362KB 
Figure 1: photo by Dean Ketelsen, October 1997.
Osservatorio Astronomico di Bologna, Via Zamboni 33, 40126 Bologna, Italy.
http://boas5.bo.astro.it/~zitelli/hawaii/paper/paper2.html
http://medusa.as.arizona.edu/lbtwww/tech/oab9702.htm
The realization of a washing procedure to remove the aluminum coating from large optical mirrors without the use of the hands is described. The adopted chemical sequence is optimized for the primary mirrors of LBT, for which washing and aluminizing are planned to be performed on board the telescope structure. Nevertheless, since the process gives positive results, it might be of interest also in conventional operations, i.e. with the mirror on the floor, with other large telescope mirrors.
2. Briefly what is LBT project?
The Large Binocular Telescope, LBT, is a telescope with two 8.4 m primary mirrors on a single mount. Table 1 shows the main characteristics of this project. More detailed informations can be found in Hill 1996 [1] and references therein.
In this telescope the light beam coming from the two mirrors can be either combined to feed a central instrument, or can feed two independent Gregorian foci and the data combined with a computer.
362KB
Figure 1: photo by Dean Ketelsen, October 1997.
Figure 1 shows the first primary into the furnace of the Mirror Lab in Arizona, at the end of the last procedures of melting The honeycomb structure is easily seen: this primary is significantly large and dramatically light. Removing the mirror from the optical support structure would be a very risky operation ( this is likely to be a common problem for all very large lightweight mirrors). LBT is thus designed to hold the main mirrors on board the telescope during all the usual operations of maintenance.
The structure, see figure 2, is completely open and leaves almost free access to the front of the mirror. The open structure allows great flexibility in handling some equipment and in particular to introduce a washing equipement or to fit the aluminizing tanks into the primary. But this strategy, more "rigid" than having the mirror on the floor, needs to know in advance each particular of the procedure and requires careful considerations and tests on all the operations to be performed on board the telescope, since a number of provisions, like pipes and seals, become ancillary parts of the primary cells and must be part of the initial design.
127K
Figure 2: rendered drawings of LBT (images by ADS Italia and
Max Planck Institut fur Astronomie - Heidelberg )
A first series of experiments were done to test the efficiency of a number of chemicals to remove aluminum. We obtained some test mirrors, slicing and polishing a few low quality chunks of the same borosilicate glass E6 used for the LBT honeycomb blank.
The test mirrors were aluminized and sprayed with chemicals chosen among the most common products used in the observatories to remove old aluminum and dirt before realuminization. The samples of borosilicate were then left for one month in a bath of the tested chemicals, to check which liquids damage the surface. The roughness of the surface was measured with a Wyko interferometer before and after the bath. The damage tests were repeated twice using two different set of samples. In the second test the measured points were recorded to point the Wyko in the same position as before the bath . The result is shown in table 2 (a and b), where the reported roughness is the mean value of 6 different lectures in each sample mirror :
| roughness (nm) | ||
| before bath | 0.92 rms 0.16 | |
| after a bath in | HNO3 | 0.83 rms 0.05 |
| HOH | 3.44 rms 0.06 | |
| HCl | 1.21 rms 0.23 | |
| NaOH | 5.82 rms 0.82 | |
Table 2a: Chemicals damage test in the first sample
| roughness (nm) | ||
| before bath | 0.52 rms 0.06 | |
| after a bath in | HNO3 | 0.52 rms 0.09 |
| HOH | 0.44 rms 0.07 | |
| HCl | 0.50 rms 0.04 | |
| NaOH | 1.86 rms 0.15 | |
Table 2b: Chemicals damage test in the second sample
In both cases the sodium hydroxide increases the initial roughness. The potassium hydroxide presents the same behaviour as the sodium in the first case. In the second case the damage is not apparent: we interpreted this results as giving evidence that bases are likely to produce microdamages on borosilicate. We then rejected them as possible sources of damages in repeated treatments.
At the end of the tests we adopted the sequence of treatments as described in the following, with a description of the effects on a mirror with a deposition of dirty aluminum about 1000 Å thick; and exposed to ambient dust for several months:
1) Dirt is removed washing the mirror with a detergent solution of no foam soap and water, laboratory and hospital grade. The washing lasts about 15 minutes
2) Rinsing with tap water to clean the mirror and (very important) the pipelines
3) Washing with a solution of Hydrochloric Acid, at 37 % pureness, diluted with bidistilled water, plus cupric sulfate (28 g per liter of solution) for 15 minutes. After few minutes this solution attacks the aluminum and the first pinholes can be seen in the coating. Since the first application most of the coating should be gone, by a visual inspection. A second application is used to be sure of the result.
4) Rinsing with tap water
5) Steps 3 and 4 are repeated once or twice to be sure of a full removal of small aluminum fragments.
6) Washing with Nitric Acid, 65 % pureness, diluted 1:1 with bidistilled water, for 15 minutes. This step is to remove anything that may have been missed on the mirror surface .
7) Rinsing with tap water
8) Repetition of steps 6 and 7.
9) Prolonged rinse with bidistilled water to clean everything having contact with chemicals.
The washing machine (see figure 3) is designed to holds the mirror in
vertical position, because this seemed the safest position for
operations at the telescope.
The box is made of rigid PVC, transparent, to allow a
continuum visual inspection of entire surafce during the treatment. The
front dimensions of the box are 80 cm x 80 cm, with depth 30 cm. It is
designed to host our dummy mirror of 60 cm of diameter.
An horizontal arm made of stainless alloy, lodge 8 nozzles, on PVC,
(spaced by about 10 cm) and spattering the mirror surface from a distance of
about 5 cm. The arm is moved up and down
in front of the mirror spraying the liquids.
After a series of failures we get convinced that residual dirt recycled from
walls and lid of the washing chamber was spoiling the operations.
Accurate and repeated cleaning of the chamber was a condition for a positive
result.
We then added a second line of nozzles, with the same geometry, at the top of
the arm to wash and clean also walls and lid of the machine.
The nozzle exit area is rectangular in shape, with a diameter of 0.8 mm.
Coupled with the linear movement
of the arm, this shape provided an homogeneous coverage of the mirror
surface than the most common circular aperture nozzles.
The fluid is sprayed fanlike with a flow rate of 0.45 l/min at the working
pressure
of 4 atm.
A pump provides the circulation of the liquid, it is a pulsed pump
working with compressed air, coupled to the liquid piping system by a
diaphragm, so it is able to transfer heavy liquids with solid suspensions.
This pump is manufactured in thermoplastic materials and can be used with
acids and corrosive fluids as well. An important peculiarity of this type of
pump is the fact that it can run dry. This adds safety in case of failure as
finishing of the used liquid or during the switch between a tank to an
other.
The weight of the pump is 4 Kg, the capacity range from 1 to 40 l/min,
the pressure in the fluid can range from 4 to 6 atm. In our test the pump
was working at a capacity of 7 l/min pressure was 4 atm.
After the washing process, the mirror is left, protected from the dust, in
the washing machine for about two hours to drip the water.
It is recomended, to have a successful result, to rinse very well all the
components of the hydraulic circuit having contact with chemicals.
In the last phase a prolonged rinse shall be done; unaccurate rinse
leaves a significant quantities of undesiderable remnants.
At the end of the accurate final rinse a thin film of water is left on the
surface and is easly seen to progressively dry from the top to the bottom,
without leaving any droplet behind.
The procedure is completed removing and placing the mirror inside the vacuum
chamber for coating. Evaluation of the quality of the cleaning is empirically
based on quality and stability of the deposition.
From an old experience, poor cleaning results in an almost immediate decay of
reflectivity and defoliation of the aluminum layer.
If in a few hours, after the deposition aluminum does not shows oxidations or
mini defoliations, the washing is considered successful.
From the experiment made on a 60 cm mirror we can try to estimate
quantities and costs for treating large size mirror.
With a capacity of 0.45 l/min, the 16 jets should spray 7.2 l/min, in 15 min
the needed amount of solution is 100 l; this can be significantly reduced
adopting circulation in close loop.
Table 3a: Costs for a 60 cm mirror. In this experiment it
is used only 4 l of solution (2 l of chemicals) for one complete washing
cycle. It is assumed 1$ = 1700 LIT.
Assuming that scaling is goes with the surface, there is a factor 150.
We obtain for an 8 m mirror:
Table 3b: The Costs in the same conditions of table 3a,
scaled for a 8 m mirror.
Costs are driven by rinse water. If the
bidistilled water is produced in loco the cost can significantly
decrease. It can be seen that the initial cost of the equipment is easly
absorbed in a few years.
ADS Italia is the company committed to design both aluminizing and
washing systems.
For the aluminizing operations the telescope is pointed to the
horizon, holding the mirror in vertical position.
The washing equipment, made by a rotating arm and a cover, is picked up
by a service crane and positioned in front of the mirror.
The cover is sealed to the mirror cell.
The rotating arm lodging the nozzles spraying the mirror; fluids are passively
collected at the bottom of the tank. Length and diameter of pipes, as well as
time needed for a full recirculation of fluids are presently under study.
They constraint the actual quantities of chemicals necessary for each step and
as a consequence the actual operating costs of the process.
The authors acknoledge the assistance of the Loiano staff and particularly
Rino Mezzini. A special thank goes to Piero Salinari, Osservatorio Astrofisico
di Arcetri, and John M. Hill, Steward Observatory, for the helpful discussion.
5. The washing machine
110 KB
Figure 3: The machine prototype is composed by: a water-proof
box with a moving arm, a pump, a compressor and a series
of tanks containing the chemicals used to clean the mirror.
7. Costs and scaling
cc LIT $
HNO3 65% 2000 38400 22.6
HCl 37% 2000 28000 16.5
bidist.water -- 50000 50000 28.4
total cost 116400 68
cc LIT $
HNO3 65% 2000 5800 K 3400
HCl 37% 2000 4200 K 2500
bidist.water -- 50000 7500 K 4400
total cost 17500 K 10300
61 KB
Figure 4: A conceptual and preliminary drawing of the
washing system is shown in figure .
