Lasers & Stimulated Emission · Guide · glidepathscience.org
GuideLasers & Stimulated Emission
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook takes them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about a quarter of an hour to read. Each section ends with a worked example that is not one of the sim’s own problems.
Part 1 · The essentials
Math you’ll need: algebra, powers of ten, exponentials ($e^{x}$) and square roots; one optional box uses a derivative.
One photon makes two
A laser pointer, a supermarket scanner, the fiber-optic link that carries this page and the surgeon’s scalpel that reshapes a cornea all run on one process that Albert Einstein predicted in 1917 and nobody used until 1960: stimulated emission. An atom that is already excited, met by a photon of exactly the right energy, gives out a second photon that is a perfect copy of the first. Each copy adds light to the mode, the color, direction and phase, of the photon that triggered it. Copying alone does not make a laser, though: the light grows only where the copies outnumber the absorptions (net gain), and it builds into a beam only if mirrors keep sending it back through the atoms (feedback). Given both, a feeble flicker of light grows into a beam in which trillions of photons march in step.
That simple idea hides three problems, and this Guide takes them in turn. Light is normally absorbed by matter, not amplified, so the atoms must first be driven into an unnatural state, a population inversion. The copies must be kept going the same way, which is the job of two mirrors, an optical cavity. And the light must gain more on each round trip than it loses, which sets a threshold. Past it, the result is light unlike any lamp’s: one color, one direction, its waves in step over meters.
Photons and energy levels
Light is absorbed and emitted in packets of energy called photons. A photon of frequency $f$ and wavelength $λ$ carries
$E = hf = hc/λ$
where $h$ = 6.626 × 10⁻³⁴ J·s is Planck’s constant and $c$ = 3.00 × 10⁸ m/s. Atomic energies are small, so they are measured in electron volts (1 eV = 1.602 × 10⁻¹⁹ J), and then the handy product $hc$ = 1,240 eV·nm turns a photon’s energy into its wavelength at once.
An atom’s electrons can have only certain energies, its energy levels. An atom changes level only by taking in or giving out exactly the difference, so a photon is absorbed only if its energy matches a gap between two levels. “More energy than the gap” does not help: an atom with a 1.96 eV gap and no level 2.50 eV up simply lets a 2.50 eV photon pass. (Knocking an electron right out of the atom, as in the photoelectric effect, is different: there the electron leaves and carries away whatever energy is left over.)
Absorption is the first of the three processes in the figure; the next section takes the two ways an excited atom gives its energy back.
Spontaneous and stimulated emission
An excited atom does not stay excited. Left alone, it falls back at a random moment and emits a photon in a random direction: spontaneous emission, the source of the light from every lamp, flame and star. Each atom has the same chance of decaying in each second, so a crowd of $N_0$ excited atoms dwindles exponentially:
$N = N_0e^{−t/τ}$
Here $τ$ is the level’s lifetime, the average time an atom stays up. After one lifetime 37 % of the atoms remain, after two 14 %. Typical allowed transitions in the visible have lifetimes of nanoseconds; some levels, called metastable, last microseconds or milliseconds, and those are the levels lasers use.
Einstein’s insight was that a photon whose energy matches the gap can also trigger an excited atom to drop. The atom’s energy leaves as a second photon, and the two travel on together: one photon in, two out. This is stimulated emission, and its second photon is an exact twin of the first. It has the same frequency, travels in the same direction, oscillates in phase with it and has the same polarization. A spontaneous photon matches only in frequency; its moment, direction, phase and polarization are random.
Stimulated emission adds photons to a beam, but absorption takes them away. Which wins depends on the atoms, as the next section shows.
Gain needs a population inversion
A matching photon passing through a crowd of atoms meets ground-state atoms, which can absorb it, and excited atoms, which can be stimulated by it. For atoms whose two levels are equally degenerate the two processes are equally likely per atom, so what happens to a beam depends only on how many atoms are in each level. Out of every 100 encounters, the share with excited atoms adds a photon and the share with ground atoms removes one:
net per 100 = 100($N_2 − N_1$)/$N$
In any ordinary material, almost every atom is in its ground state. For a 1.96 eV gap at room temperature, Boltzmann’s factor $e^{−ΔE/kT}$ puts about one atom in 10³³ in the upper level; even at the Sun’s 5,800 K it is only 2 %. Light is absorbed. A beam can grow only where more atoms are up than down, $N_2 > N_1$: a population inversion. Heating never makes one, since thermal equilibrium always favors the lower level.
The obvious way to put atoms up is to shine strong light on them at the gap itself. It fails. The pumping light lifts ground atoms, but it stimulates excited atoms back down equally often per atom, and spontaneous emission drains them further. With a pump strength $Wτ$ (the pump transitions per atom in one lifetime), the steady share excited is
two levels: $N_2/N = Wτ/(1 + 2Wτ)$
which creeps toward one half as the pump is turned up, but never reaches it. A two-level medium pumped hard becomes transparent, never a gain medium.
Getting past one half needs atoms with more than two levels.
Three and four levels
The way out is to pump at a different transition from the one that lases. In a three-level scheme the pump lifts atoms from the ground to a high, short-lived level, which empties within nanoseconds into a metastable level, the upper laser level. The laser transition runs from there back to the ground. Because the pump never sees the laser level, it never drives atoms down from it. With pump rate $W_p$ per ground atom, the upper laser level holds $W_pτ/(1 + W_pτ)$ of the atoms, so the inversion is
three levels: $(N_2 − N_1)/N = (W_pτ − 1)/(W_pτ + 1)$
Inversion needs $W_pτ > 1$: more than half the atoms must be pumped before there is any gain, because the lower laser level is the ground state, where every atom starts. Ruby, the first laser (Theodore Maiman, 1960), works this way, and needed a powerful flash lamp.
A four-level scheme adds one more step: the laser transition ends on a level above the ground that empties in nanoseconds. That lower laser level is nearly always empty, so every atom in the upper laser level counts as inversion, and the inversion is $W_pτ/(1 + W_pτ)$, positive for any pump at all. Most lasers, from the helium–neon laser to Nd:YAG, use four levels for this reason.
An inverted medium amplifies light, but only along the path the light takes through it. To build a beam, the light must be sent back through it again and again.
The cavity and the threshold
An optical cavity is two mirrors facing each other along an axis, with the gain medium between them. The pumped atoms emit spontaneously in every direction, and nearly all of those photons leave through the sides after one pass. A photon that happens to travel along the axis is sent back by each mirror and crosses the medium hundreds of times, gathering twins as it goes, and its twins travel with it. The cavity picks one direction out of all directions and amplifies it over and over. One mirror, the output coupler, lets a few percent of the light through: that is the beam.
On each round trip the light passes the gain medium twice and meets each mirror once. If one pass multiplies the light by $G$ and the mirrors reflect $R_1$ and $R_2$, a round trip multiplies it by $G^2R_1R_2$. If that is less than 1, any light dies away and the medium only glows. If it is more than 1, the light grows on every trip. The dividing line is the threshold:
threshold: $G^2R_1R_2 = 1$
Above threshold the light does not grow forever. Each stimulated photon uses up an excited atom, so as the light grows it pulls the inversion down until the gain exactly balances the loss. The inversion stays there, and every extra atom the pump supplies becomes another photon in the beam.
What makes laser light different
Laser light is made of ordinary photons. A 1 mW red laser and a 1 mW red LED of the same color emit photons of the same energy, $hc/λ$, and the same number of them each second:
photons per second = $P/(hc/λ)$
What differs is how those photons are organized. Because the beam is copies of copies, laser light is monochromatic (a spectrum far narrower than a lamp’s), directional (it spreads by a milliradian or so instead of in every direction) and coherent (its waves stay in step over meters, where a lamp’s fall out of step within micrometers). Directionality alone makes a milliwatt beam far brighter on a distant wall than a lamp of the same power, which spreads its light over a whole sphere.
Laser safety
The same directionality makes lasers dangerous to the eye. The eye focuses a parallel beam to a spot on the retina only some tens of micrometers across, concentrating its power per area about a hundred thousand times, so even a few milliwatts can do lasting harm, and infrared beams are invisible and give no warning at all. Never look into a beam or its mirror reflection, and never point a laser at a person, a vehicle or an aircraft.
Every laser product carries a hazard class on its label (the international standard IEC 60825-1). In outline: class 1 is safe in normal use; class 2 covers visible lasers up to 1 mW, where the natural reflex to look away and blink normally limits a brief accidental exposure (staring into one is still unsafe); class 3R visible lasers reach 5 mW and can injure an eye that looks into the beam; class 3B lasers, up to 0.5 W, injure the eye from the direct beam or a mirror reflection; class 4 lasers are hazardous to eyes and skin even from scattered light, and can start fires. This outline is here to help you read a label. It is not laser-safety training: anyone working with class 3B or class 4 lasers needs their institution’s laser-safety training and must follow its rules.
Part 2 · Going further
Math you’ll need: derivatives and first-order differential equations, logarithms and exponentials, and solving an equation numerically.
Einstein’s coefficients
Einstein asked how two-level atoms could sit in equilibrium with blackbody radiation of energy density $ρ(ν)$ per unit frequency. Absorption runs at $N_1B_{12}ρ$, spontaneous emission at $N_2A_{21}$, and his new process, stimulated emission, at $N_2B_{21}ρ$. In equilibrium the up and down rates balance, $N_1B_{12}ρ = N_2(A_{21} + B_{21}ρ)$, with Boltzmann’s ratio $N_2/N_1 = e^{−hν/kT}$ for equal degeneracies. Solving for $ρ$ reproduces Planck’s law at every temperature only if $B_{21} = B_{12}$ and
$A_{21}/B_{21} = 8πhν^3/c^3$
Without stimulated emission the balance gives Wien’s law, right only at high frequency. One coefficient therefore fixes the other two. Two consequences matter for lasers. First, in thermal light the ratio of spontaneous to stimulated emission is
$A_{21}/(B_{21}ρ) = e^{hν/kT} − 1$
which is enormous for visible light at any ordinary temperature: stimulated emission becomes important only with intense, non-thermal light inside an inverted medium. Second, $A/B$ grows as $ν^3$: at ten times the frequency, spontaneous emission drains an inversion a thousand times faster relative to stimulated emission, which is why X-ray lasers need extreme pumping.
Rate equations and gain clamping
A laser is described by rate equations for its populations and its photons. In an ideal four-level medium the lower laser level is always empty, so the inversion is just $N_2$. Below threshold, with a pump that fills the upper level at $R_p$ atoms per unit volume per second,
$dN_2/dt = R_p − N_2/τ$
whose solution from an empty start is $N_2 = R_pτ(1 − e^{−t/τ})$: the inversion climbs toward $R_pτ$ over a few lifetimes. The sim solves the full set of equations for two, three and four levels, with finite decay times and the photon density $φ$ in the cavity:
$dN_2/dt = R_p − N_2/τ − σcφ(N_2 − N_1)$ $dφ/dt = σcφ(N_2 − N_1) − φ/τ_c + βN_2/τ$
The last term is the small share $β$ of spontaneous emission that happens to go into the lasing mode: the seed. Above threshold the steady state needs $dφ/dt$ = 0 with $φ$ large, so $σc(N_2 − N_1) = 1/τ_c$ (in a gas, where $n ≈ 1$): the inversion is held at its threshold value, $N_{th} = 1/(σcτ_c)$, whatever the pump. This is gain clamping. With the inversion fixed, the first equation says the extra pump, $R_p − N_{th}/τ$, all goes into stimulated emission, so the photon number and the output grow in a straight line with the pump.
Gain and the threshold condition
In an inverted medium a beam grows exponentially, $I(x) = I_0e^{gx}$, with the gain coefficient $g = σΔN$, where $σ$ is the stimulated-emission cross section and $ΔN = N_2 − N_1$. One pass through a length $L$ multiplies the light by
$G = e^{σΔNL}$
Real cavities also lose light inside, to scattering and absorption, at a rate $α$ per meter. A round trip returns $R_1R_2e^{2(g − α)L}$ of the light, and threshold is where that equals 1:
$g_{th} = α + \ln(1/R_1R_2)/(2L)$
The required inversion follows, $N_{th} = g_{th}/σ$. A short cavity with a leaky output mirror needs a great deal of gain per meter; a long one with good mirrors needs very little, which is why a helium–neon laser, whose gain is only a few percent per pass, has mirrors better than 99 %.
Photon lifetime, Q and efficiency
Switch the gain off and the light in a cavity dies away exponentially. Each round trip takes $2nL/c$ and keeps $R_1R_2e^{−2αL}$ of the light, so the photon lifetime is
$τ_c = 2nL/(c\ln(1/R_1R_2))$
with no internal loss ($2αL$ joins the logarithm when there is). The cavity is a resonator, and its quality factor is $Q = 2πντ_c$: the number of optical cycles, times 2π, that the light survives. Laser cavities have $Q$ values of millions to billions.
Above threshold an ideal four-level laser turns every extra pumped atom into a laser photon. The photon leaves through the output mirror with probability $η_{out} = \ln(1/R_2)/[\ln(1/R_1R_2) + 2αL]$ and is lost inside otherwise; and each pump photon of wavelength $λ_p$ becomes at most one laser photon of the longer wavelength $λ_l$, the rest of its energy heating the crystal (the quantum defect). The output rises above threshold with the slope efficiency $η = η_{out}\,λ_p/λ_l$: $P_{out} = η(P − P_{th})$.
Modes and coherence
A cavity sustains only standing waves that fit a whole number of half-wavelengths into its length. Their frequencies, the longitudinal modes, are spaced
$Δν = c/(2nL)$
The gain medium amplifies over a band of frequencies (in a gas, the Doppler width of its moving atoms: 1.5 GHz for neon at the helium–neon laser’s temperature), and every mode whose gain beats the loss can lase. A long cavity packs many modes under the gain curve; a short one may hold only one.
The width of a source’s spectrum sets how far its light stays in step with itself, the coherence length, about $c/Δν$, or in wavelengths
$l_c = λ^2/Δλ$
A single-mode laser’s line can be a megahertz wide or less, giving hundreds of meters; a lamp’s spectrum tens of nanometers wide gives tens of micrometers. That difference is why interferometers and holography need lasers.
Gaussian beams
A laser beam is not perfectly parallel: its transverse profile is a Gaussian, and diffraction makes it spread. At its narrowest point, the waist, its radius is $w_0$; the beam stays nearly that narrow over the Rayleigh range $z_R = πw_0^2/λ$, and its radius at a distance $z$ from the waist is
$w = w_0\sqrt{1 + (z/z_R)^2}$
Far beyond $z_R$ the radius grows in a straight line at the divergence angle $θ = λ/(πw_0)$. A wider waist spreads less, which is why a beam expander, a small telescope run backward, is used to send a beam a long way.
Spikes and giant pulses
Switched on, a solid-state laser overshoots. The inversion climbs past threshold before the light, starting from a few spontaneous photons, has grown enough to stop it; then a burst of light pulls the inversion far down, the light dies, and the cycle repeats. Linearizing the rate equations about the steady state gives a damped oscillation, the relaxation oscillations, at
$ω_R = \sqrt{(r − 1)/(ττ_c)}$
with $r$ the pump over its threshold value, damped at the rate $r/2τ$. Lasers whose photon lifetime is longer than the upper level’s lifetime, such as the helium–neon laser, do not ring at all.
Q-switching exploits the same physics deliberately. A shutter in the cavity (an electro-optic cell, or a saturable absorber) keeps the losses high while the pump builds the inversion far past threshold. Opened, the shutter lets the light grow so fast that it dumps the stored inversion in one pulse a few photon lifetimes long. In an ideal four-level laser the inversion left afterward, $N_f$, satisfies $\ln(N_i/N_f) = (N_i − N_f)/N_{th}$; the pulse overshoots and takes most of the stored energy, far more than just the excess over threshold.
Reference
Common mistakes
Part 1
- Thinking a photon with more energy than a gap is absorbed: it must match the gap.
- Thinking a stimulated photon goes in a random direction: it copies its trigger exactly.
- Counting only the stimulated photons: absorption by ground atoms removes photons too.
- Expecting a hard-pumped two-level atom to invert: it never passes one half.
- Forgetting that a three-level medium needs more than half its atoms pumped.
- Thinking the mirrors trap every photon: only those along the axis return.
- Using the output mirror’s whole loss for one pass: a round trip passes the gain twice.
- Thinking laser photons carry more energy, or that there are more of them, than a lamp’s of the same power and color.
Part 2
- Expecting the inversion to rise above threshold with more pump: it is clamped, and the output rises instead.
- Dividing the mirrors’ loss by $L$ instead of $2L$, or leaving out the internal loss.
- Using $c/L$ for the mode spacing, or forgetting the center mode when counting.
- Leaving the $π$ out of a beam’s divergence, or the $2π$ out of a frequency.
The equation sheet
| Part 1 | |
|---|---|
| a photon | $E = hf = hc/λ$ |
| spontaneous decay | $N = N_0e^{−t/τ}$ |
| gain or loss | net per 100 = 100($N_2 − N_1$)/$N$ |
| two levels | two levels: $N_2/N = Wτ/(1 + 2Wτ)$ |
| three levels | three levels: $(N_2 − N_1)/N = (W_pτ − 1)/(W_pτ + 1)$ |
| four levels | $(N_2 − N_1)/N = W_pτ/(1 + W_pτ)$ |
| threshold | threshold: $G^2R_1R_2 = 1$ |
| a beam’s photons | photons per second = $P/(hc/λ)$ |
| Part 2 | |
|---|---|
| Einstein | $A_{21}/B_{21} = 8πhν^3/c^3$ |
| thermal light | $A_{21}/(B_{21}ρ) = e^{hν/kT} − 1$ |
| pumping | $N_2 = R_pτ(1 − e^{−t/τ})$ |
| single-pass gain | $G = e^{σΔNL}$ |
| threshold gain | $g_{th} = α + \ln(1/R_1R_2)/(2L)$ |
| photon lifetime | $τ_c = 2nL/(c\ln(1/R_1R_2))$ |
| slope efficiency | $P_{out} = η(P − P_{th})$, $η = η_{out}λ_p/λ_l$ |
| modes | $Δν = c/(2nL)$ |
| coherence | $l_c = λ^2/Δλ$ |
| a Gaussian beam | $w = w_0\sqrt{1 + (z/z_R)^2}$, $z_R = πw_0^2/λ$ |
| relaxation | $ω_R = \sqrt{(r − 1)/(ττ_c)}$ |
| Q-switching | $\ln(N_i/N_f) = (N_i − N_f)/N_{th}$ |
Symbols: $h$ Planck’s constant; $c$ the speed of light; $f$, $ν$ frequency; $λ$ wavelength ($λ_p$ pump, $λ_l$ laser); $N$ atoms (per volume in Part 2), $N_1$, $N_2$ the lower and upper laser levels’ populations and $ΔN = N_2 − N_1$; $τ$ the upper level’s lifetime; $W$, $W_p$ pump rates per atom; $R_p$ a pump rate per volume; $G$ the gain on one pass; $R_1$, $R_2$ the mirrors’ reflectivities; $P$ a power; $A_{21}$, $B_{21}$, $B_{12}$ Einstein’s coefficients and $ρ$ the radiation’s energy density per unit frequency; $k$ Boltzmann’s constant and $T$ the temperature; $σ$ the stimulated-emission cross section; $g$ the gain coefficient and $α$ the internal loss, both per meter; $L$ the cavity’s length and $n$ its index; $τ_c$ the photon lifetime; $η$ an efficiency; $Δν$ a frequency width or spacing, $Δλ$ a wavelength width; $l_c$ the coherence length; $w$, $w_0$ a beam’s radius and waist, $z$ the distance and $z_R$ the Rayleigh range; $r$ the pump ÷ threshold; $ω_R$ the relaxation frequency; $N_i$, $N_f$, $N_{th}$ the inversions before and after a pulse, and at threshold.
Sources and further reading
- OpenStax, University Physics Volume 3 (S. J. Ling, J. Sanny and W. Moebs), Chapter 8, “Atomic Structure,” §8.6 “Lasers,” CC BY-NC-SA 4.0 (openstax.org): the Intro track’s first ideas and terms, stimulated emission, population inversion, metastable states and coherent light. The pumping schemes, the cavity, the threshold and Part 2 go beyond that section.
- A. E. Siegman, Lasers (University Science Books, 1986); P. W. Milonni and J. H. Eberly, Laser Physics (Wiley, 2010); O. Svelto, Principles of Lasers, 5th ed. (Springer, 2010): rate equations, threshold, modes, Gaussian beams and pulses.
- A. Einstein, “Zur Quantentheorie der Strahlung,” Physikalische Zeitschrift 18, 121 (1917); T. H. Maiman, Nature 187, 493 (1960); A. Javan, W. R. Bennett and D. R. Herriott, Phys. Rev. Lett. 6, 106 (1961).
- This sim’s Field Manual (H) for the model and its limits; the Quantum Atom Lab for energy levels and spectra; Blackbody Radiation for Planck’s law; the Michelson Interferometer for coherence measured by fringes.
Lasers & Stimulated Emission v1.0 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org