# How does the temperature of water affect its refractive index?

I am aware this question has been asked many a time but i have never found a answer which satisfied my exact requirements. I recently completed an experiment on how refractive index is affected by temperature. I found that as temperature increases, refractive index decreases. I thought this was due to a density change and the fact there are less molecules that the light ray is absorbed and remitted by. However my teacher believed that i should look further into optical density. I know this is the sluggish tendency of the molecules but i don't know how to link these two definitions to give a correct scientific background context. I was also told to look into relative permittivity. If someone could give me a push in the right direction i would be very grateful.

• I would have thought that over the relatively narrow temperature range of liquid water that the change in the refractive index is mostly just due to density changes, too. Actually, it's probably the permittivity of water that scales linearly with the density of water, and the index of refraction n of water goes as the square root of the permittivity, so n should scale with the square root of the density of water. Why don't you see if your measured n versus T data agrees with that model? – Samuel Weir Jul 5 '18 at 22:44
• According to [iopscience.iop.org/article/10.1088/0022-3727/11/8/007], the temperature dependence is exponential in the range from -3 to +80 C. – S. McGrew Jul 6 '18 at 14:05
• @S.McGrew - I don't think that what they describe is really exponential behavior. They show that the dn/dT of their data can be fit to a function that includes an exponential in it, but if you look at their plot of dn/dT points it could just as well be fit to a low-order polynomial. Also, if you look at the average value of dn/dT, it's very small, just around -10^-4, which means that over a 100 ˚C interval from ice to steam, the index of refraction changes about just around 0.01 or so. (Too bad they didn't present their original data in the form of a simple n versus T plot or table.) – Samuel Weir Jul 6 '18 at 17:45
• I reckon you are probably right. Exponential wouldn't seem likely. – S. McGrew Jul 6 '18 at 18:01