This article is from the CD-Recordable FAQ, by Andy McFadden (email@example.com) with numerous contributions by others.
CD players use a near-infrared 780nm laser. The visible light spectrum
is generally considered to be 400nm to 700nm; few people can see light
past 720nm. (DVD, by contrast, uses a visible red 635nm or 650nm laser.)
The drive shines a laser through the polycarbonate (plastic) on the "bottom"
of the disc. This bounces off the reflective layer, passes back through the
polycarbonate, and is read by a photosensor in the drive head. The index
of refraction for polycarbonate is about 1.55, so laser light bends when
it enters, allowing a much finer focus for the laser (from 800um at the
bottom of the polycarbonate down to about 1.7um at the metal surface).
This minimizes the effects of dust and scratches, because the effects
of any surface gunk are reduced as the laser's focus width is reduced.
A 400um-wide piece of dust on the surface of a CD would completely block
a laser focused down to 200um at the surface, but has little effect on a
If the photosensor sees a strong beam -- the CD standard requires the
signal strength to be at least 70% when fully reflected -- it knows it's
traveling over a "land". If it sees a weaker response, it's traveling
over a "pit". Technically, it's traveling "under" a pit or land, so from
its perspective a "pit" is actually a bump. The height of the bump is 1/4
of the laser's wavelength when traveling in polycarbonate, so that light
reflected from the bump has a phase difference of one-half wavelength.
The light reflected from the pit and from the surrounding land thus cancel
each other out. (The geometries are actually such that a "pit" reflects
about 25% of the intensity rather than 0%. For example, pits are 0.5um
wide, or about 1/3 of the focused width of the laser.)
There are a lot of optical tricks involving polarization of light and the
action of diffraction gratings going on. For example, the read head uses
a three-beam auto-focus system that keeps the laser properly aligned on
the spiral track and at the correct distance from the bottom of the disc.
(Side note: if adjacent loops of the spiral are too close together -- the
"track pitch" is too small -- the laser tracking can fail. This is why
90- and 99-minute discs are harder to write and read.) It's also worth
mentioning that, because light travels more slowly in polycarbonate,
the wavelength of the laser inside the CD is closer to 500nm.
CD-R and CD-RW discs do not have pits and lands. On CD-R media, the write
laser heats the organic dye to approximately 250 degrees Celsius, causing
it to melt and/or chemically decompose to form a depression or mark in the
recording layer. The marks create the decreased reflectivity required by the
read laser. On CD-RW media, the write laser changes the material between
crystalline (25% reflectivity) and amorphous (15% reflectivity) states.
This is done by either heating it above its melting point (500C to 700C)
and letting it cool rapidly to convert it to amorphous form, or heating it
to its transition point (200C) and letting it cool slowly to return it to
the more stable crystalline state. The lower reflectivity of CD-RW makes
the discs unreadable on most older players.
The rest of this discussion refers to "pits" and "lands", but applies
equally to pressed CDs, CD-Rs, and CD-RWs.