A few changes and additions to the original SOW were made after the contract was awarded. The most important of these are: (1) the resolution of the 2 stage was increased from 1 arc minute to 5 arc seconds, and (2) an in-vacuum motorized motion, Z2, was added. Figure 1 shows all the motions for the DCM in the final design. Note that because the two crystals are not mechanically linked (as in the "boomerang" design), the DCM can be operated in a number of ways: (a) a "channel-cut" mode in which the two crystals stay fixed relative to one another during an energy scan, (b) a fixed-exit mode in which only Y1 moves during an energy scan, and (c) a fixed-exit mode in which both Y1 and Z2 move during an energy scan. Mode (a) is not a fixed-exit mode, and thus the beam moves vertically during a change in energy. Mode (b) is a fixed-exit mode, but the beam walks across the face of the second crystal. Thus, long crystals may be needed. Mode (c) is a fixed-exit mode in which the beam stays centered on the faces of both crystals.
Table 1 shows the as-built range and resolution of the translations/rotations of the DCM. Since delivery, a lot of effort has been put into interfacing and integrating the EPICS-VME software and hardware required to drive the DCM (thanks to Tim Mooney and Dave Reid). Currently, the DCM (with all three possible operation modes) and all the monochromator accessories, such as the liquid gallium pump, vacuum gauges, encoders, and thermocouples, are completely interfaced to a Sun workstation. Tests have mainly focused on the functionality of the piezoelectric (PZT) driven motions, the straightness of the Y1 and Z2 translations, and the verification of the various motion resolutions. All these tests were performed with the DCM open (at atmosphere). Technical difficulties associated with making autocollimator measurements in vacuum prevented in-vacuum measurements. The measurements were also limited by the overall stability of the autocollimator. The autocollimator was usually only stable to about 0.1-0.5 arc seconds, depending on the exact setup, and it has a slow overall drift, which varies from 1 to 5 arc seconds per half hour.
The PZT-driven stages on the first and second crystals are critical to the proper functioning of the DCM. The main function of the PZT adjustments is to correct for the inevitable misalignments and thermal drifts between the two crystals. In this design, the PZTs have to push a considerable amount of mass in both the first and second crystal -adjust stages. The maximum extension of the PZTs is about 75 microns. The lever arm from each PZT pushing point to the center of the rotation is 100 mm. Thus, the maximum expected rotation due to the PZT motion is about 155 arc seconds. A major concern is whether or not the PZT motions are smooth and reproducible. When the DCM was first delivered in January 1993, each PZT-adjust rotation stage was supported by six rollers and wheels. Figure 2 shows the 2 fine-adjust rotation due to the PZT extension as measured by an autocollimator. From the large hysteresis and the measured range of motion (about half of what is expected), it is clear that the PZT stage was "sticking." Kohzu personnel came to ANL and changed the PZT stages. The new PZT stages have conventional ball bearings, and in Figure 3 we show the measured rotations (back and forth) of the new first and second crystal PZT-driven stages. These figures show that the range of the PZT motions is close to the expected values and that the amount of hysteresis in the system is considerably reduced. Note that the measurements were made with steps covering several arc seconds. In these measurements, the resolution was limited by the autocollimator stability. The motion of the PZT in the sub-arc-second range can be seen in Figure 4, which is a (+,-) rocking curve of a Si(111) x Si(111) crystal taken with Mo K radiation. The measured width of the rocking curve is broader than the theoretical width, probably due to strains in the crystals we used. The step sizes taken varied from 0.35 to 5 arc seconds.
In order to keep the fixed-exit beam height fixed during an energy scan, the first crystal must be translated (Y1). If one is interested in keeping the beam centered on the second crystal, the second crystal must also move (Z2). When the monochromator is operated in this manner, the yaw of the Y1 and Z2 translation stages is critical if one expects to stay on the Bragg peak during an energy scan. (Here, yaw is defined as the rotation that changes the Bragg angle of the crystals.) If the yaw of these stages is large (compared to the Darwin widths of the reflection), the reflection will be lost during an energy scan.
Figure 5 shows the measured yaw of the Y1 stage with a dummy 10-kg load at = 30 degrees. The measured yaw of the Y1 stage is about 1-2 arc seconds for 10 mm of travel. Note that the actual motion required for a fixed offset is less than 3 mm between 4 and 20 keV for a Si(111) crystal. Figure 6 shows the measured yaw of the Z2 translation stage with a dummy 5-kg load at = 0 degrees. In this case, the measured yaw is about 3 arc seconds for 120 mm of travel. The data suggest that it should be possible to scan in energy while staying within the Bragg peak. Of course, a small PZT adjustment (manual or in a feedback loop) may be necessary to stay on the peak exactly.
Another aspect of the DCM that is of interest to some users is the angular error of the main -rotation stage. This error is due to fabrication imperfections in the gear-drive mechanism. In Figure 7, we show the measured angular error of the Kohzu DCM for moving from 0 degrees and 30 degrees. The total error is about 10 arc seconds.
Finally, because the monochromator has a large number of in-vacuum components, the time it takes for the chamber to pump down is of concern. Shortly after the DCM was delivered to the APS, a vacuum pump-down test was performed. With a 400-liter/sec turbomolecular pump and a 400-liter/sec ion pump, the system took about 24 hours to go from atmosphere to Torr and another 24 hours to reach Torr. After that test, the DCM was opened for other tests and modifications (including the Kohzu retrofit of the PZT stages). Thereafter, it remained open for about 10 months. The next vacuum pump-down took about 48 hours to go from atmosphere to Torr. All vacuum tests were done at room temperature without any baking, which is not recommended due to the high vapor pressures of the in-vacuum lubricants at elevated temperatures (~100 degrees C).
One possible option for dealing with the high power and power densities of the APS undulator A on the 1-ID line is to use inclined crystals with the liquid-gallium pump. The first commercial liquid-gallium pump, made by Qmax Corporation, has been checked out and is ready for use on the experimental floor. Currently, we plan to use a 78-degree inclined Si(111) crystal for the 8-20 keV range and an 85-degree inclined Si(111) crystal for the 4-9 keV range. The inclination angles and energy ranges are chosen such that the surface power density on the crystal does not exceed and the crystals are less than 250 mm long. Experiments at CHESS and NSLS and computer simulations have shown that thermal distortions are minimal for surface power densities below about . The lengths of the second crystals are chosen so that no Z2 translations are necessary for the energy range. Thus, the beam will walk across the face of the second crystal. Prototypes of both crystals have been made and are being tested in-house. However, due to the increased sensitivity to strains in the inclined crystal, the prototypes, which were epoxied together in-house, will most likely not be optimum for flux or brilliance. We are currently investigating possible bonding options to minimize the bonding-induced strains. Possibilities include frit-glass bonding, gold-based solder, and direct silicon-silicon bonding. The cooling geometry will most likely utilize either core-drilled holes or slotted cooling channels.
In summary, the limited tests we have performed on the Kohzu DCM show that it satisfies the SOW specifications. However, we note that the true test of the DCM can only be conducted on the experimental floor. Current plans are to install the DCM on the beamline in 1995.
-Wah Keat Lee
The preceding article was reprinted with permission from the January 1995 issue of SRI CAT Newsletter.