3

It's said that light always travels in straight lines. Even when reflection happens, another light ray is generated by electrons in the mirror , which cancels out the incident light. Refraction is also caused by superposition of light made by electrons in the material and incident light.

But in diffraction, on passing light through small holes, the rays bend to cover a larger area, does it violate the statement that light always travel in a straight line? Or is it something else that's going on?

Does the light bend due to interaction with electrons in outer edge of hole?

P.s.: light also travels in a straight line when general relativity is considered. Especially in this case, the curvature of space could only affect the path minimally.

Shaurya Kad
  • 603

2 Answers2

16

This is no contradiction.

Ray optics (i.e. modeling light as straight lines) is an idealization. Or more precisely: It is a good approximation for small wavelengths $\lambda$.

For longer wavelengths you need to model light as waves. Wave optics correctly predicts diffraction and interference. And in the limit of very short wavelengths ($\lambda\to 0$) it reduces to ray optics. Then light rays emerge as the lines perpendicular to the wavefronts.

light rays and wavefronts
(image from Ray optics)

Thomas Fritsch
  • 42,352
4

Light rays do not necessarily propagate in straight lines; if the refractive index, $\nu=\nu(x,y,z)$ is spatially varying then so is the direction of rays even in classical ray optics at $\lambda \to 0$ limit. A most famous and very practical example of such curved ray propagation is the Luneburg lens, see, for example, https://en.wikipedia.org/wiki/Luneburg_lens.

In fact, the rays in a general inhomogeneous medium, denoting its position vector by $\mathbf r,$ are described according to the equation $$\frac{d}{ds}\left(\nu \frac{d\mathbf r}{ds}\right)=\nabla\nu \tag{1},$$ where $ds$ is the line element along the ray, see Born&Wolf Section 3.2.

Only if the medium is homogeneous, $\nu=\rm{const},$ then $\nabla \nu=0$ and (1) reduces to $\frac{d^2\mathbf r}{ds^2}=\mathbf 0$ which is solved by $\mathbf r = \mathbf a s+ \mathbf b,$ a straight line for some $\mathbf {a,b} =\rm{const}.$

Note too that diffraction can be described by classical ray optics very well, for example, by applying the law of Malus to the various diffraction orders separately, see Keller's Geometrical Theory of Diffraction, or Toraldo di Francia's parageometrical optics although this has gone out of fashion in the age computers, see Conifold's answer to https://hsm.stackexchange.com/questions/14078/whatever-happened-to-parageometrical-optics-of-diffraction.

hyportnex
  • 21,193