1. A U.S. Department of Energy Laboratory managed by The University of Chicago Is your cold-stream working for you or against you? An in-depth look at temperature and sample motion Abstract Introduction R. W. Alkire, N. E. C. Duke and F. J. Rotella Structural Biology Center, Biosciences Division, Argonne National Laboratory, 9700 S. Cass Ave., Argonne, IL 60439 To measure the force of the cold-stream we employ a silicon single crystal cube (0.2 x 0.2 x 0.2 mm 3 ) mounted to a sample loop. Using a constant rotation speed of 1°/sec, x-rays diffracted from the Si(220) reflection are detected by a photodiode and the output recorded by an oscilloscope. We use the oscilloscope timing to correlate the reflection angle. From the data we compute a center-of-mass (centroid) for each reflection. Deviations in the centroid positions across the span of the cold-stream give a relative measure of the force exerted by the cold-stream. To ensure an accurate statistical measure, reflections are measured 5-6 times at every setting. When combined with the statistical variation in each measurement, a map of turbulence within the cold-stream emerges. To measure an instrumental baseline we use a silicon rod. Below are data from six scans using the silicon rod; centroid standard deviation for the six scans is ± 0.00035°, reflection width (FWHM) is 0.0368(3)° and the intensity variation, without correcting for beam decay in top-up mode operation is 0.7%. In this work we investigate the temperature and vibration profile for an Oxford Cryosystems Cryostream 600 (flow rate 5 l min -1, shield gas 7 l min -1 at 100 K). The first step in this study is to align the cold-stream nozzle using the manufacturer’s supplied alignment tool as shown below; the cold-stream angle is 42°. Once aligned, the temperature profile is measured by moving the cold-stream in X, Y and along the nozzle direction via heavy duty translation stages. Temperature is monitored using a type K thermocouple and digital thermometer. It is important that the thermocouple not be shielded. Otherwise the heat from nearby metal will lead to inaccurate readings. Temperature profiles The thermocouple tip is first centered on the rotation axis using the sample visualization cameras. Once centered, the cold-stream is moved vertically and the temperature recorded at each step. The vertical profile is shown below. Alignment tool 110 K avg Δ 1.27mm We will define the distance between 110 K positions as the width of the gas cold-stream and the 110 K average position as its center. Using this criteria, the nozzle alignment tool positions the cold-stream 1.27 mm below the rotation axis. This is an offset of 25% (see note below) of the total cold- stream width. The horizontal temperature profile taken at two vertical positions (0.0 mm and 1.27 mm) shows that when the cold-stream is raised the horizontal cross section gets wider by 0.8 mm, confirming the raised position is a marked improvement. Alignment tool 110 K avg Wider profile The force profiles have been combined into a single plot so direct comparisons between directions can be made. The optimal cold-stream position at 0.0 mm on the above plot corresponds to 1.27 mm in the vertical (V) direction, 0.5 mm in the horizontal (H) direction and a nozzle-to-sample distance (NSD) of 7.9 mm. The center of the force profile is also the center of the 110 K average temperature profile. The vertical force profile shows a steep slope on the positive side of the curve which, because the thermocouple is stationary, corresponds to the bottom of the cold-stream. The top region of the cold-stream has a much more gradual slope, indicating less force. Horizontally, the uniform force region is narrower than the vertical with a much steeper but symmetric profile. Only small changes occur as the nozzle moves farther away from the sample once the vertical offset has been corrected. What is clear from the above plot is that the region of constant force within the cold-stream is only about 2-2.5 mm wide, or roughly 1/3 the width of the nozzle opening. This means that in order to align the cold-stream properly you must first determine the temperature profile and not rely on the nozzle alignment tool provided by the manufacturer; this tool is very useful but only as a reference to the temperature profiles. To assess the impact of the cold-stream force on crystallographic data, we perform experiments on lysozyme. If our results are correct, the processed data should follow a pattern similar to the force profiles in each direction. Force/vibration measurements An investigation is conducted into the temperature and vibration profiles from a commercial cold-stream. Our results indicate that the often used nozzle alignment tool does not point to either the temperature or vibration center of the cold-stream. This offset can impact data quality, particularly if the sample mounting loop is at all flexible. Based on these results we make suggestions on how to align a cold-stream, the importance of cold-stream placement in terms of vibration and how to look at data processing statistics to see if loop motion is affecting data quality. In order for the measurement to provide useful information, the probe should be sensitive to the forces exerted by the cold-stream. We try three different loop mounts before we get the response we are looking for. To ensure we do not get broadening of the reflection at 100 K, the silicon cube is attached to the loop using Apiezon-T grease. Below we show the responses of our loops, along with reference data from the silicon rod. Dimensions of the loops represent the overall length of the loop including loop stem, followed by just the open area of the loop. Because the silicon crystal was mounted in the center of the lithographic loop, it showed little sensitivity to the forces of the cold-stream. To enhance sensitivity, we mounted new crystals on two nylon loops but this time put the crystals on the edges of the loops. With the 1.1mm nylon loop we achieved good sensitivity to the forces of the cold-stream. Below, we show a force profile for the vertical, horizontal and the nozzle-to-sample directions. Measuring the sample temperature while changing the nozzle-to-sample distance shows the sample is colder over a longer distance when the vertical offset is corrected. To verify the results from the silicon measurements, we collected lysozyme data from several crystals at different cold-stream positions. Data were collected at 12.66keV using a single pass, with 1° frames collected at a rate of 1 sec/frame. Each point represents 90-100° of data processed to a resolution of 1.5 Å. Our measure of data quality is the linear Rmerge computed from HKL3000 (Minor et. al, 2006) output without applying any rejections. Data were collected multiple times in some locations and these data sets show up as separate points on the plots if they do not overlap. Although present, radiation decay is not large enough in any particular series to influence the results. Below is the vertical lysozyme profile. Lysozyme data collection The temperature and force center are at 1.27 mm. This is also the area where the most flexible 1.67 mm nylon loop shows the greatest stability. In agreement with the silicon results, the positive side of the curve has the highest linear Rmerge i.e., the lowest data quality. Below 1.27 mm the data also degrade but at a much smaller rate. This agrees with the silicon results which indicated a more gradual slope of the force curve in this region of the cold-stream. As we will describe shortly, adding grease to the stem of the nylon and lithographic loop stabilizes the loop, leading to much better data. Data from the horizontal direction are shown below. Processed lysozyme data collected when moving in the horizontal direction shows a steep increase in linear Rmerge as the cold-stream moves away from center. The rapid decrease in data quality is similar to that found on the most turbulent side of the vertical plot except this time the degradation is symmetric, in good agreement with the silicon results. Most people believe it is important to move the cold-stream in close to get the best quality data. Below are the nozzle-to-sample results. If the loops used to mount samples did not move, there would be little impact from the cold-stream. So, how do I tell if loop motion is present in my data? Well, it’s complicated. There is really no direct way to tell unless you make some reasonable assumptions. If you can assume that the x-ray source is stable, goniometer-shutter synchronization is accurate and the crystal does not suffer from excessive radiation damage, then as long as the space group is correct you can use the processed data to get a window into loop movement. HKL3000 provides a normalized chi squared versus intensity plot. Since loop motion usually effects the strongest reflections, evidence of loop motion shows up in this plot when the curve is non-linear. If the curve tails up off the baseline in the high intensity area, then it is reasonable to assume that some loop motion is occurring. To illustrate this effect, we use the 1.39 mm nylon loop curve for the lysozyme data collected as the cold-stream was moved in the horizontal direction (see the middle plot in this column). Normalized chi squared plots from each data set are overlaid into a single plot and shown in the next panel. Loop motion In keeping with the rather shallow force profile determined using silicon, linear Rmerges change only gradually for even the most flexible loop out to a distance of 8 mm. Beyond this the flexibility of the loop plays a key role in how fast the data degrade. Overall, force changes in this direction are very small compared to the vertical and horizontal directions. Linear Rmerge values for each data set are listed in the plot legend next to the horizontal position of each data set. The lowest Rmerge corresponds to the flattest profile. Since cold-stream forces are reasonably symmetric in the horizontal direction and all these data are from the same crystal, the increase in the height of the curve follows the increase in linear Rmerge. To prevent loop motion, always use the smallest loop diameter that will fit the sample. When possible, always add support to the loop stems. In our case, we used Apiezon-T grease. This is a silicon free, medium temperature vacuum grease with high sticking power that is gel-like at room temperature. We covered the nylon loop completely around the stem, just short of the start of the loop. The lithographic loop was coated on both sides of the wide stem but left uncoated on the short neck just below the actual loop. Note: The crosshair box has dimensions 200  m x 200  m. The linear Rmerge values for each crystal are listed in the legend along with the length and size of the loops used. The dashed curves indicate that for these three crystals some loop motion was present. Solid curves represent data taken from loops that had been reinforced with Apiezon-T before measurement. As can be seen, these curves are essentially flat and gave the lowest linear Rmerges of all the data. Conclusions The impact of the cold-stream can be severe depending upon the flexibility of the sample loop and degree of cold-stream misalignment. To effectively align the cold-stream it is necessary to measure a complete temperature profile and set the cold-stream in the middle of the width as defined by the 110 K positions. Loop motion is something that must be dealt with and one way is to try different loop materials. There are many different loop styles and different types of loop materials. Our lithographic loop was made by Molecular Dimensions and we did not test MiTeGen loops. We have collected many good data sets from nylon loops and it would be a mistake to assume that all nylon loops exhibit motion or that all lithographic loops are motion free. Our recommendation is to try different loop materials using known crystals and look at the processing statistics to see if loop motion is present. In all cases, loop motion will be less if the cold-stream is properly aligned before the experiment begins! This work has been submitted to the Journal of Applied Crystallography, 2008. Acknowledgements The authors would like to thank Zbyszek Otwinowski for many helpful discussions regarding loop motion and Stephan Ginell for his expertise with cold-streams. References Minor, W., Cymboroski, M., Z. Otwinowski & M. Chruszcz (2006). Acta Cryst. D62, 859–866. This work was supported by the U. S. Department of Energy, Office of Biological and Environmental Research and Office of Basic Energy Sciences, under Contract DE-AC02- 06CH11357. As can be seen from the lysozyme data where the cold-stream was moved vertically, the curves labelled Apiezon-T gave excellent results at 1.27 mm, where the cold-stream should be positioned. On the high turbulence side, neither loop was completely rigid but when compared directly, the lithographic loop outperformed a similar sized nylon loop. Finally, by examining the output from the first data set collected from each of five lysozyme crystals with the cold-stream in its raised vertical position, we are able to see which data sets actually suffered loop motion and which gave us the best we could expect from our crystal. Below is the normalized chi squared plot for each of these lysozyme data sets. Nylon loop Side view lithographic loop Grease around loop stem Grease above and below flat stem Nozzle-to-sample Vertical Horizontal direction parallel to beam Note: Although these measurements are from the Cryostream 600, similar temperature profiles were also observed for two Cryostream 700’s and three (Oxford Diffraction Ltd) CryojetXL’s. Vertical offsets for these systems ranged from 20-37%. V1.27, H0.5, NSD7.9