Scintillator Detector

A crystal that screams in visible light the instant ionising radiation passes through it — and a photomultiplier that hears each scream as a cascade of a million electrons.

A scintillator detector registers individual particles or gamma-ray photons that are too energetic or too penetrating to be stopped by an ordinary photodiode. It does two jobs at once: a dense scintillating crystal stops (or substantially slows) the incoming radiation and converts a fraction of its deposited energy into visible or near-UV light; a photodetector — traditionally a photomultiplier tube (PMT), increasingly a silicon photomultiplier (SiPM) — then converts that light flash into a measurable electrical pulse. The pulse height is proportional to the energy deposited, giving both a count (how many particles arrived) and a spectrum (at what energy), in a detector that can be made from a coin-sized crystal to a multi-tonne array.

When a gamma ray or charged particle travels through the crystal lattice it excites electrons into the conduction band, creating electron-hole pairs at a rate proportional to its energy. In an activated inorganic crystal such as NaI(Tl) — sodium iodide with trace thallium dopant — the electrons migrate to thallium luminescence centres where they recombine, emitting photons at 415 nm with a decay time of about 250 nanoseconds. A typical yield is roughly 38 photons per keV deposited, meaning a 1 MeV gamma ray produces around 38,000 visible photons. These photons exit the crystal and strike the photocathode of a coupled PMT, ejecting photoelectrons with 20–30% quantum efficiency. Each photoelectron is then multiplied through 8–12 dynode stages at ~100 V each, each step producing 3–4 secondary electrons; 12 stages yields a total gain of 10⁶–10⁸. The resulting nanosecond-scale current pulse is large enough to register on standard electronics. In fast-timing applications (PET scanners, TOF experiments) LYSO crystals replace NaI: their 40-nanosecond decay time and 30,000 photons/MeV output enable coincidence timing below 200 picoseconds.

In plain terms

Imagine you're trying to catch invisible bullets flying through a room. You fill the room with soap bubbles: every invisible bullet that hits a bubble makes it pop and glow for a moment. A scintillator detector works the same way — the crystal is the room full of bubbles, and each glowing pop is a flash of light you can detect and count.

Also called: scintillation detector, NaI detector, crystal detector, scintillator-PMT, scintillator-SiPM.

Why it matters

Energy spectroscopy from portable instruments: a 3-inch NaI(Tl) detector can identify gamma-emitting radioisotopes by their characteristic peak energies in the field, enabling nuclear security screening, environmental monitoring, and medical isotope verification without a laboratory.
Scalability to enormous detector masses: ATLAS TileCal comprises roughly 465,000 scintillating plastic tiles and 10,000 PMTs; the CMS ECAL uses 75,848 lead-tungstate crystals — both measuring particle energies at the Large Hadron Collider with better than 1% resolution.
Medical imaging: LYSO and LSO scintillators in PET scanners detect the 511 keV annihilation photons produced when positron-emitting tracers meet electrons in tissue, reconstructing 3D metabolic maps of the brain, heart, and tumours.
Dark-matter and rare-decay searches: the high light yield of NaI(Tl) (38,000 photons/MeV) enables low-threshold detection of sub-keV nuclear recoils, and its chemical inertness allows multi-tonne arrays to run underground for years without degradation.

Where it lives

ATLAS TileCal hadronic calorimeter at CERN LHC — 465,000 polystyrene scintillating tiles, 10,000 Hamamatsu R5900 PMTs, measuring hadronic jet energies from ~30 MeV to ~2 TeV
LUX-ZEPLIN (LZ) outer detector — gadolinium-loaded liquid scintillator veto surrounding 10 tonnes of liquid xenon, instrumented with PMTs, located 1 mile underground at SURF in South Dakota
DAMA/LIBRA dark matter experiment — 250 kg of NaI(Tl) crystals at Gran Sasso National Laboratory, the only experiment claiming an annual modulation signal consistent with WIMP dark matter (result disputed and under active independent investigation)
Clinical PET scanners (Siemens Biograph, GE Discovery) — LYSO or LSO crystal arrays read by SiPMs or PMTs, resolving 511 keV coincidences to sub-nanosecond timing for time-of-flight PET
Compton Gamma Ray Observatory EGRET telescope — NaI(Tl) scintillator stack designed partly by Robert Hofstadter, mapped the MeV–GeV sky from 1991 to 2000
Hospital nuclear medicine gamma cameras and survey meters — 3-inch NaI(Tl) cylinders coupled to single PMTs, the workhorse radiation detector for 70 years

Key numbers

NaI(Tl) light yield: ~38,000 photons/MeV (38 photons/keVγ; one of the highest of any practical scintillator (Luxium Solutions / NRC data))
NaI(Tl) decay time: ~250 ns (Primary decay constant at room temperature; limits count rate to < ~1 MHz)
LYSO decay time: ~40 ns (Enables < 200 ps coincidence timing in TOF-PET; preferred for clinical PET scanners)
BC-408 plastic scintillator decay time: 2.1 ns (Saint-Gobain (now Luxium) BC-408 datasheet; used in large-area time-of-flight counters)
PMT current gain: 10⁶–10⁸ (8–12 dynode stages at ~100 V/stage; single-photon detection above electronic noise)
PMT quantum efficiency (photocathode): 20–30% (At peak wavelength (~400 nm bialkali cathode); SiPMs reach 50–65% PDE)
Energy resolution (NaI, 662 keV): ~7–8% FWHM (Standard metric for ¹³⁷Cs gamma peak; better than most other portable gamma spectrometers)

Surprising facts

A single 1 MeV gamma ray striking a NaI(Tl) crystal produces roughly 38,000 visible photons in about 250 nanoseconds — yet the crystal absorbs so much of its own light that reaching the PMT requires polishing every surface to optical flatness and wrapping the outside in a diffuse reflector.
The PMT coupled to a scintillator can amplify a single photoelectron (produced by a handful of visible photons) into a pulse of a million electrons in under 50 nanoseconds — a current gain of 10⁶ with a noise level measured in single-electron equivalents.
Scintillator technology was invented in 1948 by Robert Hofstadter at Stanford, who discovered that a trace of thallium added to a sodium iodide melt produced a crystal far brighter than anything previously known; Hofstadter later won the Nobel Prize — not for the scintillator but for what he discovered about proton and neutron structure using the electron scattering technique his own detector enabled.
The ATLAS TileCal contains 2,900 tonnes of steel and plastic, yet the scintillating tiles themselves weigh only 80 tonnes — roughly the mass of 13 elephants — spread across a calorimeter the size of a four-storey building.

Open questions

DAMA/LIBRA annual modulation — confirmed signal, disputed interpretation: The DAMA experiment at Gran Sasso National Laboratory has observed a statistically significant annual modulation in the rate of low-energy events in a 250 kg NaI(Tl) array, consistent with the expected dark matter signal, at > 13σ accumulated significance over more than 20 annual cycles. This is a real, reproducible experimental observation. However, results from more sensitive experiments using different target materials (LUX-ZEPLIN, XENONnT, PandaX) have not detected the complementary WIMP signal that DAMA's interpretation requires under standard WIMP models. Whether the modulation is dark matter, seasonal systematic effects, or an unexplained background process remains unresolved and is the subject of active experimental campaigns (COSINE-100, ANAIS-112) using NaI(Tl) to replicate the precise conditions.

A short history

Scintillation counting dates to 1903, when William Crookes and Julius Elster noticed that alpha particles striking a zinc sulfide screen produced visible sparks (spinthariscope). Systematic scintillation counting was practised by Rutherford and Geiger at Manchester until the Geiger–Müller tube displaced hand-counting in the 1920s. The modern era began in 1948 when Robert Hofstadter at Princeton discovered that sodium iodide crystal doped with a trace of thallium produced a scintillation light yield far exceeding any previous material, and — crucially — matched the spectral sensitivity of the bialkali PMT photocathode. Hofstadter published his findings and promptly applied them to gamma-ray spectroscopy; NaI(Tl) became the standard detector for nuclear medicine, environmental monitoring, and particle physics within a decade. In the 1990s and 2000s, demands from PET scanners and future high-luminosity colliders drove development of faster crystals: BGO (bismuth germanate) was adopted by LEP experiments and early PET scanners; LYSO (lutetium–yttrium oxyorthosilicate) then emerged as the PET standard, combining NaI-comparable light yield with BGO-comparable density and a 40 ns decay time. Since 2010, silicon photomultipliers (SiPMs) have progressively replaced PMTs in many applications, offering higher photon-detection efficiency, insensitivity to magnetic fields, and compatibility with cryogenic temperatures down to 77 K.

Sources

Luxium Solutions (Saint-Gobain successor) NaI(Tl) Scintillation Crystals — light yield 38 photons/keVγ, emission peak 415 nm (https://luxiumsolutions.com/radiation-detection-scintillators/crystal-scintillators/naitl-scintillation-crystals)
Luxium Solutions BC-400/BC-408/BC-412/BC-416 Plastic Scintillators Datasheet — BC-408 decay time 2.1 ns, emission peak 425 nm (https://luxiumsolutions.com/radiation-detection-scintillators/plastic-scintillators/bc400-bc404-bc408-bc412-bc416)
NRC Solid Scintillators course material — NaI(Tl) 52–56 × 10³ photons/MeV (high end), decay time 230 ns, PMT quantum efficiency 20–30% (https://www.nrc.gov/docs/ml1122/ML11229A680.pdf)
Hofstadter discovery — Stanford University Physics Department biography confirming 1948 NaI(Tl) discovery (https://physics.stanford.edu/people/robert-hofstadter)
ATLAS TileCal — 465,000 scintillating tiles, 10,000 Hamamatsu R5900 PMTs, energy range 30 MeV–2 TeV (https://arxiv.org/abs/2105.09099)
LZ detector overview — 494 Hamamatsu R11410 PMTs, 10 tonnes liquid xenon, SURF South Dakota (https://lz.lbl.gov/detector/)
Berkeley Nucleonics — NaI(Tl) key specs: approximately 38,000 photons/MeV light yield (https://www.berkeleynucleonics.com/ufaq/what-are-key-specifications-of-naitl-scintillation-material/)