Physics › Section S · Vision, instruments, and wave optics › Topic S.6
Thin-film interference
The swirling colours of a soap bubble and the rainbow sheen of oil on a wet road come from light reflecting off the two faces of a very thin layer. The same physics, run in reverse, lets engineers coat lenses and solar cells so that they hardly reflect at all.
Colours from a layer thinner than a hair
A soap bubble's wall is only a few hundred nanometres thick, comparable to the wavelength of visible light. When light falls on it, part reflects from the front surface and part enters the film, reflects from the back surface and comes out again. These two reflected waves overlap and interfere. Whether they reinforce or cancel depends on the film's thickness, its refractive index and the wavelength, so in white light some colours are reflected strongly and others hardly at all. As the thickness varies, the colour changes, giving the familiar bands. Films many wavelengths thick show no colours, because the conditions for different wavelengths overlap.
The extra path inside the film
Take light arriving almost perpendicular to a film of thickness \(t\). The wave reflected from the back surface crosses the film twice, so it travels an extra distance of about \(2t\). That extra distance is inside the film, where the wavelength is shorter. As the earlier topic on light as a wave showed, in a medium of refractive index \(n\) the wavelength becomes
where \(\lambda\) is the wavelength in a vacuum (air is close enough). So what matters is how many film wavelengths fit into \(2t\). If that were the only difference, a whole number of \(\lambda_n\) would give constructive interference. But there is a second effect.
The half-wavelength phase change on reflection
When light travelling in one medium reflects from a boundary with a medium of higher refractive index, the reflected wave is inverted: it gains a phase change of half a cycle, which is the same as shifting it by half a wavelength. When light reflects from a boundary with a lower refractive index, there is no phase change. Light transmitted through a boundary never gains this shift. A rope gives the picture: a pulse on a light rope that meets a heavier rope comes back upside down, while a pulse on a heavy rope meeting a lighter one comes back upright. For thin films, only the relative shift between the two reflected waves matters, so count how many of the two reflections are from a higher index.
Putting the two effects together
If neither reflection gains the half-wavelength shift, or both do, the shifts make no difference, and for near-normal incidence
If exactly one reflection is shifted, the two conditions swap:
In both cases \(m = 0, 1, 2, \ldots\) and \(\lambda_n = \lambda/n_{\text{film}}\). The swap explains a striking sight. A soap film in air has one shifted reflection, so where it is extremely thin, just before it bursts, the path difference is almost zero and every colour cancels: the film looks black.
Worked example: a soap film in white light
A soap film, \(n = 1.33\), is 300 nm thick and has air on both sides. Which visible wavelength does it reflect most strongly? 1. Count shifts: air to soap is to a higher index (shift); soap to air at the back is to a lower index (no shift). Exactly one shift. 2. Bright condition: \(2t = (m + \tfrac{1}{2})\lambda/n\), so
3. \(m = 0\) gives 1596 nm (infrared); \(m = 1\) gives 532 nm (green); \(m = 2\) gives 319 nm (ultraviolet). 4. So this part of the film looks green. 5. Check with the dark condition, \(\lambda = 798/m\): 399 nm, at the violet edge, is cancelled, consistent with green dominating.
Oil on water
A thin layer of oil on a wet road is a film between air and water. Typical oils have a refractive index near 1.45, higher than water's 1.33. Air to oil is to a higher index, so that reflection is shifted; oil to water is to a lower index, so that one is not. As with soap, exactly one reflection is shifted. The oil spreads unevenly, and each thickness reflects a different colour strongly, giving the rainbow swirls. If instead the film's index lay between those of the media on either side, both reflections would be shifted and the conditions would not swap.
Worked example: the thinnest bright oil film
Oil of refractive index 1.45 floats on water. Find the two smallest oil thicknesses that strongly reflect orange light of wavelength 600 nm at near-normal incidence. 1. One reflection is shifted (air to oil), so bright needs \(2t = (m + \tfrac{1}{2})\lambda_n\). 2. Wavelength in the oil: \(\lambda_n = 600/1.45 = 413.8\) nm. 3. For \(m = 0\): \(t = \lambda_n/4 = 103\) nm. 4. For \(m = 1\): \(t = 3\lambda_n/4 = 310\) nm. 5. Check: successive bright thicknesses differ by \(\lambda_n/2 = 207\) nm, and \(310 - 103 = 207\). Using 600 nm instead of the film wavelength would have given 150 nm, a common slip.
Anti-reflection coatings
A bare glass surface reflects about 4% of the light falling on it, and a camera with many lenses loses light and gains glare at every surface. The fix is a thin coating whose refractive index lies between air and glass. Now both reflections are from a higher index, so both are shifted and the shifts cancel. For the two reflected waves to cancel, the extra path must be half a film wavelength, \(2t = \lambda_n/2\), giving the thinnest coating
a quarter of a wavelength in the film. Cancellation is complete only for the design wavelength, usually mid-green, so a coated lens reflects a little red and blue and looks purplish.
Worked example: coating a solar cell
Silicon solar cells (refractive index about 3.9) are coated with silicon nitride, \(n = 2.00\), to reduce reflection. What is the thinnest coating that cancels the reflection of 600 nm light? 1. Count shifts: air to nitride, higher index (shift); nitride to silicon, higher again (shift). Two shifts, which cancel. 2. Dark condition: \(2t = (m + \tfrac{1}{2})\lambda_n\), thinnest with \(m = 0\). 3. So
4. Check: the film wavelength is 300 nm, and the round trip of 150 nm is exactly half of it. Such cells reflect mostly blue, which is why many look dark blue.
Common mistakes to avoid
First, using the vacuum wavelength inside the film instead of \(\lambda/n\). Second, putting the phase change on the wrong reflection: the shift happens when light reflects from a higher index, not when it enters one, and transmitted light is never shifted. Third, forgetting to count the shifts at all, which swaps every bright and dark answer. Fourth, using \(t\) instead of \(2t\) for the extra path. Fifth, using the film's index to decide the shifts but forgetting the index of what lies behind it, as with oil on water. Sixth, assuming a coating cancels all colours equally; it is tuned to one wavelength.
Key takeaways
Thin-film colours come from interference between light reflected at the front and back surfaces. At near-normal incidence the back reflection travels an extra 2t, measured in film wavelengths λ/n. Reflection from a higher index adds a half-wavelength shift; reflection from a lower index adds none. With zero or two shifts, bright needs 2t = mλₙ; with one shift, bright needs 2t = (m + ½)λₙ and the conditions swap. Soap films and oil on water have one shift. An anti-reflection coating with an index between air and glass has two, and its thinnest thickness is λ/4n.
Flashcards
Flashcard
Interference between light reflected from the front and back of a thin layer
Test your understanding
Question 1 of 5
Light travelling in water (n = 1.33) reflects from a layer of air above it. What phase change does the reflected light undergo?
Question 2 of 5
A lens is given a coating of index 1.30 (less than the glass) to cancel reflection of 520 nm light. What is the thinnest coating that works?
Question 3 of 5
A soap film (n = 1.33) has air on both sides. What is the thinnest film that strongly reflects light of wavelength 532 nm at near-normal incidence?
Question 4 of 5
Just before it bursts, part of a soap film becomes much thinner than a wavelength. How does that part look in reflected light?
Question 5 of 5
A film of oil (n = 1.50) floats on water (n = 1.33). For near-normal light, which describes its reflections?
