Skip to content

Light at an Interface

Calculate reflection and transmission between two homogeneous, lossless media. Change either refractive index, the incidence angle or polarization. The calculation uses the Fresnel equations, written in tangential-field (admittance) form; see Model and conventions.

How light divides at a boundary

Take a plane wave travelling through one medium with refractive index n1n_1 and let it hit a flat boundary with a second medium, n2n_2. Some light comes back, the rest goes through. This page asks two questions about that split: in which directions, and how much. The editor below answers both for any angle and polarization. The thin films, mirrors and cavities in the other experiments combine this boundary calculation with the phase each wave picks up between boundaries.

Directions

The fields on both sides have to fit together along the whole boundary at once. That only works if the wave crests line up along the surface: the spacing of crests along the boundary has to be the same on both sides. The reflected wave therefore leaves at the angle it arrived, and the transmitted wave bends so that

θr=θi,n1sin⁡θi=n2sin⁡θt\theta_r=\theta_i,\qquad n_1\sin\theta_i=n_2\sin\theta_t

Nothing about amplitudes is needed for this. It is a statement about phase.

How much

Amplitudes come from the boundary conditions: the tangential electric and magnetic fields are continuous. The answer depends on polarization. In TE (also called s) the electric field is perpendicular to the plane of incidence; in TM (p) it lies in that plane. The model section writes both cases with one optical admittance YY, which gives the amplitude ratio r=(Y1−Y2)/(Y1+Y2)r=(Y_1-Y_2)/(Y_1+Y_2) and the reflected fraction R=∣r∣2R=|r|^2. Roughly, YY says how much tangential magnetic field a given tangential electric field drives in that medium, at that angle and polarization. Reflection comes from the mismatch between the two sides.

The transmitted fraction is not simply ∣t∣2|t|^2. Power crossing the boundary depends on the medium and on how steeply the beam crosses it, which is what the admittance factor in T=Re⁡Y2Re⁡Y1∣t∣2T=\tfrac{\operatorname{Re}Y_2}{\operatorname{Re}Y_1}|t|^2 accounts for. With no absorption, R+T=1R+T=1.

Derivation from Maxwell's equations

Start from the macroscopic Maxwell equations for a linear, isotropic, nonmagnetic medium with no free charge. Fields vary as exp⁡(−iωt)\exp(-i\omega t), units are SI, ε(ω)\varepsilon(\omega) is the relative permittivity, k0=ω/ck_0=\omega/c is the vacuum wave number and Y0=ε0/μ0Y_0=\sqrt{\varepsilon_0/\mu_0} is the vacuum admittance.

∇×E=iωμ0H,∇⋅(εE)=0,∇×H=−iωε0εE,∇⋅H=0.\begin{aligned}\nabla\times\mathbf E&=i\omega\mu_0\mathbf H, & \nabla\cdot(\varepsilon\mathbf E)&=0,\\ \nabla\times\mathbf H&=-i\omega\varepsilon_0\varepsilon\mathbf E, & \nabla\cdot\mathbf H&=0.\end{aligned}

In a uniform region ε\varepsilon is constant, so ∇⋅E=0\nabla\cdot\mathbf E=0. Taking the curl of Faraday's law, substituting the Ampère–Maxwell law and using ∇×∇×E=∇(∇⋅E)−∇2E\nabla\times\nabla\times\mathbf E=\nabla(\nabla\cdot\mathbf E)-\nabla^2\mathbf E gives the Helmholtz equation, with n2=εn^2=\varepsilon:

∇2E+εk02E=0\nabla^2\mathbf E+\varepsilon k_0^2\mathbf E=0

Geometry. The boundary is the plane z=0z=0. Medium 1 fills z<0z<0, medium 2 fills z>0z>0, and xx–zz is the plane of incidence, so nothing depends on yy. "Forward" means travelling toward +z+z. Medium 2 has only the transmitted, forward wave, because nothing sends light back from z>0z>0. Both media are lossless here, so ε\varepsilon is real.

Plane waves and phase matching. In medium jj try exp⁡[ik0(kxx±qjz)]\exp[ik_0(k_x x\pm q_j z)], where kxk_x and qjq_j are the xx and zz components of the wave vector divided by k0k_0. The Helmholtz equation requires kx2+qj2=nj2k_x^2+q_j^2=n_j^2. Integrate Faraday's law around a thin loop straddling the boundary: as the loop's width shrinks, the magnetic flux through it goes to zero, so tangential E\mathbf E must agree on the two sides. The Ampère–Maxwell law does the same for tangential H\mathbf H when there is no surface current. These conditions must hold at every xx, so the incident, reflected and transmitted waves share the same kxk_x. For the incident wave at angle θ\theta (the θi\theta_i of the lecture), kx=n1sin⁡θk_x=n_1\sin\theta:

kx=n1sin⁡θ=n2sin⁡θt,qj=nj2−n12sin⁡2θ.\begin{aligned}k_x&=n_1\sin\theta=n_2\sin\theta_t,\\ q_j&=\sqrt{n_j^2-n_1^2\sin^2\theta}.\end{aligned}

The reflected wave has the same kxk_x and −q1-q_1, which is the law of reflection.

Admittance. For TE, E=Eyy^\mathbf E=E_y\hat{\mathbf y}, and the xx component of Faraday's law gives Hx=−1iωμ0∂zEyH_x=-\tfrac{1}{i\omega\mu_0}\partial_z E_y. For TM, H=Hyy^\mathbf H=H_y\hat{\mathbf y}, and the Ampère–Maxwell law gives Ex=−iωε0nj2∂zHyE_x=-\tfrac{i}{\omega\varepsilon_0 n_j^2}\partial_z H_y. Use ∂z=±ik0qj\partial_z=\pm ik_0q_j for the forward (+) and backward (−) waves. The constants that appear are k0/(ωμ0)=1/(cμ0)k_0/(\omega\mu_0)=1/(c\mu_0) in TE and ωε0/k0=cε0\omega\varepsilon_0/k_0=c\varepsilon_0 in TM; both equal Y0Y_0:

TE: Hx=∓Y0YjEy,Yj=qj,TM: Hy=±Y0YjEx,Yj=nj2/qj.\begin{aligned}\text{TE: }H_x&=\mp Y_0Y_jE_y, & Y_j&=q_j,\\ \text{TM: }H_y&=\pm Y_0Y_jE_x, & Y_j&=n_j^2/q_j.\end{aligned}

The backward wave has the opposite sign, because its qq is reversed. The overall minus in TE is common to every term and cancels below.

Amplitudes. Take unit incident tangential EE, reflected rr and transmitted tt. Continuity of tangential EE and HH at z=0z=0 gives

1+r=t,Y1(1−r)=Y2 t⇓r=Y1−Y2Y1+Y2,t=1+r.\begin{gathered}1+r=t,\qquad Y_1(1-r)=Y_2\,t\\ \Downarrow\\ r=\frac{Y_1-Y_2}{Y_1+Y_2},\qquad t=1+r.\end{gathered}

These are amplitudes of EyE_y or ExE_x, not of the full field vector. For TM, sources differ in how they orient the reflected field vector; relative to Byrnes's full-field convention, rr here has the opposite sign. Powers are unaffected.

Power. The time-averaged normal Poynting flux is ⟨Sz⟩=12Re⁡(ExHy∗−EyHx∗)\langle S_z\rangle=\tfrac12\operatorname{Re}(E_xH_y^*-E_yH_x^*). For a single forward wave in either polarization this is 12Y0Re⁡(Yj) ∣Et∣2\tfrac12Y_0\operatorname{Re}(Y_j)\,|E_t|^2, where EtE_t is the tangential amplitude. At z=0z=0 in medium 1 the total tangential field is 1+r1+r and the total tangential HH is proportional to Y1(1−r)Y_1(1-r), with Y1Y_1 real because medium 1 is lossless and θ<90∘\theta<90^\circ. Then Re⁡[(1+r)(1−r∗)]=1−∣r∣2\operatorname{Re}[(1+r)(1-r^*)]=1-|r|^2, because r−r∗r-r^* is imaginary. Dividing by the incident flux 12Y0Y1\tfrac12Y_0Y_1 gives

⟨Sz⟩1=12Y0Y1(1−∣r∣2),R=∣r∣2,T=Re⁡Y2Re⁡Y1 ∣t∣2.\begin{aligned}\langle S_z\rangle_1&=\tfrac12Y_0Y_1\left(1-|r|^2\right),\\ R&=|r|^2,\qquad T=\frac{\operatorname{Re}Y_2}{\operatorname{Re}Y_1}\,|t|^2.\end{aligned}

Tangential EE and HH are continuous, so ⟨Sz⟩\langle S_z\rangle is the same on both sides, which is R+T=1R+T=1.

Total internal reflection. When n2<n1sin⁡θn_2<n_1\sin\theta, q22<0q_2^2<0 and q2q_2 is imaginary. Choosing Im⁡q2>0\operatorname{Im}q_2>0 makes exp⁡(ik0q2z)\exp(ik_0q_2z) decay into medium 2 instead of growing. Then Y2Y_2 is imaginary in both polarizations, so Re⁡Y2=0\operatorname{Re}Y_2=0 and T=0T=0. With Y1Y_1 real and Y2Y_2 imaginary, the numerator of r=(Y1−Y2)/(Y1+Y2)r=(Y_1-Y_2)/(Y_1+Y_2) is the complex conjugate of the denominator, so ∣r∣=1|r|=1.

Three things to try

Straight on. At normal incidence both polarizations agree and

R=(n1−n2n1+n2)2R=\left(\frac{n_1-n_2}{n_1+n_2}\right)^2

For air and glass (n2=1.5n_2=1.5) that is 4%. The page opens at 45° instead, where the two polarizations already disagree: about 9.2% for TE and 0.85% for TM. Switch polarization in the Light panel and read RR on the Power and propagation tab, or compare both curves on Angle response.

Brewster angle. For TM light the reflection drops to zero at

tan⁡θB=n2n1\tan\theta_B=\frac{n_2}{n_1}

which is 56.3° for air to glass. At that angle the reflected and refracted directions are perpendicular. Load the Brewster example, then switch to TE: the reflection comes back. Near this angle, glare from water or a glass tabletop is mostly TE, which for a horizontal surface is horizontally polarized; polarizing sunglasses block that component.

Total internal reflection. Going from glass into air, Snell's law runs out of solutions once

sin⁡θc=n2n1\sin\theta_c=\frac{n_2}{n_1}

which is about 41.8°. Beyond it there is no propagating transmitted wave and R=1R=1. The field in the air does not vanish; it decays exponentially away from the boundary and carries no net power across it. The total internal reflection example sits at 60°. The amplitude rr there is a complex number with magnitude one: all the light returns, with a phase shift that differs between TE and TM. At 60° the Power and propagation tab shows r≈−0.10−0.99 ir\approx-0.10-0.99\,i for TE and r≈0.72+0.69 ir\approx0.72+0.69\,i for TM. The TM phase depends on the sign convention in the model section; the difference between the two does not.

UNIFORM LAYERS

Light at an interface

Loading 3D…

Loading geometry…

Physical scale
Drag to orbit · click a region to inspect

Analytic Fresnel solution · revision 0

Angle response
10.750.50.2500.00089.50Incidence angle (°)Fraction of incident normal power

R T

Sample values
Incidence angle (°)R (Fraction of incident normal power)T (Fraction of incident normal power)
0.00000.0400000.96000
3.00000.0401460.95985
6.00000.0405890.95941
9.00000.0413390.95866
12.0000.0424120.95759
15.0000.0438350.95616
18.0000.0456420.95436
21.0000.0478770.95212
24.0000.0505970.94940
27.0000.0538740.94613
30.0000.0577960.94220
33.0000.0624740.93753
36.0000.0680440.93196
39.0000.0746760.92532
42.0000.0825790.91742
45.0000.0920130.90799
48.0000.103300.89670
51.0000.116840.88316
54.0000.133130.86687
57.0000.152790.84721
60.0000.176570.82343
63.0000.205430.79457
66.0000.240540.75946
69.0000.283330.71667
72.0000.335560.66444
75.0000.399360.60064
78.0000.477290.52271
81.0000.572430.42757
84.0000.688370.31163
87.0000.829300.17070

Model and conventions

The plane of incidence is xx–zz, with +z+z pointing into the sample and time dependence exp⁡(−iωt)\exp(-i\omega t). TE uses EyE_y; TM uses tangential ExE_x. For incidence angle θ\theta in medium 1, the normal wave number (divided by k0=2π/λk_0=2\pi/\lambda) and the optical admittance of medium jj are

qj=nj2−n12sin⁡2θYj={qjTEnj2/qjTM\begin{aligned}q_j&=\sqrt{n_j^2-n_1^2\sin^2\theta}\\ Y_j&=\begin{cases}q_j & \text{TE}\\ n_j^2/q_j & \text{TM}\end{cases}\end{aligned}

YjY_j equals Hy/ExH_y/E_x for a forward TM wave and −Hx/Ey-H_x/E_y for a forward TE wave, in units of the vacuum admittance. When n2<n1sin⁡θn_2<n_1\sin\theta, q2q_2 is imaginary; the root with Im⁡q2>0\operatorname{Im}q_2>0 is used, so the field decays away from the boundary. The amplitude coefficients, and the fractions of incident power flowing through the boundary (the normal component of the Poynting vector), are

r=Y1−Y2Y1+Y2,t=1+r,R=∣r∣2,T=Re⁡Y2Re⁡Y1 ∣t∣2\begin{aligned}r&=\frac{Y_1-Y_2}{Y_1+Y_2}, & t&=1+r,\\ R&=|r|^2, & T&=\frac{\operatorname{Re}Y_2}{\operatorname{Re}Y_1}\,|t|^2\end{aligned}

This tangential-field convention changes the sign of TM reflection relative to some textbook polarization bases. Powers agree. There is no absorption, roughness, finite beam or wavelength dispersion. Wavelength therefore changes only the spatial scale of the field volume, not RR or TT. The analytic normal-incidence air/glass check gives R=0.04R = 0.04 and T=0.96T = 0.96.

History of the method

The sine law of refraction was found more than once. Ibn Sahl stated it in a treatise on burning instruments written around 984, according to Roshdi Rashed's analysis. Thomas Harriot knew it by 1602 but did not publish it, Willebrord Snell wrote it down in a 1621 manuscript, and René Descartes put it into wide circulation in 1637. Kwan, Dudley and Lantz give a short account of that sequence.

Étienne-Louis Malus had found that light reflected from water at a certain angle is completely polarized, and that the angle differs between materials, but he did not find the rule linking it to refractive index. David Brewster did, in a paper read to the Royal Society on 16 March 1815: at the polarizing angle, the tangent of the angle of incidence equals the refractive index, and the reflected and refracted rays form a right angle. The same rule held for internal reflection and for the boundary between two media.

In 1823 Augustin-Jean Fresnel presented his work on how reflection changes polarized light. According to Karam, it contains what is arguably the first physical interpretation of complex numbers, which clarified elliptical and circular polarization. The reflection and transmission amplitude equations now carry Fresnel's name.

The admittance notation used on this page is the usual thin-film form. Byrnes's derivation uses full field amplitudes instead; the tangential amplitudes here follow from his p-polarized results by projecting onto xx, which flips the sign of TM rr and scales tt by cos⁡θ2/cos⁡θ1\cos\theta_2/\cos\theta_1. Powers are the same in both.

References

Steven J. Byrnes, Multilayer optical calculations (2016; revision 5, 2020), sections 2–4 and appendices A and D: field conventions, Fresnel amplitudes in full-field form, transmitted power, p-polarization sign conventions and the branch choice for total internal reflection. A modern derivation written alongside the tmm Python package; cited for conventions, not for the origin of the Fresnel equations. The interactive editor and numerical checks here are a later browser implementation.

Alistair Kwan, John Dudley and Eric Lantz, Who really discovered Snell's law?, Physics World 15(4), 64 (April 2002; one-page column, subscriber access at IOP): the chronology of Ibn Sahl, Harriot, Snell and Descartes. The Ibn Sahl attribution rests on Roshdi Rashed, A pioneer in anaclastics: Ibn Sahl on burning mirrors and lenses, Isis 81(3), 464–491 (1990), which is cited here through Kwan et al.; its full text was not consulted.

David Brewster, On the laws which regulate the polarisation of light by reflexion from transparent bodies, Philosophical Transactions of the Royal Society of London 105, 125–159 (1815). The statements above follow the abstract printed in the Society's Abstracts of the Papers Printed in the Philosophical Transactions 2, 14–15 (doi:10.1098/rspl.1815.0009), openly available on the Internet Archive.

Ricardo Karam, Fresnel's original interpretation of complex numbers in 19th century optics, American Journal of Physics 86(4), 245–249 (2018): the source for the 1823 date and the physical interpretation of complex numbers. Only the abstract was available when this page was written; the article itself is paywalled.

Calculation source code · Implementation and validation notes