Vacuum Lifting Equipment To Handle Polished Mirrors

W. Davison, D. Ketelsen, R. Cordova; Steward Observatory
J. Williams, W. Omann, W. Kindred; Smithsonian Institution/MMTO

Large Binocular Telescope Project

Technical Memo

UA-93-09

November 17, 1993
http://medusa.as.arizona.edu/lbtwww/tech/ua9309.htm

ABSTRACT

Unlike previous mirrors the large honeycomb mirrors cannot be lifted with simple edge rings and maintain a low glass stress. A distributed lifting scheme is needed to lift the 6.5 m and 8.4 m mirrors after polishing and to install them in the telescope. Vacuum lifting equipment has been successfully used on a very large number of optics, but there are techniques that need to be tested to assure that it will meet our needs. Our planned method of lifting would use a frame similar to the one that is used for handling the casting. We use 36 vacuum pads instead of glued on pads (figure 1). Six vacuum pumps would be connected to 6 pads each in a manner that assures maximum redundancy at the lowest stress.

INTRODUCTION

Testing of the VAC-U-LIFT system that we are proposing to use has been extremely successful. The experimental tests have demonstrated that the vacuum lift system could lift all our mirrors, at any site and any surface quality, including coated, with safety and with no surface damage. Appendix A & B list the Lifting Criteria and Experimental Results. We would strongly recommend the system and propose that we begin purchasing the parts. When we install the mirror in the polishing cell and separate the lifting frame (which gets glued onto the next 6.5m mirror) we will have no means of lifting the mirror until a vacuum lift system is purchased. The first planned use of the vacuum lift fixture is to transfer the mirror from the polishing cell to the telescope cell. If we have a polishing cell problem that requires lifting the mirror we could have lengthy delays unless the system is purchased earlier.

LIFTING EXPERIMENTS

The first phase of vacuum lift equipment tests was the mechanical tests that tested the pads and vacuum pumps for pull, sealing and some failure methods. Two sets of experiments were conducted, one at the Mirror Lab and one at Mount Hopkins. The equipment works very well and exceeds our lifting criteria.

The pads performed very well in that they provided sufficient pull with an additional advantage that they separated from the glass by more than two inches before any vacuum was lost. This means that it would take much more than our 1.5 g jerk for them to come loose from the surface or it would take a leak a lot longer to fill the pad volume in case of a power failure or leak. With leaks like experienced in the tests, the mirror would probably hold for days to weeks with no power. The vacuum pump pumped to within 2.7" Hg of absolute, within seconds of attachment, and reached 1.7" Hg in ten minutes regardless of elevation.

SURFACE DAMAGE EXPERIMENTS

Our tests show that: the pads when clean will not damage polished surfaces; contamination can be cleaned off for successful coating; the pads do not physically damage the bare coating (spray silver or aluminum) but do contaminate it; and opticote provides excellent protection of a coated surface. The pad was placed on polished glass and a vacuum applied several times. The neoprene pad material left contamination that was visibly obvious but was easily cleaned to pass the black breath test. After the glass was spray silvered, we put the pad on the opticoated surface and then the surface after the opticoat was removed from one half of the mirror. We were amazed that the spray silver was not scratched, but some small patches of silver transferred to the pad in an area where the mirror did not completely pass the scotch tape test. The contamination to the mirror could be cleaned, but that would be a last resort, since it is so easy to prevent the contamination in the first place with opticoat. The aluminized surface performed even better since there was no transfer to the pad. We applied the pad to at least six coated surfaces without any problem. The only reservation is that you need to let the opticoat dry or it can wrinkle under the pad. We tried full and partial partially dried, wrinkles, drips and completely lifted opticoat; with the only problem being a slightly higher leak rate, which was still small enough to hold for a day without power.

LIFTING PADS

The vacuum pad has a very pliable outer sealing that is a couple of inches wide and rolls into engagement with the glass (figure 2). This provides a large sealing area that would bridge small imperfections (less than 1"). The central area is an aluminum casting that is lined with rubber to give distributed contact. This is extremely important because our mirrors could get elevated stresses if the full vacuum load was setting on a circumferential "line" contact. The pad totally negates this problem. The stress induced into the mirror is almost nothing but the lifting stress. Finite element calculations of the lifting stresses will be reported in a future memo.

VACUUM UNITS

The vacuum units are VAC-U-LIFT 3200 series which have a 1/3 hp oil-less 4 cfm vacuum pump as well as a gauge, reserve tank and visual indicators. We borrowed a unit that had been in use at NOAO for many years and it still worked very well. The unit has a valve that allows the flow to be reversed and air pumped into the pads so positioning is like using an air pad, very helpful.

The leak rate, when turned off, was determined more by our temporary plumbing than the pad or the unit; but the holding time would be substantially increased if we added an additional reservoir. The unit would take linger to attach (minutes instead of seconds) if we added the extra reservoirs. These units are very robust and are almost the industry standard at the optics shops that we know.

LIFTING FRAME

The lifting frame is composed of six W18 X 35 I beams that radiate from the center. At an equal distance out (i.e. equal deflection) on each I beam, a second six legged frame is attached so that it can swivel. At the end of each leg on this second frame a 24" pad is attached with rubber mounts. The attachments are adjustable to make up for fabrication tolerances, note that the two frames are planer and that tilts are the only adjustments required to fit the shape of any mirror to the required tolerances. The principal of operation is that you attach to points of a frame that have symmetry and therefore deflect equally. The actual frame will have small deflection differences so the rubber is proportioned to take up these and maintain nearly equal loads. This is safer than a whiffle tree because if a single pad of a whiffle tree fails the mirror drops. If a pad of this elastic whiffle tree fails the others just change their load to compensate. The more that fail the higher the stress in the glass. Depending on the specific pads, 6 to 25 of the 36 could fail and still not violate our handling stress criteria of 100 psi tension.

CONCLUSION

The vacuum lift system we investigated offered exceptional performance with very low probability of failure. Damage to the glass was not experienced and opticoat offered surface protection even to aluminized surfaces.

APPENDIX A

LIFTING CRITERIA

The general requirements for the vacuum lift system to handle large mirrors are as follows:

  1. Lift a 34400 pound (15600 kg) mirror at 11000' elevation (0.66 atmospheres).
  2. Be able to hold the mirror with 1/3 of the pads failed.
  3. Be able to hold the mirror with a 1.5 g jerk.
  4. Have 1,2 & 3 act simultaneously (no other safety factor is needed).
  5. Have global stresses less than 100 psi. (0.7 MPa)
  6. Have local stresses less than 100 psi. (0.7 MPa)
  7. Be able to hold the mirror for a 4 hour power failure.
  8. Not scratch the surface of the glass.
  9. Pad vacuum force be reacted uniformly (within 20%) by scuff pad.
  10. Glass curvature is 640" to 756" radius.

There are some other desired features that are not absolutely necessary, but highly desirable:

APPENDIX B

VACUUM PAD EXPERIMENT

Equipment

At Mirror Lab 10-1-93

Center of pad pulled away from floor at 3440 lbs.

Peak lifting force occurred at 3820 lbs.
Pad raised 2+ inches before breaking away from floor.

At MMT

19 in. Hg at starting point.
Held at 1000 lbs. lift, letting off slowly on the vacuum.

Lift on pad until seal broke.

Pad installed on a flat and with pump off for 77 min. at 19.99 in. Hg (no measurable leak).

Vacuum Pad Tests on Polished Surfaces
Dean Ketelsen
November 93

Several experiments were conducted to test the suitability of using vacuum lift pads on polished glass surfaces.

As a start, to be able to see if the surface had been damaged in any way, a visual inspection for sleeks and scratches was done on the large 26" diameter disk. Several dozen were mapped, using a bright light in transmission while looking for scattering sites. Certainly not all were mapped, but certainly a good subset of the total.

The two pads to be used for the tests were both disassembled and cleaned. There was some concern about using solvents, so only soap and water was used to scrub the neoprene (large pad) and silicone (small pad) vacuum pads. The pads were reassembled and the bolts tightened uniformly.

The first test was to use the large pad directly on the polished surface. The pump was turned on (vacuum off), the pad placed on the surface, pad was centered, and then the vacuum turned on. It took about 5 seconds to evacuate the space under the pad and for the green light to come on. The plan had been to run the vacuum for an hour, but it overheated after about 30 minutes, so subsequent tests only had the vacuum pump run for 10 minutes at most and the rest of the time with the residual vacuum. After 1 hour, the vacuum was released and the pad removed.

Three of these cycles were run, and the last was held to check the leak rate of the pad - pump combination. The data:

elapsed time vacuum
0 hr 23.5"
2.0 hr 22.5"
4.0 hr 22.0"
6.5 hr 21.4"
19.0 hr 17.0"
24.0 hr 16.0"

Use of the vacuum pad left residue on the polished surface. The large pad left physical debris over the center part of its contact, presumably left from the spacer pads that were used to build up the center part of the vacuum pad. Cleaning of these spacer pads seemed to dissolve part of the pads, and again generally, solvents other than water were not used for cleaning. In addition to the debris left from the spacer pads, the ring of the pad itself left marks with the "black breath" test. In this test, the tester breathed on the surface and observed marks on the resultant condensation. The smaller silicone rubber pad left no physical debris, but left marks visible with the "black breath" test.

A cursory cleaning was done and the surface was examined for new scratches. None were seen that had been caused by the vacuum pads.

It was decided to silver and aluminize the surface and see if the standard cleaning techniques removed the observed residue. For the silver spraying, Richard Barreda of the Optical Sciences Center performed the cleaning and silver spraying. The surface was first cleaned with calcium carbonate powder and distilled water, which took off all of the physical debris. However, the "black breath" test still showed a residual ring pattern from the pad. The next cleaning with acetone cleaned that up. He then used a mixture of Alconox and a polishing compound with water for a final cleaning with a distilled water rinse. The aluminizing was performed at the Sunnyside aluminizing facility and standard cleaning techniques were used. Again, no residue remained after cleaning. Both the silver and aluminum coatings were successful and passed the tape test, (none of the coating pulled off with scotch tape). It was, therefore, determined that the use of the vacuum pads did not have detrimental effects on the aluminizing process.

With the surface coated, the use of opticoat (Universal Photonics, Hicksville, NY) was used to protect the surface. The effect of the vacuum pads on the opticoat surface was next examined. With a new opticoat layer less than two hours old, significant stretching and pursing of the opticoat layer was observed. With an aged layer (overnight), use of the vacuum pad had no apparent external effects on the opticoat layer. When that opticoat layer was removed, absolutely no damage to the coating was observed. The vacuum seemed to leak more when attached to the opticoat surface than bare glass. Leak rates were about 1" per hour, compared to about a third of that as listed above.

The vacuum pad was then used over a partially removed opticoat surface. Where the bare coating was in contact with the pad, some contamination was observed. Again, this was cleaned off with solvent, but the cleaning process appeared to damage the coating from the local abrasion.

For the ultimate test, the opticoat layer was then partially removed and then replaced over the aluminized surface and the vacuum pad reapplied. The pad seemed to hold normally, though the leak rate seemed about double for that of bare opticoat (2" per hour). When removed, the opticoat seemed to be reattached except for a few bubbles under the layer. The aluminum coating appeared undamaged.

In summary, use of the vacuum pads had no effect on the aluminization process. Some debris remains on the surface by use of the pads, but this is cleaned off by normal cleaning techniques. With an aluminized surface, a properly aged opticoat layer seems to adequately protect the surface from any application of vacuum pads.

Subject: "Realuminizing" by Bill Kindred

When a mirror arrives at Sunnyside for realuminizing it undergoes the following process:

  1. two washes with industrial detergent,
  2. Three scrubs with (HCL based) stripper,
  3. three scrubs with a mild abrasive (CaCO3) and base KOH) solution,
  4. one wash with diluted HNO3 solution.

The surface is thoroughly rinsed between steps with demineralized H2O. After final rinsing the surface is vigorously hand-dried with KimWipes, the objective being to have no part of the surface evaporatively dried. The cleaning received by your discs was less rigorous owing principally to the lack of an old film to strip. Its exact treatment was as follows:

  1. two semi-rigorous washings with detergent (2 min. each) rinse,
  2. two rigorous scrubs with abrasive and base (3 min. each), rinse,
  3. one non-rigorous washing with nitric (10-15 sec.), final rinse (5 min.).

We made no effort to give these substrates any special handling or treatment of any sort. If anything their cleaning was less thorough than a typical process optic could expect. The last step involved two mild scrubbings with organic solvent. These were sufficient to give good black breath although this step is probably superfluous given our present understanding of the glow cleaning process. Any elastomeric (or related) residue would have been visible under the black breath. Immediately before closing the chamber the surfaces are cleaned with CO2 snow--enough to remove any remaining particulates but insufficient for any significant cleaning process.

The evaporation parameters were as follows:

The chamber was opened, the substrates were removed and opticoated in the minimum practical time, about 15 minutes.

Pad Tests after Aluminizing
J.T. Williams
November 8, 1993

Additional tests of the large neoprene vacuum pad were conducted 11/5/93 on a freshly aluminized surface of your large plate glass disc (~28" diameter x 3/4" thick). The test glass plate was aluminized 10/4/93 at the Sunnyside vacuum facility in the same process as telescope mirrors by Bill Kindred and Gary Rosenbaum. While very fresh, the 0.930 nm aluminum film was tested for adhesion, (Scotch-tape test), and immediately opti-coated with spray-on strippable plastic film, (Universal Photonics, Blue 333), to slow the oxidation process.

The following day, the vacuum pad was attached to the opti-coated surface with no special preparation. The vacuum pump was operated for 3 hours and reached the nominal dial pressure of 25" Hg. After the pad was removed, the opti-coated surface was examined for mechanical damage to the plastic film: tears, folds, wrinkles. The dulled print of the circular pad contact area was the evidence left. The opti-coat was then stripped in the normal manner, leaving half of the surface plastic coated. The freshly aluminized surface was examined in detail for mechanical damage and/or contamination -- none was found. A vapor/breath test showed no evidence of the pad's presence on the mirror.

The large pad was again attached to the half bare aluminum/half opti-coated mirror, with ragged opti-coat edges making two transitions, and pumped for ~2 minutes time. When the dial pressure reached 25" hg, the pad was released and removed. Inspection of the freshly aluminized (aluminum oxide) surface showed contamination from the pad edge seal zone and the central cushion zone but no mechanical damage or changes to the surface. The contamination pattern was not streaked radially, but looked "blotted" on the aluminum. A test zone of the contaminated region was cleaned with ethanol and KimWipe towels. No scratch marks or residual pad pattern was evident after simple cleaning. Vapor patterns were observed from the cleaning operation and breath test. No change in the vapor patterns were discernable on the metal film.

I agree with the conclusion that opti-coat should be used when lifting polished and aluminized mirrors. It is also gratifying to know that the clean pads render no permanent damage to the clean surface or metal film.