Replication Substrates

Almost any material that epoxy will adhere to can be used as a replica substrate. However, the surface accuracy and stability of the mirror will depend completely on that choice of substrate material. The characteristics of the materials that are most important include:

• Density
• Young's modulus
• Thermal coefficient of expansion
• Thermal conductivity
• Dimensional stability
• Cost
• Machinability

The most effective choices for replicated mirrors include aluminum, beryllium, fused silica and graphite epoxy. Other materials which can be used include aluminum oxide, silicon, silicon carbide, titanium, miralloy and stainless steel.

The density of the substrate is an important consideration when the weight of the replica is critical. The material's density (p) must be considered in relation to the it's Young's modulus, as well. A low density combined with a high Young's modulus (E) yields minimum component weight and minimum inertia. The specific modulus, E/p, is the figure of merit for minimum weight. The higher the specific modulus, the better the material for low-weight requirements.

Beryllium has the best specific modulus at 151 followed closely by boron carbide at 146, beryllia at 133, silicon carbide at 118 and aluminum at 26. Currently, use of boron carbide is impractical because it is manufactured only in sheets and is very expensive and difficult to machine. Because beryllia cannot be cast into shape, it offers no advantage over beryllium. Silicon carbide is most practical for larger size optics. Aluminum is the least expensive, but is not the best choice when weight is a critical consideration.

In addition to considering the specific modulus, density by itself is often important. For smaller optics, there is a limit to how thin face sheets can be fabricated. When minimum face sheet thickness is obtained, then density becomes the most important criteria. Beryllium has the lowest density at 1.85 g/cm3, compared to aluminum (2.7 g/cm3) and silicon carbide (3.14 g/cm3).

Often, minimum inertia is the important criteria for performance. Inertia is a function of Young's modulus (E) and density (p) given by: specific inertia = [(E/p) /p]

The higher the figure, the better the specific inertia of the material. Beryllium has a specific inertia of 6.65 followed by boron carbide at 4.83, miralloy at 4.49 and silicon carbide at 3.5. Again, for small size optics with minimum sheet thickness, density affects the choice of material. Beryllium provides the best option, followed by aluminum. Materials such as fused silica, silicon and graphite epoxy are generally not considered because either the material cannot be light-weighted easily or because they are not stable enough for precision optics.

The preferred aluminum for replicated optical components is 6061-T651 aluminum or 6061-T6. These types provide excellent homogeneity and tempering characteristics and are the best suited for the replication process. Other aluminums, including 2024, 7075, or cast aluminums such as 356 or precedent 71 can be used as well. The advantages of aluminum are:

• Relatively low density
• Inexpensive
• Can be fabricated as solid structures
• Can be light-weighted using a rib structure
• The surface to be replicated can be machined to an aspheric shape using a CNC machine.
• Since many optical benches are made of aluminum, assembly becomes temperature insensitive.

For precision optics, especially light-weighted optics, it is critical to stabilize, or temper the aluminum. Using 6061-T6 aluminum as an example, thick sections of blank material are solution heat treated and tempered to what is known as a T651 condition. The part should then be rough machined and stress relieved for four hours at 177°C (350°F) and oven cooled. Since this will cause some distortion, the critical dimensions should then be final machined.

After final machining, the part must be lapped (fine-ground) to remove machining marks and when required, to improve surface accuracy. Before replication, the part should be thermal cycled from -60°C (-76°F) to + 100°C (212°F) for a minimum of three cycles. After the part has been replicated, there should be no machining of the substrate on or near the optical surface since this would cause the mirror to distort.

Beryllium is a material that is often expensive to use in optical components. For certain applications where light weight or low inertia is crucial however, it is still the material of choice. There are many types of beryllium, each based mainly on the beryllium oxide content. Low oxide beryllium is best for polishing but it has a very low precision elastic limit. For small and thin low-oxide beryllium optics, it is possible to deform the part just in the handling process. Since the replication process does not care what the oxide content of the beryllium is, higher oxide (I-220) or instrument grade (I-400), make better substrate materials. These grades have much higher precision elastic limits and do not deform as easily.

Machining of the beryllium substrate is critical to the stability of the optical component. It must be rough machined in stages with sharp tools to minimize surface damage. It must then be heat treated at 1450°F in an argon atmosphere to relieve machining stresses. Final machining must be done with small cuts to avoid damage. After the part has been final machined, it must be etched, removing a minimum of .002 to .003 inches of material from all surfaces. This removes all subsurface damage. If the part must be lapped or figured after etching, care must be taken to minimize any subsurface damage. Before replication, the part should be thermal cycled from -60°C (-76°F) to 177°C (350°F) for a minimum of three cycles.

Graphite epoxy has several good characteristics that make it desirable as a replica substrate. It has low density (lower than beryllium) and high Young's modulus, as well as a low coefficient of thermal expansion. However, the hygroscopic nature of graphite epoxy frequently causes this material to distort over time and is therefore not widely used in replication. Still, there are applications (such as a solar collector on a satellite or in cases where low atomic particle cross section is required) when graphite epoxy is sometimes used.

Because it is difficult to get a smooth substrate surface on graphite epoxy (roughness on the order of hundreds of angstroms), print-through to the optical surface generally limits the material's use to non-imaging applications. For energy collection systems, whether in the ultraviolet, visible or infrared, graphite epoxy usually provides acceptable performance.

Unlike many other materials, graphite epoxy mirrors can be made very large. We have fabricated one meter graphite epoxy solar collector segments as part of a large solar collector.

Alumina is a very stable ceramic with moderate density and high Young's modulus. The ability to injection mold ceramic alumina in thin sections facilitates the cost-effective manufacturing of light weight, low inertia mirrors in modest quantity. Injection molding will, however, limit the size of the components, and the cost of the tooling must be amortized over the quantity of the parts being produced.

Injection molding of alumina is not limited to flat mirrors. Spherical, aspheric, hollow roofs and hollow retro-reflectors can be molded as well. Where the required optical accuracy of the mirror allows replicating on the molded surface without machining of the substrate surface, costs are comparable to aluminum mirrors. Alumina ceramic is also an attractive alternative in applications where metal substrates cannot be used.

Large, light weight silicon carbide mirrors are often fabricated for space and military applications. Several different fabrication techniques are used to create silicon carbide substrates including reaction bonding, nitrogen bonding, direct sintered and chemical vapor deposition (CVD). Some forms of silicon carbide can be polished to a very good surface accuracy and surface roughness. Others, such as reaction and nitrogen bonded, cannot. These forms must have their optical surfaces clad with either CVD silicon or silicon carbide.

For commercial applications, injection molded silicon carbide is most cost effective. Injection molding limits the size of the parts and tooling must be amortized over the quantity of the parts produced. It also can be polished and is more cost effective to replicate than other forms.