Skip to content

Fresnel Equations

When light strikes a planar interface between media with refractive indices n1n_1 and n2n_2The Amplitudes of the reflected and transmitted waves depend on the polarisation.

For an incident wave with amplitude EiE_iThe reflection and transmission coefficients are:

s-polarisation (perpendicular to the plane of incidence):

rs=n1cosθin2cosθtn1cosθi+n2cosθt,ts=2n1cosθin1cosθi+n2cosθtr_s = \frac{n_1\cos\theta_i - n_2\cos\theta_t}{n_1\cos\theta_i + n_2\cos\theta_t}, \quad t_s = \frac{2n_1\cos\theta_i}{n_1\cos\theta_i + n_2\cos\theta_t}

p-polarisation (parallel to the plane of incidence):

rp=n2cosθin1cosθtn2cosθi+n1cosθt,tp=2n1cosθin2cosθi+n1cosθtr_p = \frac{n_2\cos\theta_i - n_1\cos\theta_t}{n_2\cos\theta_i + n_1\cos\theta_t}, \quad t_p = \frac{2n_1\cos\theta_i}{n_2\cos\theta_i + n_1\cos\theta_t}

Reflectance and transmittance (energy fractions):

R=r2,T=n2cosθtn1cosθit2R = |r|^2, \quad T = \frac{n_2\cos\theta_t}{n_1\cos\theta_i}|t|^2

With R+T=1R + T = 1 (energy conservation).

At the Brewster angle θB\theta_BThe reflected beam for p-polarised light has zero amplitude: rp=0r_p = 0:

tanθB=n2n1\tan\theta_B = \frac{n_2}{n_1}

Proof. Setting rp=0r_p = 0: n2cosθi=n1cosθtn_2\cos\theta_i = n_1\cos\theta_t. Using Snell’s law n1sinθi=n2sinθtn_1\sin\theta_i = n_2\sin\theta_t:

cosθisinθi=cosθtsinθt\frac{\cos\theta_i}{\sin\theta_i} = \frac{\cos\theta_t}{\sin\theta_t}

cotθi=cotθt    θi+θt=90\cot\theta_i = \cot\theta_t \implies \theta_i + \theta_t = 90^\circ

So tanθi=tanθB=n2/n1\tan\theta_i = \tan\theta_B = n_2/n_1. \blacksquare

At Brewster’s angle, the reflected and refracted beams are perpendicular. This is why polarising Filters work at specific angles for reflected glare.

10.3 Total Internal Reflection and the Evanescent Wave

Section titled “10.3 Total Internal Reflection and the Evanescent Wave”

When n1>n2n_1 \gt n_2 and θi>θc=arcsin(n2/n1)\theta_i \gt \theta_c = \arcsin(n_2/n_1), sinθt>1\sin\theta_t \gt 1So cosθt=isin2θt1\cos\theta_t = i\sqrt{\sin^2\theta_t - 1} becomes imaginary.

The transmitted field becomes an evanescent wave:

Eteκxei(kzzωt)E_t \propto e^{-\kappa x}\, e^{i(k_z z - \omega t)}

Where κ=k0n12sin2θin22\kappa = k_0\sqrt{n_1^2\sin^2\theta_i - n_2^2} and kz=k0n1sinθik_z = k_0 n_1\sin\theta_i.

The field decays exponentially with penetration depth δ=1/κ\delta = 1/\kappa but propagates along the Interface. No energy is transported into the second medium: R=1R = 1.

Frustrated total internal reflection. If a third medium is brought within a few wavelengths of The interface, energy can tunnel across the gap (analogous to quantum tunnelling).

The Fresnel coefficients are real for θi<θc\theta_i < \theta_c (normal incidence/transmission) and may be positive or negative, indicating phase shifts:

  • External reflection (n1<n2n_1 < n_2): rs<0r_s < 0 for all θi\theta_i (phase shift of π\pi for s-polarisation). rpr_p changes sign at Brewster’s angle.
  • Internal reflection (n1>n2n_1 > n_2): For θi<θc\theta_i < \theta_c, both rsr_s and rpr_p are positive at normal incidence. rpr_p changes sign at Brewster’s angle.

At normal incidence (θi=0\theta_i = 0):

rs=rp=n1n2n1+n2r_s = r_p = \frac{n_1 - n_2}{n_1 + n_2}

The reflection coefficient is negative when n1<n2n_1 < n_2, corresponding to a π\pi phase shift. For n1>n2n_1 > n_2, the reflection coefficient is positive (no phase shift).

The reflectance RR varies with angle of incidence:

  • For s-polarisation, RsR_s increases monotonically from ((n1n2)/(n1+n2))2((n_1 - n_2)/(n_1 + n_2))^2 at normal incidence to 11 at grazing incidence.
  • For p-polarisation, RpR_p drops to 00 at Brewster’s angle, then increases to 11 at grazing incidence.

Anti-reflection coatings use destructive interference between reflections from two interfaces. For a single-layer coating of index ncn_c and thickness λ/4\lambda/4 on glass (ngn_g), the reflectance at wavelength λ\lambda is:

R=(nc2ngnc2+ng)2R = \left(\frac{n_c^2 - n_g}{n_c^2 + n_g}\right)^2

The reflectance is zero when nc=ngn_c = \sqrt{n_g}. For crown glass (ng=1.52n_g = 1.52), the optimal coating index is nc1.23n_c \approx 1.23, approximated by magnesium fluoride (n1.38n \approx 1.38), giving R1%R \approx 1\% per surface.

In total internal reflection, the reflected beam is laterally shifted relative to the geometrically predicted path. This Goos-Hanchen shift arises because the evanescent wave penetrates the second medium before being reflected:

D=λπsinθisin2θi(n2/n1)2D = \frac{\lambda}{\pi} \frac{\sin\theta_i}{\sqrt{\sin^2\theta_i - (n_2/n_1)^2}}

The shift is of order one wavelength for angles near the critical angle and decreases as θi\theta_i increases beyond θc\theta_c.

10.7 Worked Example: Reflectance at Normal Incidence

Section titled “10.7 Worked Example: Reflectance at Normal Incidence”

Problem. Calculate the reflectance of uncoated glass (ng=1.52n_g = 1.52) at normal incidence in air.

Solution

At normal incidence, r=(11.52)/(1+1.52)=0.52/2.520.206r = (1 - 1.52)/(1 + 1.52) = -0.52/2.52 \approx -0.206. The reflectance is R=r20.0425R = |r|^2 \approx 0.0425, or about 4.25%4.25\% per surface. For a lens with two surfaces, total transmission through uncoated glass is approximately T=(10.0425)20.917T = (1 - 0.0425)^2 \approx 0.917, meaning about 8.3%8.3\% of incident light is lost to reflections.

\blacksquare

Problem. Find the phase difference between s- and p-polarised components after total internal reflection in glass (n1=1.5n_1 = 1.5) at θi=60\theta_i = 60^\circ with n2=1n_2 = 1.

Solution

From the Fresnel equations with complex cosθt\cos\theta_t:

rs=cosθiisin2θi(n2/n1)2cosθi+isin2θi(n2/n1)2=eiδsr_s = \frac{\cos\theta_i - i\sqrt{\sin^2\theta_i - (n_2/n_1)^2}}{\cos\theta_i + i\sqrt{\sin^2\theta_i - (n_2/n_1)^2}} = e^{i\delta_s}

rp=(n2/n1)2cosθiisin2θi(n2/n1)2(n2/n1)2cosθi+isin2θi(n2/n1)2=eiδpr_p = \frac{(n_2/n_1)^2\cos\theta_i - i\sqrt{\sin^2\theta_i - (n_2/n_1)^2}}{(n_2/n_1)^2\cos\theta_i + i\sqrt{\sin^2\theta_i - (n_2/n_1)^2}} = e^{i\delta_p}

where δs=2arctan(sin2θi(n2/n1)2/cosθi)\delta_s = -2\arctan(\sqrt{\sin^2\theta_i - (n_2/n_1)^2}/\cos\theta_i) and δp=2arctan(sin2θi(n2/n1)2/((n2/n1)2cosθi))\delta_p = -2\arctan(\sqrt{\sin^2\theta_i - (n_2/n_1)^2}/((n_2/n_1)^2\cos\theta_i)).

For n1=1.5n_1 = 1.5, n2=1n_2 = 1, θi=60\theta_i = 60^\circ: (n2/n1)20.444(n_2/n_1)^2 \approx 0.444, sin260=0.75\sin^2 60^\circ = 0.75, so sin2θi(n2/n1)20.750.4440.553\sqrt{\sin^2\theta_i - (n_2/n_1)^2} \approx \sqrt{0.75 - 0.444} \approx 0.553, cos60=0.5\cos 60^\circ = 0.5.

δs=2arctan(0.553/0.5)=2arctan(1.106)95.9\delta_s = -2\arctan(0.553/0.5) = -2\arctan(1.106) \approx -95.9^\circ

δp=2arctan(0.553/(0.4440.5))=2arctan(2.491)136.2\delta_p = -2\arctan(0.553/(0.444 \cdot 0.5)) = -2\arctan(2.491) \approx -136.2^\circ

The relative phase difference Δ=δpδs40.3\Delta = \delta_p - \delta_s \approx -40.3^\circ, which is why TIR can convert linear to elliptical polarisation (the basis of Fresnel rhomb quarter-wave plates).

\blacksquare

At θi=θc=arcsin(n2/n1)\theta_i = \theta_c = \arcsin(n_2/n_1), we have θt=90\theta_t = 90^\circ and cosθt=0\cos\theta_t = 0. The Fresnel coefficients become:

rs=n1cosθc0n1cosθc+0=1r_s = \frac{n_1\cos\theta_c - 0}{n_1\cos\theta_c + 0} = 1

rp=n2cosθc0n2cosθc+0=1r_p = \frac{n_2\cos\theta_c - 0}{n_2\cos\theta_c + 0} = 1

Both polarisations have R=1R = 1 at the critical angle, and the transmitted wave propagates exactly along the interface with no energy flow into the second medium.