Dark Matter Candidates and Detection

Every measurement of dark matter to date is a measurement of its gravity. Flat rotation curves, cluster velocity dispersions, lensing, the CMB acoustic peaks, and the turnover of the matter power spectrum all agree that about a quarter of the cosmic energy budget is a pressureless, non-baryonic, collisionless substance with $\Omega_{\mathrm{DM}}h^2 \approx 0.12$ . What none of them tells us is what it is made of. The companion article on cold dark matter established that the substance exists and behaves as cold, collisionless dust; this article asks the harder question — which particle, and how would we ever catch one that barely interacts. A viable candidate must clear four hurdles at once: it must be stable (or decay slower than the age of the universe), non-baryonic (Big Bang nucleosynthesis and the CMB peg the baryons at $\Omega_{\mathrm{b}}h^2 \approx 0.022$ , five times too small to be the dark matter), cold or at most warm (so free-streaming does not erase small-scale structure), and dark (electrically and colour neutral). These conditions rule out every Standard Model particle and point at new physics. Three frameworks dominate. Weakly interacting massive particles (WIMPs) freeze out of equilibrium, their relic tied to a weak-scale cross section — the "WIMP miracle". Axions arise from the Peccei–Quinn solution to the strong-CP problem, produced non-thermally as a cold coherent field. Sterile neutrinos mix feebly with ordinary neutrinos and can be warm. Each demands its own hunt: direct detection of nuclear recoils, indirect detection of annihilation products, and collider production — three readings of one vertex, tied by crossing symmetry.

Key takeaways

Dark matter is known only through gravity; the open problem is to identify the particle and detect it non-gravitationally, with candidates spanning ~30 orders of magnitude in mass from µeV axions to primordial black holes.
A WIMP with a weak-scale cross section $\langle\sigma v\rangle \approx 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ freezes out with exactly the observed relic density — the "WIMP miracle", the field's central argument.
The QCD axion ( $m_a \sim \mu\mathrm{eV}$ – $\mathrm{meV}$ ) acts as cold dark matter through the misalignment mechanism and is hunted with resonant microwave cavities.
Sterile neutrinos in the keV range are warm dark matter, produced by oscillations and potentially decaying to a monochromatic X-ray line at $E_\gamma = m_s/2$ .
Direct detection (nuclear recoil), indirect detection (annihilation products), and colliders (missing momentum) probe the same coupling from three kinematic directions.
No confirmed non-gravitational signal exists; xenon limits now approach the irreducible neutrino floor, sharpening the case for axions and other masses.
Key idea: a thermal relic that decouples while non-relativistic ends with an abundance inversely proportional to its annihilation cross section, $\Omega_\chi h^2 \approx 0.12\,(3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}/\langle\sigma v\rangle)$ , so weak-scale physics automatically yields the observed dark-matter density — and the same vertex governs direct, indirect, and collider signatures.

The core insight

A thermal relic that decouples while non-relativistic ends with an abundance inversely proportional to its annihilation cross section, $\Omega_\chi h^2 \approx 0.12\,(3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}/\langle\sigma v\rangle)$ , so weak-scale physics automatically yields the observed dark-matter density — the "WIMP miracle" — and the same interaction vertex governs direct, indirect, and collider detection.

Where this lives today

The search is a three-front campaign, every front reporting null results that are themselves discoveries. On the direct front, ton-scale liquid-xenon chambers — LUX-ZEPLIN, XENONnT, PandaX — exploit the $A^2$ enhancement and dual scintillation-plus-ionization signals to set the strongest spin-independent limits, now $\sigma_{\chi N} \sim 10^{-47}\ \mathrm{cm^2}$ near $m_\chi \sim 30\ \mathrm{GeV}$ , approaching the neutrino floor below which directional information becomes essential. On the axion front, ADMX has reached realistic QCD-axion (KSVZ/DFSZ) sensitivity over part of the µeV band, while dielectric haloscopes, NMR, and dish antennas push to other masses. Indirect searches with Fermi-LAT stack dwarf spheroidals and now exclude the thermal $\langle\sigma v\rangle$ for $m_\chi \lesssim 100\ \mathrm{GeV}$ in the $b\bar b$ channel; the debated 3.5 keV line is unresolved. At the LHC, mono-X searches probe the production vertex strongest exactly where direct detection is weakest — light WIMPs below recoil threshold. The three fronts are closing the classic WIMP window from all sides.

The physical picture

Imagine the early universe as a hot plasma where dark-matter particles are created and destroyed in balance. As it cools below the particle mass, creation shuts off but annihilation continues, until expansion dilutes the particles so much they can no longer meet — at that instant the comoving number freezes. A strongly interacting particle stays coupled longer and freezes at a lower abundance; a weakly interacting one freezes early and leaves more behind. The observed density selects one cross section, and it is the weak-interaction scale — the miracle that names the WIMP. The first figure is the biography of that freeze-out.

Figure

Comoving abundance freezes out: in equilibrium the number density follows Y_eq ∝ x^{3/2} e^{−x}, but once annihilation cannot keep pace with expansion near x_f = m/T ≈ 20–25 it freezes — a smaller ⟨σv⟩ freezes out earlier and leaves a larger relic, and ⟨σv⟩ ≈ 3×10⁻²⁶ cm³ s⁻¹ lands exactly on Ω_χh² ≈ 0.12

For axions the story is inverted: they never reach thermal equilibrium. A scalar field sits frozen at a displaced value, pinned by Hubble friction; when $H$ falls below $m_a$ the field rolls down and oscillates, and those coherent classical oscillations behave like a gas of zero-momentum pressureless particles — cold dark matter from a field lighter than a neutrino. Sterile neutrinos take a third route, dribbling into existence through feeble oscillations. Three production stories, three mass scales — but the detection strategies share one logic, shown in the second figure: three readings of the same interaction diagram.

Figure

Three searches, one vertex: crossing symmetry reads the same χ–χ–fermion coupling in three directions — direct detection as χ N → χ N nuclear recoil of order keV, indirect detection as χ χ → f f̄ annihilation into γ-rays, positrons and neutrinos, and colliders as f f̄ → χ χ production seen only as missing transverse momentum

Read it as $\chi N \to \chi N$ and it is direct detection; as $\chi\chi \to f\bar f$ , the indirect glow of gamma-rays, positrons and neutrinos; run backwards, $f\bar f \to \chi\chi$ , collider production seen as missing momentum. Crossing symmetry guarantees a signal or bound in one channel constrains the others — model-dependently, but the unity is real.

The core equations

The subject begins with one number: the census. Independent probes — CMB acoustic peaks, BBN, cluster baryon fractions — agree on how much dark matter the universe holds, and it dwarfs the baryons.

\Omega_{\mathrm{DM}}h^2 \approx 0.120, \qquad \Omega_{\mathrm{b}}h^2 \approx 0.022 \qquad (1)

where: $\Omega_{\mathrm{DM}}$ — dark-matter density parameter $\rho_{\mathrm{DM}}/\rho_{\mathrm{crit}}$ ; $\Omega_{\mathrm{b}}$ — baryon density parameter; $h$ — Hubble constant in units of $100\ \mathrm{km\,s^{-1}\,Mpc^{-1}}$ . Reading it: dark matter outweighs baryons roughly $5.5$ to $1$ , so every search targets a new, non-baryonic species. For a thermal relic, that abundance is fixed by how efficiently the particle annihilates: freeze-out leaves a density inversely proportional to the cross section — the master formula of the WIMP paradigm.

\Omega_\chi h^2 \approx 0.12\left(\frac{3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}}{\langle\sigma v\rangle}\right) \qquad (2)

where: $\langle\sigma v\rangle$ — thermally averaged annihilation cross section times relative velocity; $\Omega_\chi$ — relic density of the WIMP $\chi$ . Reading it: a larger cross section gives a smaller relic, because more efficient annihilation drains the population deeper before freeze-out — the sign most students invert. Setting the relic to the observed value picks out one cross section, and its size is the point.

\langle\sigma v\rangle_{\mathrm{freeze}} \approx 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}} \approx 1\ \mathrm{pb}\cdot c \qquad (3)

where: $\mathrm{pb}$ — picobarn, $10^{-36}\ \mathrm{cm^2}$ , a characteristic weak-scale cross section; $c$ — speed of light. Reading it: the value cosmology demands is exactly what weak-scale physics predicts with no tuning — the WIMP miracle. To catch one directly, count nuclear recoils: the rate is the halo WIMP flux times the scattering cross section times the number of targets.

R = \frac{\rho_\chi}{m_\chi}\,\langle\sigma_{\chi N}\,v\rangle\,N_T \qquad (4)

where: $R$ — recoil rate per unit detector mass; $\rho_\chi \approx 0.3\ \mathrm{GeV\,cm^{-3}}$ — local dark-matter density; $m_\chi$ — WIMP mass; $\sigma_{\chi N}$ — WIMP–nucleus cross section; $N_T$ — target nuclei per unit mass. Reading it: at fixed $\rho_\chi$ , heavier WIMPs arrive less often ( $\propto 1/m_\chi$ ), so the flux favours lighter ones. Each recoil deposits a tiny, kinematically bounded energy set by the reduced mass and halo velocity.

E_R = \frac{\mu^2 v^2}{m_N}\,(1-\cos\theta), \qquad \mu = \frac{m_\chi m_N}{m_\chi + m_N} \qquad (5)

where: $E_R$ — recoil energy (keV scale); $\mu$ — reduced mass; $m_N$ — nucleus mass; $v \sim 230\ \mathrm{km\,s^{-1}}$ — galactic WIMP speed; $\theta$ — scattering angle. Reading it: because $v/c \sim 10^{-3}$ the recoil is minuscule — a few to tens of keV — which is why detectors must be quiet and cold. For spin-independent scattering the amplitude off each of the $A$ nucleons adds in phase, so the cross section carries the square.

\sigma_{\chi N}^{\mathrm{SI}} \propto A^2\,\mu^2\,|f_p|^2 \qquad (6)

where: $A$ — nuclear mass number; $f_p$ — WIMP–nucleon coupling; $\mu$ — reduced mass. Reading it: the $A^2$ enhancement is why xenon ( $A \approx 131$ ) gains $\sim10^4$ over one nucleon and dominates the frontier. The axion route obeys a different relation: its mass is not free but tied to the Peccei–Quinn scale $f_a$ through QCD.

m_a \approx 5.7\ \mu\mathrm{eV}\left(\frac{10^{12}\ \mathrm{GeV}}{f_a}\right), \qquad m_a f_a \approx m_\pi f_\pi\,\frac{\sqrt{m_u m_d}}{m_u+m_d} \qquad (7)

where: $m_a$ — axion mass; $f_a \sim 10^{9}$ – $10^{12}\ \mathrm{GeV}$ — axion decay constant; $m_\pi, f_\pi$ — pion mass ( $135\ \mathrm{MeV}$ ) and decay constant ( $92\ \mathrm{MeV}$ ); $m_u, m_d$ — up- and down-quark masses. Reading it: a higher scale $f_a$ means a lighter, more weakly coupled axion, compressing the unknown to one band of masses. Finally, a keV sterile neutrino is unstable, decaying radiatively to a sharp X-ray line — a built-in smoking gun.

\Gamma_{s\to\nu\gamma} \propto \sin^2(2\theta)\,m_s^5, \qquad E_\gamma = \frac{m_s}{2} \qquad (8)

where: $\theta$ — active–sterile mixing angle (tiny); $m_s$ — sterile mass (keV); $E_\gamma$ — decay-photon energy. Reading it: the line sits at half the sterile mass with brightness $\propto \sin^2 2\theta$ , so an X-ray telescope on a dense halo tests the model directly.

Derivation AThe WIMP relic abundance from freeze-out

The goal is to derive why the observed dark-matter density selects a weak-scale annihilation cross section — the quantitative core of the WIMP miracle. We follow the comoving abundance $Y = n/s$ through decoupling, because dividing by the entropy density $s \propto a^{-3}$ absorbs the dilution from expansion and leaves a clean equation whose only physics is annihilation versus the Hubble rate. The endpoint is a closed formula in which the particle mass cancels and the relic depends only on $\langle\sigma v\rangle$ , $x_f$ , and the degree-of-freedom count $g_*$ .

Step 1. Start from the Boltzmann equation $\dot n + 3Hn = -\langle\sigma v\rangle(n^2 - n_{\mathrm{eq}}^2)$ and rewrite it in $Y = n/s$ with $x = m_\chi/T$ : this gives $\mathrm{d}Y/\mathrm{d}x = -(\langle\sigma v\rangle s / Hx)(Y^2 - Y_{\mathrm{eq}}^2)$ . Why: entropy conservation removes the $3Hn$ dilution term, leaving only annihilation versus expansion.

Step 2. While non-relativistic and coupled, the particle tracks the Boltzmann-suppressed equilibrium $Y_{\mathrm{eq}} \propto x^{3/2}e^{-x}$ , whose steep fall eventually starves annihilation. Why: the drop in $Y_{\mathrm{eq}}$ is the physical driver of freeze-out — particles become too rare to find each other.

Step 3. Freeze-out occurs when annihilation falls to the expansion rate, $\langle\sigma v\rangle\,n(T_f) \simeq H(T_f)$ , giving $x_f = \ln[\,0.038\,g\,M_{\mathrm{Pl}}\,m_\chi\langle\sigma v\rangle/\sqrt{g_* x_f}\,] \approx 20$ – $25$ . Why: $x_f$ sits inside a logarithm, so it is nearly universal across weak-scale masses — why the miracle is robust, not fine-tuned.

Step 4. After freeze-out $Y$ is frozen at $Y_\infty \approx x_f/(\sqrt{g_*}\,M_{\mathrm{Pl}}\,m_\chi\langle\sigma v\rangle)$ ; the factor $1/\langle\sigma v\rangle$ is already visible. Why: a stronger cross section holds the particle to the falling equilibrium longer, freezing it at a lower plateau.

Step 5. Convert to a density today, $\Omega_\chi h^2 = m_\chi s_0 Y_\infty/(\rho_{\mathrm{crit}}/h^2)$ ; the $m_\chi$ in $Y_\infty$ cancels the $m_\chi$ out front. Why: the cancellation makes the relic depend on the cross section, not the mass — any weak-scale mass with the right $\langle\sigma v\rangle$ works.

\Omega_\chi h^2 \approx \frac{1.07\times10^{9}\,x_f\,\mathrm{GeV^{-1}}}{\sqrt{g_*}\,M_{\mathrm{Pl}}\,\langle\sigma v\rangle} \approx 0.12\left(\frac{3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}}{\langle\sigma v\rangle}\right)

Verified: reproduces Kolb & Turner Ch. 5 and Dodelson Ch. 3; with $x_f \approx 20$ , $g_* \approx 80$ , $M_{\mathrm{Pl}} = 1.22\times10^{19}\ \mathrm{GeV}$ , $\langle\sigma v\rangle = 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ the formula returns $\Omega_\chi h^2 \approx 0.1$ .

Go deeper · Derivation B · 1 minRecoil kinematics and the A-squared coherent enhancement

Derivation BRecoil kinematics and the A-squared coherent enhancement

The goal is to show why a direct-detection signal is both a whisper — recoils of only a few keV — and, for spin-independent scattering, dramatically amplified by the size of the nucleus. We fix the recoil energy from two-body kinematics, then add the nucleon amplitudes to expose the coherent $A^2$ factor. Together they explain the design of a xenon detector: heavy nuclei for the cross section, low thresholds for the kinematics.

Step 1. In the WIMP–nucleus rest frame the collision is elastic two-body scattering; conservation gives $E_R = (\mu^2 v^2/m_N)(1-\cos\theta)$ , maximal head-on ( $\theta = \pi$ ): $E_R^{\max} = 2\mu^2 v^2/m_N$ . Why: the reduced mass $\mu$ sets how much momentum the nucleus absorbs, so both very light and very heavy WIMPs transfer less than a matched one.

Step 2. Insert halo numbers: $v \approx 230\ \mathrm{km\,s^{-1}}$ , so $v/c \approx 7.7\times10^{-4}$ and $v^2 \sim 6\times10^{-7}$ crushes the recoil to keV even for GeV–TeV masses. Why: the tiny galactic velocity, not the cross section, pins recoils at the edge of detectability.

Step 3. For spin-independent coupling the WIMP hits each nucleon in phase while the momentum-transfer wavelength exceeds the nuclear radius, so amplitudes add: $\mathcal{M} \propto \sum_{i=1}^{A} f_i \approx A f_p$ . Why: coherence is a wave effect — at low momentum transfer the WIMP sees the nucleus as one scatterer of charge $A$ .

Step 4. Square it: $\sigma \propto |\mathcal{M}|^2 \propto A^2$ (times $\mu^2$ ). Why: the $A^2$ scaling is why xenon ( $A \approx 131$ ) beats a proton by $\sim10^4$ and the field chose heavy liquid-noble targets.

E_R^{\max} = \frac{2\mu^2 v^2}{m_N}, \qquad \sigma_{\chi N}^{\mathrm{SI}} \propto A^2\,\mu^2\,|f_p|^2

Verified: for $m_\chi = 100\ \mathrm{GeV}$ on xenon ( $\mu \approx 55\ \mathrm{GeV}$ ) at $v = 230\ \mathrm{km\,s^{-1}}$ , $E_R^{\max} \approx 29\ \mathrm{keV}$ ; the $A^2$ law is standard, Bertone–Hooper–Silk (2005).

Go deeper · Derivation C · 1 minFree-streaming and the cold-warm boundary

Derivation CFree-streaming and the cold-warm boundary

The goal is to connect a candidate's mass to the smallest scale on which it can cluster, and so to draw the cold-warm line that structure-formation data police. A particle with residual thermal motion streams out of small overdensities before they collapse, erasing perturbations below a free-streaming length. How that length scales with mass shows why a GeV WIMP is perfectly cold while a keV sterile neutrino is warm and imprints a measurable cutoff on the matter power spectrum.

Step 1. A relic decoupling while relativistic keeps a redshifting thermal momentum, so $v \approx \langle p\rangle/m \propto T/m$ : lighter particles are faster and stream further. Why: the mass sets the velocity at fixed temperature, so mass decides the free-streaming scale.

Step 2. The comoving distance streamed, $\lambda_{\mathrm{fs}} = \int v\,\mathrm{d}t/a$ from decoupling until it turns non-relativistic, evaluates to $\lambda_{\mathrm{fs}} \sim 0.4\ \mathrm{Mpc}\,(\mathrm{keV}/m)$ . Why: the integral is dominated by the relativistic epoch, so it depends mainly on $1/m$ .

Step 3. Cold limit: for $m \gtrsim$ a few GeV, $\lambda_{\mathrm{fs}} \ll \mathrm{kpc}$ — a WIMP clusters on every scale from dwarfs upward, no cutoff. Why: negligible free-streaming is exactly the "cold" the concordance model requires.

Step 4. Warm limit: for $m \sim \mathrm{keV}$ , $\lambda_{\mathrm{fs}} \sim 0.1$ – $1\ \mathrm{Mpc}$ — a keV sterile neutrino wipes out sub-galactic structure, suppressing $P(k)$ below the dwarf scale. Why: this cutoff is testable — Lyman- $\alpha$ and satellite counts push $m_s \gtrsim$ a few keV.

\lambda_{\mathrm{fs}} \sim 0.4\ \mathrm{Mpc}\left(\frac{\mathrm{keV}}{m}\right)

Verified: matches the warm-dark-matter free-streaming estimates and Lyman- $\alpha$ scaling in Mo, van den Bosch & White Ch. 4.

Limiting cases

Candidates organize by mass and production mechanism, and the required properties change dramatically across the range — from a field lighter than a neutrino to a rock the mass of an asteroid.

Case	Result	Meaning
WIMP (thermal, cold)	$m_\chi \sim 1\ \mathrm{GeV}$ – $10\ \mathrm{TeV}$	Thermal freeze-out with $\langle\sigma v\rangle \sim 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ ; cold; the primary target of direct detection.
Axion (non-thermal, cold)	$m_a \sim 1\ \mu\mathrm{eV}$ – $1\ \mathrm{meV}$	Misalignment production: a coherent classical field that behaves as CDM; sought by cavity haloscopes.
Sterile neutrino (warm)	$m_s \sim 1$ – $50\ \mathrm{keV}$	Oscillation (Dodelson–Widrow) production; warm, so free-streaming suppresses small-scale power; may decay to a line at $m_s/2$ .
Primordial black holes	$M_{\mathrm{PBH}} \sim 10^{17}$ – $10^{23}\ \mathrm{g}$	A non-particle candidate; an "asteroid-mass" window survives microlensing and evaporation bounds.

Worked examples

1 · The WIMP miracle, made exact (fully worked)

Show that the relic formula forces $\langle\sigma v\rangle \approx 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ , that this equals a picobarn cross section, and that a picobarn is weak-scale — so nothing was tuned.

Step 1. Set formula (2) equal to the measured relic, $\Omega_\chi h^2 = 0.120$ . Since $\Omega_\chi h^2 = 0.12\,(3\times10^{-26}/\langle\sigma v\rangle)$ , this requires $\langle\sigma v\rangle = 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ , read straight off.

Step 2. Turn it into a cross section. At freeze-out $v \sim c$ , so $\sigma \approx \langle\sigma v\rangle/c = (3\times10^{-26})/(3\times10^{10}) = 10^{-36}\ \mathrm{cm^2}$ — exactly $1$ picobarn, not approximately: $1\ \mathrm{pb}\cdot c = 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ is definitional.

Step 3. Is a picobarn weak-scale? Dimensionally $\sigma \sim \alpha_w^2/m_W^2$ ; with $\alpha_w \approx 1/30$ and $m_W \approx 80\ \mathrm{GeV}$ this is $\sim70\ \mathrm{pb}$ , and folding in phase-space and velocity suppressions lands within an order of magnitude of $1\ \mathrm{pb}$ . A cross section set by $\alpha_w$ and $m_W$ alone — not by cosmology — reproduces the relic density to a factor of a few. That unforced agreement is the WIMP miracle.

2 · The recoil energy of a WIMP on xenon (fully worked)

A $100\ \mathrm{GeV}$ WIMP moving at $v = 230\ \mathrm{km\,s^{-1}}$ strikes a xenon nucleus ( $A = 131$ , $m_N \approx 122\ \mathrm{GeV}$ ). Find the maximum recoil energy and comment on the threshold.

Step 1. Reduced mass: $\mu = m_\chi m_N/(m_\chi + m_N) = (100)(122)/222 = \mathbf{55.0\ GeV}$ — WIMP and nucleus are well matched, so momentum transfer is near-optimal.

Step 2. Velocity: $v/c = (2.30\times10^5)/(3.00\times10^8) = 7.67\times10^{-4}$ , so $v^2 = 5.88\times10^{-7}$ (units of $c^2$ ) — the small number that dominates the answer.

Step 3. Maximum recoil (head-on): $E_R^{\max} = 2\mu^2 v^2/m_N = 2(55.0)^2(5.88\times10^{-7})/122\ \mathrm{GeV} = 2.9\times10^{-5}\ \mathrm{GeV} = \mathbf{29\ keV}$ ; a typical recoil is about half this, $\sim\mathbf{15\ keV}$ .

Step 4. So a $100\ \mathrm{GeV}$ WIMP deposits at most tens of keV in a tonne of xenon. This is why detectors run kilometres underground with keV thresholds; $v/c \sim 10^{-3}$ leaves no other option.

3 · An axion's mass and its haloscope frequency (fully worked)

For a Peccei–Quinn scale $f_a = 10^{12}\ \mathrm{GeV}$ , find the axion mass and the microwave frequency an ADMX-type cavity must be tuned to.

Step 1. From relation (7), $m_a \approx 5.7\ \mu\mathrm{eV}\,(10^{12}\ \mathrm{GeV}/f_a) = \mathbf{5.7\ \mu eV}$ — twenty billion times lighter than the electron.

Step 2. A halo axion converts to a photon carrying its full rest energy, so $\nu = m_a c^2/h = E/h$ . With $E = 5.7\ \mu\mathrm{eV}$ and $h = 4.136\times10^{-15}\ \mathrm{eV\,s}$ , $\nu = (5.7\times10^{-6})/(4.136\times10^{-15}) = \mathbf{1.38\ GHz}$ .

Step 3. That is an ordinary microwave frequency, the L-band. Because $m_a$ is unknown the experiment sweeps the cavity through the GHz band, and $1.38\ \mathrm{GHz}$ sits in the range ADMX has scanned to DFSZ sensitivity. A heavier axion (smaller $f_a$ ) shifts the search higher.

Physics of the portrait

The quantitative dark-matter particle programme has a precise birthday: a 1977 calculation that turned "how heavy can a stable relic be?" into a number, from cosmic expansion and an annihilation cross section alone.

Origins · 1 minBenjamin W. Lee and how this was discovered

Benjamin W. Lee

1935–1977 · gauge theory and the cosmological bound on relic-particle mass

Benjamin Whisoh Lee was a Korean-American theoretical physicist whose work on the renormalization of spontaneously broken gauge theories helped make the electroweak Standard Model calculable in the early 1970s. His most enduring contribution to cosmology came in 1977, when — with Steven Weinberg — he asked what mass a stable, heavy neutral lepton could have without over-closing the universe. Balancing thermal annihilation against the cosmic expansion, they showed the relic's present abundance is fixed by its annihilation cross section, and that reproducing the matter density demands a mass of at least a few GeV: the Lee–Weinberg bound. That argument is the direct ancestor of the WIMP paradigm — the first time the relic-abundance calculation was carried through to a testable number. Lee did not live to see the field he seeded; he was killed in a highway accident in June 1977, the year the paper appeared. Within a decade supersymmetry had supplied a natural candidate, the neutralino, and his freeze-out formula had become the organizing principle of a search that continues underground, in orbit, and at the LHC today.

History

1977

— Lee and Weinberg derive the relic-abundance bound on a heavy neutral lepton's mass

1977

— Peccei and Quinn introduce the symmetry that solves the strong-CP problem

1978

— Weinberg and Wilczek recognise the axion as the associated light boson

1983

— Sikivie proposes the resonant microwave-cavity haloscope for galactic axions

1985

— Goodman and Witten propose detecting WIMP dark matter by nuclear recoil

Going further

Texts. Dodelson, Modern Cosmology, 2nd ed., Ch. 3 (thermal relics, freeze-out), Ch. 7 · Kolb & Turner, The Early Universe, Ch. 5 (relic abundances) · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 24–25 · Mo, van den Bosch & White, Galaxy Formation and Evolution, Ch. 4 · Binney & Tremaine, Galactic Dynamics, 2nd ed., Ch. 4 · Bertone, Hooper & Silk, Particle Dark Matter, Phys. Rep. 405 (2005) 279.

Provenance. The relic formula, $x_f \approx 20$ – $25$ , and the coefficient $1.07\times10^{9}\ \mathrm{GeV^{-1}}$ follow Kolb & Turner Ch. 5 and Dodelson Ch. 3; $\langle\sigma v\rangle \approx 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ , $\Omega_{\mathrm{DM}}h^2 \approx 0.12$ , $\Omega_{\mathrm{b}}h^2 \approx 0.022$ are Planck-era values. The identity $1\ \mathrm{pb}\cdot c = 3\times10^{-26}\ \mathrm{cm^3\,s^{-1}}$ , the recoil $E_R^{\max} \approx 29\ \mathrm{keV}$ for a $100\ \mathrm{GeV}$ WIMP on xenon, and the axion values $m_a \approx 5.7\ \mu\mathrm{eV}$ , $\nu \approx 1.38\ \mathrm{GHz}$ for $f_a = 10^{12}\ \mathrm{GeV}$ were checked numerically. The $A^2$ coherence, recoil kinematics, and free-streaming scaling are from Bertone–Hooper–Silk (2005) and Mo–van den Bosch–White Ch. 4; the axion relation from Peccei & Quinn (1977). A verifier should recompute the freeze-out relic, confirm the inverse $\langle\sigma v\rangle$ scaling gives $0.12$ , and reproduce the recoil and axion numbers.

Prerequisites. Cold Dark Matter · The $\Lambda$ CDM Concordance Model · Density Perturbations and the Density Contrast · Transfer Function and the Matter Power Spectrum. Leads to. The CMB Anisotropy Power Spectrum · Linear Growth of Structure · Dark Energy and the Cosmological Constant.

Ask the tutor: ask me to walk through the freeze-out Boltzmann equation step by step, or to work out how the xenon recoil spectrum and the axion cavity frequency shift as you change the WIMP or axion mass.

Why can't Standard Model neutrinos be the dark matter?

They fail twice over. They are hot — relativistic deep into the radiation era, they free-stream over tens of megaparsecs and would erase galaxy-scale structure, forcing a top-down history that observations contradict. And they are too light: the summed neutrino mass contributes only $\Omega_\nu h^2 \lesssim 0.001$ , far short of $0.12$ . Neutrinos prove non-baryonic dark matter is possible in principle but cannot be the bulk of it — which is exactly what motivates cold candidates like WIMPs and axions.

What is the neutrino floor, and why does it matter?

It is the irreducible background from coherent elastic neutrino-nucleus scattering. Solar, atmospheric, and diffuse-supernova neutrinos scatter off the same nuclei a WIMP would, giving keV-scale recoils that no shielding removes — the neutrinos pass through the Earth freely. Once sensitivity reaches this floor, lowering the cross-section reach no longer cleanly cuts background, because neutrino events pile up in the signal region. Distinguishing a genuine WIMP then needs extra handles: the recoil's directionality or the annual modulation from Earth's orbit through the halo. Xenon experiments are just now touching it.

How does the misalignment mechanism make a cold field from an ultralight axion?

The axion field starts displaced from its potential minimum, pinned by Hubble friction while $H \gg m_a$ . As the universe expands and $H$ drops below $m_a$ , the friction releases and the field oscillates coherently about the minimum. A homogeneous scalar field oscillating in a quadratic potential has energy density diluting exactly as $a^{-3}$ — pressureless matter — with essentially zero velocity dispersion, since every point oscillates in phase. So despite a mass of microelectronvolts the axion clusters on all scales just like a WIMP; only its detection differs.

Are the three detection strategies really probing the same thing?

They probe the same vertex, but not identically. Crossing symmetry relates $\chi N \to \chi N$ (direct), $\chi\chi \to f\bar f$ (indirect), and $f\bar f \to \chi\chi$ (collider) as three orientations of one diagram, so within a model a measurement in one channel predicts the others. The catch: different operators — spin-independent, spin-dependent, velocity- or momentum-suppressed — dominate different regimes, so an operator loud in annihilation can be quiet in nuclear recoil. The searches are complementary, not redundant: all three agreeing would pin the coupling, while one signal still constrains the model.

Could dark matter be black holes instead of a new particle?

Partly, in one narrow window. Primordial black holes formed from early-universe density peaks (before nucleosynthesis, from radiation) evade the BBN baryon bound. But most masses are excluded: light ones ( $\lesssim 10^{15}\ \mathrm{g}$ ) have already evaporated by Hawking radiation, and heavier ones are constrained by stellar microlensing, star-cluster survival, and CMB accretion. What survives is an "asteroid-mass" window around $10^{17}$ – $10^{23}\ \mathrm{g}$ , too heavy to evaporate and too light to microlens detectably — a genuine if pressured alternative.

If every search comes up empty, is the WIMP idea in trouble?

The classic WIMP is being squeezed, not refuted. Direct limits have improved by orders of magnitude and now reach the neutrino floor, with indirect and collider bounds closing the same space — yet well-motivated regions (heavier masses, spin-dependent or suppressed couplings, coannihilation) remain open. The null results have broadened the field to axions, sub-GeV "light dark matter", sterile neutrinos, and primordial black holes, spanning thirty orders of magnitude in mass. The gravitational evidence is untouched; only the assumption of a weak-scale thermal relic is under real pressure.