Physics · Book 3 · Bachelor Year 1

University Physics — Year 1

University Physics — Year 1 · Bachelor Year 1

2Geometric Optics: Rays, Reflection, Refraction

A straw standing in a glass of water looks broken at the surface; a swimmer looking up from the bottom of a pool sees the whole sky squeezed into a bright disk, ringed by a mirror of the pool floor; a hair-thin glass fiber carries a telephone call across an ocean with less loss than a copper wire suffers across a street. Three laws — light travels straight, reflects at equal angles, bends by Snell’s rule — account for all of it, and this chapter states them precisely, marks where they stop being true, and puts them to work.

A prism bends a white beam toward its base and spreads it: the index of glass depends on the wavelength, and violet is deviated most.
A prism bends a white beam toward its base and spreads it: the index of glass depends on the wavelength, and violet is deviated most.

2.1 The geometric-optics model

Definition 2.1 (Light ray; homogeneous medium)

In the model of geometric optics, light propagates along light rays: oriented curves that are straight lines in a homogeneous medium (same properties at every point) and that carry energy independently of one another. A beam is a bundle of rays; a point source emits rays in every direction.

Definition 2.2 (Refractive index)

A transparent medium lets light through; light travels in it at a speed v<cv < c. Its refractive index is

n=cv1.n = \frac{c}{v} \geq 1 .

Vacuum: n=1n = 1; air: 1.00031.0003; water: 1.331.33; ordinary glass: 1.51.5 to 1.61.6; diamond: 2.422.42. A wave of frequency ν\nu keeps its frequency when it enters a medium, so its wavelength shrinks from λ0=c/ν\lambda_0 = c/\nu in vacuum to λ=λ0/n\lambda = \lambda_0/n.

Remark 2.3 (Dispersion)

In matter the index depends slightly on the wavelength — the medium is dispersive. For glass in the visible, Cauchy’s empirical law n(λ)=A+B/λ2n(\lambda) = A + B/\lambda^2 fits well: nn is larger for blue than for red by about 0.010.01 to 0.020.02. The prism and the rainbow, at the end of this chapter, live on that difference.

Remark 2.4 (Limits of the model)

Light is a wave, and a wave passing through an opening of width aa spreads by an angle of order λ/a\lambda/a (diffraction, Chapter 5). Rays are a good description as long as every aperture, lens and obstacle is much larger than the wavelength (0.40.4\, to 0.8µm0.8\,\text{µ}\mathrm{m} for visible light): a 1mm1\,\mathrm{mm} hole spreads a beam by 10310^{-3} radians, negligible on a bench; a 1µm1\,\text{µ}\mathrm{m} hole spreads it over the whole half-space, and no ray picture survives. The 9µm9\,\text{µ}\mathrm{m} core of a single-mode telecom fiber is already beyond the model.

Proposition 2.5 (Return of light)

The path of a ray does not depend on the direction in which the light travels it: if light goes from AA to BB along a path, light from BB follows the same path to AA.

Proof. Admitted at this level.

2.2 Reflection and refraction

Definition 2.6 (Incidence)

A ray meeting the surface separating two media (a diopter) at a point II makes with the normal to the surface at II the angle of incidence i1i_1. The plane of incidence contains the incident ray and the normal.

Theorem 2.7 (Laws of reflection and refraction (Snell–Descartes))

At the surface between media of indices n1n_1 (incident side) and n2n_2:

  1. the reflected and refracted rays lie in the plane of incidence;
  2. the reflected ray makes with the normal the angle i1=i1i_1' = i_1, on the other side of the normal;
  3. the refracted ray makes with the normal an angle i2i_2 such that

    n1sini1=n2sini2.n_1 \sin i_1 = n_2 \sin i_2 .

Proof. Admitted at this level.

Remark 2.8 (Where the laws come from)

Both laws follow from Fermat’s principle: between two points, light follows the path of least travel time. Reflection at equal angles is the shortest broken path touching the mirror; Snell’s law is the path that trades a longer route in the fast medium against a shorter one in the slow medium (Exercise 2.12). In the Year 2 volume both laws are derived from the wave nature of light at an interface.

Reflection and refraction at the surface between two media. The refracted ray bends toward the normal when n_2 > n_1; all three rays and the normal lie in one plane, the plane of the figure.
Reflection and refraction at the surface between two media. The refracted ray bends toward the normal when n2>n1n_2 > n_1; all three rays and the normal lie in one plane, the plane of the figure.

Proposition 2.9 (Which way the ray bends)

Entering a more refractive medium (n2>n1n_2 > n_1), a ray bends toward the normal (i2<i1i_2 < i_1); entering a less refractive one, away from it (i2>i1i_2 > i_1). A ray along the normal goes through undeviated.

Proof. sini2=(n1/n2)sini1\sin i_2 = (n_1/n_2)\sin i_1 with sin\sin increasing on [0,π/2][0, \pi/2].

Example 2.10 (Apparent depth)

A coin lies at depth hh under water (n=1.33n = 1.33), seen from almost straight above. A ray from the coin reaching the surface at small angle i2i_2 (in water) leaves at i1i_1 with sini1=nsini2\sin i_1 = n\sin i_2; for small angles i1ni2i_1 \approx n\,i_2. The eye extends the emerging ray back: it seems to come from a depth hh' with tani1x/h\tan i_1 \approx x/h' and tani2x/h\tan i_2 \approx x/h for the same surface point at distance xx from the vertical, so h=hi2/i1=h/nh' = h\,i_2/i_1 = h/n. A pool 2.0m2.0\,\mathrm{m} deep looks 1.5m1.5\,\mathrm{m} deep; a fish sees the fisherman 1.331.33 times higher than he is.

Theorem 2.11 (Total internal reflection)

Light going from a medium of index n1n_1 toward a less refractive one (n2<n1n_2 < n_1) is refracted only if i1ici_1 \leq i_c, where the critical angle satisfies

sinic=n2n1.\sin i_c = \frac{n_2}{n_1} .

For i1>ici_1 > i_c there is no refracted ray: the surface reflects all the light — total internal reflection.

Proof. Snell’s law asks for sini2=(n1/n2)sini1>sini1\sin i_2 = (n_1/n_2)\sin i_1 > \sin i_1; this exceeds 11 as soon as sini1>n2/n1\sin i_1 > n_2/n_1, and then no angle i2i_2 exists. At i1=ici_1 = i_c the refracted ray grazes the surface (i2=90i_2 = 90^\circ). That all the energy is then reflected is a statement about intensities, which geometric optics does not handle; the wave treatment of the Year 2 volume confirms it.

Rays from a source S under water. Below the critical angle (i_c = 48.8 for water–air) part of the light leaves; at i_c the refracted ray grazes the surface; beyond, the surface is a perfect mirror — the swimmer’s bright disk of sky is bounded by that angle.
Rays from a source SS under water. Below the critical angle (ic=48.8i_c = 48.8^\circ for water–air) part of the light leaves; at ici_c the refracted ray grazes the surface; beyond, the surface is a perfect mirror — the swimmer’s bright disk of sky is bounded by that angle.

Example 2.12 (Diamonds and swimmers)

Glass–air: sinic=1/1.5\sin i_c = 1/1.5, ic=41.8i_c = 41.8^\circ, which is why a 4545^\circ prism reflects totally and replaces a mirror in binoculars. Diamond–air: ic=24.4i_c = 24.4^\circ; a cut diamond traps light that enters its top and sends almost all of it back out through the top — the brilliance is total reflection, the fire is dispersion. Water–air: ic=48.8i_c = 48.8^\circ; looking up, a diver sees the entire sky inside a cone of half-angle 48.848.8^\circ (“Snell’s window”) and the reflected pool floor outside it.

2.3 The optical fiber

Definition 2.13 (Step-index fiber)

A step-index optical fiber is a cylindrical glass core of index n1n_1 sheathed in a cladding of slightly lower index n2n_2. Light entering the core within a cone of half-angle θa\theta_a around the axis is guided by successive total internal reflections at the core–cladding surface. The numerical aperture is NA=sinθa\mathrm{NA} = \sin\theta_a.

Proposition 2.14 (Acceptance cone and intermodal spread)

For a fiber whose entry face is in air,

sinθa=n12n22.\sin\theta_a = \sqrt{n_1^2 - n_2^2} .

Over a length LL, the guided ray along the axis and the steepest guided ray arrive with a delay

Δt=Ln1c(n1n21)\Delta t = \frac{L\,n_1}{c}\left(\frac{n_1}{n_2} - 1\right)

between them (intermodal dispersion).

Proof. A ray entering at angle θ\theta to the axis is refracted to θ\theta' with sinθ=n1sinθ\sin\theta = n_1\sin\theta' and meets the cladding at incidence i=90θi = 90^\circ - \theta'. Guiding requires sinin2/n1\sin i \geq n_2/n_1, i.e. cosθn2/n1\cos\theta' \geq n_2/n_1, i.e. sinθ1n22/n12\sin\theta' \leq \sqrt{1 - n_2^2/n_1^2}; multiply by n1n_1. The axial ray travels LL at speed c/n1c/n_1: t0=Ln1/ct_0 = Ln_1/c. The steepest ray zigzags at angle θmax\theta'_{\max} with cosθmax=n2/n1\cos\theta'_{\max} = n_2/n_1; its path is L/cosθmax=Ln1/n2L/\cos\theta'_{\max} = L n_1/n_2, so t1=Ln12/(n2c)t_1 = L n_1^2/(n_2 c), and Δt=t1t0\Delta t = t_1 - t_0.

A step-index fiber. A ray entering within the acceptance cone of half-angle _a meets the cladding at incidence i ≥ i_c and is guided by total reflections; the axial ray takes the shortest route, the steepest guided ray the longest — a pulse spreads in time.
A step-index fiber. A ray entering within the acceptance cone of half-angle θa\theta_a meets the cladding at incidence iici \geq i_c and is guided by total reflections; the axial ray takes the shortest route, the steepest guided ray the longest — a pulse spreads in time.

Example 2.15 (A telecom fiber’s numbers)

n1=1.480n_1 = 1.480, n2=1.465n_2 = 1.465: NA=2.1902.146=0.21\mathrm{NA} = \sqrt{2.190 - 2.146} = 0.21, θa=12\theta_a = 12^\circ; ic=81.8i_c = 81.8^\circ, so guided rays stay within 8.28.2^\circ of the axis. Intermodal spread: Δt/L=(1.48/c)(1.48/1.4651)=51ns/km\Delta t/L = (1.48/c)(1.48/1.465 - 1) = 51\,\mathrm{ns}/\mathrm{km}. Over 2km2\,\mathrm{km} a pulse spreads by 0.1µs0.1\,\text{µ}\mathrm{s}, which caps the bit rate near 5Mbit/s5\,\mathrm{Mbit}/\mathrm{s}: the weekend problem follows this limitation to its remedy, the single-mode fiber.

2.4 The plane mirror

Proposition 2.16 (Image in a plane mirror)

All rays from a point AA reflected by a plane mirror seem to come from the point AA' symmetric to AA with respect to the mirror’s plane: AA' is the image of AA. It is virtual (no light passes through it) and the mirror is rigorously stigmatic: every point has an exact point image.

Proof. A ray from AA hitting the mirror at II with incidence ii leaves at ii on the other side of the normal. The reflected line and the normal at II make the angle ii, as does the line IAIA; by symmetry with respect to the mirror plane, the reflected line is the mirror image of line IAIA, hence passes through the symmetric point AA' — for every II.

Proposition 2.17 (Rotating a mirror)

Turning a plane mirror by an angle α\alpha about an axis in its plane, perpendicular to the plane of incidence, turns the reflected ray by 2α2\alpha.

Proof. The normal turns by α\alpha, so the incidence changes from ii to i+αi + \alpha; the reflected ray, at i+αi + \alpha on the other side of the new normal, has moved by α\alpha (normal) +α+ \alpha (angle) =2α= 2\alpha from its old direction.

Example 2.18 (The galvanometer’s lever)

A tiny mirror glued to the coil of a galvanometer, lit by a lamp and throwing a spot on a scale 2m2\,\mathrm{m} away: a rotation of 1mrad1\,\mathrm{mrad} of the coil moves the spot by 2×103×2m=4mm2 \times 10^{-3} \times 2\,\mathrm{m} = 4\,\mathrm{mm} — an optical lever that magnifies without friction, the principle of the mirror galvanometer and of the laser scanner.

2.5 The prism

Proposition 2.19 (Prism formulas)

A ray crossing a prism of apex angle AA and index nn in air, in the plane perpendicular to the edge, with angles ii (entry), rr, rr' (inside) and ii' (exit) measured from the normals, obeys

sini=nsinr,r+r=A,sini=nsinr,\sin i = n \sin r, \qquad r + r' = A, \qquad \sin i' = n \sin r',

and is deviated by D=i+iAD = i + i' - A. The deviation is minimal when the path is symmetric, i=ii = i', r=r=A/2r = r' = A/2; then

n=sinA+Dmin2sinA2.n = \frac{\sin\frac{A + D_{\min}}{2}}{\sin\frac{A}{2}} .

Proof. Snell’s law at each face; r+r=Ar + r' = A because the two normals make the angle AA (the quadrilateral formed by the apex, the two incidence points and the intersection of the normals). The total deviation is (ir)+(ir)=i+iA(i - r) + (i' - r') = i + i' - A. By the return of light, the deviation is the same function of ii and of ii'; if the minimum occurred at i0i0i_0 \neq i_0' it would also occur at the reversed incidence i0i_0', giving two minima, which the monotonic behavior of D(i)D(i) on each side of its minimum forbids: the minimum is at i=ii = i'. Then r=r=A/2r = r' = A/2 and Dmin=2iAD_{\min} = 2i - A; substituting i=(A+Dmin)/2i = (A + D_{\min})/2 into sini=nsin(A/2)\sin i = n\sin(A/2) gives the formula.

A ray through a prism of apex angle A: refracted toward the base at entry and again at exit, it is deviated by D = i + i' - A from its original direction (dashed). Drawn at minimum deviation, the path is symmetric.
A ray through a prism of apex angle AA: refracted toward the base at entry and again at exit, it is deviated by D=i+iAD = i + i' - A from its original direction (dashed). Drawn at minimum deviation, the path is symmetric.

Example 2.20 (Measuring an index)

A 6060^\circ prism turned until the deviation of a yellow beam is smallest: Dmin=38.9D_{\min} = 38.9^\circ. Then n=sin49.45/sin30=0.760/0.500=1.52n = \sin 49.45^\circ/\sin 30^\circ = 0.760/0.500 = 1.52. Minimum deviation is the standard way to measure the index of a glass — the sharp minimum, found by turning the prism back and forth, is insensitive to small misadjustments.

Remark 2.21 (Thin prisms and spectroscopes)

For small AA and small incidence, the sines equal the angles and D(n1)AD \approx (n - 1)A, independent of ii: a thin prism deviates every ray by the same angle — a wedge of glass is a “prism” that bends light, and a lens will turn out to be a stack of such wedges. Because nn depends on λ\lambda, the deviation does too: a prism spreads white light into a spectrum (blue deviated most), the heart of the prism spectroscope.

2.6 The rainbow

Example 2.22 (Descartes’s rainbow)

A ray of sunlight enters a spherical raindrop at incidence ii, is refracted (sini=nsinr\sin i = n\sin r), reflects once on the back surface, and exits after a second refraction. Each refraction deviates it by iri - r, the internal reflection by π2r\pi - 2r: total deviation D(i)=π+2i4rD(i) = \pi + 2i - 4r. Rays fill the drop at every ii; the observer sees the brightest light where many rays leave in nearly the same direction, i.e. where DD is stationary:  ⁣dD/ ⁣di=24 ⁣dr/ ⁣di=0\dd D/\dd i = 2 - 4\,\dd r/\dd i = 0. Differentiating Snell’s law, cosi=ncosr ⁣dr/ ⁣di\cos i = n\cos r\,\dd r/\dd i, so the condition reads 2cosi=ncosr2\cos i = n\cos r; squaring and using n2cos2r=n2sin2i=n21+cos2in^2\cos^2 r = n^2 - \sin^2 i = n^2 - 1 + \cos^2 i gives

cos2i=n213.\cos^2 i = \frac{n^2 - 1}{3} .

For water, n=1.333n = 1.333: i=59.4i = 59.4^\circ, r=40.2r = 40.2^\circ, Dmin=137.9D_{\min} = 137.9^\circ. The light comes back toward the Sun’s side at 180137.9=42180^\circ - 137.9^\circ = 42^\circ from the anti-solar direction: the rainbow is a cone of half-angle 4242^\circ around the shadow of the observer’s head. Red (n=1.331n = 1.331) sits at 42.442.4^\circ, violet (n=1.343n = 1.343) at 40.640.6^\circ — red outside, violet inside.

Descartes’s ray through a raindrop: two refractions and one internal reflection. Near i = 59 the deviation D is stationary at 138, so rays pile up there: the primary rainbow, 42 from the point opposite the Sun.
Descartes’s ray through a raindrop: two refractions and one internal reflection. Near i=59i = 59^\circ the deviation DD is stationary at 138138^\circ, so rays pile up there: the primary rainbow, 4242^\circ from the point opposite the Sun.

2.7 Exercises

Exercise 2.1

Compute the speed of light in water (n=1.33n = 1.33), in glass (1.501.50) and in diamond (2.422.42). How long does light take to cross 1.0km1.0\,\mathrm{km} of a fiber of index 1.481.48?

Solution

Solution of Exercise 2.1.

v=c/nv = c/n: water 2.26×108m/s2.26 \times 10^{8}\,\mathrm{m}/\mathrm{s}, glass 2.00×108m/s2.00 \times 10^{8}\,\mathrm{m}/\mathrm{s}, diamond 1.24×108m/s1.24 \times 10^{8}\,\mathrm{m}/\mathrm{s}. Fiber: t=nL/c=1.48×103/3.00×108=4.9µst = nL/c = 1.48 \times 10^3/3.00\times10^8 = 4.9\,\text{µ}\mathrm{s}.

Exercise 2.2

A ray in air meets water (n=1.33n = 1.33) at 3030^\circ, then at 6060^\circ from the normal: find the refracted angles. A ray in water meets the surface at 5050^\circ: what happens?

Solution

Solution of Exercise 2.2.

sini2=sini1/1.33\sin i_2 = \sin i_1/1.33: 3022.130^\circ \to 22.1^\circ; 6040.660^\circ \to 40.6^\circ. From water at 5050^\circ: 1.33sin50=1.02>11.33\sin 50^\circ = 1.02 > 1, no refracted ray — total reflection (ic=48.8i_c = 48.8^\circ).

Exercise 2.3

Compute the critical angle for glass–air (n=1.50n = 1.50), diamond–air (2.422.42) and water–air (1.331.33). What is the angular diameter of the disk of sky a diver sees?

Solution

Solution of Exercise 2.3.

ic=arcsin(1/n)i_c = \arcsin(1/n): glass 41.841.8^\circ, diamond 24.424.4^\circ, water 48.848.8^\circ. The disk of sky spans 2×48.8=97.52 \times 48.8^\circ = 97.5^\circ.

Exercise 2.4

A person 1.70m1.70\,\mathrm{m} tall, eyes 10cm10\,\mathrm{cm} below the top of the head, stands before a vertical wall mirror. Find the smallest mirror (height, and position of its edges above the floor) in which the whole body is visible. Does the answer depend on the distance to the mirror?

Solution

Solution of Exercise 2.4.

Eyes at 1.60m1.60\,\mathrm{m}. The ray from the feet reaches the eye after reflecting at half the eye height, 0.80m0.80\,\mathrm{m}; the ray from the top of the head reflects midway between top and eyes, at 1.65m1.65\,\mathrm{m}. Mirror from 0.80m0.80\,\mathrm{m} to 1.65m1.65\,\mathrm{m}: height 0.85m0.85\,\mathrm{m}, half the body height — independent of the distance (the midpoint rule holds at any distance).

Exercise 2.5 ★★

A coin lies 1.2m1.2\,\mathrm{m} deep in a pool; at what depth does it appear to someone looking from above? A fisherman’s eyes are 1.8m1.8\,\mathrm{m} above the water: how high do they appear to a fish looking up? Justify both with small-angle rays.

Solution

Solution of Exercise 2.5.

Small angles: the apparent position is h=h/nh' = h/n when looking from the rarer medium, h=nhh' = nh from the denser one. Coin: 1.2/1.33=0.90m1.2/1.33 = 0.90\,\mathrm{m}. Fisherman: 1.33×1.8=2.4m1.33 \times 1.8 = 2.4\,\mathrm{m}. Justification: for a surface point at horizontal distance xx, tani2i2=x/h\tan i_2 \approx i_2 = x/h and tani1i1=x/h\tan i_1 \approx i_1 = x/h' with i1=ni2i_1 = n i_2.

Exercise 2.6 ★★

A prism has A=60A = 60^\circ and n=1.52n = 1.52. Compute the minimum deviation. Show that a ray can emerge from the second face only if A2icA \leq 2 i_c, and check this prism qualifies.

Solution

Solution of Exercise 2.6.

sinA+Dmin2=1.52sin30=0.760\sin\frac{A + D_{\min}}{2} = 1.52\sin 30^\circ = 0.760, so A+Dmin2=49.5\frac{A + D_{\min}}{2} = 49.5^\circ, Dmin=38.9D_{\min} = 38.9^\circ. Emergence needs ricr' \leq i_c; also sinr=sini/n1/n\sin r = \sin i/n \leq 1/n gives ricr \leq i_c; so A=r+r2icA = r + r' \leq 2i_c. Here ic=arcsin(1/1.52)=41.1i_c = \arcsin(1/1.52) = 41.1^\circ, 2ic=82.3>602i_c = 82.3^\circ > 60^\circ: fine.

Exercise 2.7 ★★

A step-index fiber has n1=1.50n_1 = 1.50, n2=1.48n_2 = 1.48. Compute its numerical aperture, its acceptance angle, the intermodal delay per kilometer, and the resulting maximum bit rate over 1km1\,\mathrm{km} (take Bmax1/(2Δt)B_{\max} \approx 1/(2\Delta t)).

Solution

Solution of Exercise 2.7.

NA=1.5021.482=0.0596=0.244\mathrm{NA} = \sqrt{1.50^2 - 1.48^2} = \sqrt{0.0596} = 0.244, θa=14.1\theta_a = 14.1^\circ. Δt/L=(1.50/c)(1.50/1.481)=5.0×109×0.0135=68ns/km\Delta t/L = (1.50/c)(1.50/1.48 - 1) = 5.0\times10^{-9} \times 0.0135 = 68\,\mathrm{ns}/\mathrm{km}. Over 1km1\,\mathrm{km}: Bmax1/(2×68ns)=7.4Mbit/sB_{\max} \approx 1/(2 \times 68\,\mathrm{ns}) = 7.4\,\mathrm{Mbit}/\mathrm{s}.

Exercise 2.8 ★★

A thin prism of apex 1010^\circ has n=1.510n = 1.510 for red and 1.5301.530 for blue. Compute the deviation of each color and the angular width of the spectrum it produces.

Solution

Solution of Exercise 2.8.

D(n1)AD \approx (n - 1)A: red 0.510×10=5.100.510 \times 10^\circ = 5.10^\circ, blue 5.305.30^\circ; spectrum width 0.200.20^\circ.

Exercise 2.9 ★★

A ray meets a parallel-faced glass plate (n=1.50n = 1.50, thickness e=2.0cme = 2.0\,\mathrm{cm}) at 4545^\circ. Show that it emerges parallel to its incident direction, shifted sideways by d=esin(ir)/cosrd = e\,\sin(i - r)/\cos r, and compute dd.

Solution

Solution of Exercise 2.9.

Entry: sinr=sin45/1.5=0.471\sin r = \sin 45^\circ/1.5 = 0.471, r=28.1r = 28.1^\circ. The two faces are parallel, so the incidence on the second face is rr and sini=nsinr=sini\sin i' = n\sin r = \sin i: i=ii' = i, the emerging ray is parallel to the incident one. Along the inside path of length e/cosre/\cos r, the perpendicular offset is (e/cosr)sin(ir)(e/\cos r)\sin(i - r): d=2.0×sin16.9/cos28.1=0.66cmd = 2.0 \times \sin 16.9^\circ/\cos 28.1^\circ = 0.66\,\mathrm{cm}.

Exercise 2.10 ★★★

The secondary rainbow comes from rays reflected twice inside the drop. Show that D=2π+2i6rD = 2\pi + 2i - 6r, that the stationary deviation satisfies cos2i=(n21)/8\cos^2 i = (n^2 - 1)/8, and compute the angle of the secondary bow from the anti-solar direction for n=1.333n = 1.333. Where is it relative to the primary, and in which order are its colors?

Solution

Solution of Exercise 2.10.

Two refractions (2(ir)2(i - r)) and two reflections (2(π2r)2(\pi - 2r)): D=2π+2i6rD = 2\pi + 2i - 6r.  ⁣dD/ ⁣di=26cosi/(ncosr)=0\dd D/\dd i = 2 - 6\cos i/(n\cos r) = 0 gives 3cosi=ncosr3\cos i = n\cos r, 9cos2i=n21+cos2i9\cos^2 i = n^2 - 1 + \cos^2 i, cos2i=(n21)/8\cos^2 i = (n^2 - 1)/8. n=1.333n = 1.333: cos2i=0.0971\cos^2 i = 0.0971, i=71.8i = 71.8^\circ, sinr=0.950/1.333\sin r = 0.950/1.333, r=45.5r = 45.5^\circ, D=360+143.7272.8=230.9D = 360^\circ + 143.7^\circ - 272.8^\circ = 230.9^\circ: the bow is at 230.9180=51230.9^\circ - 180^\circ = 51^\circ from the anti-solar point, outside the primary (4242^\circ). Red (n=1.331n = 1.331) at 50.450.4^\circ, violet (1.3431.343) at 53.653.6^\circ: the colors are reversed, violet outside.

Exercise 2.11 ★★★

On a hot road the air just above the asphalt has index nhot=1.00026n_{\text{hot}} = 1.00026, the cooler air above it ncool=1.00029n_{\text{cool}} = 1.00029. Model the layers as a sharp interface: find the critical grazing angle (measured from the road) below which a ray from the sky is totally reflected upward. A driver’s eye is 1.2m1.2\,\mathrm{m} above the road: beyond what distance does the road look like water?

Solution

Solution of Exercise 2.11.

sinic=1.00026/1.00029=0.99997\sin i_c = 1.00026/1.00029 = 0.99997, ic=89.56i_c = 89.56^\circ: rays within 0.44=7.7mrad0.44^\circ = 7.7\,\mathrm{mrad} of the road surface are totally reflected (generally, grazing angle 2Δn/n\approx \sqrt{2\Delta n/n}). From 1.2m1.2\,\mathrm{m} up, the line of sight grazes at this angle beyond 1.2/0.0077160m1.2/0.0077 \approx 160\,\mathrm{m}: the road there reflects the sky — the “puddle”.

Exercise 2.12 ★★★

Points AA (height aa above an interface) and BB (depth bb below it) are separated horizontally by dd; light travels at v1v_1 above and v2v_2 below. Write the travel time of the broken path through the interface point at horizontal position xx, and show that the time is stationary exactly when sini1/v1=sini2/v2\sin i_1/v_1 = \sin i_2/v_2Snell’s law.

Solution

Solution of Exercise 2.12.

t(x)=a2+x2v1+b2+(dx)2v2t(x) = \dfrac{\sqrt{a^2 + x^2}}{v_1} + \dfrac{\sqrt{b^2 + (d - x)^2}}{v_2}; t(x)=xv1a2+x2dxv2b2+(dx)2=sini1v1sini2v2t'(x) = \dfrac{x}{v_1\sqrt{a^2 + x^2}} - \dfrac{d - x}{v_2\sqrt{b^2 + (d - x)^2}} = \dfrac{\sin i_1}{v_1} - \dfrac{\sin i_2}{v_2}, zero exactly when sini1/v1=sini2/v2\sin i_1/v_1 = \sin i_2/v_2, i.e. n1sini1=n2sini2n_1\sin i_1 = n_2\sin i_2 with n=c/vn = c/v.

A double rainbow: the primary bow at 42 with red outside, the fainter secondary at 51 with the colours reversed, and Alexander’s dark band between them — refraction, reflection and dispersion in every drop.
A double rainbow: the primary bow at 4242{}^{\circ} with red outside, the fainter secondary at 5151{}^{\circ} with the colours reversed, and Alexander’s dark band between them — refraction, reflection and dispersion in every drop.

2.8 Problem: The optical fiber link

Problem 2.1

Weekend problem — a glass thread under the sea: how far and how fast can a pulse of light carry a bit before geometry, glass and dispersion blur it away

A link uses a step-index fiber with core n1=1.480n_1 = 1.480, cladding n2=1.465n_2 = 1.465, core radius a=25µma = 25\,\text{µ}\mathrm{m}, at the wavelength λ0=1550nm\lambda_0 = 1550\,\mathrm{nm}. Data: c=3.00×108m/sc = 3.00 \times 10^{8}\,\mathrm{m}/\mathrm{s}.

Part I — Light in glass.

  1. Define the index and compute the speed of light in the core.
  2. How long does a pulse take to travel 50km50\,\mathrm{km} along the axis?
  3. Compute the frequency of the light and its wavelength inside the core.
  4. Why is the core clad in glass of lower index rather than left bare in air, which would give a larger index step?

Part II — Guiding.

  1. Compute the critical angle at the core–cladding surface.
  2. Deduce the largest angle a guided ray can make with the axis.
  3. A ray enters the flat end face from air at angle θ\theta to the axis. Show that it is guided if sinθn12n22\sin\theta \leq \sqrt{n_1^2 - n_2^2}, and compute the acceptance angle θa\theta_a.
  4. Give the numerical aperture. Why does a larger index step make the fiber easier to feed with light?
  5. The fiber is bent along a circle of radius RbR_b. Consider a ray travelling parallel to the axis along the inner edge of the core: show that it meets the outer wall at incidence ii with sini=Rb/(Rb+2a)\sin i = R_b/(R_b + 2a), and find the smallest RbR_b for which it is still guided.

Part III — Intermodal dispersion.

  1. Express the travel times of the axial ray and of the steepest guided ray over a length LL.
  2. Deduce the spread Δt\Delta t per kilometer, and its value over L=2kmL = 2\,\mathrm{km}.
  3. A bit is a short pulse; two successive pulses must stay separated by at least 2Δt2\Delta t to be told apart. Estimate the maximum bit rate over 2km2\,\mathrm{km}.
  4. Show that the product (bit rate) ×\times (length) is a constant for this fiber, and compute it in Mbits1km\mathrm{Mbit}\,\mathrm{s}^{-1}\,\mathrm{km}.
  5. In a graded-index fiber the index decreases smoothly from the axis outward. Explain qualitatively why this reduces the intermodal spread.
  6. A single-mode fiber has a core diameter of about 9µm9\,\text{µ}\mathrm{m}. Compare with λ0\lambda_0 and explain why the ray model cannot describe it; what is gained?

Part IV — Attenuation and chromatic dispersion. The fiber attenuates the power by α=0.20dB/km\alpha = 0.20\,\mathrm{dB}/\mathrm{km} at 1550nm1550\,\mathrm{nm}: P(L)=P010αL/10P(L) = P_0\,10^{-\alpha L/10}. The transmitter launches P0=1.0mWP_0 = 1.0\,\mathrm{mW}; the receiver needs at least Pmin=1.0µWP_{\min} = 1.0\,\text{µ}\mathrm{W}. Because nn depends on λ\lambda, two wavelengths travel at slightly different speeds: a source of spectral width Δλ\Delta\lambda gives a pulse spread Δtc=DΔλL\Delta t_c = D\,\Delta\lambda\,L with D=17ps/(nmkm)D = 17\,\mathrm{ps}/(\mathrm{nm}\,\mathrm{km}).

  1. What fraction of the power survives 100km100\,\mathrm{km}?
  2. Find the maximum length before an amplifier is needed.
  3. Amplifiers are placed every 100km100\,\mathrm{km}: what margin, in dB and as a power ratio, does this leave for connectors and splices?
  4. Compute Δtc\Delta t_c over 100km100\,\mathrm{km} for an LED source (Δλ=40nm\Delta\lambda = 40\,\mathrm{nm}) and for a laser (Δλ=0.10nm\Delta\lambda = 0.10\,\mathrm{nm}); deduce the maximum bit rate of each over that length (same criterion as above).
  5. At 850nm850\,\mathrm{nm} silica attenuates 2.0dB/km2.0\,\mathrm{dB}/\mathrm{km}. What distance would the same power budget allow there? Why is 1550nm1550\,\mathrm{nm} the telecom wavelength?
  6. Copper coaxial cable attenuates about 30dB/km30\,\mathrm{dB}/\mathrm{km} at the frequencies needed for 1Gbit/s1\,\mathrm{Gbit}/\mathrm{s}. Compare the amplifier spacing with the fiber’s.

Part V — The whole link.

  1. For the step-index fiber over 2km2\,\mathrm{km}, which effect limits the bit rate: intermodal spread, chromatic spread (laser source) or attenuation? Give the limiting bit rate.
  2. For a single-mode fiber over 100km100\,\mathrm{km} with the laser source: which effect limits, and at what bit rate?
  3. The single-mode link carries 8080 lasers at wavelengths 0.8nm0.8\,\mathrm{nm} apart (wavelength multiplexing), each at the bit rate of the previous question. Total capacity?
  4. Form the bit-rate–length product of the step-index link and of one single-mode channel; by what factor does the single-mode fiber win, and which single physical fact is responsible?
Solution

Solution of Problem 2.1.

1. n=c/vn = c/v; v1=3.00×108/1.480=2.03×108m/sv_1 = 3.00\times10^8/1.480 = 2.03 \times 10^{8}\,\mathrm{m}/\mathrm{s}.

2. t=Ln1/c=5.0×104×1.48/3.00×108=247µst = Ln_1/c = 5.0\times10^4 \times 1.48/3.00\times10^8 = 247\,\text{µ}\mathrm{s}.

3. ν=c/λ0=1.94×1014Hz\nu = c/\lambda_0 = 1.94 \times 10^{14}\,\mathrm{Hz}, unchanged in glass; λ=λ0/n1=1047nm\lambda = \lambda_0/n_1 = 1047\,\mathrm{nm}.

4. A bare core guides only while its surface is perfectly clean and untouched: dust, water or a grip locally replaces air by a higher index and light escapes, and scratches scatter it. The cladding is a sealed, controlled interface with an index step chosen on purpose.

5. sinic=1.465/1.480=0.9899\sin i_c = 1.465/1.480 = 0.9899, ic=81.8i_c = 81.8^\circ.

6. 9081.8=8.290^\circ - 81.8^\circ = 8.2^\circ.

7. sinθ=n1sinθ\sin\theta = n_1\sin\theta'; guided iff cosθn2/n1\cos\theta' \geq n_2/n_1, i.e. sinθ1n22/n12\sin\theta' \leq \sqrt{1 - n_2^2/n_1^2}, i.e. sinθn12n22=2.19042.1462=0.210\sin\theta \leq \sqrt{n_1^2 - n_2^2} = \sqrt{2.1904 - 2.1462} = 0.210: θa=12.1\theta_a = 12.1^\circ.

8. NA=0.210\mathrm{NA} = 0.210. A larger step widens the cone, so more of a diverging source’s light (an LED) is captured.

9. The ray is tangent to the circle of radius RbR_b (inner edge) and meets the circle of radius Rb+2aR_b + 2a (outer edge); in the right triangle center–tangent point–impact point, the angle at the impact point between the ray and the radius (the normal) satisfies sini=Rb/(Rb+2a)\sin i = R_b/(R_b + 2a). Guided iff sinin2/n1\sin i \geq n_2/n_1: Rb2an2/n11n2/n1=50µm×97.6=4.9mmR_b \geq 2a\,\dfrac{n_2/n_1}{1 - n_2/n_1} = 50\,\text{µ}\mathrm{m} \times 97.6 = 4.9\,\mathrm{mm}. Tighter bends leak.

10. t0=Ln1/ct_0 = Ln_1/c; steepest ray path L/cosθmax=Ln1/n2L/\cos\theta'_{\max} = Ln_1/n_2, so t1=Ln12/(n2c)t_1 = Ln_1^2/(n_2 c).

11. Δt/L=(n1/c)(n1/n21)=4.93×109×0.01024=50.5ns/km\Delta t/L = (n_1/c)(n_1/n_2 - 1) = 4.93\times10^{-9} \times 0.01024 = 50.5\,\mathrm{ns}/\mathrm{km}; over 2km2\,\mathrm{km}: 101ns101\,\mathrm{ns}.

12. Bmax1/(2Δt)=1/202ns5Mbit/sB_{\max} \approx 1/(2\Delta t) = 1/202\,\mathrm{ns} \approx 5\,\mathrm{Mbit}/\mathrm{s}.

13. BmaxL=1/(2Δt/L)=1/(2×50.5ns/km)10Mbits1kmB_{\max} L = 1/(2\,\Delta t/L) = 1/(2 \times 50.5\,\mathrm{ns}/\mathrm{km}) \approx 10\,\mathrm{Mbit}\,\mathrm{s}^{-1}\,\mathrm{km}.

14. Rays far from the axis travel longer paths but in lower index, hence faster; with a well-chosen profile the two effects cancel and all rays arrive together.

15. 9µm6λ09\,\text{µ}\mathrm{m} \approx 6\lambda_0: the core is not large compared with the wavelength, so diffraction dominates and rays have no meaning. The wave treatment shows a single guided pattern: no intermodal spread at all; only chromatic dispersion and attenuation remain.

16. 100km100\,\mathrm{km} ×\times 0.2dB/km0.2\,\mathrm{dB}/\mathrm{km} =20dB= 20\,\mathrm{dB}: P/P0=102P/P_0 = 10^{-2}, one percent.

17. 10log(P0/Pmin)=30dB10\log(P_0/P_{\min}) = 30\,\mathrm{dB}, so Lmax=30/0.20=150kmL_{\max} = 30/0.20 = 150\,\mathrm{km}.

18. 100km100\,\mathrm{km} costs 20dB20\,\mathrm{dB}: margin 10dB10\,\mathrm{dB}, a factor 1010 in power.

19. LED: Δtc=17×40×100=68000ps=68ns\Delta t_c = 17 \times 40 \times 100 = 68\,000\,\mathrm{ps} = 68\,\mathrm{ns}, B1/136ns=7Mbit/sB \approx 1/136\,\mathrm{ns} = 7\,\mathrm{Mbit}/\mathrm{s}. Laser: Δtc=170ps\Delta t_c = 170\,\mathrm{ps}, B1/340ps=2.9Gbit/sB \approx 1/340\,\mathrm{ps} = 2.9\,\mathrm{Gbit}/\mathrm{s}.

20. 30/2.0=15km30/2.0 = 15\,\mathrm{km}, ten times less. Silica is most transparent near 1550nm1550\,\mathrm{nm} (its attenuation minimum), and optical amplifiers exist there.

21. 30/30=1km30/30 = 1\,\mathrm{km} between amplifiers, against 150km150\,\mathrm{km}: the fiber wins by a factor 150150.

22. Intermodal: 101ns101\,\mathrm{ns} (\to 5Mbit/s5\,\mathrm{Mbit}/\mathrm{s}); chromatic (laser): 17×0.1×2=3.4ps17 \times 0.1 \times 2 = 3.4\,\mathrm{ps}; attenuation 0.4dB0.4\,\mathrm{dB}, negligible. The intermodal spread limits: 5Mbit/s5\,\mathrm{Mbit}/\mathrm{s}.

23. No intermodal spread; chromatic 170ps170\,\mathrm{ps} \to 2.9Gbit/s2.9\,\mathrm{Gbit}/\mathrm{s}; attenuation 20dB20\,\mathrm{dB}, within budget. Chromatic dispersion limits, at about 3Gbit/s3\,\mathrm{Gbit}/\mathrm{s}.

24. 80×2.9230Gbit/s80 \times 2.9 \approx 230\,\mathrm{Gbit}/\mathrm{s}.

25. Step index: 10Mbits1km10\,\mathrm{Mbit}\,\mathrm{s}^{-1}\,\mathrm{km}; single-mode channel: 2.9×100=290Gbits1km2.9 \times 100 = 290\,\mathrm{Gbit}\,\mathrm{s}^{-1}\,\mathrm{km} — a factor of about 3×1043\times10^4. One fact: a single guided path, hence no geometric (intermodal) spread of arrival times.

Terms defined in this chapter

See all 393 terms in the glossary