Back-n Facility: White Neutron Source

• Back-n Facility is a white neutron source at CSNS that uses back-streaming neutrons from a tungsten spallation target to support precise nuclear measurements.
• The facility employs time-of-flight (TOF) methods with sub-nanosecond timing and a PXIe-based digital system for accurate neutron-energy determination.
• Integrated beam shaping, shielding, and versatile detector systems enable high flux and low background for cross-section, irradiation, and imaging experiments.

Searching arXiv for recent and foundational papers on the Back-n facility to ground the article in published work. Back-n is the white neutron experimental facility at the China Spallation Neutron Source (CSNS), built by exploiting back-streaming neutrons emitted from a thick tungsten spallation target irradiated by a 1.6 GeV, 25 Hz pulsed proton beam. The facility was completed in March 2018 and is configured primarily for neutron-induced cross-section measurements, while also supporting irradiation testing, detector calibration, and neutron imaging. Its central measurement paradigm is time of flight (TOF), enabled by the time structure of the primary proton beam and by a PXIe-based digital electronics system that performs waveform digitization and sub-nanosecond timestamping (Collaboration et al., 2021, Yu et al., 2018).

1. Source concept and beam-line geometry

Back-n uses the neutrons that stream backward along, and upstream of, the incident proton beam after spallation in tungsten. In the RTBT (Ring-to-Target Beam Transport) line, a 15° bending magnet is installed approximately 20 m from the target; this deflects charged particles and allows back-streaming neutrons to enter the dedicated neutron beamline. A neutron beam window is located at 26 m, a combined Cu–Fe shutter at 31 m, and two further Cu–Fe collimators are placed near 56 m and 76 m. Two experimental endstations are provided: ES#1 at 55 m in the facility overview, or 56 m in the design study, and ES#2 at 76 m from the target (Collaboration et al., 2021, Zhang et al., 2017).

The spallation target is described as an 11-slice tungsten assembly clad in tantalum and separated by water-cooling layers. The facility overview further notes a Be reflector and three moderators above and below the target for neutron-scattering instruments, while the Back-n beam itself is characterized as using the back-streaming component without a dedicated moderator on the beam exit. In parallel, the overview states that the multiple tungsten slices have modest moderation from the cooling water through the slices. Taken together, these descriptions distinguish Back-n from cold or thermal beamlines while also indicating that the emitted spectrum is not strictly devoid of moderation effects (Collaboration et al., 2021, Qi et al., 2019).

Shielding is integral to the layout. The line includes thick concrete walls after the shutter and collimators, a steel block wall separating RTBT and ES#1, 50 mm boron-polyethylene liners in ES#1 walls and ceiling, and a compact in-hall neutron dump in ES#2 with multilayer boron-polyethylene and lead shielding. The design study emphasizes that the first 20 m of beamline is shared by proton transport and the neutron beam, making the downstream shielding and separation scheme a defining part of the facility architecture (Collaboration et al., 2021, Zhang et al., 2017).

2. Neutron spectrum, flux, and TOF basis

Back-n is reported as providing a continuous white neutron spectrum from 0.5 eV to approximately 200 MeV in the facility overview, while the beam-line design study summarizes the operational range as 1 eV to 100 MeV. Both descriptions place the spectral maximum near 1 MeV and identify the source as a broad-spectrum white neutron facility rather than a narrow-band beamline (Collaboration et al., 2021, Zhang et al., 2017).

At 55 m and 100 kW proton power, the on-axis flux is reported as approximately $2\times10^7\ \mathrm{n/cm^2/s}$ for a $\Phi 50$ mm spot. The overview further tabulates four standard beam-spot modes and gives ES#2 fluxes up to $8.6\times10^7\ \mathrm{n/cm^2/s}$ for the imaging mode at 100 kW. A separate capture measurement at ES#2 reports an instantaneous neutron flux at the sample position of $\simeq 10^7\ \mathrm{n/cm^2/s}$ over the full spectrum, under conditions that included a cadmium filter to remove $<0.5$ eV neutrons during capture measurements (Collaboration et al., 2021, Liang et al., 27 Aug 2025).

The TOF relation used throughout Back-n is

$E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$

with the fractional resolution written as

$\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$

The overview gives $L\simeq 55$–$77$ m and attributes $\Delta t$ to proton-bunch width, target moderation broadening, and detector timing uncertainty. It reports a typical $\Phi 50$0 at the few-per-mille level over much of the 10 keV–1 MeV region, with degradation at the spectral extremes (Collaboration et al., 2021).

The accelerator time structure is mode dependent. The beam-line design study lists Normal Mode as two 13 ns (rms) bunches per pulse separated by 410 ns, Dedicated Mode 1 as a single 13 ns bunch at 50% power, and Dedicated Mode 2 as a single 1.5 ns bunch at 15% power. In the 2025 rhodium measurement, the double-bunch configuration is specified differently: each proton pulse has a full-width-at-half-maximum of 41 ns with the same 410 ns inter-bunch spacing. The same measurement reports a timing resolution of approximately 10 ns for the proton pulse plus electronics in its PXIe-based 12-bit, 1 GS/s capture setup (Zhang et al., 2017, Liang et al., 27 Aug 2025).

3. Instrumentation and detector systems

Back-n is equipped with a heterogeneous instrument suite oriented toward reaction-channel coverage. The facility overview lists capture spectrometers based on $\Phi 50$1 scintillators and GTAF-II, a 4$\Phi 50$2 BaF$\Phi 50$3 array with 40 units and approximately 90% solid angle; the FIXM multi-cell ionization chamber for fission; the NTOX transmission setup for total cross sections; LPDA for light charged particles; a gated CMOS camera for resonant imaging; and user-supplied HPGe detectors. The electronics paper adds FINDA and GAEA to the common readout framework and states that the general-purpose electronics was designed for all of the seven detectors in Back-n (Collaboration et al., 2021, Yu et al., 2018).

Detector outputs span a wide dynamic and temporal range. The electronics description states that signals extend from a few mV for light particles up to approximately 2 V for high-rate BaF$\Phi 50$4, with pulse widths from tens to hundreds of ns. These analog signals are conditioned in Signal Conditioning Modules (SCMs), which provide gain stages, anti-aliasing filters matched to the ADC input range, DC offsets for unipolar or bipolar signals, differential drivers, programmable gain from $\Phi 50$5 to $\Phi 50$6, anti-aliasing low-pass filtering with an approximately 200 MHz cutoff, AC/DC coupling, and baseline restoration (Yu et al., 2018).

Digitization is performed in Field Digitizer Modules (FDMs) within a PXIe chassis. The electronics paper specifies 8-bit, 500 MSPS flash ADCs for C4D6, FIXM, LPDA, FINDA, and GAEA, and 12-bit, 1 GSPS digitizers for GTAF-II channels because of stricter SNR and dynamic-range requirements. The facility overview describes the common chassis more generally as a 12-bit, 1 GS/s folding-ADC plus FPGA platform with fully digital, programmable trigger logic, 20–64 signal channels, and sustained DAQ rates up to approximately 400 Mbps input and approximately 200 Mbps disk write (Yu et al., 2018, Collaboration et al., 2021).

Sample environments and station logistics are likewise differentiated by instrument class. DAQ racks are located in both endstations and synchronized via GPS-derived timestamps. NTOX and LPDA sample changers are fully automated, FIXM can hold up to 8 samples simultaneously, GTAF-II uses manual exchange with an internal source changer, and the environments range from vacuum chambers for LPDA and TPC configurations to atmospheric operation for C$\Phi 50$7D$\Phi 50$8 and GTAF (Collaboration et al., 2021).

4. TOF electronics and timestamp reconstruction

The dedicated TOF electronics is based on a PXIe platform composed of FDMs, a Trigger and Clock Module (TCM), and SCMs. The TCM occupies the slot-0 timing position in the chassis, receives the asynchronous $\Phi 50$9 signal from the CSNS accelerator front-end, generates a global 125 MHz clock phase-locked to $8.6\times10^7\ \mathrm{n/cm^2/s}$0, and distributes synchronized timing to all FDMs through PXIe differential star pairs. It also interfaces to the White Rabbit timing system for absolute UTC tagging (Yu et al., 2018).

In the Back-n timing model, $8.6\times10^7\ \mathrm{n/cm^2/s}$1 represents neutron emission from the target and is the start of the timestamp. On the TCM, the interval

$8.6\times10^7\ \mathrm{n/cm^2/s}$2

between the $8.6\times10^7\ \mathrm{n/cm^2/s}$3 edge and the next 125 MHz clock rising edge is measured by a 17-tap carry-chain FPGA TDC. After a fixed buffering delay

$8.6\times10^7\ \mathrm{n/cm^2/s}$4

the TCM distributes a synchronized $8.6\times10^7\ \mathrm{n/cm^2/s}$5 on the backplane star trigger buses. Each FDM receives the common $8.6\times10^7\ \mathrm{n/cm^2/s}$6 and measures its local delay

$8.6\times10^7\ \mathrm{n/cm^2/s}$7

between $8.6\times10^7\ \mathrm{n/cm^2/s}$8 and the local sampling-clock edge (Yu et al., 2018).

The stop timestamp is derived from detector waveform sampling. The electronics paper defines the stop as the time of the first sample point above zero crossing, or above a programmable threshold, in the digitized pulse. On the FDM, the FPGA records $8.6\times10^7\ \mathrm{n/cm^2/s}$9 and also

$\simeq 10^7\ \mathrm{n/cm^2/s}$0

the first ADC-sample time offset from the clock edge, using a matching carry-chain TDC. For offline over-threshold timing, a Constant Fraction Discriminator (CFD) algorithm is applied to the digitized waveform, yielding an additional fine delay correction

$\simeq 10^7\ \mathrm{n/cm^2/s}$1

in software (Yu et al., 2018).

The same source writes

$\simeq 10^7\ \mathrm{n/cm^2/s}$2

with fixed calibrated delays $\simeq 10^7\ \mathrm{n/cm^2/s}$3 from cables, backplane, and FPGA routing, and gives the total TOF as

$\simeq 10^7\ \mathrm{n/cm^2/s}$4

The section also provides the LaTeX-style expression

$\simeq 10^7\ \mathrm{n/cm^2/s}$5

Once TOF is obtained, the neutron kinetic energy is written as

$\simeq 10^7\ \mathrm{n/cm^2/s}$6

or in practical units

$\simeq 10^7\ \mathrm{n/cm^2/s}$7

The electronics paper reports a TCM TDC average bin of approximately 63 ps, an FDM TDC average bin of approximately 46 ps, and end-to-end TOF reproducibility of approximately 280 ps RMS over a 10 ms full scale. Its stated conclusion is that the TOF accuracy is sub-nanosecond and meets the Back-n requirement (Yu et al., 2018).

A notable architectural property is that this method does not require additional modules. The same description characterizes the combined system as fully trigger-less: every pulse carries its own timestamps and can be reassembled offline on a common time axis for physics analysis. This places the timing problem largely in the domain of calibrated digital reconstruction rather than external coincidence hardware (Yu et al., 2018).

5. Beam shaping, profile control, and background suppression

The beam delivered to experiments is defined by a multi-stage collimation system rather than by the raw source divergence. The layout comprises a motorized shutter with interchangeable apertures and two downstream collimators. The design study states that the optical design is based on three neutron-ray lines to control core spot, halo, and intensity, and reports beam-core uniformity of at most 10% over the nominal spot with beam-halo fraction kept below a few percent for the nominal settings (Zhang et al., 2017).

Background suppression is treated as a beam-line design problem and an in-hall shielding problem simultaneously. For ES#2, the design study reports absolute beam-off background levels of at most $\simeq 10^7\ \mathrm{n/cm^2/s}$8 for both neutrons and gammas, while the overview gives approximately $\simeq 10^7\ \mathrm{n/cm^2/s}$9 and $<0.5$0 in ES#2 for a $<0.5$1 mm spot and approximately $<0.5$2 and $<0.5$3 in ES#1 for a $<0.5$4 mm spot. The same sources attribute suppression to the RTBT shield wall, boron-polyethylene liners, local concrete shielding, Cu–Fe collimators, and the compact ES#2 dump with boron-polyethylene bushings and lead outer shell (Zhang et al., 2017, Collaboration et al., 2021).

The beam profile has been measured directly with a custom 2D Micromegas detector. In the reported ES#1 configuration, both shutter and Collimator-1 were set to $<0.5$5 mm. The reconstructed central-slice full widths at half maximum were approximately 58 mm in the $<0.5$6 projection and approximately 54 mm in the $<0.5$7 projection, in agreement with Monte Carlo expectations of 58 mm and 54 mm. The same measurement attributes a slight ellipticity of the neutron profile to the upstream proton-beam spot on the target, described as approximately $<0.5$8 (Qi et al., 2019).

These beam-profile data are operationally significant because sample placement and detector alignment depend on the true spot size and centroid. The Micromegas study emphasizes that knowledge of the spatial distribution is essential for avoiding edge effects and flux-gradient errors in cross-section measurements. A common simplification is to describe Back-n only as an intense back-streaming neutron beam; the facility data instead show a deliberately engineered phase space in which collimation, shielding, and dump design are part of the measurement system (Qi et al., 2019, Zhang et al., 2017).

6. Experimental use cases and demonstrated measurements

From April 2018 onward, Back-n has been used for over 40 nuclides in $<0.5$9, $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$0, $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$1, and $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$2 channels, spanning basic nuclear data, reactor engineering isotopes, and astrophysical $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$3-process targets. In favorable cases, overall cross-section uncertainties of at most 3% have been reported. The same overview identifies additional use cases in single-event-effects testing of electronics, displacement-damage and total-dose studies, neutron-response calibration, resonant imaging, and dual-mode neutron/$E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$4 imaging (Collaboration et al., 2021).

A representative neutron-capture application is the direct measurement of the $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$5 cross section at Back-n from 1 eV to 1 MeV. That experiment used two C$E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$6D$E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$7 liquid-scintillator detectors positioned at approximately 125° relative to the beam, a flight path of $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$8 m, pulse-height weighting technique (PHWT), and background runs with an empty frame, natC, and natPb. In the resolved resonance region from 0.3 to 4 keV, the weighted yield $E_n=\frac{1}{2}m_n\left(\frac{L}{t}\right)^2,$9 was fitted with the Reich–Moore $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$0-matrix code SAMMY, including Doppler broadening, self-shielding, and multiple scattering corrections (Liang et al., 27 Aug 2025).

The rhodium measurement reports several new resonances at 26.5, 79.9, 86.8, 102.3, 941.9, and 976.8 eV that were not present in ENDF/B-VIII.1, JEFF-3.3, or JENDL-5. It also identifies impurity resonances at 11.9 eV and 33.0 eV as arising from $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$1 and $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$2 contaminants. In the unresolved resonance region from 10 to 1000 keV, average capture cross sections were extracted in 34 neutron-energy bins, with total systematic uncertainty below 10%. Using TALYS interpolation, the corresponding Maxwellian-averaged cross sections were calculated for $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$3–100 keV, giving $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$4 mb at $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$5 keV, in excellent agreement with ENDF/B-VIII.1 at 811 mb and KADoNiS 1.0 at 837 mb (Liang et al., 27 Aug 2025).

The facility’s non-cross-section applications rely on the same white-spectrum and long-flight-path characteristics. The overview reports a fully parasitic irradiation facility upstream of the shutter with $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$6 total flux, single-event-effects testing with $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$7 above 1 MeV at ES#1, and a time-gated CMOS imaging system with spatial resolution of approximately 50 $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$8m and time windows of 10–100 ns. This range of use indicates that Back-n functions not only as a measurement beamline but also as a general neutron-effects and neutron-imaging platform (Collaboration et al., 2021).

7. Comparative position and development trajectory

The design study compares Back-n with other white neutron sources including n_TOF at CERN, WNR at LANSCE, GELINA at IRMM, and the RPI LINAC. Within that comparison, Back-n is summarized as combining a broad energy range, high flux up to $\frac{\Delta E}{E}\approx 2\frac{\Delta t}{t}.$9, good timing with $L\simeq 55$0 in D2 mode, low backgrounds, and flexible beam spots from $L\simeq 55$1 mm to 90 mm $L\simeq 55$2 90 mm. The facility overview goes further and states that Back-n’s on-axis flux of $L\simeq 55$3 at 55 m is the highest among 0.5 eV–200 MeV white sources worldwide, with overall performance among that of the best white neutron sources in the world (Zhang et al., 2017, Collaboration et al., 2021).

Its measurement record supports that positioning. The rhodium study explicitly attributes the discovery of new resonances to the combination of high flux, long flight path, cadmium filtering, and a two-detector PHWT geometry with signal-to-background ratios greater than 10:1 across 1 eV–1 MeV. It also states that high flux and long flight path allow sub-meV energy resolution in the resolved region and percent-level accuracy in the unresolved resonance region (Liang et al., 27 Aug 2025).

The reported future directions are still centered on nuclear-data production, but they broaden the facility’s methodological scope. These include extension of TOF capture measurements to other $L\simeq 55$4-process branch-point isotopes such as $L\simeq 55$5 and $L\simeq 55$6, optimization of moderator assemblies to enhance thermal-neutron flux for activation studies, combined TOF–activation experiments to reduce systematic uncertainties in Maxwellian-averaged cross sections, and use of the Back-n white beam for differential scattering and inelastic cross-section measurements. A plausible implication is that Back-n is evolving from a high-flux white source optimized for differential nuclear-data work into a more integrated platform linking differential measurements, activation, irradiation effects, and application-oriented isotope studies (Liang et al., 27 Aug 2025).