I have two questions on mirrors.

  1. I’ve read that in the past quality mirrors were coated with silver but that today vacuum evaporated coatings of aluminum are the accepted standard. When I look at the reflectance vs. wavelength plot,

enter image description here

I see that silver has a higher reflectance than aluminum. So why use aluminum instead of silver? If one wants the highest quality mirrors I assume that cost is secondary, so what is the physics that I am missing here?

  1. Why are the more “technical mirrors” (I am not sure what this means but I assume more precise?) front-surfaced instead of back-surfaced?

Telescope mirrors and other mirrors used by scientists telescopes regularly do use a silver coating. See for instance here. However, aluminum coating are the norm (certainly for the large primary mirrors deployed in telescopes) because of durability reasons. I quote from the text linked to above:

The challenge with using silver as a coating material is that, unlike aluminum, it tarnishes with exposure to air, specifically to sulfur.


"Like the family silver set," explains Tom Geballe, Gemini Senior Astronomer, "which slowly develops brown tarnish spots over time and must be regularly polished, the shiny silver on a telescope mirror also tarnishes rapidly reducing reflectivity and increasing emissivity. The observatory's engineers, however, can't just grab a cloth and some polish when the tarnish spots appear."

On your second question: a back surface reflective coating implies an additional reflective surface: the air-glass interface. This leads to increased light losses and the need for anti-reflective coatings.

  • $\begingroup$ Thanks Johannes, I never thought about tarnishing of the silver. Loved the links. $\endgroup$ – Carlos Jun 2 '14 at 13:55
  • $\begingroup$ A bit of history here. $\endgroup$ – uhoh Dec 20 '18 at 8:25

Johannes makes a good point about durability. As a footnote, I'll add that aluminum has another nice property over silver, at least as far as your plot shows: constant reflectance over the visible spectrum.

Look at the slopes of the lines from $400\ \mathrm{nm}$ to $700\ \mathrm{nm}$ - silver varies from $80\%$ to $95\%$ reflectance, while aluminum stays between $90\%$ and $93\%$. So yes, a primarily red image will be slightly dimmer in aluminum than in silver, but all images are slightly reddened (the blues are suppressed more heavily than the reds) in silver, whereas aluminum gives a more faithful representation of the colors. (Plus primarily blue images will actually be darker in a silver mirror than an aluminum mirror.) For everyday use, it's easy to compensate for a darker image (just light up the room more), but not so easy to compensate for an image that has the wrong colors.

  • 1
    $\begingroup$ So if I read you correctly, if two parallel plane mirrors have an object in-between them and I was looking at a numerous reflections of the object, I would expect the image should “appear” slight more blue-green with aluminum and slight more green-red with silver. I would then assume that I could tell the difference between these two coatings from the image produced – is this correct? $\endgroup$ – Carlos Jun 3 '14 at 12:46
  • $\begingroup$ @Carlos That's pretty much correct. Note that the reason why gold appears gold-colored is due to this variable reflectance – it's just more apparent than with silver. $\endgroup$ – ntoskrnl Jun 3 '14 at 15:07

Dr Chuck mentioned one reason to avoid a rear surface mirror, but there are a few more.

If the light goes through the glass, then you need to design it to:

  1. Minimize refraction effects such as chromatic aberration
  2. Maximize transmission
  3. Minimize bubbles

From a viewpoint of the material used, a front surface mirror has fairly simple, mostly mechanical requirements, such as the ability to polish it to the required smoothness, and stability to maintain the intended shape across temperature changes. Even so, the tolerances involved are so fine that meeting requirements is fairly non-trivial.

The "wave" of telescopes that (sort of1) culminated in the Hale 200-inch at Mount Palomar was based (in large part) on the invention of Pyrex. Earlier glass was enough weaker that it required a considerably thicker mirror and/or better support structure so the center didn't "sag" under its own weight. The Hale's 200-inch mirror uses a honeycomb design to reduce its mass (to around 14.5 tons), and still requires a fairly elaborate structure to support it without distorting the mirror.

If you wanted to transmit light through the mirror, you almost certainly could not use such a honeycomb design, which (in its case) would roughly double the mirror's mass.

Worse, glass used in most lenses (for example) is much softer and weaker than the Pyrex used in those telescopes, so it would almost certainly require a mirror that was still thicker and more massive and/or more elaborate support structure.

Doing a quick back-of-the-envelope calculation, I'd guess the primary mirror in such a design would be somewhere around 100 tons.

If anything, actual development has tended to go the opposite direction: if I'm not mistaken, the "Zerodur" ceramic used in the mirrors for the Keck telescopes is nearly (fully?) opaque. Although it's roughly double the diameter (and therefore quadruple the area) the primary mirror in each of the Keck telescopes totals only about 18 tons (but note the "totals"--each mirror is actually built of 36 segments, each of which is around half a ton).

1. You could argue that it really "culminated" in the BTA-6. This is a 6 meter telescope located in Russia. Although its designers (apparently) believed they could successfully fabricate a mirror this size, its mirror cracked after a fairly short time, and had to be replaced. Although larger than the Hale telescope and therefore theoretically more sensitive and capable of higher resolution, its contributions have been comparatively minimal. In any case, its mirror design is enough like the Hale's that it doesn't change the overall situation to any significant degree.


Also, regarding the front/back surface point, if the back of the mirror is surfaced, you need to get the right shape on each surface, rather than just the front.


Nobody has yet mentioned the one aspect of silver that the graph depicts. Silver absorbs strongly near 315 nm. If you were interested in UV light from the stars, you would have a hole in your data (between 310 and 320 nm) if you used silver. Not only that, any light gathered with a wavelength less than 310 nm would require longer acquisition times.

  • 1
    $\begingroup$ You would be talking about space telescopes then? $\endgroup$ – Rob Jeffries Jul 31 '15 at 16:31

It ain't necessarily so. Some large telescopes are using silver coatings. For example the Gemini telescopes (Boccas et al. 2006).

The reason for going with silver rather than aluminium is a slightly improved reflectivity in the near- and mid-infrared (99.1 per cent at 10 microns), but also that the emissivity of the silver coating is around 38% that of aluminium, which reduces background in infrared observations.

The paper reports on durable coatings that reduces the tarnishing that favours an aluminium coating (as well as aluminium's superior performance shortward of 400nm). I guess that these are rather expensive, so aluminium is still the norm unless you want to be re-silvering the mirror very frequently.

  • $\begingroup$ I've asked a related question and linked to here. $\endgroup$ – uhoh Jul 12 '16 at 15:45

Not the answer you're looking for? Browse other questions tagged or ask your own question.