If you hold a good camera lens or filter under a light and tilt it around. Depending on the coating, you may see purple, green, blue, bronze or brown reflections, and the colour often changes as the angle changes.
Those colours are not decorative, and they are not simply a tint applied to the glass. They are a visible result of the way extremely thin optical coatings interact with different wavelengths of light. The layers can be only tens or hundreds of nanometres thick, yet their thickness, refractive index and order can determine which wavelengths are reflected, which are transmitted, and how the coating behaves as the angle changes.
The interesting part is that these coatings are usually there to reduce reflections, even though the coating itself is what produces the coloured reflection you can see. In some cases, adding several microscopic layers to a piece of glass can actually make more light pass through it than through the bare glass.
Why glass reflects light
Whenever light crosses between materials with different refractive indices, some of it is reflected. Plain glass therefore loses a small amount of light at each air-to-glass surface. In a camera system with several optical elements, those reflections can contribute to flare, ghosting and reduced contrast.
Anti-reflection coatings reduce those losses, and they do it in a slightly counterintuitive way: by adding more material to the surface.
How adding a coating can make glass more transparent
Light behaves as a wave. When it reaches a very thin coating, part of the light reflects from the top surface while another part travels through the coating and reflects from the boundary below it. If the coating thickness and refractive index are chosen correctly, those reflected waves return out of phase and partially cancel each other.
A useful analogy is noise-cancelling headphones. They reduce unwanted sound by generating another wave that interferes with it. An optical coating does something similar with light, except the distances involved are measured in nanometres.
The diagram below shows the basic idea. The two reflected waves represent two paths taken by light of the same wavelength: one reflects from the upper surface of the coating, while the other travels through the coating before reflecting from the coating-to-glass boundary. That extra distance changes its phase. With the right coating thickness, the peaks of one reflected wave line up with the troughs of the other, so much of the reflected light cancels while most of the light continues through the glass.
For a simple single-layer coating, a common starting point is a quarter-wave optical thickness:
d = λ / 4n
Here, d is the coating thickness, λ is the wavelength of light being targeted, and n is the refractive index of the coating material. In simple terms, the coating thickness is chosen so that the extra distance travelled by the second reflection puts the two reflected waves at roughly opposite points in their cycles.
For visible light, that usually means layers only tens or hundreds of nanometres thick. A human hair is tens of thousands of nanometres across, so these coatings are extremely thin, yet optically they can have a large effect.
Why the reflection has colour
A coating cannot suppress reflection equally across every visible wavelength and every angle. What remains is residual reflection, and that often appears coloured. Depending on the design, you may see purple, green, blue, gold, bronze or brown.
The colour can also change as you tilt the filter because the optical path through the coating changes with angle. This is the same general thin-film interference that produces colours in soap bubbles and oil films, although optical coatings are carefully engineered rather than random.
The colour of the reflection can tell you something about how the coating behaves, but it does not directly tell you what colour the image will be. The reason is partly one of scale. The coating may be reflecting only a small percentage of the incoming light. If that small reflected portion contains noticeably more purple than green, for example, it can look vividly purple against the dark surface of a lens even though the much larger amount of light passing through remains very close to neutral.
A coating can still contribute to colour shift if its transmission varies enough across the visible spectrum, so reflection and transmission are related. The important point is that the apparent colour of the reflection does not tell you how large that difference is in the transmitted light. When colour accuracy matters, the useful measurement is the transmission spectrum, which is why we evaluate filters spectrally rather than simply judging them by eye. We cover that in more detail in our ND filter quality article.
Why multilayer coatings get complicated quickly
Modern optical coatings usually use several layers with different refractive indices. Every boundary creates another small reflected wave, and the designer controls the thickness and order of those layers so the reflections interact in useful ways across a broader range of wavelengths and viewing angles.
This is why the order of the layers matters. A stack arranged as glass-A-B-C-D-air will generally behave differently from glass-D-C-B-A-air, even though the same materials are present. The phase relationships between the reflected waves are different.
It is also why coating count by itself is not a very useful specification. Claims such as “16-layer” or “28-layer coating” sound precise, but the number does not tell you how well the coating was designed, how accurately the layers were deposited, what wavelengths were optimized, or even exactly what the manufacturer is counting as a layer. A well-designed seven-layer coating can outperform a poorly designed twenty-layer coating, so the meaningful specification is the resulting optical performance rather than the layer count.
How the coatings are applied
Making these coatings is closer to manufacturing a computer chip than painting a surface. The glass is placed inside a vacuum chamber, and the coating material is turned into a vapour or a stream of energetic particles. Those atoms travel through the chamber and settle onto the glass, gradually building a layer that may be only a few dozen or a few hundred nanometres thick.
Different systems use different ways of creating that stream. In vacuum evaporation the material is heated until it evaporates. Electron-beam deposition uses a focused electron beam to heat the coating material. Sputtering knocks atoms from a target using energetic ions, while ion-assisted processes use additional ions to help produce denser, more durable layers. Whatever the method, the important part is controlling exactly how much material reaches the glass and keeping the thickness uniform across the entire surface.
That control matters because these layers are part of the optical design. Change their thickness enough and you change the wavelengths and angles at which the coating suppresses or reinforces reflections.
The same physics can work in reverse
Thin-film interference can also be used to make a surface highly reflective. A dielectric mirror, commonly used in lasers and other precision optical equipment, is made from multiple transparent layers arranged so that their small individual reflections reinforce each other rather than cancel. With enough carefully designed layers, reflectivity at the target wavelengths can become extremely high. The same underlying physics can therefore make one piece of glass transmit more light and another reflect almost all of it.
Nature has developed another solution to the reflection problem. The surface of a moth's eye is covered with nanostructures smaller than the wavelengths of visible light. These create a more gradual transition between the refractive index of air and that of the eye, reducing the light that is reflected away. For a nocturnal animal, that has a direct benefit: more of the limited available light enters the eye and reaches the light-sensitive structures inside, improving vision in very low light. The reduced reflection also makes the eyes less conspicuous to predators. Similar “moth-eye” surfaces are now being engineered for displays, solar cells and other optical devices.
What the colour actually tells you
A purple, green, blue, bronze or brown reflection tells you that the coating is interacting differently with different wavelengths and angles. It does not, by itself, tell you whether the filter is colour neutral, whether its ND rating is accurate, how much flare it produces, or whether one coating is better than another. Those properties have to be measured rather than inferred from the colour of the surface.
There is something slightly deceptive about optical coatings in that respect. The coloured reflection is the easiest part of the coating to see, but it is often the least useful part for judging how well the filter will actually perform. Most of the important work is happening in the light that you don't see reflected: the light that makes it through to the camera.
So two filters can flash completely different colours under the same desk lamp and still transmit almost identical colour to the camera. Conversely, two coatings that look similar can perform quite differently. The only reliable way to know is to measure what comes through the glass, which is considerably less fun than tilting filters around under a lamp, but much more useful.