Skip to content

Lasers: Einstein Coefficients, Rate Equations and Resonator Modes

CSIR NET Physical Sciences Part C: Einstein A and B coefficients, pumping, population inversion, rate equations, resonator modes, coherence length.

  • 18 min read
  • Updated
On this page · 16 sections
  1. Prerequisites
  2. The Hook: Why a Hot Body Is Not a Laser
  3. Einstein Coefficients
  4. Gain, Absorption and Population Inversion
  5. Pumping and Level Schemes
  6. Rate Equations
  7. Threshold Condition
  8. Resonator Modes
  9. Coherence
  10. Memory Tricks
  11. Don’t Confuse
  12. Common Mistakes
  13. Practice
  14. Quick Revision Table
  15. Key Takeaways
  16. What’s Next

Prerequisites

The Hook: Why a Hot Body Is Not a Laser

Einstein Coefficients

Consider two levels, lower 1 and upper 2, with degeneracies \( g_1, g_2 \), populations \( N_1, N_2 \), separated by \( h\nu \), in radiation of spectral energy density \( \rho(\nu) \) (energy per unit volume per unit frequency). Three processes:

ProcessRate (per unit volume)Needs radiation?
Absorption \( 1\to2 \)\( B_{12}\rho(\nu)N_1 \)yes
Stimulated emission \( 2\to1 \)\( B_{21}\rho(\nu)N_2 \)yes; photon cloned in direction, frequency, phase, polarisation
Spontaneous emission \( 2\to1 \)\( A_{21}N_2 \)no; random direction and phase

In thermal equilibrium the populations obey Boltzmann, \( N_2/N_1 = (g_2/g_1)e^{-h\nu/k_BT} \), and the radiation obeys Planck’s law. Requiring the upward and downward rates to balance at every temperature gives the Einstein relations:

$$\boxed{g_1B_{12} = g_2B_{21}, \qquad \frac{A_{21}}{B_{21}} = \frac{8\pi h\nu^3}{c^3}}$$

(in a medium of refractive index \( n \), \( c^3 \to c^3/n^3 \)). These are properties of the atom, so they hold out of equilibrium too. Three consequences:

  1. Spontaneous emission grows as \( \nu^3 \). At short wavelengths (UV, X-ray) the upper level empties quickly by spontaneous emission, which makes inversion hard to maintain: X-ray lasers are difficult.
  2. Ratio of stimulated to spontaneous emission in thermal radiation:
$$\frac{B_{21}\rho(\nu)}{A_{21}} = \frac{1}{e^{h\nu/k_BT} - 1}$$

This is large only when \( h\nu \ll k_BT \) (microwaves at room temperature) and tiny for visible light from any ordinary source.

  1. Lifetime. If spontaneous emission is the only decay, the upper level lives \( \tau = 1/A_{21} \), which sets the natural line width \( 1/(2\pi\tau) \).

The full derivation from detailed balance is on Blackbody Radiation and Detailed Balance.

Gain, Absorption and Population Inversion

A beam of intensity \( I \) at frequency \( \nu \) travelling through the medium changes as \( dI/dz = \gamma(\nu)I \), with

$$\boxed{\gamma(\nu) = \sigma(\nu)\left(N_2 - \frac{g_2}{g_1}N_1\right), \qquad \sigma(\nu) = \frac{\lambda^2}{8\pi n^2}A_{21}\,g(\nu)}$$

where \( \sigma(\nu) \) is the stimulated-emission cross-section and \( g(\nu) \) is the normalised line shape (Lorentzian or Gaussian from the previous page). If \( N_2 > (g_2/g_1)N_1 \), \( \gamma > 0 \) and the beam grows: population inversion. In thermal equilibrium at any positive temperature, \( N_2 < (g_2/g_1)N_1 \) and \( \gamma < 0 \) (ordinary absorption). An inverted population is sometimes described as a “negative temperature”.

The cross-section shows why long wavelengths are easier: \( \sigma\propto\lambda^2A_{21}g(\nu) \), and a narrow line (large peak \( g \)) helps.

Why two levels are not enough

Pump a two-level system optically with intense light at \( \nu \). Absorption and stimulated emission have the same coefficient (for \( g_1 = g_2 \)), so as the pump intensity rises, \( N_2 \to N_1 \) but never exceeds it: the transition saturates and the medium becomes transparent, not amplifying. Inversion needs at least a third level so that the pump and the laser transition are different transitions.

Pumping and Level Schemes

Pumping is any process that drives atoms into the upper laser level:

PumpExamples
Optical (flashlamp, arc lamp, another laser, diode laser)ruby, Nd:YAG, Ti:sapphire, dye lasers
Electrical discharge (electron impact)He–Ne, argon ion, CO\( _2 \)
Resonant collisional energy transferHe to Ne in the He–Ne laser; N\( _2 \) to CO\( _2 \)
Current injection in a p–n junctionsemiconductor diode lasers
Chemical reactionHF and DF chemical lasers
graph TD
    subgraph TL["Three-level, e.g. ruby"]
    A1[Level 3: pump band] -->|fast non-radiative decay| A2[Level 2: metastable upper laser level]
    A2 -->|laser transition| A0[Level 1: ground = lower laser level]
    A0 -->|pump| A1
    end
    subgraph FL["Four-level, e.g. Nd:YAG"]
    B3[Level 3: pump band] -->|fast decay| B2[Level 2: upper laser level]
    B2 -->|laser transition| B1[Level 1: lower laser level]
    B1 -->|fast decay| B0[Level 0: ground]
    B0 -->|pump| B3
    end

Three-level laser (ruby: Cr\( ^{3+} \) in Al\( _2 \)O\( _3 \), 694.3 nm). The lower laser level is the ground state, which starts full. More than half the atoms (for equal degeneracies) must be pumped out of the ground state before there is any gain. Threshold is high; ruby is usually pulsed.

Four-level laser (Nd:YAG, 1064 nm; He–Ne, 632.8 nm; CO\( _2 \), 10.6 μm). The lower laser level lies well above the ground state and empties quickly, so it is nearly empty. Any population in the upper level is an inversion. Threshold is much lower, and continuous operation is easy.

Requirements on the upper laser level: long lifetime (metastable, so population accumulates); on the lower laser level: short lifetime (so it empties). The quantum efficiency \( h\nu_{\text{laser}}/h\nu_{\text{pump}} < 1 \) is a basic limit on efficiency.

He–Ne in brief. An electrical discharge excites helium into its metastable \( 2\,^3S_1 \) and \( 2\,^1S_0 \) levels (see Electron States, Spin, Helium and Alkali Spectra). These lie very close in energy to excited levels of neon, and collisions transfer the energy to neon selectively. Neon is the lasing atom; helium is the energy store.

Rate Equations

Four-level laser

Assume the pump band and the lower laser level empty instantly, so only the upper-level population \( N_2 \) (per unit volume) and the photon density \( \phi \) in the cavity matter. With pump rate \( R_p \) (per unit volume per second), upper-level lifetime \( \tau \), cross-section \( \sigma \), photon lifetime in the cavity \( \tau_c \), and \( n = 1 \):

$$\boxed{\frac{dN_2}{dt} = R_p - \frac{N_2}{\tau} - \sigma c\,\phi N_2, \qquad \frac{d\phi}{dt} = \sigma c\,\phi N_2 - \frac{\phi}{\tau_c}}$$

(plus a small spontaneous-emission seed in the second equation that starts oscillation). Steady state:

  • Below threshold, \( \phi\approx0 \) and \( N_2 = R_p\tau \): the inversion grows with pumping.
  • The photon equation has a non-zero solution only if \( \sigma cN_2 = 1/\tau_c \). So above threshold the inversion is clamped at
$$N_{\text{th}} = \frac{1}{\sigma c\tau_c}$$
  • The threshold pump rate is \( R_{\text{th}} = N_{\text{th}}/\tau \). Above it, the extra pumping all goes into photons:
$$\phi = \tau_c\left(R_p - R_{\text{th}}\right)$$

and the output power rises linearly with \( R_p - R_{\text{th}} \).

Saturation

A strong beam depletes the inversion it feeds on. For a homogeneously broadened four-level medium,

$$\gamma = \frac{\gamma_0}{1 + I/I_{\text{sat}}}, \qquad I_{\text{sat}} = \frac{h\nu}{\sigma\tau}$$

where \( \gamma_0 \) is the small-signal gain. In an inhomogeneously broadened medium (Doppler), a beam saturates only the atoms in resonance with it and burns a spectral hole in the gain curve; the gain falls more slowly, as \( \gamma_0/\sqrt{1 + I/I_{\text{sat}}} \), and several longitudinal modes can oscillate independently.

Three-level laser

With the ground state as the lower laser level and total density \( N = N_1 + N_2 \) (fast decay from the pump band), inversion requires \( N_2 > N_1 \), that is \( N_2 > N/2 \). Even with no laser oscillation, the pump has to work against spontaneous decay of more than half the atoms: \( R_p \gtrsim N/(2\tau) \). This is why three-level lasers have very high thresholds compared with four-level ones, where \( N_{\text{th}} \) can be a tiny fraction of \( N \).

Threshold Condition

A laser starts when the round-trip gain balances the round-trip loss. For a gain medium of length \( L \) filling a cavity with mirror reflectivities \( R_1, R_2 \) and distributed loss coefficient \( \alpha \):

$$\boxed{R_1R_2\,e^{2(\gamma_{\text{th}} - \alpha)L} = 1 \quad\Longrightarrow\quad \gamma_{\text{th}} = \alpha + \frac{1}{2L}\ln\frac{1}{R_1R_2}}$$

The photon lifetime of the passive cavity is the time for the stored energy to fall by \( 1/e \):

$$\tau_c = \frac{2L}{c\left[2\alpha L + \ln(1/R_1R_2)\right]}$$

and the cavity resonance width and quality factor are

$$\Delta\nu_c = \frac{1}{2\pi\tau_c}, \qquad Q = \frac{\nu_0}{\Delta\nu_c} = 2\pi\nu_0\tau_c$$

Combining with the clamping result: \( \gamma_{\text{th}} = \sigma N_{\text{th}} \), the same statement from the gain side.

Resonator Modes

Longitudinal (axial) modes

A standing wave in a cavity of optical length \( nL \) needs an integer number \( q \) of half-wavelengths:

$$\boxed{\nu_q = q\frac{c}{2nL}, \qquad \Delta\nu_{\text{FSR}} = \frac{c}{2nL}}$$

(the free spectral range). \( q \) is enormous for optical cavities (about \( 10^6 \)). The laser oscillates on every mode for which the gain exceeds the loss. The number of oscillating modes is roughly

$$N_{\text{modes}} \approx \frac{\text{gain bandwidth above threshold}}{c/2nL}$$

A short cavity and a narrow gain line give a single longitudinal mode; a long cavity and a broad gain line (Doppler-broadened gas, dye, Ti:sapphire) give many.

Transverse modes

The field pattern across the beam is also quantised: the transverse electromagnetic modes TEM\( _{mn} \), which for spherical mirrors are Hermite–Gaussian functions with \( m \) and \( n \) nodes in the two transverse directions. TEM\( _{00} \) is a Gaussian beam with the smallest divergence, \( \theta\approx\lambda/\pi w_0 \) for beam waist \( w_0 \). Higher transverse modes have slightly different frequencies (through the Gouy phase) and are suppressed with an aperture when a clean beam is needed.

Stability

With mirror radii \( R_1, R_2 \) (positive for concave towards the cavity) and spacing \( L \), define \( g_i = 1 - L/R_i \). A ray stays trapped near the axis if

$$\boxed{0 \leq g_1g_2 \leq 1}$$
ResonatorCondition\( g_1g_2 \)Comment
Plane-parallel\( R_1 = R_2 = \infty \)1marginally stable, sensitive to tilt
Confocal (symmetric)\( R_1 = R_2 = L \)0most tolerant of misalignment
Concentric (spherical)\( R_1 = R_2 = L/2 \)1marginally stable
Hemispherical\( R_1 = \infty \), \( R_2 = L \)0easy to align; tight focus on flat mirror

Coherence

Temporal coherence measures how long the wave keeps a predictable phase. For a source of bandwidth \( \Delta\nu \):

$$\boxed{\tau_{\text{coh}} \approx \frac{1}{\Delta\nu}, \qquad l_c = c\,\tau_{\text{coh}} = \frac{c}{\Delta\nu} = \frac{\lambda^2}{\Delta\lambda}}$$

This is the largest path difference in a Michelson interferometer that still gives fringes. (Some texts include a factor such as \( 1/2\pi \) or \( 2\ln2/\pi \) depending on the line shape and definition; for exam estimates use \( c/\Delta\nu \).)

SourceBandwidthCoherence length
White lightabout \( \Delta\lambda\sim300 \) nm around 550 nmabout 1 μm
LED at 850 nm, \( \Delta\lambda = 30 \) nmabout 24 μm
Single-longitudinal-mode laser, \( \Delta\nu = 1 \) MHzabout 300 m

A multimode laser has a bandwidth roughly equal to the span of its oscillating modes and a correspondingly shorter coherence length.

Spatial coherence measures phase correlation across the beam. A laser on the TEM\( _{00} \) mode is spatially coherent across its whole cross-section, which is why it can be focused to a near diffraction-limited spot and why its divergence is so small.

Pulsed operation (brief). Q-switching holds the cavity loss high while the pump builds a large inversion, then switches the loss off: a giant pulse of nanosecond duration. Mode-locking fixes the phases of many longitudinal modes: the output is a train of pulses separated by the round-trip time \( 2L/c \), each about \( 1/\Delta\nu_{\text{gain}} \) long. A broad gain bandwidth gives short pulses.

Memory Tricks

Short anchors for the results on this page. Say each one aloud before a mock test; if one does not trigger the formula, reread its section.

Don’t Confuse

ThisvsThat
Spontaneous emission: random phase and direction, \( A_{21} \)Stimulated emission: copies the incident photon, \( B_{21}\rho \)
Three-level: lower laser level = groundFour-level: lower laser level empties fast
Small-signal gain \( \gamma_0 \)Saturated gain \( \gamma_0/(1 + I/I_{\text{sat}}) \)
Longitudinal mode spacing \( c/2nL \)Cavity resonance width \( 1/2\pi\tau_c \) (much narrower)
Temporal coherence: set by bandwidthSpatial coherence: set by transverse mode structure
Homogeneous gain: one mode tends to winInhomogeneous gain: hole burning, several modes
Q-switching: one giant ns pulseMode-locking: pulse train at \( 2L/c \), short pulses

Common Mistakes

Practice

All questions below are original practice questions written for this page. They are not from any CSIR NET paper. Take \( h = 6.626\times10^{-34} \) J s, \( c = 3.00\times10^8 \) m s\( ^{-1} \), \( k_B = 1.381\times10^{-23} \) J K\( ^{-1} \), refractive index 1 unless stated.

Level 1: Foundation

  1. How does \( A_{21}/B_{21} \) depend on frequency?
  2. Why can a ruby laser not operate with weak pumping, whereas Nd:YAG can?
  3. What is the longitudinal mode spacing of a 1.5 m cavity?
Answers
  1. As \( \nu^3 \) (\( 8\pi h\nu^3/c^3 \)).
  2. Ruby is a three-level laser: the lower laser level is the ground state, so more than half the ions must be excited before gain appears. Nd:YAG is four-level, with an almost empty lower laser level.
  3. 100 MHz. \( c/2L = 3\times10^8/3 \).

Level 2: Part B Style

Practice Q1. A gas laser has a 50 cm cavity, and its gain exceeds the loss over a 1.5 GHz band. The number of longitudinal modes that can oscillate is about:

(a) 1 (b) 3 (c) 5 (d) 10

Answer: (c) 5

\( c/2L = 3\times10^8/1.0 = 300 \) MHz; \( 1500/300 = 5 \). Option (b) uses \( c/L \); option (d) uses \( c/4L \).

Practice Q2. For a thermal source at 3000 K, the ratio of stimulated to spontaneous emission at 600 nm is about:

(a) \( 3.4\times10^{-4} \) (b) \( 3.4\times10^{-2} \) (c) 1 (d) \( 2.9\times10^{3} \)

Answer: (a)

\( h\nu/k_BT = hc/(\lambda k_BT) = (6.626\times10^{-34}\times3\times10^8)/(600\times10^{-9}\times1.381\times10^{-23}\times3000) = 7.99 \); ratio \( 1/(e^{7.99} - 1) = 3.4\times10^{-4} \). Option (d) is the reciprocal, \( e^{7.99} \).

Practice Q3. A laser rod 25 cm long fills a cavity with mirrors of reflectivity 1.00 and 0.95; other losses are negligible. The threshold gain coefficient is about:

(a) 0.026 m\( ^{-1} \) (b) 0.051 m\( ^{-1} \) (c) 0.103 m\( ^{-1} \) (d) 0.205 m\( ^{-1} \)

Answer: (c) 0.103 m⁻¹

\( \gamma_{\text{th}} = \frac{1}{2\times0.25}\ln\frac{1}{0.95} = 2\times0.0513 = 0.103 \) m\( ^{-1} \). Option (d) drops the factor 2 in \( 2L \).

Practice Q4. A single-mode laser has a line width of 1 MHz. Its coherence length is about:

(a) 0.3 m (b) 3 m (c) 30 m (d) 300 m

Answer: (d) 300 m

\( l_c = c/\Delta\nu = 3\times10^8/10^6 = 300 \) m.

Level 3: Part C Style

Practice Q5. A four-level laser medium (\( \lambda = 1064 \) nm, emission cross-section \( 3\times10^{-23} \) m\( ^2 \), upper-level lifetime 230 μs) fills a 10 cm cavity with mirror reflectivities 1.00 and 0.90 and no other losses. Consider:

  1. The threshold gain coefficient is about 0.53 m\( ^{-1} \).
  2. The threshold inversion density is about \( 1.8\times10^{22} \) m\( ^{-3} \).
  3. The saturation intensity is about \( 2.7\times10^7 \) W m\( ^{-2} \).
  4. Above threshold, the steady-state inversion keeps increasing in proportion to the pump rate.

Which are correct?

(a) 1 and 2 only (b) 1, 2 and 3 only (c) 3 and 4 only (d) 1, 2, 3 and 4

Answer: (b) 1, 2 and 3 only
  1. \( \gamma_{\text{th}} = \frac{1}{0.2}\ln\frac{1}{0.9} = 0.527 \) m\( ^{-1} \). True.
  2. \( N_{\text{th}} = \gamma_{\text{th}}/\sigma = 0.527/(3\times10^{-23}) = 1.76\times10^{22} \) m\( ^{-3} \). True.
  3. \( h\nu = hc/\lambda = 1.87\times10^{-19} \) J; \( I_{\text{sat}} = h\nu/\sigma\tau = 1.87\times10^{-19}/(3\times10^{-23}\times2.3\times10^{-4}) = 2.7\times10^7 \) W m\( ^{-2} \) (2.7 kW cm\( ^{-2} \)). True.
  4. False: the inversion is clamped at \( N_{\text{th}} \); the photon number grows instead.

Practice Q6. A resonator uses two concave mirrors with radii of curvature \( R_1 = 2 \) m and \( R_2 = 1 \) m. Consider:

  1. A spacing of 0.5 m gives a stable resonator.
  2. A spacing of 1.5 m gives an unstable resonator.
  3. A spacing of 2.5 m gives a stable resonator.
  4. The resonator is stable for every spacing between 0 and 3 m.

Which are correct?

(a) 1 and 2 only (b) 1, 2 and 3 only (c) 1 and 4 only (d) 1, 2, 3 and 4

Answer: (b) 1, 2 and 3 only

\( g_1 = 1 - L/2 \), \( g_2 = 1 - L \).

  1. \( L = 0.5 \): \( g_1g_2 = 0.75\times0.5 = 0.375 \). Stable.
  2. \( L = 1.5 \): \( 0.25\times(-0.5) = -0.125 < 0 \). Unstable.
  3. \( L = 2.5 \): \( (-0.25)\times(-1.5) = 0.375 \). Stable.
  4. False: \( g_1g_2 \geq 0 \) needs \( L \leq 1 \) or \( L \geq 2 \), and \( g_1g_2 \leq 1 \) needs \( L \leq 3 \). The stable ranges are \( 0 < L \leq 1 \) m and \( 2 \leq L \leq 3 \) m.

Practice Q7. A 0.5 m He–Ne cavity (632.8 nm) has mirror reflectivities 1.00 and 0.98, with no other losses. Consider:

  1. The photon lifetime of the passive cavity is about 165 ns.
  2. The passive cavity resonance width is about 1 MHz.
  3. The longitudinal mode spacing is 300 MHz.
  4. The passive cavity resonance width is larger than the longitudinal mode spacing.

Which are correct?

(a) 1 and 3 only (b) 1, 2 and 3 only (c) 2 and 4 only (d) 1, 2, 3 and 4

Answer: (b) 1, 2 and 3 only
  1. \( \tau_c = 2L/[c\ln(1/R_1R_2)] = 1.0/(3\times10^8\times0.0202) = 1.65\times10^{-7} \) s. True.
  2. \( \Delta\nu_c = 1/(2\pi\tau_c) = 0.96 \) MHz. True.
  3. \( c/2L = 300 \) MHz. True.
  4. False: 1 MHz is far narrower than 300 MHz; the modes are well resolved (finesse about 300).

Quick Revision Table

IdeaFormula or fact
Einstein relations\( g_1B_{12} = g_2B_{21} \); \( A_{21}/B_{21} = 8\pi h\nu^3/c^3 \)
Stimulated / spontaneous (thermal)\( 1/(e^{h\nu/k_BT} - 1) \)
Gain\( \gamma = \sigma(N_2 - \frac{g_2}{g_1}N_1) \); \( \sigma = \frac{\lambda^2}{8\pi n^2}A_{21}g(\nu) \)
Two-levelcannot invert; saturates at \( N_2 = N_1 \)
Three-level (ruby, 694.3 nm)need \( N_2 > N/2 \)
Four-level (Nd:YAG 1064 nm, He–Ne 632.8 nm, CO\( _2 \) 10.6 μm)low threshold
Rate equations\( \dot N_2 = R_p - N_2/\tau - \sigma c\phi N_2 \); \( \dot\phi = \sigma c\phi N_2 - \phi/\tau_c \)
Clamping\( N_{\text{th}} = 1/\sigma c\tau_c \); \( \phi = \tau_c(R_p - R_{\text{th}}) \)
Saturation\( \gamma_0/(1 + I/I_{\text{sat}}) \), \( I_{\text{sat}} = h\nu/\sigma\tau \)
Threshold\( \gamma_{\text{th}} = \alpha + \frac{1}{2L}\ln\frac{1}{R_1R_2} \)
Photon lifetime\( \tau_c = 2L/c[2\alpha L + \ln(1/R_1R_2)] \); \( \Delta\nu_c = 1/2\pi\tau_c \)
Modes\( \Delta\nu = c/2nL \); TEM\( _{00} \) Gaussian
Stability\( 0 \leq g_1g_2 \leq 1 \), \( g = 1 - L/R \)
Coherence\( l_c = c/\Delta\nu = \lambda^2/\Delta\lambda \)

Key Takeaways

What’s Next

Advertisement
Advertisement