The advent of these very low expansion materials generated the need for more precise measuring techniques. Plummer and Hagy (ref. 2) have described Corning's work with two techniques. One uses a Fizeau interferometer, and the other involves a rod type of dilatometer. The accuracy of the dilatometer has been further improved since that paper was published.

A third method for measuring the overall expansion of a blank will be described by the Itek Corporation in a future issue of "Applied Optics." This involves the use of a holographic interferometer.

A comparison of the current status of these measurements is shown in table 2.

High Thermal Diffusivity

This factor becomes very much less important as the thermal expansion coefficient approaches zero. it is, of course, one of the most important advantages of metals over glass. Selection of a glass for this property would not be significant because of the narrow range of values for most glasses (ref. 3). It is interesting that fused silica is at the top of the range.

High Elastic-Modulus-to- Density Ratio

The spread in this ratio for glasses is not very great; i.e., about 4.43 (x106 psi/gr/cc)

for ULE to 5.38 for a low-expansion glassceramic. A ratio of the order of 7.5 can be obtained with some glass-ceramics but at great sacrifice of low thermal expansion. On the other hand, beryllium has a ratio of 23.

Homogeneity and Clarity for Inspection

These properties go well together, particularly for a transparent material such as ULE glass. Small differences in composition, such as cords, produce large distortions in transmitted light paths, and birefringence studies can quickly reveal differences in mechanical properties throughout a single piece. The difficult problem, however, is relating these to the actual performance in a mirror blank. Even though some early work demonstrated that satisfactory figures could be obtained on blanks having over 80 millimicrons per centimeter birefringence, much progress has been made in the control of our process to 20 millimicrons per centimeter or less.

Size Capability

Boules, such as the one shown in figure 2, are about 6 feet in diameter as they come from the furnace in which the material

is produced from the vapor state. These can be machined and fused together to make larger sizes. As shown in table 1, blanks up to 157 inches in diameter by 25 inches thick have been delivered in fused silica. The same technique and apparatus work for ULE glass.

Lightweight Capability

There is no technical limitation on size capability.

Structure of ULE Glass

Zachariesen (ref. 4) and Warren (ref. 5), using x-ray techniques, developed the onedimensional model for crystalline silica, shown in figure 3a, and for fused silica, shown in figure 3b. The black dots represent silicon ions, and the circles represent oxygen. Actually, in three dimensions the silicon ions are bonded to four oxygen ions, and each oxygen ion is linked to two silicon ions. When the crystalline silica is heated above the liquidus, the structure opens up and becomes irregular.

Harold Smyth (ref. 6) of Rutgers University has made calculations that account for the zero expansion of fused silica at low temperatures. This property is based on the open structure permitting more lateral vibration of the oxygen ions than when they are

(a) Crystalline silica (b) Fused silica Figure 3. One-dimensional models developed by Zachariesen and Warren.

confined in the crystal structure. The temperature around which the expansion changes from positive to negative and the general shape of the expansion-versus-temperature curve can be changed by the addition of titanium. The titanium ions substitute for some of the silicon ions in the random glassy network and permit different vibrational characteristics. This accounts for the zero expansion of ULE titanium silicate around room temperature.

The vibrational concept to explain zero expansion seems to be further confirmed by the large increase in thermal expansion when soda is added to fused silica. According to Warren (ref. 7), the sodium ions occupy holes in the network, as shown in figure 4. These ions limit the lateral vibration of the oxygen so that, when heat is applied, the network expands.

Summary

This great similarity of structure between ULE titanium silicate and fused silica adds all of the advantages of fused silica to the very low thermal expansion of ULE to make it an ideal mirror material.

The fact that this material can be fused with a flame or in a furnace without destroying the near-zero expansion makes possible the construction of any size required.

References

1. Dietz, R. W.; Bennet, J. M.: Bowl Feed Technique for Producing Supersmooth Optical Surfaces. Applied Optics, vol. 5, no. 881, 1966. 2. Plummer, W.A.; Hagy, H.E.: Precision Thermal Expansion Meausrements on Low Expansion Optical Materials. Applied Optics, vol. 7, no. 824, 1968.

3. Shand, E.B.: Glass Engineering Handbook. McGraw-Hill Book Company, Inc., New York, 1958, p. 30.

4. Zachariesen, W.H.: Atomic Arrangement in Glass. Journal of the American Chemical Society, vol. 54, no. 3841, 1932.

5. Warren, B. E.: X-ray Determination of the Structure of Liquids and Glass. Journal of Applied Physics, vol. 8, no. 645, 1937.

6. Smyth, Harold: Booklet published by NSF, Grant No. G24276, February 1966.

7. Warren, B.E.: Summary of Work on Atomic Arrangement in Glass. Journal of the American Ceramic Society, vol. 24, no. 256, 1941.

Four of the important design considerations for high-resolution astronomical mirrors are the thermal stability, thermal expansion, ability to obtain and to retain surface finishes, and the mechanical characteristics of the material used to fabricate these mirrors. These properties are basic to the material and present the limiting factors to distortion-free imaging. In addition, optical surfaces can be deformed by thermal stresses in the mirror material, resulting in a distorted image (ref. 1). The combination of good thermal stability, low thermal expansion, ease of obtaining good optical finishes and the ability of the material to retain these finishes indefinitely, high modulus of both rigidity and elasticity, and a minimum of thermal stresses within the material are of critical importance in the proper selection of the material for optimal performance. Two materials, one the standard in high-resolution mirror blanks and the other a ceramic possessing these characteristics in comparably high quantities, are compared in table 1.

After World War II, the introduction of commercially available fused quartz made possible considerably more effective mirror blanks. A most significant point of progress came with the manufacture and shipment of the 158-inch-diameter mirror blank to the Kitt Peak National Observatory. This 15-ton blank, shown in figure 1, was formed by fusing together several hundred hexagonal quartz ingots and consists of a substrate made from two layers of 6-inch ingots 12 inches high and a cap 5 inches thick made from 21-inch hexagonal ingots. Since the manufacture of this blank, great strides have been made in increasing the ingot sizes for large mirror manufacture. Figure 2 represents the original 6-inch ingots used in the Kitt Peak blank as well as the 21-inch ingots used for the surface. In addition, the newer manufacturing facilities have produced this 72-inchdiameter giant.

This manufacturing accomplishment demonstrates that there now exists no foreseeable theoretical limitation on size for