Guided light
A Socratic walk-through of guided light — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can light be made to follow a bending glass fibre instead of leaking out of its side?
Light travels in straight lines. That is the one fact most of us keep from school, and it is very nearly true in a uniform medium. Yet a hair-thin glass fibre wound onto a drum carries a signal round every one of those turns and delivers it, faint but intact, tens of kilometres away. So either light does bend, or something at the fibre's edge keeps turning it back. Which would you rather believe — and what would have to be happening at that edge?
Reasoning it through
REASONING #Start with the thing light does do at a boundary: it changes direction. Light travels more slowly in glass than in air, and the ratio of the two speeds is the refractive index — about 1.5 for ordinary glass. Cross a boundary at an angle and the ray kinks, by Snell's rule that n1 sin(theta1) equals n2 sin(theta2), with both angles measured from the perpendicular to the surface.
Now follow that rule in the direction that matters here: from the glass outwards, from slow medium to fast. The transmitted ray bends away from the perpendicular — further from it than the incoming ray. So tilt the incoming ray more and more toward grazing, and the transmitted ray swings ever flatter against the surface. What happens when it lies flat?
At that point sin(theta2) has reached 1 and cannot go further. Push the incoming angle past it and Snell's equation has no solution at all: there is no transmitted ray, and everything goes back into the glass. That threshold is the critical angle, arcsine of the index ratio — for glass against air, about 42 degrees from the perpendicular. Two features are worth pausing on. It is a genuine threshold, not a gradual fading; and the reflection is total, unlike a metal mirror that swallows a percent or two per bounce. A ray may hit the wall thousands of times per kilometre, and a mirror losing even 0.1% each time would be hopeless.
Then why not simply draw a bare glass rod and let air do the work? Because the trick depends entirely on what sits on the other side of the boundary. Lay a finger on the rod, or let dust, water or a protective coating touch it, and locally the outside index rises, the critical angle changes and light pours out. So a real fibre buries the boundary inside the glass: a core of slightly higher index surrounded by a cladding of slightly lower index, the difference typically a few tenths of one percent. That tiny step puts the critical angle up near 83 or 84 degrees, so only rays within a narrow cone of the axis stay guided — which is why coupling light into a fibre is fiddly.
One more consequence follows without any new physics. If several ray paths are guided, the one bouncing steeply travels further than the one running straight down the axis, and therefore arrives later. A sharp pulse launched at one end emerges smeared at the other — modal dispersion — and smeared pulses run into their neighbours, which caps how fast you may send bits over a given distance. Two remedies exist: grade the index so it falls smoothly from the centre, letting the longer off-axis paths run through faster glass and roughly keep pace; or shrink the core to about 8 or 9 micrometres so that only one path is supported at all. Long-haul cable is single-mode for that reason.
What is left is loss in the glass itself, and it is astonishingly small: around 0.2 decibels per kilometre at a wavelength of 1550 nanometres, meaning roughly half the light survives 15 km and about a hundredth survives 100 km. That wavelength is not arbitrary. Scattering from frozen-in density ripples in the glass falls steeply as wavelength grows, while absorption by the silica's own bond vibrations climbs at longer wavelengths — and the two curves cross near 1550 nm.
The analogy
THE ANALOGY #Think of a flat stone skipped across a lake. Thrown at a shallow enough angle it refuses to enter the water at all and skims onward; thrown any steeper it plunges straight in and is gone. There is an angle that separates the two behaviours, and either side of it the stone's fate is completely different.
the stone loses energy at every skip and gravity eventually claims it, whereas total internal reflection takes nothing from the beam at the wall — and the stone's threshold depends on its speed and spin, while light's depends only on the ratio of two refractive indices.
Clarifying the model
THE MODEL #The bouncing-ray picture is a useful fiction, and it is worth knowing where it stops being true. Light in a fibre is properly described as a set of guided modes — solutions of Maxwell's equations for that geometry — and in single-mode fibre the core is only a handful of wavelengths across, far too narrow for a drawn ray to mean anything. The mode's field does not stop dead at the core boundary either; an evanescent tail reaches into the cladding, which is why the cladding must be good clean glass rather than merely "not core", and why bending a fibre too tightly lets light leak away.
Two smaller corrections. "Total" describes the boundary, not the fibre: the 0.2 dB per kilometre is lost in transit through the glass, not at the reflections. And single-mode fibre has not abolished pulse spreading — it removed modal dispersion, while chromatic dispersion, the fact that different colours travel at slightly different speeds, remains and must be managed.
A picture of it
THE PICTURE #How to readStart at the black dot: light enters the guided state only if it is launched within the acceptance cone. The loop back onto Guided is a reflection at the core-cladding wall, taken whenever the ray meets that wall steeper than the critical angle — it costs nothing, so it can repeat thousands of times per kilometre. The branch to Escaped is the one failure of guiding itself: a shallower ray refracts into the cladding and never returns. The branch to Delivered is the different, gentler loss that remains even when guiding works perfectly — absorption and scattering in the glass, at the rate marked on that arrow.
What became clearer
WHAT CLEARED #The fibre does not bend light. It offers light a boundary it cannot cross, so long as the light meets that boundary steeply enough — and a threshold, unlike a mirror, can be crossed for free. Everything else in a fibre's design follows from protecting that threshold and paying for the distance: cladding so the boundary is never touched, a tiny index step so the guided cone is narrow and clean, and a wavelength chosen where the glass happens to be most nearly transparent.
Where to go next
ONWARD #- Why hollow-core and photonic-crystal fibres guide light by an entirely different mechanism, without relying on total internal reflection at all.
- How erbium-doped amplifiers restore a signal without converting it back to electricity, and why they work best at 1550 nm.
Key terms
TERMS #| Term | What it means |
|---|---|
| Refractive index | the ratio of light's speed in vacuum to its speed in a material. |
| Critical angle | the incidence angle, measured from the perpendicular, beyond which no light is transmitted out of the denser medium. |
| Total internal reflection | reflection of all incident light at such a boundary, with no transmitted ray. |
| Cladding | the outer glass layer of slightly lower index that forms the guiding boundary and keeps it clean. |
| Modal dispersion | pulse spreading caused by different guided paths having different lengths, eliminated in single-mode fibre. |
Every term the collection defines is gathered in the glossary.