Modeling Light Signals Using Data from the First Pulsed Neutron Source Program at the DUNE Vertical Drift ColdBox Test Facility at CERN Neutrino Platform

This paper presents a quantitative validation of Fluka simulations against experimental data from a pulsed neutron source test at CERN's DUNE ColdBox facility, demonstrating strong agreement in photoelectron counts and time constants for a small vertical drift LArTPC while identifying systematic effects for future large-scale detectors.

The Gist

The Big Picture: A "Night Vision" Test for a Giant Ice Cube

Imagine the Deep Underground Neutrino Experiment (DUNE) as a massive, high-tech underwater camera buried deep inside a mountain. Its job is to take pictures of ghostly particles called neutrinos that pass through the Earth. To do this, the camera is filled with 40,000 tons of liquid argon (basically super-cold, frozen air).

When a neutrino hits the liquid argon, it creates a tiny spark of light and an electrical signal. The scientists need to see that light perfectly to understand what happened. To make sure their "night vision" cameras work, they built a smaller, test version of the camera called the ColdBox at CERN (the European particle physics lab).

This paper is a report card on how well that test camera worked when they shined a pulsed neutron flashlight at it.

---

The Setup: The "Dark Room" and the "Flashlight"

1. The Test Chamber (The ColdBox):
Think of the ColdBox as a giant, insulated cooler filled with liquid argon. Inside, there are four special light-sensing cameras (called X-ARAPUCA detectors) mounted on the bottom. These cameras are incredibly sensitive; they can detect a single photon (a particle of light), like a firefly blinking in a pitch-black room.

2. The Light Source (The Pulsed Neutron Source):
Instead of neutrinos (which are hard to control), the scientists used a commercial machine that shoots out neutrons in short, rhythmic bursts.

• The Analogy: Imagine a strobe light that flashes 80 times a second. Every time it flashes, it sends a burst of invisible "neutron bullets" into the liquid argon.

3. The Goal:
When these neutron bullets hit the argon or the walls of the container, they make the liquid argon glow (scintillate). The scientists wanted to see if their computer simulations could predict exactly how bright that glow would be and how long it would last, and then compare that to what the real cameras actually saw.

---

The Experiment: "Blind" vs. "Real"

The scientists ran two types of tests:

1. The Simulation (The "Blind" Guess): They used a powerful computer program (Fluka) to simulate the physics. They told the computer: "Here is the neutron gun, here is the argon, here are the cameras. Now, calculate exactly how much light should hit the cameras."
2. The Real Data (The "Eyes"): They turned on the real neutron gun and recorded what the cameras actually saw.

The Calibration (The Ruler):
Before comparing the two, they had to make sure their "rulers" matched.

• They used cosmic rays (natural particles raining down from space) to check the cameras. Since cosmic rays hit everything equally, they used them to calibrate how sensitive each of the four cameras was relative to the others. It's like making sure all four eyes on a face see the same brightness.

---

The Results: A Good Match with a Mystery

1. The Good News (The "Sweet Spot"):
For the majority of the signals (specifically, when the light was between 0 and 650 "units" of brightness), the computer simulation and the real data matched perfectly.

• The Metaphor: It's like predicting the weather. If the computer says "it will rain 1 inch," and the rain gauge actually measures 1 inch, the model is working great. This proves their physics models for how neutrons interact with liquid argon are solid.

2. The Mystery (The "High Brightness" Excess):
However, when the light got very bright (above 650 units), the real cameras saw more light than the computer predicted.

• The Metaphor: The computer predicted a gentle drizzle, but the rain gauge showed a sudden, heavy downpour. The scientists asked: "Where is this extra light coming from?"

3. Solving the Mystery:
They investigated several suspects:

• Cosmic Rays? No, they subtracted those out.
• Pile-up (Two events happening at once)? No, that only explained a small part.
• The Electric Field? Bingo. The liquid argon has an electric field running through it to pull signals to the cameras. The scientists realized that in the "dead" zones (areas where they aren't reading data), the electric field might be weaker or messy.
  • The Analogy: Imagine the electric field is like a wind blowing the light toward the camera. If the wind is stronger or more turbulent in the corners of the room than the computer thought, it would blow more light into the camera than expected. This "messy wind" is the most likely reason for the extra brightness.

4. The Timing (The "Echo"):
They also looked at when the light arrived. After the neutron gun stopped flashing, the light didn't stop instantly; it faded away slowly, like a bell ringing after you stop hitting it.

• The computer predicted a "ringing" time of about 255 microseconds.
• The real data showed a "ringing" time of 257 microseconds.
• Verdict: They are almost identical. This confirms that the computer understands how long the light lasts.

---

Why Does This Matter?

This might sound like a small test, but it's huge for the future of physics.

1. Calibrating the Giant Camera: DUNE needs to measure the energy of particles very precisely. Neutrons are a great "standard candle" (like a known weight on a scale) to calibrate the light sensors. This paper proves we can use neutrons to calibrate the DUNE camera.
2. Cleaning the Signal: In the real DUNE experiment, there will be background noise from neutrons bouncing around the cavern. Understanding exactly how these neutrons behave in light helps scientists filter out the noise and find the real neutrino signals.
3. Future Proofing: The "extra light" mystery taught them that the electric fields in the big, complex DUNE detector need to be modeled very carefully. If they don't, they might misinterpret the data later.

The Bottom Line

The scientists built a test chamber, shot neutrons at it, and compared the real light to a computer prediction. They matched perfectly for most of the data. The small mismatch at the high end taught them a valuable lesson about how electric fields affect light in liquid argon. This success gives them confidence that the giant DUNE detector, currently being built, will work exactly as planned when it starts hunting for neutrinos deep underground.

Technical Summary

1. Problem and Motivation

The Deep Underground Neutrino Experiment (DUNE) aims to utilize large Liquid Argon Time Projection Chambers (LArTPCs) for neutrino oscillation studies. A critical component of the DUNE physics program is the MeV-scale energy calibration of the photon detection system (PDS), which is essential for low-energy physics measurements (e.g., supernova neutrinos and solar neutrinos).

• The Challenge: Calibrating the light yield of LArTPCs at the MeV scale is difficult. A promising "standard candle" is the neutron capture on Argon-40 ($^{40}\text{Ar}$), which releases a gamma cascade with a total energy of 6.1 MeV.
• The Goal: To validate simulation models (specifically Fluka) against real data regarding light signals produced by neutron interactions in liquid argon. This validation is necessary to develop reliable neutron capture tagging techniques for future DUNE operations.
• Context: This study utilizes the DUNE Vertical Drift (VD) ColdBox test facility at CERN, equipped with X-ARAPUCA photon detectors, and a Pulsed Neutron Source (PNS).

2. Methodology

Experimental Setup

• Facility: The VD ColdBox cryostat (internal volume $\approx 3.89 \times 3.91 \times 1$ m) filled with Liquid Argon (LAr) to a height of $\sim 70$ cm.
• Detectors: Four X-ARAPUCA photon detectors (C1–C4) are mounted on the cathode plane. They utilize a mix of SiPMs (FBK and Hamamatsu) and are read out via Power-over-Fiber (PoF).
• Neutron Source: A commercial Thermo Scientific MP 320 Deuterium-Deuterium Generator (DDG). It operates in a special burst mode: 5 bursts of neutrons every 12.5 ms (80 Hz), with each burst lasting 60 $\mu$s.
• Data Acquisition: The system recorded over 160,000 PNS triggers and 250,000 cosmic ray triggers. Cosmic data were used for background estimation and relative calibration.

Simulation (Monte Carlo)

• Framework: The Fluka simulation code (version 2024.1) was used to model the ColdBox geometry, the neutron source, and particle interactions.
• Physics Processes: The simulation tracks neutron transport, including elastic/inelastic scattering and capture on $^{40}\text{Ar}$ and surrounding materials (shielding, cryostat walls).
• Optical Modeling:
  • Scintillation light yield: $2.55 \times 10^4$ photons/MeV (for MIPs at 454 V/cm).
  • Time constants: Fast (6 ns) and Slow (1.5 $\mu$s).
  • Photon Detection Efficiency (PDE): Assumed 3% for all XA modules in the baseline simulation.
  • Note: Only 1/10 of optical photons were tracked to save CPU time, with scaling factors applied later.

Calibration and Analysis

• ADC to PE Conversion: Single photoelectron (SPE) templates derived from LED calibration runs were used to convert Analog-to-Digital Converter (ADC) counts to Photoelectrons (PE).
• Relative PDE Calibration: Cosmic ray tracks passing through the center of the detector were used to normalize the response of the four XA modules relative to each other (C3 was used as the reference).
• Event Selection:
  • Signals required a peak amplitude between 100 PE and 2100 PE to avoid baseline fluctuations and saturation.
  • Time coincidence between the two channels of each XA module was required (within 80 ns).
  • Cosmic backgrounds were subtracted using a data-driven method normalized to the PNS run duration.

3. Key Contributions

1. First Quantitative Test: This is the first study to quantitatively compare detected light signals from a pulsed neutron source in a vertical drift LArTPC against a detailed Fluka simulation.
2. Validation of Neutron Interaction Modeling: The study validates the simulation of neutron interactions not just in active LAr, but crucially in inactive regions (shielding, cryostat walls), which contribute significantly to the detected signal.
3. Time Constant Extraction: The paper successfully extracts the decay time constant of the light signal from the neutron beam-off period, comparing data and simulation to verify the scintillation model.
4. Systematic Error Analysis: A comprehensive study of systematic uncertainties, including electric field variations in buffer regions, detector displacement, and absolute PDE variations.

4. Results

Light Signal Amplitude

• Agreement: There is good agreement between data and simulation for photoelectron counts up to ~650 PE across all four XA modules.
• High PE Excess: An excess of data over simulation is observed for signals >650 PE.
  • Investigation: This excess was ruled out as cosmic background (timing profile differs) or pile-up (shapes are similar).
  • Potential Causes: The paper suggests the excess may stem from:
    • Electric Field Uncertainty: The electric field in the "buffer" LAr (outside the active readout volume) is likely lower than the nominal 454 V/cm. Simulations with a zero-field buffer region showed a significant increase in high-PE events, potentially explaining the tail.
    • Absolute PDE: The actual absolute PDE of the prototype detectors may be higher than the assumed 3%, or cross-talk effects may be under-modeled.
    • Cherenkov Light: Estimated to contribute <2% to the signal, insufficient to explain the excess.

Light Signal Timing

• Decay Constant: An exponential fit to the light signal decay (after the neutron beam stops) yielded a time constant of $257 \pm 8$ $\mu$s in data.
• Simulation Consistency: The Fluka simulation predicted a decay constant of $255 \pm 68$ $\mu$s, which is consistent with the data within one standard deviation.
• Source of Decay: The simulation indicates that the long decay tail is primarily driven by neutron interactions occurring outside the active LAr volume (e.g., in shielding or the cryostat walls), highlighting the importance of modeling the entire facility geometry.

Systematic Effects

• Electric Field: The most significant systematic effect. Changing the field in the buffer LAr from 454 V/cm to 0 V/cm nearly doubled the total PE count on the module closest to the source (C4).
• Detector Position: Shifting the neutron source by 5 cm resulted in minor variations in the photon yield.
• PDE: Increasing the assumed PDE to 4.44% (accounting for cross-talk) increased the high-PE tail, suggesting absolute PDE calibration is a key area for future refinement.

5. Significance and Outlook

• DUNE Readiness: This work provides a validated baseline for modeling neutron interactions in LAr, which is crucial for the MeV physics program of DUNE.
• Neutron Tagging: The ability to model and tag neutron captures (via the 6.1 MeV gamma cascade) will improve energy resolution for low-energy events and help reject cavern neutron backgrounds.
• Future Applications: The analysis framework and validation methods developed here will be applied to larger prototypes (e.g., ProtoDUNE-VD) and the final DUNE Far Detector.
• Calibration Strategy: The results confirm that neutron capture can serve as a reliable "standard candle" for PDS calorimetry, provided that systematic effects like electric field variations in non-active volumes are carefully modeled.

In conclusion, the paper demonstrates that current simulation tools (Fluka) can accurately model the bulk of light signals from neutron interactions in a LArTPC environment, while identifying specific systematic areas (buffer field, absolute PDE) that require further investigation to fully explain high-energy tails.