First, a quick review of the basics for those not familiar with the coating process. The three materials used in astronomical coatings are aluminum, silicon dioxide and titanium dioxide. A standard coating consists of just enough aluminum to achieve the maximum reflection that the material can provide. As the aluminum is applied the reflectivity increases to a maximum of ~88% in the visible for bare aluminum. A standard protected aluminum coating includes one layer of SiO2 on top of the aluminum. The basic protective SIO2 overcoat is only thick enough to “seal” the aluminum from the elements and provide hard, scratch resistant layer. It’s not intended to enhance the reflectivity.

If you can control the thickness of the SiO2 layer you can enhance the reflectivity by applying 1/4-wave thickness of the wavelength you are most interested in. This is referred to as one QWOT (quarter wave optical thickness) of SiO2. In our case that would be _-wave of 5500 Angstroms or 1375 Angstroms of SiO2. By applying the proper thickness, the reflectivity can be enhanced to about 91% at 5500 Angstroms. The region of 91% reflectivity is fairly narrow. In other words, as you move away from 5500 Angstroms on the spectrum the reflectivity drops. Many coating labs refer to this single _-wave layer as Semi-enhanced Aluminum.

If you take the semi-enhanced aluminum coating and add a QWOT of TiO2 you can enhance the reflectivity to about 96% in the visible. Now you have one High index/Low index stack or HL stack. Most labs refer to this coating as Enhanced Aluminum. To continue to increase the reflectivity you add additional HL stacks. You can increase the reflectivity to about 98% with two HL stacks, in other words, Al, SiO2, TiO2, SiO2, TiO2. These are typically branded as proprietary enhanced aluminum coatings.

All of the above are subject to fiddling and fine tuning depending on several factors; evaporation method, process temperature, ion assist and the addition of process gas. This is where it gets complicated. I’ll start with the most basic process.

Thermal Evaporation
Thermal evaporation of aluminum is accomplished by wrapping aluminum around a tungsten coil. Current is applied to the coil to melt the Al (wet the coil) then the current is increased to evaporate the aluminum. This is fairly straight forward. Next the SiO2 is evaporated but this is more of a challenge. SiO2 does not melt and evaporate from the liquid state. It sublimates. In other words it goes straight from solid to vapor. In addition, the plume of evaporant coming off the source is not regular. This makes it difficult to control the rate of deposition and the uniformity of the deposit on the glass. To combat this problem SiO2 is evaporated from a container with internal baffles and an opening in the top from which the evaporant escapes. This provides a more uniform plume and allows for a uniform coating. TiO2 is evaporated from a crucible that is heated by a tungsten coil.

During the evaporation process some disassociation occurs. Some if the SiO2 breaks down to become SiO and just Si. What is deposited on the glass is a mixture of these elements. Similarly for the TiO2. This changes the refractive index of the layers as they are deposited on the glass. What you get is not necessarily what theory predicts. If you measure the refractive index of your deposits with an ellipsometer you can adjust your coating design according to your actual indices.

Other factors that affect the refractive index of the coating layers are substrate temperature and coating layer structure. As the material condenses on the mirror in has to maintain some of its energy so that it can settle into the material matrix. If the substrate is cold the condensate accumulates like frost. What you’re after is something more like freezing rain. Hence, the substrate needs to be heated to a certain temperature specific to each material as each material is deposited. You want the coating layers to be amorphous and dense. Such a coating has a more predictable and repeatable refractive index and it is moisture stabile. A porous, frost-like coating absorbs moisture which changes the refractive index of the coating and results in premature deterioration of all of the layers, including the aluminum. This type of coating should not pass QC.

E-beam Evaporation
In an E-beam system the material is evaporated by sweeping a high energy electron beam across the material. Modern e-beam systems allow for control of beam energy, focus, sweep pattern and sweep speed. They provide so much control that they can generate a uniform evaporant plume, even when evaporating SiO2. The typical e-beam system has several indexable pockets. This allows the system to apply all of the different materials in as many layers as you like.

When integrated with a deposition rate monitoring system, such as an oscillating crystal monitor, the e-beam can be set to deliver very specific deposition rates and thickness for each material in the coating design.

Ion Assist
Ion assist provides several benefits to the coating process. The ion source provides a plume of high energy inert particles (ionized argon gas) that impinge upon the coating materials as they are deposited on the glass. They literally hammer the coating molecules into place to provide a more dense, moisture stabile matrix. Some added benefits are; the process is effective at lower substrate temperatures thus allowing more gentle heating of the mirror.

Process Gas
As mentioned, some disassociation of the SiO2 and TiO2 occurs during the evaporation process. This happens regardless of the evaporation method used. This is where a slight partial pressure of process gas comes in. In this case, oxygen is injected into the process. The background of oxygen gas helps to keep the SiO2 and TiO2 at stabile equilibrium during the deposition process.

Coating Uniformity
Coating uniformity is achieved by rotating the mirror above the evaporant source or rotating three or four mirrors in a planetary holder system. A single rotating mirror will typically require what’s called uniformity mask. The mask blocks some of the evaporant such that the mirror receives a uniform thickness of coating from center to edge. Finding the right shape for the mask is non-trivial and requires several trial runs for each evaporant material. With care however, thickness uniformity of a few percent can be achieved over a large mirror. The better alternative is a planetary rotation system where the mirror rotates around it’s axis as well as around the center of the chamber. Uniformity as low as 1% can be achieved with a planetary rotation system.

Will an enhanced coating change a mirrors figure?
With regard to the coating changing the figure of the mirror, consider the application of each layer of material. The layer most likely to change the mirrors figure is the aluminum layer as it forms the surface from which the light is reflected. The SiO2 and TiO2 layers are transparent. They act only to enhance the reflectivity. If they lack uniformity their primary effect will be a variation in the reflectivity across the mirror.

The aluminum layer is about 900 Angstroms or approximately 1/6-wave thick at 5500 Angstroms. A loose tolerance on coating uniformity is 5%. This is equivalent to 1/122-wave at 5500 Angstroms. A more realistic target for uniformity is 2% or 1/300-wave. A properly applied coating will not change the figure of the mirror by any significant amount. It would have to be a very poor job in deed to ruin a mirrors figure.

Summary
There are many details that go into producing a high quality coating. I haven’t even listed all of them here. There are many other factors on the input side of the coating equation. The trick is to have a way of measuring the out put and an understanding of which inputs to adjust to achieve your process goals. At the same time you need good process monitoring and real time control of the process. Modern coating machines typically include e-beam evaporation, ion assist with process gas and deposition rate monitors with all of the machines functions controlled by a computer connected a programmable logic controller (PLC) for each of the systems on the coating machine. The machines are complex, fully automated and self monitoring and they can reliably and repeatably run a well designed and de-bugged process to produce a desired result. Much of this technology was developed by the semi-conductor industry where coating designs can include hundreds of layers with very tight tolerances. The coatings we use on telescope mirrors are exceedingly simple compared to the capability of modern coating machines.

Coating deterioration

Need description for this image Need description for this image

I've worked with many different coating vendors. The ones we've had longevity issues with did not use ion assist. They all look fine out of the box but non-ion assist failures exhibit signs of moisture absorption in the over-coat layers early in their useable life. When a porous coating absorbs moisture it starts to look hazy. As time goes on the moisture in the over-coat layers breaks them down further and accelerates the deterioration process. This allows the elements to get through to the aluminum. Eventually, you'll see blotches in the aluminum. If you clean the mirror, the spots where the blotches are will eventually wash away leaving bare glass. By this time, the rest of the coating will look very hazy. It will have deteriorated to the point where it scatters an incredible amount of like. I have null Ronchigrams of failed coatings that exhibit this scattering. If find some and post them to the photos section. I'll let you know when they are posted.

It's not that all non-ion assist coatings will inevitably fail this way. This only occurs when something goes amiss in the process. It's tough to detect this problem right out of the chamber because the coating looks fine to the eye. You can detect the problem by measuring witness samples that are coated with each batch of mirrors. If you expect your reflectivity to be X and it's significantly less than X you probably didn't apply exactly what you wanted. In this case you should run samples of each material as if you were coating a mirror but run only a single layer of each material. With the proper instruments you can measure the actual thickness you applied and the refractive index of that particular material. If the index is off by too much then you have a problem to solve.

There are many factors in a process that can drift and affect the quality of the coating. That's why it's important to measure witness samples from every run. In addition to the reflectivity and index measurements you can do destructive testing on the samples like; the tape peel test, eraser rub test, salt spray test or leave the sample exposed to the elements for long term tests. Modern coating machines keep a log of everything they did for each run. If you see a problem developing in your witness samples you can look at the logs to see what is drifting. With this information you can set things right and keep the product within your QC parameters.