HETG Vacuum Storage Gratings

HETG Vacuum Storage Gratings, Version 2.0, November 20, 1998

http://space.mit.edu/HETG/vsg/vsg.html go to HETG Calibration

Busy-person Summary

The grating properties fall into Period properties and Efficiency properties. These have now been monitored over a two year baseline.

All grating periods and period variations appear stable (periods within 20 ppm and dp/p within 20%); the only unexplained exception is grating HA2003 which varied by -110 ppm but has returned to its original period.

The HEG efficiencies appear stable or reasons have been found for the several anomalies seen.

The MEG efficiencies continue to show a drift with time. At 4 keV at least the three of six gratings show a trend towards lower efficiency with time at a rate of roughly -3% per year. At 1 keV, the MEGs show an increase in efficiency with time at a rate of about +4% per year.

Investigating the "MEG drift with time" has lead to clear evidence that the X-GEF system detectors are being very slowly coated with a contamination layer, changing their detection efficiency properties. The detector "contamination rate" is ~ 1 micron of polyimide-equivalent per year for the SSD, ~ 0.4 microns/year for the PSPC and negligible accumulation on the SMD. Analysis and re-analysis of data-in-hand is on-going to assess the effects on the VSG X-GEF measurements. Future testing will monitor and decontaminate the SSD and PSPC windows.

Finally, while the details and effects of the detector changes are under study, we consider the possibility that the observed MEG efficiency drift is a real sample change. Assuming as a worst case that all of the drift seen is due to real sample changes, the MEG 4 keV efficiency will remain above the Level 1 requirement for over 10 years and the 1 keV efficiency is actually improving with time!


Appendix A. X-GEF Contamination and Effects

Version 1.0, June 1998 is available at http://space.mit.edu/HETG/vsg9806/vsg.html

1. Introduction

Selected gratings were put into vacuum storage and their properties are being periodically measured. This vacuum storage grating (VSG) program will allow us to detect or set upper limits on changes in the grating properties with time. The properties being measured for each sample are the period and period variation ( "LR Measurements") and X-ray diffraction efficiency ( "X-GEF Measurements"). Familiarity with these quantities and test setups is assumed in the following.

The figure below shows a time line and events relevant to the HETG flight and VSG gratings. The flight gratings were X-GEF tested in the laboratory as they were produced between mid-1995 and September 1996 (date codes 950615 to 960914). Due to LR system upgrades and the relative ease of the test, all but a handful of the flight gratings were LR (re-)tested in the April to September 1996 time frame. These detailed, facet-by-facet laboratory data sets provide the basis of the instrument model and calibration.

The HETG was next tested as a full system at XRCF from December 1996 through April 1997. These system-level measurements represent an independent set of data to complement the laboratory measurements. The result of analyzing and synthesizing these two data sets will yield the final pre-flight calibration of the HETG and HETGS (HRMA-HETG-ACIS-S).

As the time line indicates, flight data will not be taken until more than two years after the XRCF calibration measurements were made! Thus, it is important to estimate the degree to which the HETG properties might change with time - especially during the period from XRCF to flight first-light. In order to accomplish this, a total of five vacuum storage retest cycles are planned to cover the XRCF to flight first-light time period, as indicated. We plan to continue the tests at six (TBR) month intervals throughout flight operations.

Figure vsg.1: Time line of Flight and Vacuum Storage Gratings Activities

2. Summary Results to Date

Currently four retests have been carried out, indicated by the vertical lines in the diagram above with date codes 9611, 9707, 9801, and 9810. Ideally plots of the measured facet properties with time ("trend plots") would show unchanging horizontal lines; variations are, however, visible. The following sections summarize the data and list significant variations that are seen. In most cases possible explanations for the variations have been identified and traced to a "measurement problem" rather than a true sample variation. In these cases a "problem code" (e.g., LRmp1) is given along with relevant remarks; details and explanations of the measurement problem codes are given in the LR Measurement Notes and X-GEF Measurement Notes sections below.

2.1 Facet Period Properties
These two plots and table summarize the LR period data on the facets, both the average period of the facet as well as the level of period variations within the facet.

Anomalies seen in these plots are listed and described here:

Period Plot Anomalies
  • MA1019 at 9707 test: LRmp1 error due to test setup and MA1019 frame twists.
  • HA2003 shows period drifting - reason is not clear. Unlikely a humidity or temperture effect (9611, 9707, 9801, 9810 have humidities: 31%, 38%, 1%, 24%; temp.s: 21.7C, 22.2C, 24.4C, 23.3C). The dp/p map looks very similar from 9611 to 9810, except for variability in the corners, 9810 has well-behaved corners. Laliberte reports no visible mechanical issues with the grating.
  • HF2503 at 9707 test: LRmp1 error likely: the LR Camera Plots show the beams shifted 1.5 mm in X and reflected beam in Y.

Period Variation (dp/p) Plot Anomalies
  • MA1019 at all tests: LRmp2, LRmp6 - the tests include more or less of the "-X bad edge region" whose spots often fall off the camera. The dp/p in active region is dominated by these false values.
  • HC2251 at 9611 test: LRmp2 - the facet has a very bad region in the -X, +Y quadrant. For the 9611 test this region did not fall into the LR active region, hence the 9611 dp/p is lower than the others.

2.2 Ensemble Period Changes

The individual facet LR data can be combined to produce average period and the dp/p values for MEG and HEG grating sets as is done for the flight HETG, using the "active region" of the facets to approximate the flight-illuminated region on the facets. The four gratings with non-flight-like periods (HD2334, HA2045, MC1212, and ME1408) have not been included.

         Vacuum Storage LR Results: MEG Ensemble

       Test Date :      9611xx     9707xx     9801xx     9810xx 
           P_ave :     4000.36   [-68ppm]   [-11ppm]   [-30ppm]
           dp/p  :      192.07     [-22%]     [-18%]     [+19%]

         Vacuum Storage LR Results: HEG Ensemble

       Test Date :      9611xx     9707xx     9801xx     9810xx 
           P_ave :     2001.95   [-70ppm]   [-37ppm]    [-8ppm]
           dp/p  :      229.48     [-14%]      [+5%]      [-2%]


2.3 Facet Diffraction Efficiency Properties
The plot and table below summarize the diffraction efficiency data on the facets.

HEG First-order Efficiency Plot Anomalies
  • HA2045 at 9707 test and 1 keV trend: Shows deviant efficiencies at 4 and 8 keV, due to X-GEFmp1 effects and a bad region identified on LR plot. HA2045 also shows a 1 keV trend to higher efficiencies but all of the change is due to the bad "p4" region; considering only the "m4" and "p0" regions the 1 keV efficiency changed by only +1.7% at 1 keV from 9511 to 9810.
  • HC2251 tests: X-GEFmp1 : large changes in efficiencies in the the "p4" region of the facet dominate, other regions are stable. A "silvery" region is seen on the grating in the p4 area. Recalibration of the original (reference) and the 9707 data sets may be indicated, X-GEFmp2.
  • HF2503 at 1 keV: Shows trend to higher efficiencies through 9801 test, 9810 test shows it not continuing and reversing to within 2% of its 9611 value; it is possible that the 9605 test needs reanalysis. The facet does not appear to have any serious blemishes or erratic regions, regions 'p0 and 'p4 have the trend more so than 'm4. The key 960502 dataset is not available so calibration can be suspect, X-GEFmp2.
  • H2102 at 1 keV: an almost 11% change from 9801 to 9810 is seen. This grating has only ~ 0.015 efficiency at Cu (0.93 keV) - the other VSGs have values of 0.025 to 0.080 - and so the Cu measurement here can be very sensitive to the calibration and fitting process, i.e. it inherently has higher measurement noise. For example recalibration can be a 3% effect and the fitting uncertainty leads to a larger than root-N statistical error for an overall ~ 4% 1-sigma error for the 1 keV value.
MEG First-order Efficiency Plot Anomalies
  • MEGs at 4 keV: At least the three gratings MD1367, ME1408, and MB1165 show a trend towards lower efficiency with time; the other three MEGs may also take part in this drift. The rate is roughly -3% per year. MD1367, the most deviant, shows a similar drift in all 5 regions of the grating -- it is not a blemish effect.
  • MEGs at 1 keV: All the MEGs seem to show an increase in 1 keV efficiency with time at a rate of about +4% per year. Need to investigate whether the initial measurements of ME1408 and MB1165 account for their separation from the pack. MB1165 shows the trend in all 5 test regions, not a blemish effect.

In tracking down the sources for these drifting efficiencies, it was discovered that the SSD and PSPC windows are very slowly coated with a thin, nonuniform layer of hydrocarbon from the X-GEF system, see X-GEFmp3 in the "X-GEF Measurement Notes" section below and Appendix A.

The common trend among samples also suggests the possibility of "reference variations" - that the X-GEF MEG reference grating's properties are changing with time, possibly due to the same contamination that is seen on the detector windows, X-GEFrv1. These possibilities are under investigation, see the "X-GEF Measurement Notes" section below and Appendix A.

Finally, while the details and effects of the above are under study, we must consider the possibility that the observed drift is a real sample change. Assuming as a worst case that all of the drift seen is due to real sample changes, the MEG 4 keV efficiency will remain above the Level 1 requirement for over 10 years and the 1 keV efficiency is actually improving with time!

3. The Vacuum Storage Gratings

3.1 General Information

General information on the 14 grating facets which make up the Vacuum Storage Gratings are given in the tables below. As Table vsg.4 indicates, the VSGs come from a subset of flight batches that together yielded 180 of the full 336 flight gratings. The Period and Diffraction properties of the flight gratings and VSGs are compared in Figures vsg.5 and vsg.6 - showing that the VSGs cover a wide range of flight grating types.

3.2 Specific Facet Notes

Specific information and notes regarding each grating especially as it effects test results are collected here, organized by the facet name.

  • Large frame twisting leads to LR beams covering full LR Camera range and hence sensitive to LRmp1 and LRmp6 errors.
  • Bad dp/p or twisting at -X edge in active region enhances sensitivity to sample positioning, LRmp2.
  • Shows period drift without clear instrument-related reason for the changes.
  • Period is large indicating membrane stress; excluded from ensemble period analysis.
  • Period is large indicating membrane stress; excluded from ensemble period analysis.
  • Period is large indicating membrane stress; excluded from ensemble period analysis.
  • dp/p map shows bad region at X~0, +Y just outside of active region. Could lead to LRmp2 effects if mispositioned.
  • Large change in HA2045 4, 8 keV efficiencies at 9707 tracked down to a strong variation of the efficiency with position on the facet. On the plus side ('p4) there is a blemish region (shows up in dp/p map) that is more transmitting to Mg K than normal - causing bizarre fits to the X-GEF data.
  • The 1 keV efficiency trend for HA2045 is also dominated by this same blemish as the region-by-region percent changes of 1 keV efficienciy indicate:
             9511  9611  9707  9801
        'p4   0 %  20 %  33 %  42 %  <-- blemish region
        'p0   0 %  2.8%  0.0%  3.6%
        'm4   0 %  4.3% -0.2%  0.5%
  • dp/p blemish in the -X, +Y quadrant produces LRmp2.
  • The X-GEF 'p4 region has large variations in X-ray efficiency (7of 9 points greater than 6%) as opposed to the 'p0 and 'm4 regions for which 17 of 18 points have less than 6% deviation from the 9512 measurement values. Need to check correlation of dp/p blemish with X-GEF bad region...
  • Period is large indicating membrane stress; excluded from ensemble period analysis.
  • Has the largest drift to higher efficiency at 1 keV.
  • Drift to lower efficiency at 4 keV.
  • Has next-to-highest drift to higher efficiency at 1 keV.
  • Drift to lower efficiency at 4 keV.
  • Shows 1 keV trend to increasing efficiency with 2/3 of the increase between first and second X-GEF tests (first test is not on line). No obvious blemishes in X-GEF region and no erratic efficiency behavior in the test regions; regions 'p0 and 'p4 each have the 1 keV trend.
  • Bar parameters (0.70 width/period, 4000 A thick) are outside of the flight HEG range.
  • Drift to lower efficiency at 4 keV.
  • Low efficiency at 0.93 keV, can be sensitive to fitting and has higher statistical error level.

4. Why Do Measurements Change?

When a sample is re-measured there are several general categories of effects that can be responsible for a change in the measured quantity - these are discussed in general as well as with respect to the LR and X-GEF setups themselves.

measurement noise (mn)
  • Definition: Measurement noise represents the random variations inherent in the measurement process due to true physical noise processes as well as low-level systematic effects (e.g., repositioning tolerance.)
  • Trend Behavior: Produces a spread in the measurement values that is uncorrelated with sample and time.
sample variation (sv)
  • Definition: Sample variation represents actual changes in the properties of the sample being measured.
  • Trend Behavior: The simplest assumption for true sample variation is a monotonic variation with time that may continue to drift or possibly converge. Similar samples may show similar behavior, however exact tracking is not generally expected.
reference variation (rv)
  • Definition: Reference variation represents an actual change in the reference components of the measuring system.
  • Trend Behavior: The resulting trend would be similar to the sample variation - however, because it occurs in the measuring system it will show up as a correlated and similar variation among all samples.
measurement problem (mp)
  • Definition: Measurement problem represents gross changes in the measurement system which effect the resulting measured value. Examples are operator setup errors, equipment malfunctions, contamination effects, etc.
  • Trend Behavior: Produces large (compared to noise), non-monotonic-in-time variations which may or may not be corelated sample-to-sample.
  • What can be done?: By studying data sets with obvious measurement problems the source of the measurement problem can often be identified. Once identified, data sets already taken can be checked for the same (perhaps at a lower level) problem and procedures can be modified to detect or prevent the problems.

5. LR Measurement Notes

A variety of measurement problems can be caught by reviewing the LR data sheets:

LR reference variation, that is a change in the period of the LR reference samples built into the LR system , is unlikely. However illumination of different regions of the samples may create apparent changes - this could be investigated further. The general stability of the LR system can be monitored with the Reference Stability data and plots - however, changes in these plots (not shown) do not immediately translate into sample period changes.

LR sample variations may show up as clear changes in the dp/p pattern or clear changes in the frame flatness (LR surface plot.)

Other ideas and actions for improved LR testing are noted here:

6. X-GEF Measurement Notes

Identified sources of X-GEF measurement changes are listed here:

7. VSG Retest Procedure Summary

The following is a summary of the tasks that need to be carried out in order to perform a VSG retest. The retest period generally refers to the period during which X-GEF measurements are made - for the 14 VSGs this is about two weeks. Note, however, that there are preliminary and follow up activities as well.

8. VSG Software

The analysis and presentation of the VSG measurement results is carried out by the IDL routine vsg_info.pro (in ~dd/idl/meta/). This routine has multiple sections starting with "if 0 EQ 1 then begin" which may be changed to "if 1 EQ 1 then begin" to produce that section's product. In this way a large variety of VSG-related analyses are assembled in one location.

This web page is: http://space.mit.edu/HETG/vsg/vsg.html.
Please send any comments and updates to: dd@space.mit.edu.

MIT Accessibility