Nab Experiment: Precision Neutron Beta Decay Study

• Nab Experiment is a precision study of free-neutron decay focused on measuring the electron–antineutrino correlation coefficient (a) and the Fierz term (b) to test the Standard Model.
• It employs an asymmetric magnetic spectrometer with advanced time-of-flight techniques to reconstruct decay kinematics and control systematic uncertainties.
• The experiment integrates state-of-the-art segmented silicon detectors, innovative thermal stabilization, and reconfigurable data-acquisition systems to achieve sub-nanosecond timing precision.

Nab is a precision neutron $\beta$-decay experiment at the Fundamental Neutron Physics Beamline of the Spallation Neutron Source, built to measure the electron–antineutrino correlation coefficient $a$ and the Fierz interference term $b$ in unpolarized free-neutron decay by recording the decay electron energy and the recoil-proton time of flight in coincidence (Fry et al., 2018). Through the relation between $a$ and the weak-coupling ratio $\lambda=g_A/g_V$, and in combination with the neutron lifetime, Nab is intended to provide an independent determination of $V_{ud}$ and a first-row CKM unitarity test that is free of nuclear-structure corrections, while also probing non-$(V-A)$ scalar and tensor contributions through $b$ (Broussard et al., 2018).

1. Physics motivation and decay observables

The experimental program is anchored in the kinematics of neutron decay,

$n \to p + e^- + \bar\nu_e,$

for which the fully differential rate may be written, neglecting recoil-order and radiative corrections, as

$$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$

In the Standard Model, the electron–antineutrino correlation coefficient depends only on $a$0 through

$a$1

while the Fierz term vanishes at leading order; a nonzero $a$2 would therefore signal scalar or tensor admixtures to the charged-current weak interaction (Broussard et al., 2018).

Nab is motivated by the role of neutron decay in determining $a$3 without nuclear-structure corrections. The first-row unitarity condition,

$a$4

has been highlighted as a precision test because recent improvements in theory and experiment have pushed the sum of squares below unity by $a$5 in the Particle Data Group 2024 assessment summarized in collaboration reporting. In this framework, a measurement of $a$6 fixes $a$7, and together with the neutron lifetime $a$8 determines $a$9 through

$b$0

or equivalently

$b$1

Collaboration documents state precision targets of $b$2 and $b$3, together with $b$4 or $b$5, depending on the analysis context (Fry et al., 2018, Broussard et al., 10 Nov 2025).

2. Spectrometer concept and kinematic reconstruction

The apparatus is a long asymmetric electromagnetic spectrometer. Collaboration overviews describe a $b$6-tall system in which the magnetic field is about $b$7 in the decay volume, rises to $b$8 in a magnetic filter region, and then expands adiabatically to $b$9 over a $a$0 time-of-flight region leading to the upper detector; a second detector is located below the decay volume in a $a$1 region (Fry et al., 2018). Electric potentials are superposed on this magnetic transport so that the upper detector is held at $a$2 in the primary $a$3-measurement configuration, accelerating recoil protons with birth energies below $a$4 to about $a$5 and allowing them to traverse the detector dead layer (Broussard et al., 2016).

The basic reconstruction strategy is coincidence kinematics. The electron deposits essentially its full kinetic energy in silicon, while the proton momentum is inferred from its time of flight. In the simplest approximation,

$a$6

with $a$7 an effective flight length, and energy–momentum conservation gives

$a$8

This permits inference of $a$9 event by event from the measured pair $\lambda=g_A/g_V$0, or equivalently from $\lambda=g_A/g_V$1, whose distribution in electron-energy slices carries the coefficient $\lambda=g_A/g_V$2 (Broussard et al., 2018).

A more complete transport relation uses the measured field and potential profiles,

$\lambda=g_A/g_V$3

and in practice Monte Carlo response functions map $\lambda=g_A/g_V$4 onto $\lambda=g_A/g_V$5 (Fry et al., 2018). Recent descriptions emphasize that the asymmetric field geometry captures nearly all decay-particle pitch angles above threshold and yields full coverage of the allowed $\lambda=g_A/g_V$6 phase space for $\lambda=g_A/g_V$7 and all proton trajectories (Broussard et al., 10 Nov 2025).

3. Detection system, readout, and thermal infrastructure

Nab uses thick, large-area, highly segmented silicon detectors developed with Micron Semiconductor. The detector format common to the collaboration’s instrumentation papers consists of a roughly $\lambda=g_A/g_V$8-diameter active area divided into 127 hexagonal pixels, with thicknesses of $\lambda=g_A/g_V$9 or $V_{ud}$0 and dead layers at or below the $V_{ud}$1 scale (Broussard et al., 2016). The segmentation supports position reconstruction, backscatter identification, and coincidence logic, while the thickness is sufficient to stop electrons up to the neutron-decay endpoint. Prototype operation with 19 of the 127 pixels instrumented demonstrated coincident detection of $V_{ud}$2 particles and recoil protons from neutron decay at LANSCE, with energy thresholds below $V_{ud}$3, energy resolution of about $V_{ud}$4 FWHM, and rise time of about $V_{ud}$5 (Broussard et al., 2016).

The full experiment employs two detector assemblies and 256 analog channels. The data-acquisition architecture is based on National Instruments PXIe-5171 reconfigurable oscilloscope modules, arranged as sixteen 8-channel modules per detector with a dedicated timing/synchronization module in each remote chassis. At the FPGA level, raw ADC samples are filtered and converted into L1 triggers; on the host, LabVIEW time-orders the L1 records and applies L2 logic for coincidence, calibration, random, and periodic triggers. The architecture is explicitly reconfigurable at runtime through configuration files and controls, and reported performance includes an L1 trigger identification rate of at least $V_{ud}$6, a peak raw-signal readout rate of approximately $V_{ud}$7, and a sustained commissioning readout of approximately $V_{ud}$8 from roughly $V_{ud}$9 L2 triggers (Mathews et al., 2024).

Thermal stabilization is an integral part of the detector design. The collaboration’s gaseous-helium cooling paper describes two identical closed recirculating helium loops, one per detector, intended to cool the silicon detectors and associated electronics below $(V-A)$0 with $(V-A)$1 stability (Richburg et al., 8 Jan 2026). Each loop uses a pump, mass-flow controller, counter-flow heat exchanger, cryocooler-mounted primary heat exchanger, in-line heater and sensors, vacuum-jacketed transfer lines, and a detector-end OFHC-copper “ring” exchanger. Reported operating points include flow rates of $(V-A)$2–$(V-A)$3 in steady operation, cool-down from $(V-A)$4 to about $(V-A)$5 in roughly $(V-A)$6, detector-end temperatures near $(V-A)$7, and $(V-A)$8 variation over $(V-A)$9 with the heater off (Richburg et al., 8 Jan 2026).

4. Calibration, detector modeling, and systematic control

A defining feature of Nab is the degree to which detector response is modeled and calibrated as part of the physics analysis. Detailed semiconductor transport simulations were developed to understand proton pulse formation at the sub-nanosecond level. These simulations include drift–diffusion, weighting fields via Ramo’s theorem, Coulomb self-repulsion, plasma delay, and the electronics transfer function. They showed that fitting all pulses with a single center-pixel template can produce mean timing biases up to $b$0–$b$1 near pixel edges and a net $b$2 bias averaged over the pixel, whereas pulse-shape discrimination and template matching can reduce the mean bias to approximately zero at high signal-to-noise ratio (Hayen et al., 2022).

Dedicated proton-beam and conversion-electron measurements were then used to fix the detector material model and timing response. One such study exposed pixels to $b$3, $b$4, and $b$5 protons and to $b$6Cd and $b$7Sn conversion electrons over multiple cooling cycles spanning one year. The proton peak position was stable within the energy-calibration uncertainty of $b$8, and the observed energy loss of about $b$9–$n \to p + e^- + \bar\nu_e,$0 was reproduced by dead-layer models corresponding to thicknesses of approximately $n \to p + e^- + \bar\nu_e,$1 or $n \to p + e^- + \bar\nu_e,$2, well below the $n \to p + e^- + \bar\nu_e,$3 specification. The same study extracted a radial impurity profile rising from $n \to p + e^- + \bar\nu_e,$4 at the center to $n \to p + e^- + \bar\nu_e,$5 at the edge, and concluded that proton timing systematic uncertainties from pulse-shape effects were below $n \to p + e^- + \bar\nu_e,$6, sufficient for Nab (Taylor et al., 19 Nov 2025).

Electrostatics in the decay and filter regions constitute another major systematic. For the $n \to p + e^- + \bar\nu_e,$7 measurement, the collaboration required the potential difference between the decay volume and the magnetic filter to satisfy

$n \to p + e^- + \bar\nu_e,$8

corresponding to an electric field on the order of $n \to p + e^- + \bar\nu_e,$9 or less in the low-field region (Li, 2024). To control contact-potential differences, the titanium electrode surfaces were coated with a silver-copper conductive paint, and Kelvin-probe measurements were used to map work-function variations. COMSOL-based field calculations using the measured work-function inputs yielded an electric field in the decay/filter region below the $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$0 specification, and Monte Carlo studies found that $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$1 implies $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$2 (Li, 2024).

For the Fierz term, the dominant systematics arise from detector calibration and response modeling. A dedicated analysis of the main electrode system and $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$3 measurement identified requirements on gain, offset, nonlinearity, response-tail knowledge, coincidence window, and geometric acceptance. In that budgeting, floating the gain parameter in the spectrum fit removes the first-order gain bias, while the remaining dominant contributions can be held to a total

$$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$4

below the design goal of $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$5 (Li, 2024). An independent approximation method subsequently reproduced Monte Carlo slopes for calibration-induced biases, including $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$6 and $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$7 at fixed gain, providing a fast cross-check of the response-model systematics (Li, 2024).

5. Commissioning history and operational improvements

Commissioning exposed several nontrivial operational limitations that directly affected the physics reach. In 2023, the collaboration observed proton deposits of only about $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$8 rather than the expected $$\frac{d^3\Gamma}{dE_e\,d\Omega_e\,d\Omega_\nu} \propto F(Z,E_e)\,p_eE_e\,(E_0-E_e)^2 \Bigl[ 1 +a\,\frac{\mathbf p_e\!\cdot\!\mathbf p_\nu}{E_eE_\nu} +b\,\frac{m_e}{E_e} +\cdots \Bigr].$$9, traced to surface contamination or ice formation that effectively increased the dead layer. The reported mitigations included an improved ultrahigh-vacuum test environment, more stringent detector-qualification procedures, and the use of silicon witness samples for surface analysis. After these interventions, proton deposits were restored to the expected $a$00–$a$01 and the proton detection efficiency exceeded $a$02 across the active area (Broussard et al., 10 Nov 2025).

Electronics stability was also substantially improved. Scaling the front-end preamplifier cards from a few channels to 127 channels had produced rail-to-rail oscillations, initially recoverable only by depowering whole boards and thereby losing central pixels. Subsequent redesign of the segment-to-board mapping, removal of optional shaping amplifiers, and introduction of new adapter boards with thicker ground layers, shorter traces, and optimized power/ground planes improved the equivalent noise charge from about $a$03 rms to about $a$04 rms, corresponding to less than or equal to $a$05 energy noise, while eliminating the oscillations and allowing all 127 channels to run stably (Broussard et al., 10 Nov 2025).

Segment availability and timing synchronization underwent similar refinement. Connector damage, accidental disconnects, and unpowered preamplifier boards had led to as much as $a$06 of detector segments not reporting in 2023, with direct consequences for electron-backscatter reconstruction and solid-angle coverage. By 2025, after redesigned cable harnesses, ruggedized connectors, and a fully integrated electronics package, more than $a$07 of pixels reported reliably in physics runs. The same report describes corrected timestamp-distribution firmware stabilizing sub-nanosecond inter-board timing, optical survey and shimming of the electrode eliminating $a$08 systematic shifts in proton time-of-flight spectra, and bake-outs to below $a$09 together with improved high-voltage conditioning protocols stabilizing the $a$10 potential (Broussard et al., 10 Nov 2025).

6. Measurements, projected precision, and broader significance

Nab’s early physics output has moved beyond apparatus commissioning. A 2025 report presented the first full Dalitz-plot representation of neutron $a$11-decay phase space in the Nab spectrometer for electrons above $a$12 (Gonzalez et al., 22 Aug 2025). That analysis used $a$13 decays recorded over $a$14 of live time, reconstructed a well-populated “teardrop” envelope in the $a$15 plane, and compared the phase-space edges with Geant4-based simulations. The same work reported an earliest-front-edge timing difference

$a$16

and, from both proton-front-edge and electron-endpoint analyses, found no excess endpoint consistent with an exotic excited neutron; at $a$17 confidence it set

$a$18

excluding the excited-neutron explanation of the neutron-lifetime anomaly for de-excitation lifetimes in the range $a$19 (Gonzalez et al., 22 Aug 2025).

The longer-term physics case remains the precision extraction of $a$20, $a$21, and new-physics constraints from neutron decay. Collaboration projections state that Nab aims for $a$22, corresponding to $a$23, and that, when combined with a neutron-lifetime uncertainty of $a$24, the expected precision on $a$25 is less than or equal to $a$26 (Broussard et al., 10 Nov 2025). Earlier design documents phrased the same program as $a$27, $a$28, and $a$29 (Fry et al., 2018). In either formulation, Nab is intended to supply a neutron-based determination of weak couplings that does not depend on nuclear-structure corrections and therefore directly complements superallowed nuclear $a$30 decay.

Within the broader landscape of precision weak-interaction experiments, Nab occupies the niche of an unpolarized-neutron correlation measurement that couples a long-flight-path spectrometer to segmented silicon calorimetry and nanosecond-level detector-response control. Its significance lies not only in the target observables $a$31 and $a$32, but also in the fact that the entire neutron-decay phase space can be reconstructed with a degree of redundancy that supports cross-checks through electron energy, proton time of flight, backscatter tagging, phase-space imaging, and dedicated calibration campaigns (Broussard et al., 2018). This suggests that the experiment’s final impact will depend as much on systematic closure across these subsystems as on raw counting statistics.