Q-Switching

Q-switching is what you get when you refuse to let a laser lase. The laser rate equations say that the moment round-trip gain reaches the loss line, stimulated emission clamps the inversion: every extra atom the pump lifts is knocked straight back down as low-power light, and the inversion never climbs past the threshold value fixed by the cavity Q factor and photon lifetime. Q-switching breaks that clamp on purpose. Spoil the cavity — misalign a mirror, insert an absorber, rotate a prism out of line — and the loss grows so large that the threshold inversion is pushed far above anything the pump can supply. Oscillation is now forbidden, and the pump has nowhere to put its energy but into the inversion itself, which fills like a charging capacitor, limited only by the upper-state lifetime and parasitic spontaneous emission. Then, at a chosen instant, the switch is thrown: the cavity Q snaps from low to high, the threshold collapses far below the accumulated inversion, and a medium that stored energy for microseconds suddenly has a round-trip gain vastly exceeding loss. A handful of spontaneous photons seeds an avalanche; the intracavity field grows exponentially, tears the stored inversion down through threshold, and empties the reservoir in a single burst — the giant pulse — a few nanoseconds long carrying megawatts of peak power. The event obeys the same two coupled rate equations as steady-state lasing, read in a new regime: pumping and spontaneous decay freeze on the pulse timescale, leaving a nearly conservative system whose first integral yields the peak power, extracted energy, and pulse width analytically — all functions of one dimensionless number, the inversion ratio $r=N_i/N_t$ . Where mode-locking phases many modes for femtosecond brevity, Q-switching stores and dumps for nanosecond ferocity.

Key takeaways

Q-switching makes giant nanosecond pulses by holding the cavity at low Q (high loss) so the pump builds a population inversion far above the ordinary threshold, storing energy in the upper level instead of releasing it.
Restore high Q abruptly and round-trip gain hugely exceeds loss: a spontaneous-emission-seeded photon avalanche dumps the stored inversion in a pulse lasting a few cavity photon lifetimes $\tau_c$ .
One dimensionless number organizes the pulse dynamics — the inversion ratio $r=N_i/N_t$ — setting the available inversion relative to threshold and therefore strongly influencing the peak photon number, extraction efficiency $\eta_E$ , and pulse evolution, together with the cavity and material parameters.
Typical solid-state numbers: energy stored over a $\sim$ 200 μs lifetime is released in $\sim$ 20 ns, turning stored power into multi-megawatt peaks — a $10^4$ – $10^5$ concentration in time, not extra energy.
Switches range from spinning mirrors and Pockels cells to acousto-optic modulators and passive saturable absorbers; the upper-state lifetime must outlast the pump-plus-switching time or the inversion leaks away.
Key idea: keep the cavity lossy to store energy in the inversion, then suddenly restore high $Q$ so the accumulated inversion, now far above threshold, discharges as one short intense pulse whose peak power is strongly influenced by the inversion-to-threshold ratio $r=N_i/N_t$ together with the cavity and gain parameters.

The core insight

Store energy in the inversion by keeping the cavity lossy (low Q), then suddenly restore high Q so the accumulated inversion, now far above threshold, discharges as a short intense pulse whose peak power is set by the inversion-to-threshold ratio $r=N_i/N_t$ .

Where this lives today

The giant pulse is everywhere concentrated light is needed. A Q-switched Nd:YAG at 1.064 μm — often frequency-doubled to 532 nm — is the workhorse of rangefinders and lidar, of tattoo and pigment removal (the pulse shatters ink particles by photoacoustic stress before heat spreads), and of marking and micromachining, where megawatt peaks vaporize material cleanly while depositing little heat. In laser-induced breakdown spectroscopy (LIBS) a single pulse ablates a microplasma whose emission fingerprints a material's elements — Mars rovers carry exactly this instrument. At the largest scale, Q-switched oscillators seed the amplifier chains of inertial-confinement-fusion drivers. The modern frontier is miniaturization and self-timing. Passively Q-switched microchip lasers bond a thin gain crystal to a saturable absorber such as Cr⁴⁺:YAG: the absorber holds the cavity lossy until the building intensity bleaches it transparent, self-switching with no electronics. These millimetre-scale devices emit sub-nanosecond, kilowatt-peak pulses, and now sit inside handheld LIBS analyzers, ceramic-igniter "laser spark plugs," and compact lidar. The tradeoff is exactly the jitter derived above — the pulse fires whenever intensity reaches the bleaching point — which is why timing-critical systems still pay for a Pockels cell.

The physical picture

Picture the gain medium as a reservoir and the cavity output as a drain. In continuous-wave (CW) operation the drain is wide open: the instant the level reaches the rim — the threshold inversion — it spills out as low-power light and never climbs higher. Q-switching plugs the drain, so the pump builds a huge head of stored energy above the rim; pull the plug and that whole head discharges at once, the giant pulse dwarfing the CW trickle for the same reason a burst dam dwarfs a garden tap — the same water, released in a hundred-thousandth of the time. Two wildly separated clocks make this work. At low Q the inversion fills linearly, capped by the upper-state lifetime — microseconds in a good crystal. At high Q the photon number grows exponentially and the discharge is over in nanoseconds. That ratio is essentially the factor by which peak power beats average power: store for 200 μs, dump in 20 ns, and the same energy is concentrated in time by $\sim10^4$ .

Figure

The Q-switch cycle: during ~200 μs of low-Q pumping the inversion N climbs to N_i far above the high-Q threshold N_t while the output stays dark; the switch then restores high Q and the stored inversion dumps in a ~20 ns giant pulse peaking near 4 MW as N collapses to N_f

The pulse builds from noise: when the switch opens only a handful of spontaneous photons occupy the cavity mode, amplified through many round trips until they dominate. Because buildup depends only logarithmically on that tiny seed, the delay to the peak is long — tens of photon lifetimes — and jitters slightly shot to shot. The pulse then self-terminates: the avalanche drains the inversion, the gain falls, and the pulse peaks at the exact instant $N=N_t$ , where $\mathrm{d}\phi/\mathrm{d}t=0$ . Thereafter the inversion is below threshold and leftover photons leak out at the photon lifetime — a fast rise, a slower tail. The second figure turns this discharge into one universal curve: how much stored inversion comes out as light, versus the single number $r=N_i/N_t$ .

:::svg[Energy-extraction efficiency η_E as a universal function of the inversion ratio $r = N_i/N_t$ , from the transcendental relation $r\,\eta_E = \ln[1/(1-\eta_E)]$ : it climbs past 0.80 at $r = 2$ and 0.98 at $r = 4$ , so efficient giant-pulse extraction wants $r$ of at least about 3 to 5]

The core equations

Everything begins with the two coupled laser rate equations, restricted to the giant-pulse window. On the nanosecond timescale the pump and spontaneous decay are negligible, so their terms drop, leaving only stimulated emission and cavity loss — a nearly conservative pair, which is what makes Q-switching analytically tractable.

\frac{\mathrm{d}\phi}{\mathrm{d}t}=\left(\frac{N}{N_t}-1\right)\frac{\phi}{\tau_c},\qquad \frac{\mathrm{d}N}{\mathrm{d}t}=-\gamma\,\frac{N}{N_t}\,\frac{\phi}{\tau_c} \qquad (1)

where: $\phi$ — photons in the lasing cavity mode; $N$ — instantaneous population inversion (atom count); $N_t$ — threshold inversion in the high-Q state; $\tau_c$ — cavity photon lifetime at high Q; $\gamma$ — inversion-reduction factor per photon, $\gamma=1$ (four-level), $\gamma=2$ (three-level). Physically: photons grow whenever $N>N_t$ and stop growing exactly at $N=N_t$ ; each stimulated event mints one photon while costing the inversion $\gamma$ atoms. The threshold inversion is set not by the medium alone but by the cavity loss — the knob Q-switching turns. Equating round-trip gain and loss fixes $N_t$ , so spoiling the cavity (large loss, small $\tau_c$ ) inflates $N_t$ and restoring it collapses $N_t$ .

N_t=\frac{\alpha_{loss}\,V}{\sigma\,\ell}=\frac{V}{c\,\sigma\,\tau_c} \qquad (2)

where: $\sigma$ — stimulated-emission cross-section; $V$ — mode volume; $\ell$ — gain-medium length; $\alpha_{loss}$ — fractional round-trip loss; $c$ — speed of light in the medium. Both forms say the same thing: threshold scales with loss, inversely with photon lifetime. In the low-Q state $\alpha_{loss}$ is deliberately enormous, so $N_t^{\text{low}}\gg N_i$ and oscillation is forbidden while the pump fills the reservoir. Dividing the two rate equations eliminates time: the common factor $\phi/\tau_c$ cancels, leaving a first-order relation between photon number and inversion — the trajectory the system traces in the $(\phi,N)$ phase plane.

\frac{\mathrm{d}\phi}{\mathrm{d}N}=-\frac{1}{\gamma}\left(1-\frac{N_t}{N}\right)\;\;\Longrightarrow\;\; \phi(N)=\frac{1}{\gamma}\left[(N_i-N)-N_t\ln\frac{N_i}{N}\right] \qquad (3)

where: $N_i$ — stored inversion at switching, with $\phi\approx0$ (the negligible seed sets the constant to zero). This one curve contains the entire pulse: $\phi$ rises as $N$ falls from $N_i$ , peaks, and returns to zero at a residual inversion $N_f$ . The pulse peaks where net gain vanishes, $N=N_t$ . Substituting into the trajectory gives the peak intracavity photon number — the most useful quantity in Q-switching, since peak power follows directly.

\phi_{peak}=\frac{1}{\gamma}\left[\,N_i-N_t-N_t\ln\frac{N_i}{N_t}\right]=\frac{N_i}{\gamma}\left[1-\frac{1}{r}\left(1+\ln r\right)\right] \qquad (4)

where: $r=N_i/N_t$ — inversion ratio. For $r\gg1$ the bracket approaches unity, so $\phi_{peak}\to N_i/\gamma$ : essentially one output photon per stored atom, the near-unit efficiency of the giant pulse. Peak output power is the photon energy carried through the output coupler at the peak. Photons leave at rate $\phi/\tau_c$ , and only the fraction of loss that is useful transmission (not scatter or absorption) shows up as beam.

P_{peak}=\frac{h\nu\,\phi_{peak}}{\tau_c}\,\frac{\delta_{out}}{\delta} \qquad (5)

where: $h\nu$ — photon energy; $\delta_{out}$ — output-coupling loss per round trip; $\delta$ — total round-trip loss; $\delta_{out}/\delta$ — useful-output fraction (unity when the output coupler dominates). With $\phi_{peak}\sim N_i$ and $\tau_c\sim10$ ns, a reservoir of $10^{17}$ – $10^{18}$ atoms delivers megawatts. The pulse ends when the field decays to zero at $N_f$ . Setting $\phi(N_f)=0$ and nondimensionalizing gives a transcendental relation in which extraction efficiency depends on $r$ and nothing else — no atom counts, no material constants survive.

r\,\eta_E=\ln\frac{1}{1-\eta_E},\qquad r=\frac{N_i}{N_t},\qquad \eta_E=1-\frac{N_f}{N_i} \qquad (6)

where: $N_f$ — inversion left after the pulse; $\eta_E$ — fraction of stored inversion extracted as light. This is the curve plotted above: $\eta_E(1)=0$ , $\eta_E(2)\approx0.80$ , $\eta_E(4)\approx0.98$ , $\eta_E\to1$ as $r\to\infty$ — the reason designers push for $r$ of a few.

Derivation APeak photon number of the giant pulse

The goal is to pin down the height of the pulse — the peak intracavity photon number — without solving the rate equations in time. During the giant pulse the system is nearly conservative, so it has a first integral: a fixed relation between $\phi$ and $N$ along the whole trajectory. We build it by dividing one rate equation by the other, integrate from the stored state, locate the peak where photon growth stops, and evaluate — the strong-pumping limit then shows why a Q-switched laser converts nearly one stored atom into one output photon.

Step 1. Divide the photon equation by the inversion equation; the shared factor $\phi/\tau_c$ cancels, leaving $\mathrm{d}\phi/\mathrm{d}N=-\gamma^{-1}(1-N_t/N)$ , a separable relation with time gone. Why: the fast field and slow inversion collapse into a single trajectory once you ask only how $\phi$ and $N$ move together, not how fast.

Step 2. Integrate from $(\phi\approx0,\,N=N_i)$ to a running $N$ : $\phi(N)=\gamma^{-1}[(N_i-N)-N_t\ln(N_i/N)]$ . Why: the seed of a few spontaneous photons is negligible against the $10^{17}$ to come, so the lower limit is set to zero.

Step 3. Locate the peak: $\mathrm{d}\phi/\mathrm{d}t=0$ precisely when $N=N_t$ , the inversion fallen back to threshold. Why: photons grow only while gain exceeds loss, so the turning point is where gain and loss balance — the definition of $N_t$ .

Step 4. Substitute $N=N_t$ : $\phi_{peak}=\gamma^{-1}[N_i-N_t-N_t\ln(N_i/N_t)]$ , tending to $N_i/\gamma$ for $r\gg1$ . Why: that limit is the whole promise of Q-switching — nearly every pumped atom returns as a beam photon, so stored and pulse energy are almost equal. *Note: this is a pure photon count; $r\gg1$ gives $\eta_E\to1$ , confirming near-unit extraction. For $r=4,\ \gamma=1$ : $\phi_{peak}=0.403\,N_i$ .*

Go deeper · Derivation B · 1 minExtraction efficiency as a universal function of r

Derivation BExtraction efficiency as a universal function of r

The aim is to show that the fraction of stored inversion a pulse extracts is not a messy function of the laser's size, doping, or cross-section, but one universal curve depending only on $r=N_i/N_t$ . We reach it by asking when the pulse ends: the field decays to zero at a residual inversion $N_f$ the avalanche could not consume in time. Imposing $\phi=0$ at $N_f$ in the first integral, then dividing out the absolute atom numbers, leaves a transcendental relation in $r$ and $\eta_E$ alone — a curve every Q-switched laser lies on, from milliwatt microchip to kilojoule fusion driver.

Step 1. Impose termination: set $\phi(N_f)=0$ in equation (3), giving $0=(N_i-N_f)-N_t\ln(N_i/N_f)$ . Why: the pulse ends when the photons are gone; the leftover $N_f$ is what the finite-speed avalanche stranded, generally nonzero.

Step 2. Nondimensionalize: divide by $N_i$ and write $x=N_f/N_i=1-\eta_E$ , so $1-x=r^{-1}\ln(1/x)$ . Why: dividing by $N_i$ deletes every extensive quantity, leaving a relation between two pure ratios — hence universal.

Step 3. Rearrange to $r\,\eta_E=\ln[1/(1-\eta_E)]$ : monotonic, with $\eta_E\to0$ as $r\to1$ and $\eta_E\to1$ as $r\to\infty$ . Why: the endpoints are sanity checks — a laser barely above threshold wastes its inversion, a hugely overpumped one empties it.

Step 4. Solve numerically: $r=1.5\Rightarrow0.58$ ; $2\Rightarrow0.80$ ; $3\Rightarrow0.94$ ; $4\Rightarrow0.98$ ; $5\Rightarrow0.99$ . Why: the design table — useful extraction starts near $r=2$ and returns flatten past $r\approx4$ – $5$ , so more pumping buys shortening, not energy. *The asymptotes $\eta_E(1)=0$ and $\eta_E(\infty)=1$ are exact; the tabulated values follow from root-finding on $r\eta_E+\ln(1-\eta_E)=0$ .*

Go deeper · Derivation C · 1 minPulse buildup time from the spontaneous-emission seed

Derivation CPulse buildup time from the spontaneous-emission seed

The purpose is to explain a phenomenon every experimentalist meets: after the switch opens the giant pulse does not appear instantly but emerges after tens to hundreds of nanoseconds, and that delay wobbles slightly shot to shot. Both follow from the pulse growing out of noise. Early on the inversion is barely touched, sitting at $N_i$ , so the photon equation is linear with fixed net gain and the field grows exponentially from a seed of order one photon. Solving for the time to the peak gives a delay scaling as $\tau_c/(r-1)$ and depending only logarithmically on the tiny, fluctuating seed — exactly why it is long and slightly jittery.

Step 1. Linearize: while depletion is negligible, $N\approx N_i$ , so $\mathrm{d}\phi/\mathrm{d}t\approx(r-1)\,\phi/\tau_c$ . Why: the pulse spends most of its buildup with the inversion nearly untouched, so the coupling freezes and growth is a clean exponential.

Step 2. Integrate: $\phi(t)=\phi_0\exp[(r-1)t/\tau_c]$ , with $\phi_0\sim1$ spontaneous photon in the mode. Why: the seed cannot be zero — spontaneous emission always feeds the mode — and its smallness is what makes buildup take many e-foldings.

Step 3. Solve for the peak time: $t_{build}\approx\dfrac{\tau_c}{r-1}\ln\dfrac{\phi_{peak}}{\phi_0}$ . With $\phi_{peak}\sim10^{17}$ , $\phi_0\sim1$ the log is $\approx40$ , so at $r=4,\ \tau_c=10$ ns the delay is $\sim130$ ns and at $r=2$ it stretches to $\sim390$ ns. Why: only $\ln\phi_0$ enters, so a factor-of-two seed fluctuation shifts $t_{build}$ by under a nanosecond — the observed timing jitter. *The result $t_{build}\gg t_p$ explains the experimentally observed delayed pulse, and the logarithmic seed dependence accounts for the residual timing jitter; larger $r$ shortens the delay.*

Limiting cases

The pulse's entire character is set by the inversion ratio $r=N_i/N_t$ and by how the storage and switching times compare with the upper-state lifetime.

Case	Result	Meaning
Barely above threshold, $r\to1$	$\eta_E\to0$ , $P_{peak}\propto(r-1)^2$	Weak, broad, low-contrast pulse; almost no inversion above threshold, so extraction is poor.
Strongly pumped, $r\gg1$	$\eta_E\to1$ , $\phi_{peak}\to N_i/\gamma$ , $t_p\sim\tau_c\,\dfrac{\ln r}{r}$	Nearly all inversion becomes light; the pulse shortens toward the photon lifetime — the giant-pulse regime.
Slow switch, $t_{sw}\gtrsim t_{build}$	multiple / distorted pulses	Oscillation begins before the inversion fully accumulates, splitting the output; fast electro-optic switching cures it.
Short lifetime, $\tau\lesssim t_{store}$	$N_i\approx N_i^{max}\!\left(1-e^{-t_{store}/\tau}\right)$	Spontaneous emission drains the reservoir before the switch fires — why nanosecond-lifetime media store poorly.
Three-level medium	$\gamma=2$ , threshold and stored energy doubled	Each photon empties the upper level and fills the lower one, changing the inversion by two — halving efficiency at equal $r$ .

Common trapThe giant pulse proves Q-switching wrings more energy out of the laser than the same laser running CW

It does not. Q-switching does not create energy — it repackages it. The energy in the giant pulse is at most the energy stored in the inversion, capped by the upper-state lifetime and generally comparable to or less than what the same medium, continuously pumped, would emit as CW light over the same interval. What Q-switching multiplies is peak power, by squeezing that fixed budget into nanoseconds instead of releasing it steadily. Store 100 mJ over 200 μs and dump it in 20 ns and the peak is megawatts, but the time-averaged power at a modest repetition rate is only watts — often lower than the CW output, because the cavity spends most of each cycle dark and lossy while charging. The giant pulse is concentration in time, not energy generation; expecting more total joules than CW confuses peak power with energy.

Worked examples

1 · Peak power of a Q-switched Nd:YAG (the concrete one)

A four-level Nd:YAG laser ( $\gamma=1$ , $\lambda=1.064$ μm) stores $E_{st}=100$ mJ in the inversion, then is Q-switched with inversion ratio $r=4$ . The high-Q photon lifetime is $\tau_c=10$ ns and the output coupler dominates the loss ( $\delta_{out}/\delta\approx1$ ). Find the stored inversion, peak photon number, peak power, extracted energy, and pulse duration.

Step 1. Photon energy: $h\nu=hc/\lambda=(6.626\times10^{-34})(2.998\times10^8)/(1.064\times10^{-6})=1.87\times10^{-19}$ J $=1.17$ eV. Stored inversion $N_i=E_{st}/(\gamma\,h\nu)=0.100/1.87\times10^{-19}=\mathbf{5.36\times10^{17}}$ atoms — the reservoir filled over the $\sim$ 200 μs lifetime.

Step 2. Peak photons from equation (4), $r=4$ : $\phi_{peak}=N_i[1-\tfrac14(1+\ln4)]=5.36\times10^{17}\times0.403=\mathbf{2.16\times10^{17}}$ photons — only 40% of the stored atoms are present as photons at the peak; the rest already left on the rising edge or remain in the un-consumed inversion.

Step 3. Peak power from equation (5): $P_{peak}=h\nu\,\phi_{peak}/\tau_c=(1.87\times10^{-19})(2.16\times10^{17})/(10\times10^{-9})=\mathbf{4.0\ MW}$ — a hundred millijoules through a ten-nanosecond aperture, four orders of magnitude above the tens-of-watts CW level.

Step 4. Extracted energy: $\eta_E(r{=}4)\approx0.98$ , so $E_{pulse}\approx\mathbf{98\ mJ}$ (ideal output coupler). Duration by energy conservation: $t_p\sim E_{pulse}/P_{peak}=0.098/(4.0\times10^6)=\mathbf{24\ ns}$ — a few photon lifetimes. Peak power ≈ 4 MW, energy ≈ 98 mJ, duration ≈ 24 ns.

2 · Buildup delay, jitter, and the range-resolution reality check

For the same laser ( $\tau_c=10$ ns, $\phi_{peak}\sim2\times10^{17}$ , seed $\phi_0\sim1$ ): (a) compare the buildup delay at $r=4$ and $r=2$ and explain the jitter; (b) if used for ranging, what range resolution does the width imply?

Step 1. The logarithm is $\ln(2\times10^{17})\approx40$ . At $r=4$ : $t_{build}=\tau_c\ln(\cdots)/(r-1)=10\ \text{ns}\times40/3=\mathbf{133\ ns}$ ; at $r=2$ : $10\ \text{ns}\times39/1=\mathbf{390\ ns}$ . Lower $r$ means weaker net gain $(r-1)/\tau_c$ , so the avalanche climbs out of noise far more slowly — harder pumping gives a faster, more punctual pulse.

Step 2. Jitter: since only $\ln\phi_0$ enters, a factor-of-two seed fluctuation shifts the delay by $\Delta t_{build}\approx\tau_c\ln2/(r-1)\approx\mathbf{2\ ns}$ at $r=4$ — the noise-seeded signature, and why passively switched lasers (which also let $N_i$ drift) jitter more.

Step 3. Range check: the pulse-width resolution is $\Delta R=c\,t_p/2=(3.0\times10^8)(24\times10^{-9})/2=\mathbf{3.6\ m}$ ; even a 10 ns pulse gives only $\approx1.5$ m. So a giant pulse does not itself deliver centimetre ranging — that comes from leading-edge sub-pulse timing, which a fast rise and high peak power make possible. Buildup ≈ 133 ns (r = 4) vs ≈ 390 ns (r = 2), jitter ≈ 2 ns; pulse-width resolution ≈ 3.6 m, sharpened by leading-edge timing.

Physics of the portrait

Q-switching was developed as a way to store energy in the laser medium and then release it in a short, intense pulse by rapidly changing the cavity losses. Hellwarth developed the concept in 1961, and McClung and Hellwarth demonstrated giant optical pulses from a ruby laser using a Kerr-cell switch in 1962.

Origins · 1 minRobert W. Hellwarth and how this was discovered

Robert W. Hellwarth

b. 1930 · energy storage in the inversion and the giant optical pulse

Within months of Maiman's first ruby laser in 1960, Robert W. Hellwarth, at Hughes Research Laboratories, asked a sharp question: what if you stopped the laser from lasing while you pumped it? In a 1961 paper he reasoned from the rate equations that holding the cavity $Q$ artificially low during pumping — "Q-spoiling" — could drive the inversion far past its normal threshold, storing energy ordinary lasing would have bled away, then release it as a single "giant pulse" by suddenly restoring high $Q$ . It was a purely theoretical prediction of a regime no one had seen. In 1962 Hellwarth and Fred McClung confirmed it, placing a Kerr-cell shutter in a ruby cavity and watching the output collapse from a long feeble train into one towering nanosecond spike — the first giant pulse. The technique, soon renamed Q-switching, enabled everything that needs concentrated optical energy: ranging, machining, frequency conversion, and pumping other lasers. Hellwarth went on to foundational work in nonlinear optics, but it is the deliberate storing and dumping of a laser's inversion that carries his name into every optics course.

History

1958

— Schawlow and Townes lay out the principle of the optical maser

1960

— Maiman operates the first laser, in ruby

1961

— Hellwarth proposes Q-spoiling to store the inversion and release a giant pulse

1962

— McClung and Hellwarth demonstrate the giant-pulse ruby laser with a Kerr-cell shutter

1964

— passive Q-switching by saturable-absorber dyes is demonstrated

1965

— electro-optic Pockels cells and acousto-optic modulators become standard Q-switches

Going further

Texts. Saleh & Teich, Fundamentals of Photonics, 3rd ed., Ch. 16 (Pulsed Lasers; Q-switching) · Svelto, Principles of Lasers, 5th ed., Ch. 8 · Siegman, Lasers (1986), Ch. 26 · Koechner, Solid-State Laser Engineering, Ch. 8 · Milonni and Eberly, Laser Physics, pulsed-laser chapter. Key results to keep in mind: the first integral is $\phi(N)=\gamma^{-1}[(N_i-N)-N_t\ln(N_i/N)]$ ; the peak photon number is $\phi_{peak}=\gamma^{-1}[N_i-N_t-N_t\ln(N_i/N_t)]$ , giving $0.403\,N_i$ for $r=4$ ; the efficiency relation $r\,\eta_E=\ln[1/(1-\eta_E)]$ yields $\eta_E(2)\approx0.80$ , $\eta_E(4)\approx0.98$ , $\eta_E(5)\approx0.99$ ; $\gamma=1$ for four-level and $\gamma=2$ for three-level media from the inversion bookkeeping. Nd:YAG anchor values: $\lambda=1.064$ μm, $h\nu=1.17$ eV, upper-state lifetime $\sim230$ μs. The worked example with $E_{st}=100$ mJ, $\tau_c=10$ ns, $r=4$ gives $P_{peak}=4$ MW, $E_{pulse}=98$ mJ, $t_p\approx24$ ns. Pulse-width and buildup expressions are order-scaling relations.

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

Ask the tutor: Ask me to walk through how the first integral of the rate equations turns two coupled ODEs into a single giant-pulse trajectory, or to plug your own crystal's numbers into the peak-power and efficiency formulas.

Why does spoiling the cavity store energy instead of just wasting it?

Because a low-Q cavity has such high loss that its threshold inversion is enormous — beyond anything the pump can supply. With threshold out of reach the medium cannot oscillate, so stimulated emission never fires and never bleeds the inversion away; the pump keeps lifting atoms and they stay there, limited only by spontaneous decay over the upper-state lifetime. The high loss is not wasting the stored energy but guarding it, forbidding the low-power CW leakage that would otherwise clamp the inversion at threshold. The energy releases only when high Q is restored and the now-overwhelming gain discharges it.

What sets the pulse duration, and why nanoseconds rather than microseconds?

The cavity photon lifetime $\tau_c$ , not the pump or the upper-state lifetime. Once high Q is restored the field rises and falls on the timescale over which photons circulate and leak from the cavity — a few $\tau_c$ , of order 10 ns for a modest resonator (round-trip time a few ns divided by the fractional loss). The trailing edge is literally the passive decay of leftover photons at rate $1/\tau_c$ . Harder pumping (larger $r$ ) shortens the pulse toward $\tau_c\ln r/r$ . The microsecond storage time never enters the width — it only sets how much energy is available.

Active or passive Q-switching — which should I use?

Active switches (Pockels cells, acousto-optic modulators, rotating mirrors) are driven externally, so you choose exactly when the pulse fires: low jitter, synchronization, a clean single pulse — at the cost of a high-voltage driver and insertion loss. Passive switches (saturable absorbers such as Cr⁴⁺:YAG) need no electronics — the absorber holds the cavity lossy until the building intensity bleaches it — making compact, cheap, self-synchronized microchip lasers, but the pulse fires whenever intensity reaches the bleaching point, so it jitters more. Choose active for precision, passive for size, simplicity, and cost.

Why does the pulse build from noise, and what causes the jitter?

When the switch opens there is no established field — only a few photons spontaneous emission scattered into the mode. The giant pulse is the amplification of that noise through many round trips, so its start time depends on how many seed photons were present. Because buildup depends only on the logarithm of the seed, the delay is long (tens to hundreds of nanoseconds) and only weakly sensitive to its exact value — but "weakly" is not "not at all." The seed fluctuates by order unity shot to shot, and $\ln$ of that shifts the delay by a nanosecond or two: the residual timing jitter, fundamental because spontaneous emission cannot be switched off.

How is Q-switching different from mode-locking?

Two strategies for different regimes. Q-switching works by energy storage: it holds the inversion high and dumps it as one energetic nanosecond pulse, peak power set by how much was stored. Mode-locking works by phase coherence: it fixes the phases of many longitudinal modes so they interfere into ultrashort (picosecond–femtosecond) pulses, one per round trip. Q-switching gives high energy per pulse at low repetition rate; mode-locking gives extreme brevity at high rate but modest per-pulse energy — and the two can combine. In short: Q-switching is about when the energy comes out, mode-locking about how briefly.

Does a larger inversion ratio r always help?

Up to a point, after which it mostly buys pulse-shortening, not energy. Efficiency climbs steeply at first — 0 at $r=1$ , about 0.80 at $r=2$ , 0.98 at $r=4$ — then flattens, since you already extract nearly all the inversion. Beyond $r\approx4$ – $5$ extra pumping shortens the pulse and cuts buildup delay and jitter but adds little energy. Worse, very high $r$ pushes the intracavity intensity toward the optical-damage threshold and can trigger parasitic amplified spontaneous emission that drains the reservoir before you switch. Designers target $r$ of a few.