1. IBD Detection Efficiencies and Uncertainties
Liang Zhan, Institute of High Energy Physics
1st Workshop on Reactor Neutrino Experiments, Seoul, Oct , 2016

2. The number of IBD events
Reactor antineutrino flux
IBD cross section
Baseline
Detection efficiency
Number of target protons
This talk will focus on the detection efficiency and number of target protons.

3. Breakdown of detection efficiency
Break into sub efficiencies according the physical effects
Efficiencies of delayed energy cut, Gd capture fraction and spill-in correction are equal for Daya Bay identically designed antineutrino detectors (ADs)
Other efficiency: multiplicity cut (reviewed in the IBD selection talk)
Delayed energy cut (>6 MeV) efficiency for Gd capture events
Gd capture for neutrons from inside and outside of GdLS target.
Gd capture fraction for IBD neutrons generated in GdLS target.
Extra IBD events due to IBD neutrons generated outside of GdLS target.

4. Detection efficiency overview
arXiv: 1610.XXXX
Correlated uncertainties have no impact in the relative measurement for oscillation.
They are important in the absolute reactor flux measurement.
Uncorrelated uncertainties are propagated to the relative measurement of far/near deficit.
Three other talks will cover
Flasher cut
Gd capture fraction and spill-in
Multiplicity cut and livetime
This talk will focus on other aspects.

5. Number of target protons
Target mass was measured with an uncertainty of 0.015% (NIM A811, 133(2016)) and precisely monitored by the liquid level in the overflow tank.
Np uncertainty dominated by the hydrogen mass fraction in GdLS.
Combined results of two independent combustion measurements. (one in NIM A811, 133(2016))
Results of GC-MS (Gas Chromatography-Mass Spectrometer) measurement are served as a cross check.
Uncertainty of GC-MS measurement may be smaller but is hard to estimate because some components (such as impurities) in the liquid scintillator have no or very different response in the GC-MS measurement and the fractions of these components are hard to estimate.

6. Data-driven efficiency studies
Usually, It is 1 for well tuned MC
The selected IBD data sample never knows what is the selection efficiency.
Perform calibration runs to study the efficiency. However, the calibration data is similar but not identical to IBD data (such as spallation neutrons vs. IBD neutrons).
Perform calibration MC and compare it with the calibration data. Good agreement gives our confidence on the MC.
Determine the IBD detection efficiency from the validated MC.
Correlated efficiency uncertainty determined by the comparison between data and MC.
Uncorrelated efficiency uncertainty determined by the comparison of data between ADs.

7. Capture time cut
IBD neutrons will scatter on the nucleus and be captured on the nucleus after thermalized.
Capture time cut: 1 μs < ΔT < 200 μs
Efficiency: 98.7%
Inefficiency: 0.2% for ΔT < 1 μs and 1.1% for ΔT > 200 μs
By default, Geant4 uses free gas model (hydrogen is free), which does not agree with data.
We modified it to include the effect of the binding of scattering nucleus
Am-C neutron source data

8. Free gas model Binding model
Good agreement achieved by taking into account the binding effect.
0.1% uncertainty estimated from difference between Data and updated MC.

9. Uncorrelated uncertainty of capture time cut
Almost identical capture time distributions among all ADs.
Uncorrelated uncertainty is negligible (0.01%) determined by the variation of Gd concentration (<0.5%).
IBD neutron

10. Prompt energy cut Prompt energy cut: E > 0.7 MeV
Efficiency = 99.81%
Inefficiency due to energy loss in inner acrylic vessel
Variation on energy scale near 0.7 MeV and IAV thickness ( mm) induce a efficiency variation ~ 0.1%

11. Uncorrelated uncertainty of prompt energy cut
Relative energy scale uncertainty determined to be 0.2% (refer to energy response talk)
This relative energy scale variation introduce a 0.01% variation on the prompt energy cut efficiency.

12. Delayed energy cut Delayed energy cut: 6 MeV < E < 12 MeV
6 MeV cut
Delayed energy cut: 6 MeV < E < 12 MeV
Efficiency = 92.71%
Inefficiency determined by the energy tail due to the energy leakage.
The tail shape depends on the energy spectrum of individual gamma emitted from neutron capture on Gd.

13. Individual gamma spectrum
Three models are studied
M13A, Old uses the gamma spectrum based on spectroscopic measurements (Nuclear Data Tables 5 (1968))
M14A, Caltech uses the gamma spectrum based on the measurement at Caltech using a benchtop HPGe detector
M14A, Geant uses the gamma spectrum from Geant4 packages with updates requiring the energy conservation of gamma emission (7.94 MeV for 157Gd capture, 8.54 MeV for 155Gd capture)

14. M14A, Geant model has the best agreement with the data.
To quantify the agreement, we define a tail shape metric.
Rts difference between Geant model and data is < 0.1%
Other contribution to the energy cut uncertainty
Statistics: 0.1%
Background subtraction: 0.1%
lack of constraint on low-energy regions (<3 MeV), 100% uncertainty conservatively estimated for R<3MeV: 0.9%
Non-uniformity and non-linearity difference between data and MC: 0.3%
Old
Geant
Dominate uncertainty

15. Uncorrelated uncertainty for delayed energy cut
A 0.2% relative energy scale variation between detectors corresponds to a 0.05% variation in the cut efficiency. Consider the dependence of the energy scale variation on the vertex in GdLS, the variation of the cut efficiency is 0.07%
As a cross check, direct comparison of the delayed energy spectrum above 3.6 MeV shows good agreement between ADs. Variation on the spectrum introduces 0.08% variation on the efficiency

16. Summary Absolute efficiency estimation based on well tuned MC.
Correlated uncertainty determined by data and MC comparison.
Uncorrelated uncertainty determined by the comparison between ADs.

17. Back up slides

18. Target mass

19. Energy peak variation versus the vertex pixel

20. Ratio of N(6-12 MeV)/N(3.6-12 MeV) calculated for each AD.
Good agreement on the ratio demonstrates small uncorrelated uncertainty of delayed energy cut.