Laserlight that marches in step

Population inversion

—

Photons in the cavity

—

Output power

—

Pump threshold (this scheme)

—

Simplified model: two rate equations (upper-level population and the number of photons in the cavity) computed in the background; the animation shows only a sample of the photons and atoms. Time is measured in upper-level lifetimes τ (ruby: about 3 ms, Nd:YAG: about 0.23 ms). Power is in arbitrary units (a.u.).

Below threshold the cavity only glows faintly with spontaneous emission. Once the gain catches up with the losses, the number of photons grows like an avalanche, and the output power rises linearly with the pump — that's the “knee” in the graph. Right after switch-on you can also see the characteristic relaxation oscillations (spikes).

Where the name comes from, and why 4 levels are easier

Stimulated emission: a photon of just the right energy passes an excited atom and “persuades” it to give up its energy. The new photon is an exact copy of the first — same direction, phase, polarization and wavelength. Each copy can trigger more — hence the avalanche and the gain.

Population inversion: gain beats absorption only when there are more atoms in the upper laser level than in the lower one. That's why the laser has to be pumped — with a flash lamp, a diode or an electrical discharge.

3 levels (ruby): the lower laser level is the ground state, where all the atoms sit at the start. So you have to excite more than half of them before any gain appears. 4 levels (Nd:YAG, He-Ne): the lower laser level lies a little above the ground state and empties almost instantly — a handful of excited atoms is enough for inversion, so the threshold is many times lower.

Mirrors: one reflects almost 100%, the other e.g. 98% — it lets a fraction of the light out. A photon bounces back and forth many times and is amplified on every pass. If the output mirror lets too much through, the losses exceed the gain and the laser goes dark.

A short history

• 1917Albert Einstein describes stimulated emission (in his work on the quantum theory of radiation, 1916–1917).
• 1954Charles Townes and his team run the first maser, using ammonia — a “laser” for microwaves. Nikolay Basov and Alexander Prokhorov developed similar ideas independently (they shared the 1964 Nobel Prize).
• 1960On May 16, Theodore Maiman (Hughes Research Laboratories) runs the first laser — a pulsed ruby laser at 694.3 nm. In December, Ali Javan, William Bennett and Donald Herriott run the first gas laser (He-Ne, initially in the infrared).
• 1962The first semiconductor lasers (GaAs) — at General Electric and IBM labs, among others.
• 1963The first lasers built in Poland — including a gas laser at the Military University of Technology in Warsaw, by Zbigniew Puzewicz's team.

Back then people joked that the laser was “a solution looking for a problem.” Today it's hard to find a field of technology without one.

Peak power P ≈ E/τ

—

Intensity at a ⌀10 µm focus

—

For comparison

—

How to make a petawatt pulse

CPA — chirped pulse amplification (Donna Strickland and Gérard Mourou, 1985; Nobel Prize 2018). A short pulse is first stretched (different colors travel separately, like in a bird's chirp), which lowers its power so it doesn't destroy the amplifier. After amplification, diffraction gratings squeeze it back to femtoseconds — and the peak power shoots up.

Attoseconds: 1 as = 10⁻¹⁸ s. Pierre Agostini, Ferenc Krausz and Anne L'Huillier (Nobel Prize 2023) learned to make pulses so short that you can “film” electrons moving in atoms.

Record lasers: ELI-NP in Romania (Măgurele) — 10 PW in pulses of about 20 fs; NIF in the USA — about 2 MJ in a pulse lasting a few nanoseconds (peak power around 500 TW). On December 5, 2022, NIF for the first time got more energy out of fusion (about 3.15 MJ) than the laser delivered to the target (2.05 MJ) — although powering the lasers themselves took about 300 MJ from the grid.

A simplification for a continuous beam based on the IEC 60825-1 standard; the exact limits depend on wavelength, exposure time and beam diameter. Laser pointers sold in shops are sometimes stronger than their label claims.

30 years of waiting for blue

1. 1962Nick Holonyak (General Electric) demonstrates the first visible light-emitting diode — a red one (GaAsP). Yellow and greenish ones follow, but no blue — and without blue there's no white LED light and no full-color screens.
2. 1970–90Attempts with SiC (which gave only very dim blue diodes, because its gap is indirect) and with ZnSe (the diodes degraded quickly). Gallium nitride, GaN, has an ideal gap (3.4 eV), but there is no matching substrate: it is grown on sapphire, whose crystal lattice differs by about 16% — the crystal is riddled with billions of defects. On top of that, nobody knew how to make p-type GaN.
3. 1986Isamu Akasaki and Hiroshi Amano (Nagoya) hit on a thin AlN buffer layer on sapphire, grown at low temperature — GaN grows on it as a smooth, high-quality crystal.
4. 1989Akasaki and Amano obtain p-type GaN: they “wake up” the magnesium dopant by irradiating the crystal with an electron beam (LEEBI). The first GaN p–n junction is born.
5. 1990–92Shuji Nakamura (Nichia) builds his own two-flow MOCVD reactor and discovers that simply annealing the crystal in nitrogen is enough. He explains why: hydrogen from the process gases binds to magnesium in Mg–H complexes and “puts it to sleep.” Annealing drives the hydrogen out — cheaply, and across whole wafers.
6. 1993Nichia announces a bright blue InGaN LED. Adding indium narrows the GaN gap down to blue or green.
7. 1995–96Nakamura demonstrates the first blue-violet InGaN laser diodes — pulsed at first, then running continuously at room temperature.
8. 2014Akasaki, Amano and Nakamura receive the Nobel Prize in Physics “for the invention of efficient blue light-emitting diodes which has enabled bright and energy-saving white light sources.”

Fun fact: Nakamura's company paid him a bonus for the invention that is often said to have been about 20,000 yen. After a high-profile lawsuit, the dispute ended in 2005 with a settlement of about 840 million yen.

A white LED is a blue diode (about 450 nm) coated with a yellow phosphor. The phosphor absorbs some of the blue photons and re-emits them as a broad yellow band. Blue + yellow = white. The color is computed from the spectrum using the eye's sensitivity curves (CIE 1931, approximation).

Savings: an incandescent bulb gives about 10–15 lumens per watt, a good LED lamp over 100. The light output of a 60 W bulb (about 800 lm) comes from an LED lamp drawing about 8 W.

Spot ≈ λ/(2·NA)

—

From optics alone

—

Format

—

Capacity grows roughly as (NA/λ)². Going from CD to Blu-ray, optics alone gives about 13 times more, while the real difference is about 36 times (0.7 → 25 GB). The rest came from more tightly packed tracks, a thinner protective layer, and better coding and error correction.