires custom software development, which we planning to begin toward the end of the first year of this project.
