Collider experiments

A collider experiment turns an invisible quantum event into a measured number. Two beams cross inside a detector the size of a house; a few particles interact, spraying fresh particles outward through a nested set of instruments — each layer engineered to make a different species betray itself. Nothing in the debris is labelled. A photon, an electron, a pion, a muon and a neutrino all leave the interaction point at nearly light speed, and the detector's art is to read the pattern of signals they scatter across its layers and work back to the Feynman diagram that produced them. This is event reconstruction: from ionisation dots and calorimeter blobs to "a Z boson decayed to two muons here, at 91 GeV." Three ideas do almost all the work. A solenoid wraps the collision in an axial field, so every charged track curves and its curvature is its momentum — the tighter the curl, the softer the particle. Calorimeters outside the tracker stop electrons, photons and hadrons and measure their energy by the size of the shower they trigger. And conservation of energy and momentum lets the summed four-momenta of a few tracks reveal the mass of a parent that lived $10^{-22}\,\mathrm{s}$ and was gone before the first sensor. This article assembles these into the modern $4\pi$ detector: how it is built, what each layer measures, and how raw signals become the discoveries — the W, the Z, the top quark, the Higgs — the collider was built to find.

Key takeaways

A collider detector is concentric layers — tracker → electromagnetic calorimeter (ECAL) → hadronic calorimeter (HCAL) → muon chambers — each responding to a different interaction, giving near-complete ( $4\pi$ ) coverage.
A solenoid ( $B \approx 1\text{–}4\,\mathrm{T}$ ) bends charged tracks; momentum follows from the curvature by $p_T[\mathrm{GeV}/c]=0.2998\,B[\mathrm{T}]\,R[\mathrm{m}]$ .
Tracker resolution degrades with momentum, $\sigma(p_T)/p_T \propto p_T$ ; calorimeter resolution improves, $\sigma_E/E \propto 1/\sqrt{E}$ — complementary, crossing over near $100\,\mathrm{GeV}$ .
Particle ID is a logic table across layers: photon = ECAL blob, no track; electron = track + ECAL shower; hadron = track + HCAL shower; muon = track through everything to the outer chambers.
Neutrinos escape invisibly, inferred from missing transverse momentum $\vec p_T^{\,\mathrm{miss}}=-\sum_i \vec p_{T,i}$ , since only transverse momentum balances at a hadron collider.
A parent is reconstructed from its decay products' invariant mass, $M^2=(\sum E_i)^2-|\sum \vec p_i|^2$ , which peaks at the rest mass — the signature of every resonance discovery.
Key idea: By layering subdetectors that each respond to a different interaction mechanism, a collider experiment converts invisible fundamental particles into a recognisable pattern of signals, so the original event — and the mass of a particle that already decayed — can be reconstructed.

The core insight

By layering subdetectors that each respond to a different interaction mechanism, a collider experiment turns invisible fundamental particles into a recognisable pattern of signals, letting the original event — and even the mass of a particle that already decayed — be reconstructed.

Where this lives today

The archetypes are ATLAS and CMS at the LHC — cylindrical detectors $\sim25\,\mathrm{m}$ and $21\,\mathrm{m}$ long, each with of order $10^8$ readout channels, wrapped around a point where proton bunches cross forty million times a second. Every layer of the physical-picture figure is real hardware: silicon pixel and strip trackers resolving b-quark vertices to tens of microns; a lead-tungstate or liquid-argon ECAL; a brass or steel HCAL; superconducting solenoids at $3.8\,\mathrm{T}$ (CMS) or $2\,\mathrm{T}$ (ATLAS, plus a toroid); and muon chambers on the outside. Modern reconstruction runs particle-flow algorithms that fuse tracker and calorimeter object by object, taking equation (2) at low momentum and equation (3) at high energy. Because writing out $40\,\mathrm{MHz}$ is impossible, a two-tier trigger decides keep-or-discard in microseconds, discarding $99.998\%$ of events and recording only the $\sim1\,\mathrm{kHz}$ that look interesting — the rate that built the Higgs discovery. The frontier now is pileup: at the High-Luminosity LHC up to $\sim140$ collisions overlap in one crossing, so new upgrades add precision timing detectors ( $\sim30\,\mathrm{ps}$ ) and finer silicon to disentangle them. Proposed machines — the FCC, a linear $e^+e^-$ collider, a muon collider — each demand a detector re-optimised around the same three principles used here.

The physical picture

Picture the beam pipe running through the centre of a stack of nested cylinders, with a collision at the middle. The particles fan outward and first meet the tracker: thin layers of silicon (historically, gas wire chambers) that record a dot wherever a charged particle passes but are so thin the particle sails through almost undisturbed. In the solenoid's axial field each track curls into a helix, and the tracker measures that curvature without spoiling the particle for the calorimeters behind. Just outside sits the ECAL, dense enough to force electrons and photons into an electromagnetic shower and absorb them; then the coarser, larger HCAL, which does the same for hadrons; and on the very outside, beyond the magnet's return yoke, the muon chambers — since only muons and neutrinos survive the calorimeters, a hit out here is a near-certain muon tag. The first figure is this cross-section, one of each species drawn in so the signatures read directly.

Figure

A collision in cross-section: each species writes a different signature across the concentric layers, the axial field B = 3.8 T bending the charged tracks — photon γ an ECAL deposit with no track, electron e a track plus ECAL shower, charged hadron h a track plus HCAL shower, muon μ crosses everything to the outer chambers, neutrino ν escapes invisibly

The tracker and calorimeters do more than identify particles — they measure them, in opposite regimes, which is why a detector needs both. A track's curl radius is proportional to its momentum, so a soft particle spirals tightly and a hard one flies almost straight; measuring the tiny bow (the "sagitta") of a nearly straight track is hard, so tracker precision worsens as momentum rises. Calorimeters are the opposite: a higher-energy shower contains more particles, so fractional resolution improves with energy. But the deepest point of a collider is not any single measurement — it is that conservation laws let you reconstruct what you never directly saw. Sum the four-momenta of a couple of muons and the invariant length is the mass of the particle they came from, though it decayed before the first silicon layer. The second figure is the payoff: histogram that mass over millions of collisions and a short-lived particle materialises as a sharp peak.

Figure

The payoff of event reconstruction: histogram the two-muon invariant mass M² = 2E₁E₂(1 − cos θ₁₂) over millions of collisions and the Z boson appears as a sharp peak at M = 91.19 GeV, width Γ ≈ 2.5 GeV, on a smooth combinatorial background

The core equations

The tracker's most important relation is startlingly clean. In the axial field a charged particle moves on a helix; projected onto the transverse plane it is a circle whose radius $R$ fixes $p_T$ . In the units an experimenter uses, the constant is the speed of light in disguise.

p_T = 0.2998\;B\,R \quad\Longleftrightarrow\quad p_T[\mathrm{GeV}/c] = 0.2998\;B[\mathrm{T}]\;R[\mathrm{m}] \qquad (1)

where: $p_T$ — momentum transverse to the beam; $B$ — axial field; $R$ — radius of the projected circle; $0.2998 = c/10^{9}$ carries the units (from $p_T=eBR$ ). Reading it: a straighter track (larger $R$ ) is a stiffer particle; a $1\,\mathrm{GeV}$ track in $1\,\mathrm{T}$ curls with $R=3.34\,\mathrm{m}$ , while at $1000\,\mathrm{GeV}$ the same field barely bends it. A tracker does not measure $R$ directly — it measures the small displacement of the track from a straight line, the sagitta $s$ , over a chord $L$ . Inverting gives the momentum, and propagating the position error $\sigma_x$ gives the Gluckstern resolution, the master formula of tracking.

s=\frac{L^2}{8R}=\frac{0.2998\,B\,L^2}{8\,p_T},\qquad \frac{\sigma(p_T)}{p_T}=\frac{\sigma_x\,p_T}{0.2998\,B\,L^2}\sqrt{\frac{720}{N+4}} \qquad (2)

where: $s$ — sagitta; $L$ — lever arm; $\sigma_x$ — single-hit resolution; $N$ — number of points. Reading it: because $s\propto1/p_T$ , a fixed position error becomes a fractional momentum error growing linearly with $p_T$ — why trackers struggle with the stiffest particles. Calorimeters obey the opposite trend. A shower of energy $E$ carries $N_{\text{shower}}\propto E$ particles, so the signal fluctuates by $\sqrt{E}$ and the fractional resolution falls as energy rises — a stochastic term, a constant (calibration/leakage) floor, and a noise term in quadrature.

\frac{\sigma_E}{E}=\frac{a}{\sqrt{E}}\;\oplus\; b \;\oplus\; \frac{c}{E} \qquad (3)

where: $a$ — stochastic term; $b$ — constant term (dominates at high $E$ ); $c$ — noise term (dominates at low $E$ ); $\oplus$ — quadrature sum. Reading it: a good ECAL has $a\approx 3\%\sqrt{\mathrm{GeV}}$ , $b\approx0.5\%$ ; a hadron calorimeter is coarser, $a\approx 100\%\sqrt{\mathrm{GeV}}$ . Against equation (2), the roles reverse: calorimetry wins at high energy, tracking at low. Once the particles are measured, conservation does the reconstruction. Decay products' four-momenta sum to the parent's, whose invariant length is the parent's rest mass — a Lorentz invariant, the same in every frame.

M^2=\Big(\sum_i E_i\Big)^2-\Big|\sum_i \vec p_i\Big|^2 \;\xrightarrow{\text{2 massless}}\; M^2=2E_1E_2\big(1-\cos\theta_{12}\big) \qquad (4)

where: the sums run over the decay products; $\theta_{12}$ — opening angle between two (massless) daughters. Reading it: histogram $M$ and a real parent piles up as a peak at its rest mass while random combinations spread into a smooth background — how the Z, $J/\psi$ and Higgs were found. At a hadron collider the partons carry unknown shares of longitudinal momentum, so only the transverse plane is constrained. Angles there are described by pseudorapidity, and invisible particles show up as a transverse-momentum imbalance.

\eta=-\ln\tan\frac{\theta}{2},\qquad \vec p_T^{\,\mathrm{miss}}=-\sum_i \vec p_{T,i} \qquad (5)

where: $\theta$ — polar angle from the beam; $\eta$ — pseudorapidity ( $\eta=0$ at $90^\circ$ , $|\eta|\to\infty$ down the pipe); the sum runs over all visible objects. Reading it: $\Delta\eta$ is invariant under boosts along the beam, so detectors are built uniform in $\eta$ ; and since the initial transverse momentum is essentially zero, any leftover $\vec p_T^{\,\mathrm{miss}}$ is the fingerprint of neutrinos (or new invisible particles).

Derivation AHow a curved track becomes a momentum

The goal is to derive the tracker's master relation from first principles and expose the awkward consequence buried inside it. We start from the Lorentz force in an axial field, show the transverse motion is exactly circular with $p_T=eBR$ , recast it in GeV/tesla/metre units, connect the radius to the sagitta silicon actually measures, and propagate the position error to see why the tracker uniquely gets worse as the particle gets stiffer.

Step 1. A charge $e$ with velocity $\vec v$ in $\vec B=B\hat z$ feels $\vec F=e\,\vec v\times\vec B$ , perpendicular to both $\vec v$ and $\hat z$ . So $v_z$ is constant and the transverse velocity keeps constant magnitude, tracing a circle: the magnetic force is centripetal, $\gamma m v_T^2/R=e v_T B$ , giving $p_T=\gamma m v_T=eBR$ . Why: a magnetic force does no work, so it cannot change the speed — it only curls the transverse motion, making "momentum" and "radius" the same information.

Step 2. In usable units, $p_T=eBR\,c/(10^9 e)=BR\,c\times10^{-9}$ , and with $c=2.998\times10^8\,\mathrm{m/s}$ the factor is $0.2998$ , so $p_T[\mathrm{GeV}/c]=0.2998\,B[\mathrm{T}]\,R[\mathrm{m}]$ . Why: the " $0.3$ " everyone memorises is just the speed of light in these units.

Step 3. The tracker samples $N$ points along a short arc, not the whole circle, so it measures the sagitta $s=R-\sqrt{R^2-(L/2)^2}\approx L^2/8R$ for a chord $L$ . Substituting $R=p_T/0.2998B$ gives $s=0.2998\,B L^2/(8p_T)$ : the bow shrinks as momentum grows. Why: over a finite lever arm a stiff track is nearly straight, and the whole momentum measurement lives in the last tens of microns of departure from that line.

Step 4. Propagate the error. With $N$ points the fitted sagitta has error $\sigma_s=\sigma_x\sqrt{720/(N+4)}/8$ (Gluckstern), and since $p_T\propto1/s$ , $\sigma(p_T)/p_T=\sigma_x\,p_T\sqrt{720/(N+4)}/(0.2998\,BL^2)$ . Why: the punchline — a constant position error becomes a fractional momentum error rising linearly with $p_T$ , so tracking must hand off to calorimetry at high energy. Multiple Coulomb scattering (owned by the ionisation-energy-loss article) adds a momentum-independent floor at low $p_T$ .

Go deeper · Derivation B · 1 minReconstructing a particle from its debris — the invariant mass

Derivation BReconstructing a particle from its debris — the invariant mass

The aim is to justify the most powerful move in experimental particle physics: taking a fistful of tracks and calorimeter deposits and recovering the mass of a particle that decayed $10^{-22}\,\mathrm{s}$ after birth, before it could reach a sensor. We build the parent's four-momentum from conservation, take its invariant length, specialise to the two-body case that dominates discovery plots, and explain why real parents make peaks while accidents make a smooth background.

Step 1. Four-momentum is conserved at the decay vertex, so $P^\mu=\sum_i p_i^\mu$ . Its invariant length is the rest mass squared, $M^2=P^\mu P_\mu=(\sum_i E_i)^2-|\sum_i \vec p_i|^2$ (with $c=1$ ). Why: $P^\mu P_\mu$ is a Lorentz scalar — every observer computes the same $M$ — so the answer is independent of how fast the parent moved through the lab.

Step 2. For two daughters of negligible mass, $E_i=|\vec p_i|$ , so $M^2=(E_1+E_2)^2-(\vec p_1+\vec p_2)^2=2E_1E_2-2\vec p_1\!\cdot\!\vec p_2=2E_1E_2(1-\cos\theta_{12})$ . Why: mass is manufactured entirely from the daughters' energies and their opening angle — a wide-angle, high-energy pair reconstructs a heavy parent, a collinear soft pair a light one.

Step 3. Plot $M$ event by event. Daughters that genuinely came from one parent always give the same $M$ (smeared only by resolution), stacking into a peak; accidental pairings give a broad continuum. Why: this contrast — resonant peak over combinatorial background, exactly the second figure — is the operational definition of "discovering a particle," from the $J/\psi$ in 1974 to the $125\,\mathrm{GeV}$ Higgs (owned by the Higgs-discovery article).

Go deeper · Derivation C · 1 minThe transverse plane — pseudorapidity and missing momentum

Derivation CThe transverse plane — pseudorapidity and missing momentum

The goal is to explain why hadron-collider physics is done almost entirely in the transverse plane and in the variable $\eta$ , not the full momentum vector, and how that same restriction turns an undetectable neutrino into a measurable quantity. We start from rapidity and its behaviour under a boost, take the high-energy limit to reach the geometric pseudorapidity, then use transverse-momentum conservation to catch the invisible.

Step 1. Define rapidity $y=\tfrac12\ln[(E+p_z)/(E-p_z)]$ . A boost of rapidity $y_\beta$ along the beam gives $y\to y+y_\beta$ — merely a constant shift. Hence differences $\Delta y$ and the shapes of $y$ -distributions are boost-invariant. Why: at a hadron collider the collision's rest frame is boosted by an unknown amount along the beam, so physics must use quantities that ignore that boost.

Step 2. For $E\gg m$ , $p_z=E\cos\theta$ gives $y\to\eta=-\ln\tan(\theta/2)$ , the pseudorapidity — a function of polar angle alone, needing no mass. It runs $\eta=0$ at $90^\circ$ , $\eta\approx2.4$ at $10^\circ$ , diverging down the beam pipe. Why: $\eta$ is read directly from geometry and inherits rapidity's near-boost-invariance, so detectors are segmented uniformly in $(\eta,\phi)$ .

Step 3. The partons carry unknown longitudinal momenta (and beam remnants vanish down the pipe), so $\sum p_z$ is unusable — but the incoming transverse momentum is $\approx0$ , so $\sum_i \vec p_{T,i}\approx0$ over everything. Any imbalance among visible objects means something invisible left, with $\vec p_T^{\,\mathrm{miss}}=-\sum_i \vec p_{T,i}^{\,\mathrm{vis}}$ . Why: the only handle on neutrinos at a hadron collider — the W boson was discovered in 1983 as a high- $p_T$ lepton recoiling against large missing transverse momentum.

Limiting cases

Case	Result	Meaning
Low- $p_T$ track	$\sigma(p_T)/p_T \to$ const	multiple scattering dominates; resolution set by material, not curvature
High- $p_T$ track	$\sigma(p_T)/p_T \propto p_T$	nearly straight track; tiny sagitta hard to measure
Low- $E$ shower	$\sigma_E/E \approx a/\sqrt{E}$	sampling statistics dominate; resolution improves as $\sqrt{E}$
High- $E$ shower	$\sigma_E/E \to b$	calibration and leakage set a floor calorimetry cannot beat
Muon	track everywhere + outer-chamber hit	minimum-ionising, non-showering — the one charged particle reaching the outside
Neutrino	no signal anywhere	seen only as missing transverse momentum $\vec p_T^{\,\mathrm{miss}}$
Quark or gluon	a collimated jet of hadrons	confinement forbids a free colour charge; the parton is inferred

Common trapAssuming a collider "sees" quarks and gluons directly

It never does. QCD confinement means a struck quark or gluon cannot exist free for more than about $10^{-24}\,\mathrm{s}$ ; it immediately hadronises into a collimated spray of colourless hadrons — a jet (owned by the jets-and-fragmentation article). The detector records the jet, and from its energy and direction you infer the parton. Worse, you usually cannot tell which flavour it was: an up-jet and a down-jet look the same. The one exception is a b-quark jet, tagged because b-hadrons live long enough ( $c\tau\approx0.5\,\mathrm{mm}$ ) to decay at a displaced secondary vertex hundreds of microns from the collision, resolved by the pixel detector. So "we detected a top quark" always means "we reconstructed the jets, leptons and missing energy of the top's decay chain," never a bare quark in a sensor.

Worked examples

1 · Weighing a muon by its curl (fully worked)

A muon crosses the CMS tracker ( $B=3.8\,\mathrm{T}$ ) and its helix is reconstructed with projected radius $R=8.78\,\mathrm{m}$ over a lever arm $L=1.1\,\mathrm{m}$ , from $N=13$ silicon hits each of resolution $\sigma_x=15\,\mu\mathrm{m}$ . Find its transverse momentum, the sagitta, and the intrinsic resolution.

Step 1. Transverse momentum from equation (1): $p_T=0.2998\times3.8\times8.78=\mathbf{10.0\ GeV}/c$ . A $10\,\mathrm{GeV}$ muon in $3.8\,\mathrm{T}$ curls with a radius of nearly nine metres — almost straight across a one-metre tracker.

Step 2. The sagitta, from equation (2): $s=L^2/8R=(1.1)^2/(8\times8.78)=1.21/70.2=\mathbf{17.2\ mm}$ — the transverse bow the silicon must catch. It shrinks as $1/p_T$ , so a $1\,\mathrm{TeV}$ muon would bow only $\sim0.17\,\mathrm{mm}$ .

Step 3. Intrinsic resolution from the Gluckstern term: $\dfrac{\sigma(p_T)}{p_T}=\dfrac{\sigma_x\,p_T}{0.2998\,B L^2}\sqrt{\dfrac{720}{N+4}}=\dfrac{(15\times10^{-6})(10)}{0.2998\times3.8\times1.21}\sqrt{\dfrac{720}{17}}=\dfrac{1.5\times10^{-4}}{1.379}\times6.51=\mathbf{7.1\times10^{-4}\approx0.07\%}$ .

Step 4. Read the trend. This is the measurement-only limit; real resolution at $10\,\mathrm{GeV}$ is worse because multiple scattering adds a floor. But note the scaling: at $100\,\mathrm{GeV}$ the same formula gives $0.7\%$ , at $1\,\mathrm{TeV}$ about $7\%$ — degrading tenfold for each tenfold rise in momentum, exactly why the outer muon chambers add a second long-lever measurement and calorimetry takes over the highest energies. $p_T=10\,\mathrm{GeV}$ , $s=17\,\mathrm{mm}$ , $\sigma(p_T)/p_T\approx0.07\%$ , growing $\propto p_T$ .

2 · Rebuilding a Z boson from two muons (fully worked)

Two oppositely-charged muons are reconstructed, each with energy $E_1=E_2=50\,\mathrm{GeV}$ (mass negligible), with opening angle $\theta_{12}=131.5^\circ$ . What did they come from?

Step 1. Two-body invariant mass, equation (4): $M^2=2E_1E_2(1-\cos\theta_{12})$ , with $\cos131.5^\circ=-0.663$ , so $1-\cos\theta_{12}=1.663$ .

Step 2. Insert the energies: $M^2=2\times50\times50\times1.663=8313\ \mathrm{GeV}^2$ , so $M=\sqrt{8313}=\mathbf{91.2\ GeV}$ .

Step 3. Compare with the table: $M_Z=91.19\,\mathrm{GeV}$ . This pair is a $Z\to\mu^+\mu^-$ decay — the peak of the second figure. One event proves nothing (it could be accidental), but millions stack into the sharp resonance, whose width folds the Z's natural lifetime ( $\Gamma_Z\approx2.5\,\mathrm{GeV}$ ) with the detector's momentum resolution. $M=91.2\,\mathrm{GeV}$ → the Z boson.

3 · Does the jet fit inside the detector? (acceptance)

A jet emerges at polar angle $\theta=10^\circ$ ; a typical detector instruments $|\eta|<2.5$ . Is it measured, and what about one at $\theta=1^\circ$ ?

Step 1. From equation (5): $\eta=-\ln\tan5^\circ=-\ln(0.0875)=\mathbf{2.44}$ — just inside $|\eta|<2.5$ , so the $10^\circ$ jet is caught (barely) by the endcaps.

Step 2. At $\theta=1^\circ$ : $\eta=-\ln\tan0.5^\circ=\mathbf{4.74}$ , far outside coverage — that jet vanishes down the beam pipe. Its lost energy is one reason $\vec p_T^{\,\mathrm{miss}}$ is defined transversely: forward energy escapes, but transverse balance still holds.

Physics of the portrait

Modern collider experiments were made possible by a quiet revolution in the late 1960s: the moment particle tracks stopped being photographed and started being read out electronically — the ancestor of every wire chamber and, in spirit, every silicon tracker today.

Origins · 1 minGeorges Charpak and how this was discovered

Georges Charpak

1924–2010 · multiwire chambers and the electronic revolution in particle detection

Georges Charpak was born in 1924 in Dąbrowica (then Poland), emigrated to France as a child, joined the Resistance, and survived deportation to Dachau before completing a doctorate under Frédéric Joliot-Curie. He joined CERN in 1959 and in 1968 invented the multiwire proportional chamber (MWPC) — a plane of closely spaced anode wires in a gas volume, each an independent proportional counter with its own electronic readout. It was a decisive break from the bubble chamber and photographic emulsion, which recorded beautiful images but had to be developed and scanned by eye at perhaps one event per second. The MWPC gave millimetre resolution with microsecond response and, crucially, an electronic signal that could trigger on interesting events and log thousands per second. Charpak and collaborators generalised it into drift chambers and time-projection chambers, the technologies that instrumented the detectors which found the W and Z. He received the 1992 Nobel Prize in Physics — unusually, as sole laureate — "for his invention and development of particle detectors, in particular the multiwire proportional chamber." Every large tracker since inherits his insight: to measure the sub-atomic, make the track speak in electronics, not film.

History

1932

— Cockcroft and Walton split lithium with an accelerated proton beam

1952

— Glaser invents the bubble chamber, imaging particle tracks in superheated liquid

1968

— Charpak builds the multiwire proportional chamber at CERN

1983

— UA1 and UA2 discover the W and Z bosons at CERN's proton–antiproton collider

1995

— CDF and DZero discover the top quark at the Tevatron

2012

— ATLAS and CMS discover the Higgs boson at the LHC

Going further

Texts. Thomson, Modern Particle Physics, §1.3 (collider experiments and detection of quarks) — closest to this article's scope · Perkins, Introduction to High Energy Physics, Ch. 4 · Grupen & Shwartz, Particle Detectors, 2nd ed. · Peskin & Schroeder, QFT, §1 · Halzen & Martin, Quarks and Leptons, Ch. 1 · Griffiths, Introduction to Elementary Particles, 2nd ed., Ch. 1.

Provenance. Each layer's internal physics is owned by the prerequisites and recalled by name: single-track measurement and gas/silicon detectors from aq-pp-tracking-detectors; the electromagnetic shower from aq-pp-electromagnetic-showers; and the ionisation and multiple-scattering physics behind the low- $p_T$ resolution floor from aq-pp-ionisation-energy-loss. Downstream, jets and hadronisation are developed in aq-pp-jets-fragmentation, the rate law $\mathrm{d}N=\mathcal{L}\,\sigma$ in aq-pp-luminosity-event-rates, and the flagship application of the invariant-mass method in aq-pp-higgs-discovery. A verifier should confirm: $p_T=0.2998\,BR$ from $p_T=eBR$ ; the sagitta $s=L^2/8R$ and Gluckstern $\sqrt{720/(N+4)}$ scaling; $M^2=2E_1E_2(1-\cos\theta_{12})$ giving $91.2\,\mathrm{GeV}$ for the Z example; $\eta=-\ln\tan(\theta/2)=2.44$ at $10^\circ$ . Reference values ( $M_Z=91.19\,\mathrm{GeV}$ , $\Gamma_Z=2.49\,\mathrm{GeV}$ , b-hadron $c\tau\approx0.5\,\mathrm{mm}$ , CMS $B=3.8\,\mathrm{T}$ ) are from the Particle Data Group.

Prerequisites. Tracking Detectors · Electromagnetic Showers · Ionisation Energy Loss. Leads to. Jets and Fragmentation · Luminosity and Event Rates · The Higgs Discovery.

Ask the tutor: Want to walk a CMS event display particle by particle, or see how the same invariant-mass trick that found the Z uncovered the Higgs?

If calorimeters measure energy directly, why bother with a tracker at all?

Because they measure different things in complementary regimes, and only together give particle identification. The calorimeter destroys the particle and reports its energy with resolution that improves at high energy, but cannot distinguish an electron from a photon (both shower) and gives no sign of charge. The tracker measures momentum non-destructively, improves at low momentum, records the charge sign from the bend, and pins vertices to microns for b-tagging. Matching a track to a deposit separates an electron (track + shower) from a photon (shower, no track), and combining the two (particle flow) beats either alone across the momentum range.

Why use missing transverse momentum, not missing total momentum, for neutrinos?

Because only the transverse component is known to balance. The two partons carry unknown fractions of their protons' longitudinal momenta, and the beam fragments disappear down the pipe with large unmeasured $p_z$ , so $\sum p_z$ of the visible particles need not be zero even when nothing invisible escaped — it is simply unknown. The incoming transverse momenta, however, are essentially zero, so the visible transverse momenta must sum to zero unless something invisible carried $p_T$ away. That residual, $\vec p_T^{\,\mathrm{miss}}=-\sum\vec p_{T,i}$ , is the neutrino (or dark-matter) signature; there is no usable "missing $p_z$ ."

How do you tell an electron from a charged pion, since both leave a track?

By where and how they deposit energy. An electron is light, so it showers early and completely in the ECAL, with shower energy matching track momentum ( $E/p\approx1$ ). A charged pion is a hadron: it usually punches through the ECAL leaving little, then showers deep in the larger HCAL, with no such $E/p$ match. The transverse shape differs too — electromagnetic showers narrow, hadronic ones broad and ragged. Layering the ECAL in front of the HCAL is what makes this separation possible: the order of the layers encodes identity.

What actually is a "jet," and why can't the detector just show me the quark?

A jet is the collimated spray of hadrons a quark or gluon turns into. Colour confinement forbids an isolated colour charge, so within about $10^{-24}\,\mathrm{s}$ the parton pulls quark–antiquark pairs from the vacuum and dresses into a cone of pions, kaons and other colourless hadrons moving roughly along its original direction. The detector sees that cone — a cluster of tracks and deposits — and a jet algorithm groups them, summing four-momenta to estimate the parton's energy and direction. You never see the quark, only its hadronic shadow. The internal dynamics belong to the jets-and-fragmentation article.

How does b-tagging identify a jet as coming from a bottom quark?

Through lifetime. B-hadrons live long enough that, near light speed, they travel a measurable distance — $c\tau\approx0.5\,\mathrm{mm}$ , boosted to a few millimetres in the lab — before decaying. The silicon pixel detector, centimetres from the beam with tens-of-microns resolution, reconstructs the primary vertex and a displaced secondary vertex where the b-hadron fell apart. Tracks pointing back to that vertex with a large impact parameter flag the jet as a b-jet — the rare case where the detector names a quark's flavour, essential for the top quark and Higgs, which decay preferentially to b-quarks.

Why does the trigger throw away almost every collision — isn't that lost physics?

It is a hard necessity. Storing $40\,\mathrm{MHz}$ of hundred-megabyte events would mean petabytes per second — impossible. The overwhelming majority are low-momentum-transfer "soft" QCD events already understood in detail. The trigger makes a tiered decision — a hardware layer in microseconds, then a software farm — keeping events with rare-process signatures (high- $p_T$ leptons, large missing energy, high-mass pairs) and discarding the rest, cutting $40\,\mathrm{MHz}$ to $\sim1\,\mathrm{kHz}$ . The art is never quietly dropping the signal you are hunting; a badly designed trigger is lost physics, which is why trigger menus are studied as carefully as the detectors.