Population Inversion and Optical Gain

Send a beam of light into ordinary matter and it fades: ground-state atoms absorb photons and the intensity dies away as $I(z)=I_0e^{-\alpha z}$ . Population inversion is the trick that turns that decay into growth, and it takes two results from the articles before it one step further. From aq-optics-stimulated-emission-gain-coefficient it inherits the deepest fact of light–matter coupling: a single Einstein coefficient $B$ governs both absorption and stimulated emission, so the net effect of a beam is fixed not by how many atoms there are but by the difference between the upper- and lower-level populations, $\Delta N=N_2-(g_2/g_1)N_1$ . From aq-optics-laser-rate-equations it inherits the bookkeeping — pump in, spontaneous decay out — that decides where that difference settles in steady state. Put those together and a simple rule emerges: the absorption coefficient is proportional to $\Delta N$ , so making $\Delta N$ positive — more atoms upstairs than downstairs — turns absorption into a gain coefficient, and the beam grows exponentially as coherent, in-step photons are added by stimulated emission. That inverted, non-equilibrium state is the amplifying medium at the heart of every laser and optical amplifier. Reaching it is the hard part: thermodynamics forbids inversion at any positive temperature, and a closed two-level atom, however hard you pump it, only bleaches to transparency. This article builds the quantitative machinery — the sign of the gain, the gain cross-section, the impossibility proof, saturation, and the threshold where an amplifier becomes a laser.

Key takeaways

Population inversion is the non-equilibrium condition $N_2>(g_2/g_1)N_1$ — more atoms upstairs than down, per unit degeneracy — which reverses the sign of the net absorption.
It is the microscopic prerequisite for optical gain: only a positive inversion lets a beam grow, $I(z)=I_0e^{+\gamma z}$ , instead of decaying.
The gain coefficient is $\gamma(\nu)=\Delta N\,\sigma(\nu)$ , inversion density times stimulated-emission cross-section, and $\gamma$ inherits the sign of $\Delta N$ .
A thermal population always has $N_2<N_1$ , so inversion cannot be reached by heating — it needs a pump and at least a three-level scheme, because a closed two-level system saturates at transparency.
Gain saturates as the beam grows, $\gamma(I)=\gamma_0/(1+I/I_s)$ , and competes with cavity loss, so a laser turns on only once $\Delta N$ passes the threshold $\Delta N_{\rm th}$ .
Key idea: a beam is amplified rather than absorbed precisely when the population difference $\Delta N=N_2-(g_2/g_1)N_1$ is positive; this inverted, non-equilibrium state makes stimulated emission dominate and turns the absorption coefficient into a negative one — optical gain.

The core insight

A beam of light is amplified rather than absorbed precisely when the population difference $\Delta N=N_2-(g_2/g_1)N_1$ is positive; this inverted, non-equilibrium condition makes stimulated emission dominate over absorption and turns the absorption coefficient negative — that negative absorption is optical gain, $\gamma=\Delta N\,\sigma$ .

Where this lives today

The chain from inversion to gain to threshold is the design language of every coherent-light source. In a fibre backbone the signal crossing an ocean is re-amplified every fifty to a hundred kilometres by an erbium-doped fibre amplifier (EDFA): a $980\ \mathrm{nm}$ pump holds an inversion in $\mathrm{Er^{3+}}$ ions so that $\gamma=\Delta N\,\sigma$ boosts the $1.55\ \mu\mathrm{m}$ telecom band by tens of decibels with no electronics in the light path — a mirror-free single-pass gain medium that saturates gracefully at its own $I_s$ . The same equations size the Nd:YAG and fibre lasers of the worked examples, while titanium:sapphire crystals invert across a huge bandwidth to make the femtosecond pulses behind frequency combs. Laser diodes and semiconductor amplifiers swap optical pumping for current injection but obey identical bookkeeping, and cavity-free gain media — from astrophysical masers in molecular clouds to tabletop soft-X-ray lasers — show the same principle from radio frequencies to the X-ray edge.

The physical picture

Picture a beam of photons walking through a crowd of atoms. Every atom in the lower level $E_1$ can catch a passing photon and jump up, removing it from the beam (absorption). Every atom already in the upper level $E_2$ can be struck and induced to emit a second photon in lockstep with the first — same frequency, phase, and direction (stimulated emission). Because the same Einstein coefficient $B$ sets both rates, catching and cloning cancel exactly when the levels are equally populated. Tip the balance toward the lower level and the crowd absorbs; tip it toward the upper level and it amplifies. The whole subject is which way that balance tips. In everyday matter the balance is never in doubt. Thermal equilibrium stacks atoms into the lowest states, so $N_1$ exceeds $N_2$ and light is absorbed. Heating cannot fix it: the Boltzmann factor $e^{-h\nu/k_BT}$ stays below one at every positive temperature, and even at $3000\ \mathrm{K}$ a red transition ( $h\nu\approx1.9\ \mathrm{eV}$ ) has an upper-to-lower ratio of only $e^{-7.35}\approx6\times10^{-4}$ . An inverted population would require a negative absolute temperature — a state no equilibrium can occupy. Inversion is therefore engineered and driven: you pump atoms upstairs faster than they fall, and arrange the levels so the lower laser level empties quickly.

Figure

Thermal equilibrium versus population inversion: in the thermal medium the lower level E₁ is crowded so ΔN is negative and the beam decays as I₀ e^(−αz), while pumping inverts the populations so ΔN is positive and the same beam grows as I₀ e^(+γz) with γ = ΔN σ ≈ 2.8 m⁻¹ for Nd:YAG

Gain does not run away forever. Two things rein it in. First, the beam feeds on the very inversion that produces it: each stimulated event removes an atom from the upper level, so a strong beam depletes $\Delta N$ and the gain sags — gain saturation, the effect that stabilises a laser's output power. The gain a weak probe sees, $\gamma_0$ , is the small-signal gain; the intensity at which the inversion is half-depleted and the gain drops to $\gamma_0/2$ is the saturation intensity $I_s$ . Second, every mirror and scattering site subtracts photons, so whether a beam blows up into a laser oscillation or dies out is a race between gain per pass and loss per pass. The second figure shows the saturation curve for a real medium.

Figure

Gain saturation in Nd:YAG at 1064 nm: the small-signal gain γ₀ holds while the beam is weak, then falls as the intensity I approaches the saturation intensity I_s ≈ 2.9×10⁷ W m⁻² where the inversion is half-depleted and the gain has dropped to γ₀/2

The core equations

The beam's fate is set by a first-order growth law: when the gain coefficient $\gamma(\nu)$ is positive the intensity grows exponentially, and when it is negative the same equation is Beer–Lambert absorption. Everything else here is a computation of $\gamma$ or a statement about its sign.

\frac{\mathrm{d}I}{\mathrm{d}z}=\gamma(\nu)\,I(z),\qquad I(z)=I_0\,e^{\gamma(\nu)z} \qquad (1)

where: $I(z)$ — beam intensity at depth $z$ ( $\mathrm{W\,m^{-2}}$ ); $\gamma(\nu)$ — gain coefficient at frequency $\nu$ ( $\mathrm{m^{-1}}$ ), negative for a lossy medium; $z$ — distance through the medium. Over one gain length $1/\gamma$ the intensity multiplies by $e$ ; flip the sign and you recover attenuation. The gain coefficient factorises into an atomic part and a population part: the cross-section $\sigma(\nu)$ is the effective photon-catching area of one atom, and the inversion density $\Delta N$ counts the net emitters per unit volume. Their product is the gain.

\gamma(\nu)=\Delta N\,\sigma(\nu),\qquad \Delta N=N_2-\frac{g_2}{g_1}N_1 \qquad (2)

where: $\Delta N$ — population inversion density ( $\mathrm{m^{-3}}$ ); $N_1,N_2$ — lower/upper populations; $g_1,g_2$ — degeneracies; $\sigma(\nu)$ — stimulated-emission cross-section ( $\mathrm{m^{2}}$ ). The degeneracy-weighted combination, not the bare $N_2-N_1$ , sets the sign — a subtlety that trips up many first calculations. To see why equilibrium can never supply inversion, compare the populations at temperature $T$ : their ratio is a Boltzmann factor times the degeneracy ratio, and the Boltzmann factor is bounded above by one.

\frac{N_2}{N_1}=\frac{g_2}{g_1}\exp\!\left(-\frac{h\nu}{k_BT}\right) \qquad (3)

where: $h\nu=E_2-E_1$ — transition energy; $k_B$ — Boltzmann constant; $T$ — absolute temperature. Because $e^{-h\nu/k_BT}\le1$ , always $N_2/N_1\le g_2/g_1$ , hence $\Delta N\le0$ : an equilibrium medium absorbs. The inverted ratio would demand $T<0$ — inversion is literally a negative-temperature state. The cross-section is not free; it is tied to the spontaneous decay of the same transition. The degeneracy-weighted $B$ coefficients are equal, and the $A/B$ ratio is fixed by the mode density of the field.

g_1B_{12}=g_2B_{21},\qquad \frac{A_{21}}{B_{21}}=\frac{8\pi h\nu^3 n^3}{c^3} \qquad (4)

where: $B_{12},B_{21}$ — absorption and stimulated-emission coefficients; $A_{21}=1/\tau_{\rm sp}$ — spontaneous rate; $n$ — host refractive index; $c$ — vacuum speed of light. The first identity is why absorption and stimulated emission share one strength; the second lets a measured fluorescence lifetime predict a gain. Combining those relations with the lineshape gives the cross-section in terms of measurable quantities — the wavelength, the upper-level lifetime, and the linewidth.

\sigma(\nu)=\frac{c^2}{8\pi n^2\nu^2\,\tau_{\rm sp}}\,g(\nu),\qquad \sigma_{\rm peak}\sim\frac{\lambda^2}{\tau_{\rm sp}\,\Delta\nu} \qquad (5)

where: $\tau_{\rm sp}$ — spontaneous (radiative) lifetime of the upper level; $g(\nu)$ — normalised lineshape, $\int g(\nu)\,\mathrm{d}\nu=1$ ; $\Delta\nu$ — linewidth. The scaling is the design rule: a short lifetime and narrow line give a large cross-section, so a little inversion buys a lot of gain. A real amplifier loses its small-signal gain as the beam grows, because stimulated emission drains the inversion feeding it. The gain rolls off as a Lorentzian in intensity, halving at $I_s$ .

\gamma(I)=\frac{\gamma_0(\nu)}{1+I/I_s(\nu)},\qquad I_s=\frac{h\nu}{\sigma(\nu)\,\tau} \qquad (6)

where: $\gamma_0$ — small-signal (unsaturated) gain coefficient; $I_s$ — saturation intensity, the flux that de-excites atoms as fast as pumping refills them; $\tau$ — effective upper-level recovery time. $I_s$ is one photon energy per cross-section per recovery time — where the beam starts competing with the pump for the inversion. Finally, an amplifier becomes a laser only when round-trip gain overcomes round-trip loss; setting them equal defines the threshold inversion the pump must reach.

\Delta N_{\rm th}=\frac{\gamma_{\rm th}}{\sigma}=\frac{1}{\sigma}\left(\alpha_i+\frac{1}{2L}\ln\frac{1}{R_1R_2}\right) \qquad (7)

where: $\Delta N_{\rm th}$ — inversion density at threshold; $\alpha_i$ — internal (distributed) loss; $L$ — gain-medium length; $R_1,R_2$ — mirror power reflectivities. The logarithm is the output-coupling loss spread over a round trip; below $\Delta N_{\rm th}$ the medium amplifies but cannot oscillate.

Derivation ASign of the gain from Einstein rate bookkeeping

The goal is to prove the fact on which the whole subject rests: the net rate at which a monochromatic beam gains photons — and hence the gain coefficient — is proportional to the degeneracy-weighted inversion $\Delta N=N_2-(g_2/g_1)N_1$ , so the beam grows if and only if $\Delta N>0$ . We track only the coherent, in-mode photons — absorption removes them, stimulated emission adds them, spontaneous emission scatters elsewhere and is set aside — and the rest is pure counting once the Einstein relation between the two $B$ coefficients is used.

Step 1. In a field of spectral energy density $\rho(\nu)$ , atoms are lifted out of the beam at rate $R_{\rm up}=B_{12}N_1\rho(\nu)$ and induced to emit into it at rate $R_{\rm down}=B_{21}N_2\rho(\nu)$ . Why: these two processes exchange energy coherently with the mode; spontaneous emission radiates into $4\pi$ and does not add to the directional beam, so it is dropped.

Step 2. The net photon-addition rate is $R_{\rm net}=R_{\rm down}-R_{\rm up}=\rho(\nu)\,(B_{21}N_2-B_{12}N_1)$ , and the beam grows only if $R_{\rm net}>0$ . Why: the shared factor $\rho(\nu)$ shows both rates scale with the field, so their competition is decided entirely by the population-weighted coefficients.

Step 3. Insert the Einstein relation $g_1B_{12}=g_2B_{21}$ , i.e. $B_{12}=(g_2/g_1)B_{21}$ , giving $R_{\rm net}=\rho(\nu)\,B_{21}\big(N_2-\tfrac{g_2}{g_1}N_1\big)$ . Why: this substitution collapses two coefficients into one and reveals the degeneracy-weighted inversion $\Delta N$ as the sole factor whose sign matters.

Step 4. Each net transition deposits energy $h\nu$ ; converting the volumetric rate to a spatial intensity derivative gives $\mathrm{d}I/\mathrm{d}z=h\nu\,R_{\rm net}(n/c)=\Delta N\,\sigma(\nu)\,I$ , the atomic constants absorbed into $\sigma(\nu)$ . Why: this promotes the rate equation into the propagation law (1), and the sign of $\gamma$ follows the sign of $\Delta N$ — invert the populations and absorption becomes gain. *Verified: reproduces Loudon ch. 1 and Saleh and Teich §14.1; recompute by checking $g_1B_{12}=g_2B_{21}$ and that $\mathrm{d}I/\mathrm{d}z=\gamma I$ integrates to $I_0e^{\gamma z}$ .*

Go deeper · Derivation B · 1 minWhy a driven two-level system never inverts

Derivation BWhy a driven two-level system never inverts

The goal is to show, with the rate equations rather than a slogan, that no amount of optical pumping can invert a closed two-level atom: the steady-state inversion falls monotonically toward zero and stops at transparency, never crossing into gain — which is why real lasers cannot be built on a single pair of levels. We solve for the populations of a two-level system of total density $N=N_1+N_2$ driven by a pump that couples the levels at rate $W=B\rho$ , and take the strong-pump limit.

Step 1. With $g_1=g_2$ the upper-level rate equation is $\mathrm{d}N_2/\mathrm{d}t=W(N_1-N_2)-N_2/\tau$ , with $\tau$ the spontaneous lifetime. Why: the same $W$ drives absorption ( $N_1\to N_2$ ) and stimulated emission ( $N_2\to N_1$ ) because one $B$ governs both — a pump term symmetric in the populations, which is what forbids inversion.

Step 2. In steady state set $\mathrm{d}N_2/\mathrm{d}t=0$ and use $N_1=N-N_2$ , giving $0=W(N-2N_2)-N_2/\tau$ . Why: conservation of total population turns two unknowns into one and lets us solve the split directly.

Step 3. Solving, $N_2=\dfrac{W\tau N}{1+2W\tau}$ , so $\Delta N=N_1-N_2=\dfrac{N}{1+2W\tau}$ . Why: the factor of two in the denominator — both stimulated channels — guarantees $\Delta N$ stays on the lower-level side for every finite pump.

Step 4. As $W\to\infty$ the populations equalise, $N_1=N_2=N/2$ , so $\Delta N\to0$ and $\gamma\to0$ : the medium bleaches to transparency but never amplifies. Why: a two-level atom can at best be made transparent, so genuine gain needs a third or fourth level and asymmetric relaxation that keeps the lower laser level empty. *Verified: matches the saturation result in Foot §7.3 and Demtröder Vol. 1 §2; recompute the steady state and take $W\to\infty$ to confirm $\Delta N\to0^+$ .*

Limiting cases

Case	Result	Meaning
Thermal equilibrium	$N_2/N_1=(g_2/g_1)e^{-h\nu/k_BT}<g_2/g_1$	$\Delta N<0$ at every $T>0$ : the medium absorbs. Inversion is impossible in equilibrium — it is a negative-temperature state.
Two-level saturation	$\Delta N=N/(1+2W\tau)\to0$	Hard optical pumping of a closed two-level system drives the populations equal; the medium bleaches to transparency ( $\gamma\to0$ ) but never amplifies.
Full inversion	$N_2\to N_t,\ N_1\to0,\ \gamma\to N_t\,\sigma$	Idealised four-level limit, every active atom upstairs: gain hits its ceiling, set by doping density and cross-section.
Small signal vs. saturated	$\gamma(I)=\gamma_0/(1+I/I_s)$	A weak probe sees the full $\gamma_0$ ; as $I\to I_s$ the beam depletes the inversion and the gain halves, then falls as $1/I$ .
Line centre vs. wings	$\sigma(\nu)\propto g(\nu)$	Gain is largest at line centre where $g(\nu)$ peaks and falls off in the wings; the amplified spectrum narrows toward $\nu_0$ .

Common trapA strong enough pump will invert any two-level atom

This is the most common misconception in laser physics, and the rate equations kill it cleanly. Because one coefficient $B$ governs both absorption and stimulated emission, pumping on the lasing transition drives the populations toward equality, not inversion: the steady state $\Delta N=N/(1+2W\tau)$ shrinks toward zero as the pump grows but stays on the absorbing side forever. You can bleach a two-level medium to transparency, never to gain. Real gain needs auxiliary levels — a third to pump through, or a fourth below the transition that empties fast — so population piles up upstairs while the lower laser level stays nearly empty. A quieter trap lives in the same formula: the inversion is the degeneracy-weighted $N_2-(g_2/g_1)N_1$ , not the naive $N_2-N_1$ , and ignoring the weights can flip a real absorber into an imagined amplifier.

Worked examples

1 · Single-pass gain of a Nd:YAG amplifier (confidence builder)

A neodymium:YAG rod amplifies at $\lambda=1064\ \mathrm{nm}$ , where the stimulated-emission cross-section is $\sigma=2.8\times10^{-23}\ \mathrm{m^2}$ . A cw lamp holds a modest inversion density $\Delta N=1.0\times10^{23}\ \mathrm{m^{-3}}$ over a rod of length $L=0.10\ \mathrm{m}$ . Find the small-signal gain coefficient, the single-pass power gain, and express it in decibels.

Step 1. The gain coefficient is inversion times cross-section (equation 2): $\gamma_0=\Delta N\,\sigma=(1.0\times10^{23})(2.8\times10^{-23})=\mathbf{2.8\ m^{-1}}$ . The gain length $1/\gamma_0\approx0.36\ \mathrm{m}$ is a few times the rod — the regime where cw solid-state amplifiers operate.

Step 2. Integrate the growth law (equation 1): the exponent is $\gamma_0 L=(2.8)(0.10)=0.28$ , so $G=e^{\gamma_0 L}=e^{0.28}=\mathbf{1.32}$ — the beam leaves about $32\%$ stronger on a single pass.

Step 3. In decibels, using $10\log_{10}e\approx4.34$ : $G_{\rm dB}=4.34\times0.28=\mathbf{1.22\ dB}$ ( $12.2\ \mathrm{dB\,m^{-1}}$ ). Fold the rod through a cavity a few dozen times and this modest per-pass gain becomes enormous — the point of mirrors. γ₀ = 2.8 m⁻¹, single-pass gain G = 1.32 (1.22 dB).

2 · The threshold inversion for lasing

Put that same rod ( $\sigma=2.8\times10^{-23}\ \mathrm{m^2}$ , $L=0.10\ \mathrm{m}$ ) in a cavity with a perfect back mirror $R_1=1.00$ , an output coupler $R_2=0.90$ , and a distributed internal loss $\alpha_i=0.20\ \mathrm{m^{-1}}$ . What inversion density must the pump reach before the laser turns on?

Step 1. Find the threshold gain (equation 7). The mirror loss over a round trip is $\dfrac{1}{2L}\ln\dfrac{1}{R_1R_2}=\dfrac{1}{0.20}\ln\dfrac{1}{0.90}=5\times0.1054=0.527\ \mathrm{m^{-1}}$ — the $10\%$ tapped off each round trip, which gain must replace.

Step 2. Add the internal loss: $\gamma_{\rm th}=\alpha_i+0.527=0.20+0.527=\mathbf{0.73\ m^{-1}}$ , the gain needed just to break even.

Step 3. Divide by the cross-section: $\Delta N_{\rm th}=\gamma_{\rm th}/\sigma=0.73/(2.8\times10^{-23})=\mathbf{2.6\times10^{22}\ m^{-3}}$ . The working inversion $1.0\times10^{23}\ \mathrm{m^{-3}}$ of example 1 sits comfortably above this, so the rod lases and clamps its gain back to $\gamma_{\rm th}$ . ΔN_th = 2.6×10²² m⁻³ — about a quarter of the operating inversion.

3 · The saturation intensity

For the same $1064\ \mathrm{nm}$ Nd:YAG transition the upper-level lifetime is $\tau=230\ \mu\mathrm{s}$ . Find the saturation intensity at which the gain halves.

Step 1. Photon energy: $h\nu=hc/\lambda=(6.63\times10^{-34})(3.00\times10^8)/(1.064\times10^{-6})=\mathbf{1.87\times10^{-19}\ J}$ (about $1.17\ \mathrm{eV}$ ).

Step 2. Apply $I_s=h\nu/(\sigma\tau)$ : $I_s=\dfrac{1.87\times10^{-19}}{(2.8\times10^{-23})(2.30\times10^{-4})}=\mathbf{2.9\times10^{7}\ W\,m^{-2}}\approx2.9\ \mathrm{kW\,cm^{-2}}$ — the value on the saturation figure. A weaker beam sees the full small-signal gain; a few times stronger and the inversion is heavily depleted, the feedback that pins a laser's steady-state power. I_s = 2.9×10⁷ W m⁻² ≈ 2.9 kW cm⁻².

Physics of the portrait

Population inversion lay latent in Einstein's 1917 coefficients for four decades before anyone saw how to exploit it. The person who turned "negative absorption" from a spectroscopic curiosity into a working amplifier — first for microwaves, then light — was Charles Townes.

Origins · 1 minCharles Townes and how this was discovered

Charles Townes

1915–2015 · the maser, the laser, and population inversion

Charles Hard Townes reasoned his way to amplification by stimulated emission on a park bench in Washington in 1951, sketching how a beam of molecules sorted into an excited state could be made to give up their energy coherently. The Einstein relations had said since 1917 that stimulated emission must exist and share a coefficient with absorption; Ladenburg and Kopfermann had detected "negative dispersion" in neon in 1928, and Fabrikant had proposed amplifying light with an engineered inversion in 1939 — but no one had built a device. Townes saw that the obstacle was thermodynamic: at any positive temperature the lower state wins, so you must physically sort or pump atoms to put more upstairs than down. In 1954, with Gordon and Zeiger, he realised the idea as the ammonia maser, electrostatically selecting excited molecules into a microwave cavity — the first working device built on a population inversion. In 1958 he and Arthur Schawlow extended the idea to optical frequencies in the paper that launched the laser, and he shared the 1964 Nobel Prize with Basov and Prokhorov. His enduring lesson is the one this article is built around: gain is not about how many atoms you have, but about the sign of the difference between two populations — and getting that sign positive is engineering against thermodynamics.

History

1917

— Einstein derives the A and B coefficients, defining stimulated emission and the shared absorption strength

1928

— Ladenburg and Kopfermann detect negative dispersion in neon, an early signature of stimulated emission

1939

— Fabrikant proposes amplifying light by an engineered population inversion

1954

— Gordon, Zeiger, and Townes operate the first maser on an inverted ammonia beam

1958

— Schawlow and Townes extend the inversion principle to optical frequencies, framing the laser

1960

— Maiman fires the first laser from population-inverted ruby

Going further

Texts. Hecht, Optics, 5th ed., §13.1 (lasers and gain) · Foot, Atomic Physics, §7.3–7.4 (Einstein coefficients, saturation) · Loudon, The Quantum Theory of Light, 3rd ed., ch. 1 (absorption and stimulated emission) · Saleh and Teich, Fundamentals of Photonics, 3rd ed., ch. 14–15 (laser amplifiers and oscillators) · Siegman, Lasers, ch. 3 and 7 (cross-sections, gain saturation) · Demtröder, Laser Spectroscopy, Vol. 1, ch. 2.

Provenance. The Einstein-coefficient sign argument and $\gamma=\Delta N\,\sigma$ follow Loudon ch. 1 and Saleh and Teich §14.1; the two-level saturation $\Delta N=N/(1+2W\tau)$ and the three/four-level requirement follow Foot §7.3 and Demtröder §2; the cross-section $\sigma=c^2g(\nu)/(8\pi n^2\nu^2\tau_{\rm sp})$ follows the Einstein $A/B$ inversion in Saleh and Teich §14.2 and Siegman ch. 3. A verifier should recompute the net-rate gain, the $W\to\infty$ transparency limit, the dB factor $10\log_{10}e\approx4.34$ , and the worked numbers $\gamma_0=2.8\ \mathrm{m^{-1}}$ , $G=e^{0.28}=1.32$ , $\Delta N_{\rm th}=2.6\times10^{22}\ \mathrm{m^{-3}}$ , $I_s=2.9\times10^{7}\ \mathrm{W\,m^{-2}}$ . The Nd:YAG constants ( $\sigma\approx2.8\times10^{-23}\ \mathrm{m^2}$ , $\tau\approx230\ \mu\mathrm{s}$ , $n\approx1.82$ ) are standard tabulated values.

Prerequisites. Stimulated Emission and the Gain Coefficient · Laser Rate Equations. Leads to. Laser Threshold Condition · Three-Level Laser Systems · Four-Level Laser Systems · Cavity Q Factor and Photon Lifetime.

Ask the tutor: ask me why your favourite laser is a three- or four-level system and how its inversion, cross-section, and threshold set the pump power it needs.

Is population inversion the same thing as optical gain?

They are cause and effect, not synonyms. Inversion is a statement about the atoms — $\Delta N=N_2-(g_2/g_1)N_1>0$ — while gain is a statement about the beam, that its intensity grows ( $\gamma>0$ ). The bridge is $\gamma=\Delta N\,\sigma$ : inversion is the necessary microscopic condition, and the cross-section $\sigma$ converts it into a growth rate at a given frequency. You can have inversion yet little useful gain if $\sigma$ is tiny or the light is off resonance. So "inverted" describes the medium; "gain" describes what it does to a beam.

Why does the degeneracy ratio g₂/g₁ appear in the inversion?

Because degenerate sub-levels multiply the ways an atom can sit in a level, and absorption and stimulated emission count those ways. The Einstein relation $g_1B_{12}=g_2B_{21}$ encodes it: absorption carries an extra factor $g_2/g_1$ , so the fair comparison is $N_2$ against $(g_2/g_1)N_1$ , not $N_2$ against $N_1$ . A high-degeneracy lower level is a bigger photon sink, so you need proportionally more upper-level population to break even. Dropping the ratio is a frequent error that can make an absorber look like an amplifier on paper.

Can heating a material ever create inversion?

No — even though it feels like it should work. Heating populates higher levels through the Boltzmann factor $e^{-h\nu/k_BT}$ , which is always less than one for positive $T$ . Raise the temperature toward infinity and the best you reach is equal populations ( $N_2/N_1\to g_2/g_1$ ) — transparency, not gain. To pass that point you would need a negative absolute temperature, which no equilibrium can occupy; even at $3000\ \mathrm{K}$ a visible transition has only about one atom upstairs per $1500$ down. Inversion is intrinsically driven: you must pump selectively, not heat uniformly.

What is the saturation intensity, and why does it matter?

The saturation intensity $I_s=h\nu/(\sigma\tau)$ is the intensity at which stimulated emission de-excites atoms about as fast as pumping and relaxation refill the upper level. At $I=I_s$ the gain has dropped to half its small-signal value; above it, it falls as $1/I$ . It matters because it is the feedback that stabilises a laser: as the intracavity field builds it saturates the gain down until gain equals loss, clamping the output power. It also sets how much energy an amplifier can extract before it stops amplifying efficiently. A large-cross-section, long-lifetime medium like Nd:YAG has a low $I_s$ and saturates easily; a short-lifetime medium needs a much stronger beam.

Why do real lasers need three or four levels instead of two?

Because a two-level system cannot be inverted by pumping on its own transition — the shared coefficient $B$ drives it only to transparency. The fix is to pump through a different transition. In a three-level laser (ruby) atoms are pumped to a high band that decays into a metastable upper laser level, but the lower laser level is the ground state, so you must invert more than half of all atoms before $\Delta N$ turns positive — hence ruby's ferocious threshold. A four-level laser (Nd:YAG) adds a lower laser level above the ground state that empties almost instantly, so it stays nearly empty, $\Delta N\approx N_2$ , and even weak pumping gives gain. The extra levels exist purely to break the symmetry that dooms the two-level scheme.

If gain is "negative absorption," is it the same physics as the Beer–Lambert law?

Yes — it is literally the same equation with a sign flipped. Beer–Lambert $I=I_0e^{-\alpha z}$ describes a beam losing intensity to a medium whose net population sits low. Write $\alpha=-\gamma$ and you have $I=I_0e^{+\gamma z}$ , exponential growth, the moment the inversion turns positive. The underlying process — stimulated transitions between two levels — is identical; only the population balance differs. That is why the gain coefficient carries units of inverse length like an absorption coefficient, and why the same spectrophotometer that measures absorption measures gain in an inverted medium: it simply reads a negative $\alpha$ .