Calibration was performed by measuring two reference mirrors: one is an Edmund Scientific aluminum coated mirror (87.4%), the second one is an FSS-99 silver coated plate glass provided by Denton Vacuum (98.6%). Absolute reflectivity of the references was measured in June 1999 by Optical Data Associates and is traceable to an NIST standard. Both references were measured prior to taking mirror reflectivity measurements. Data at five different locations on the reference mirrors were taken and averaged together. This determined the calibration formula, as the μScan™ reflectometer seems to exhibit some non-linearity in its response. A simple multiplicative factor was calculated for both references and linearly interpolated in between. The estimated accuracy of the calibration is ~0.5%. Details of the calibration are given on the various spreadsheets. At least ten measurements were taken at different locations on the mirrors under evaluation. Anomalous low data points are sometimes excluded from the statistical sample to avoid biasing the results.

The reference samples were not available prior to the June 18, 1999 cleaning. Nevertheless an aluminum coated mirror and a large FSS-99 coated plate glass were available. Since the reference samples were available after cleaning was performed, the reflectivity of the aluminum coated mirror and the large FSS-99 plate were checked against the ODA’s calibrated references. Details of the first cleaning calibration is given on spreadsheet 3.


3. Results

3a. Pristine Segments

On May 25, 1999 segments 6 and 10 were replaced by mirrors SN 085 and SN 094. These mirrors were part of the last mirror shipment to the HET. They remained in the crate stored in the HET loading bay until installation. Right after the mirrors were positioned on the support frame, reflectivity and scattering measurements were taken. Results are shown on sheets 1 and 2 for reflectivity and scattering respectively. The left side of sheet 1 gives reference measurements and the calibration used, while the right side shows mirror reflectivity. The first column gives raw measurements, while the second column shows calculated reflectivity. Statistics are given at the far right. Both mirrors have high reflectivity giving us a pristine mirror coating baseline against which to evaluate later degradation in reflectivity performance.

For scattering, no calibration is applied. The reference mirrors were measured anyway to provide some baseline against which to evaluate the performance of the brand new “out of the box” mirrors. The reference mirrors have very good coating. The three columns give surface roughness in angstroms, forward and backward scattering in percent respectively. Surface roughness is calculated by the μScan™ portable scatterometer software from the signal sensed by all three detectors assuming a random distribution of surface irregularities. As can be seen, both SN 085 and SN 094 give very good performance with surface roughness a factor of two to three lowers than the reference mirrors. The increased surface roughness of the reference mirrors can be attributed to repeated positioning of the reflectometer detector head onto the reference mirrors.

3b. Cleaning Improvements

Spreadsheet 3 gives calibration details for the June 18 mirror cleaning evaluation. Sheet 4 gives the results of reflectivity measurements performed on six mirrors prior (left column) and after (right column) CO2 snow cleaning. Statistics and improvement calculations are given in the shaded areas to the right of each tables. The history of the various segments is as follows:

- Segment 1 Installed 27 October 1998. Has been cleaned “weekly” with isopropanol.

- Segment 6 This is SN 094 installed 3 weeks prior to CO2 snow cleaning.

- Segment 10 This is SN 085 installed 3 weeks prior to CO2 snow cleaning.

- Segment 14 Installed on 22 September 1997. Seven weeks without cleaning.

- Segment 89 Installed 30 October 1998. Seven weeks since previous cleaning.

- Segment 90 Installed spring 1997. Seven weeks since previous cleaning.

A further note on segment 6: since installation, water has dripped from the dome onto a substantial portion of its surface. Measurements of this segment have been restricted to the cleaned portion of its surface. The selection is such that two newly installed, two “middle aged”, and two of the oldest segments are measured. This gives us an adequate sampling to evaluate the effect of CO₂ snow cleaning on the reflectivity of coatings with different level of degradation. All segments selected are situated in the lower portion of the mirror array, such that dust “retention” rate differentials, from variation in elevation angle of the segments, is likely to be minimized.

Results clearly show that dust accumulation has a significant effect on mirror reflectivity performance. SN 085 and SN 094 have lost from 1.8% to 2.6% in only three weeks of exposure. CO₂ cleaning significantly improves mirror reflectivity of all six segments. The three segment “families” show different levels of improvement.

For the new segments (6 and 10), CO₂ snow cleaning restored their reflectivity halfway back to their previous pristine level. More data is needed to determine whether the reflectivity degradation rate is halved over the long run. If dust accumulation is a significant contributing factor in tarnishing the silver coating, we might expect reflectivity degradation rates to decrease by more than a factor of two. “Middle aged” segments show the highest level of improvement from CO2 snow cleaning. Segment 89 had been accumulating dust for seven weeks and recovered a “whopping” 4.1% in reflectivity. Even segment 1, which had been cleaned with isopropanol only ten days earlier, recovered 1.3% which is as much as most other segments. The old segments show a level of improvement intermediate between the new and “middle aged” segments.

These results can be understood as follows. With degradation, the surface becomes rugged and the FSS-99 coating develops porosity. That the surface becomes rugged is evident when wiping the mirrors with isopropanol. On new segments, the TexWipe glides on the surface; for older segments, friction is sensed. Coating porosity became clear when removing the FSS-99 silver coating from the corner of segments to be repaired. Old segment coating came out easily. The HCl acid had to be left on the surface longer to properly strip the coating of newer segments. For new segments, the surface smoothness probably leads to smaller dust retention rates and less chemisorbtion (i.e. dust slides more easily across the surface and does not stick to it as readily). As the coating ages, the dust retention rate increases significantly due to increased surface roughness, but the surface is still smooth enough that CO₂ snow crystals are still removing dust efficiently. For older segments, surface roughness becomes such that, not only is dust retention rate higher than for a pristine surface, but CO₂ snow cleaning becomes less efficient as it has a harder time dislodging dust particles from the surface. Data covering a six to nine months period are needed to determined whether the above scenario holds true.


4. Conclusions

There is no doubt that regular monthly CO₂ cleaning of the primary mirror array is highly desirable. Not only does it improve mirror reflectivity by a significant amount, but it might also reduce coating tarnishing rate. The later can only be assessed with a regular and systematic program of monthly CO₂ snow cleaning and reflectivity monitoring lasting at least six to nine months. With only one cleaning performed so far as part of this monitoring program all conclusions reached in this document are necessarily preliminary.


5. Acknowledgments