Search for an anomalous excess of inclusive charged-current ${\nu }_e$ interactions in the MicroBooNE experiment using Wire-Cell reconstruction

We report a search for an anomalous excess of inclusive charged-current (CC) ${\nu }_e$ interactions using the Wire-Cell event reconstruction package in the MicroBooNE experiment, which is motivated by the previous observation of a low-energy excess (LEE) of electromagnetic events from the MiniBooNE experiment. With a single liquid argon time projection chamber detector, the measurements of ${\nu }_{\mu }$ CC interactions as well as ${\pi }^0$ interactions are used to constrain signal and background predictions of ${\nu }_e$ CC interactions. A data set collected from February 2016 to July 2018 corresponding to an exposure of $6.369\times {10}^{20}$ protons on target from the Booster Neutrino Beam at FNAL is analyzed. With $x$ representing an overall normalization factor and referred to as the LEE strength parameter, we select 56 fully contained ${\nu }_e$ CC candidates while expecting $69.6\pm 8.0$ (stat.) $\pm 5.0$ (sys.) and $103.8\pm 9.0$ (stat.) $\pm 7.4$ (sys.) candidates after constraints for the absence (${\mathrm{eLEE}}_{x=0}$) of the median signal strength derived from the MiniBooNE observation and the presence (${\mathrm{eLEE}}_{x=1}$) of that signal strength, respectively. Under a nested hypothesis test using both rate and shape information in all available channels, the best-fit $x$ is determined to be 0 (${\mathrm{eLEE}}_{x=0}$) with a 95.5% confidence level upper limit of $x$ at 0.502. Under a simple-vs-simple hypotheses test, the ${\mathrm{eLEE}}_{x=1}$ hypothesis is rejected at $3.75\sigma$, while the ${\mathrm{eLEE}}_{x=0}$ hypothesis is shown to be consistent with the observation at $0.45\sigma$. In the context of the eLEE model, the estimated 68.3% confidence interval of the ${\nu }_e$ CC hypothesis to explain the LEE observed in the MiniBooNE experiment is disfavored at a significance level of more than $2.6\sigma$ ($3.0\sigma$) considering MiniBooNE’s full (statistical) uncertainties.

Figure 1

Expected ${\nu }_e$ CC events in the MicroBooNE detector for ${\mathrm{eLEE}}_{x=1}$ hypothesis and ${\mathrm{eLEE}}_{x=0}$ hypothesis as a function of true neutrino energy at $6.4\times {10}^{20}$ proton-on-target (POT) exposure, assuming 100% detection efficiency.

Figure 2

Predicted neutrino flux from different decay modes at the MicroBooNE detector: (a) flux of different neutrino flavors, (b) ${\nu }_{\mu }$ flux of different decay modes (the peak at 236 MeV corresponds to Kaon decay at rest), and (c) ${\nu }_e$ flux of different decay modes.

Figure 3

Taken from Ref. [57]. (a) The MicroBooNE detector cryostat with the field cage shown inside. (b) Inside the cryostat of the MicroBooNE detector, visualized with the VENu software [62]. The maximum drift distance is 2.56 m with a drift electric field of $273\mathrm{V}/\mathrm{cm}$. The light-collection system, which consists of 32 PMTs, is located behind the three anode wire planes, which detect ionization charge.

Figure 4

A ${\nu }_e$ CC interaction candidate from MicroBooNE data. The X axis is the drift electric field direction from the TPC anode to the cathode. The Y axis is vertical up, and the Z axis is along the neutrino beam direction. Panel (a) shows three 2D projections of reconstructed 3D clusters in the full TPC readout window before charge-light matching. Each cluster is shown in a different color. The gray box represents the TPC active volume while the two ends along the X axis correspond to the trigger time and the maximum drift time relative to the trigger. Panel (b) shows the 2D projections of the ${\nu }_e$ CC candidate cluster after applying the charge-light matching. The red (green) circles represent the observed (predicted) number of photoelectrons (PEs) at each PMT, where the area of the circle is proportional to the number of PEs. The effective detector boundary as a result of the space charge effect is indicated by the red dashed lines in the corner of the TPC active volume as shown in the “Y-X view” and “X-Z view”.

Figure 5

Displays of the Wire-Cell pattern recognition results at different stages. (a) In-beam candidate neutrino cluster selected by the generic neutrino selection. The color scale represents the reconstructed charge associated with each space point, where blue and cyan are lower in charge and yellow and orange higher in charge. (b) Identified tracks and EM showers, which are displayed in blue and red, respectively. (c) Identified particles (or track segments), which are displayed in different colors. (d) Fitted $dQ/dx$ associated with each piece ($\sim 6\mathrm{mm}$) along the trajectories. The blue, cyan, green, yellow, and red colors roughly correspond to $1/3$, 1, 2, 3, and 4 times the $dQ/dx$ of a minimum ionizing particle (MIP). (e) Reconstructed particle flow starting from the primary neutrino interaction vertex, which is displayed in a rainbow-colored wheel.

Figure 6

Reconstructed neutrino energy vs true neutrino energy [(a)–(d)] and reconstructed neutrino energy resolution and its bias [(e)–(h)] for FC and PC ${\nu }_{\mu }$ CC and FC and PC ${\nu }_e$ CC candidates. The black points in the energy resolution plots represent the peak positions for each bin indicating the typical bias, and the error bars represent 68.3% quantiles from each bin’s peak position.

Figure 7

(a) ${\nu }_e$ CC selection efficiency and purity at different BDT cut values, with the finalized cut value of 7.0 indicated. (b) ${\nu }_e$ BDT score distribution for events with BDT $\mathrm{score}>0$.

Figure 8

The final ${\nu }_e$ CC selections as a function of reconstructed neutrino energy [(a) and (c)] and reconstructed shower cos $\theta$ [(b) and (d)]. (a) and (b) are categorized by event types and (c) and (d) by interaction types. The number of events correspond to the range shown in the plot. The bottom subpanels show both the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with the addition of detector systematic uncertainty. The selection efficiencies are shown as a function of (e) true neutrino energy and (f) true shower cos $\theta$ with only statistical uncertainty considered. The other dimensions are integrated in calculating these efficiencies.

Figure 9

(a) ${\nu }_{\mu }$ CC selection efficiency and purity at different BDT cut values with the finalized cut value of 0.9 indicated. (b) ${\nu }_{\mu }$ BDT score distribution for events with BDT $\mathrm{score}>0$.

Figure 10

The final ${\nu }_{\mu }$ CC selections as a function of reconstructed neutrino energy [(a) and (c)] and reconstructed muon cos $\theta$ [(b) and (d)]. (a) and (b) are categorized by event types and (c) and (d) by interaction types. The number of events correspond to the range shown in the plot. The bottom subpanels present both the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with the addition of the detector systematic uncertainty. The selection efficiencies are shown as a function of (e) true neutrino energy and (f) true muon cos $\theta$ with only statistical uncertainty considered. The other dimensions are integrated in calculating these efficiencies.

Figure 11

Distributions of the reconstructed ${\pi }^0$ invariant mass [(a) and (b)], reconstructed ${\pi }^0$ kinematic energy [(c) and (d)], and selection efficiency [(e) and (f)] as a function of reconstructed ${\pi }^0$ kinetic energy for $\mathrm{CC}{\pi }^0$ and $\mathrm{NC}{\pi }^0$. Only the statistical uncertainty was considered in the efficiency plots (e) and (f). The bottom subpanels of plots [(a)–(d)] present both the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with the addition of the detector systematic uncertainty. A consistency is observed between the data and simulation validating the energy scale reconstruction for EM showers.

Figure 12

Summary of relative uncertainties ($\frac{\text{absolute error}}{\text{prediction}}$) of all systematic uncertainty sources for the seven channels. Spikes at very low- or high-energy regions for detector systematic uncertainty are largely attributed to statistical errors stemming from the finite size of related MC samples.

Figure 13

Fraction ($\frac{{\sigma }_i^2}{{\sigma }_{\mathrm{flux}}^2}\times 100$) of uncertainties of total flux systematics for the seven channels.

Figure 14

Correlations of flux systematics for the seven channels.

Figure 15

Correlations of cross-section systematics for the seven channels.

Figure 16

Correlations of hadron-argon interaction systematics for the seven channels.

Figure 17

Correlations of total detector systematics for the seven channels. The correlations between the FC and PC ${\nu }_{\mu }\mathrm{CC}$ channels are dominated by the components of light yield and recombination.

Figure 18

Fractional contribution ($\frac{{\sigma }_i^2}{{\sigma }_{\det }^2}\times 100$) to the total detector systematic uncertainty for the seven channels.

Figure 19

Correlations of all systematic uncertainties for the seven channels.

Figure 20

Summary of fractional contributions to the overall systematic uncertainty ($\frac{{\sigma }_i^2}{{\sigma }_{\text{total}}^2}\times 100$) for the seven channels.

Figure 21

Event distributions of FC ${\nu }_{\mu }$ CC, PC ${\nu }_{\mu }$ CC, FC $\mathrm{CC}{\pi }^0$, PC $\mathrm{CC}{\pi }^0$, and $\mathrm{NC}{\pi }^0$ in (a)–(e), respectively. These five selections serve as constraints to reduce the systematic uncertainties in searching for eLEE, thus maximizing the physics sensitivity. The breakdown of each component for different final states for both signal and background events is shown in the legend. The bottom subpanels present the data-to-prediction ratios as well as the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with an addition of detector systematic uncertainty.

Figure 22

Event distributions of reconstructed muon energy (a), muon angle (b), and hadronic energy (c), for selected ${\nu }_{\mu }\mathrm{CC}$ events with fully contained (FC) events on the left half and partially contained (PC) events on the right half in each subfigure. Black points are from data measurements. Red (blue) histograms and error bands are for the MC prediction before (after) constraint. The last bin is the overflow bin for FC or PC muon energy or hadronic energy distributions.

Figure 23

Event distributions of FC ${\nu }_e$ CC [(a), (c), and (e)] and PC ${\nu }_e$ CC [(b), (d), and (f)] candidates as a function of reconstructed shower energy [(a) and (b)], shower cos $\theta$ [(c) and (d)], and hadronic energy [(e) and (f)]. No constraint is used. Data-to-prediction ratios, ${\chi }^2$, and error bands are calculated based on the ${\mathrm{eLEE}}_{x=0}$ hypothesis. The MC expectation of the ${\mathrm{eLEE}}_{x=1}$ component is added on top of the energy spectrum as represented by the dashed red curve. The breakdown of each component for different final states for both signal and background events is shown in the legend. The bottom subpanels present the data-to-prediction ratios as well as the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with an addition of detector systematic uncertainty.

Figure 24

Event distributions of (a) FC ${\nu }_e$ CC and (b) PC ${\nu }_e$ CC candidates as a function reconstructed $E_{\nu }$. No constraint is used. Data-to-prediction ratios, ${\chi }^2$, and error bands are calculated based on the ${\mathrm{eLEE}}_{x=0}$ hypothesis. The MC expectation of the ${\mathrm{eLEE}}_{x=1}$ component is added on top of the energy spectrum as represented by the dashed red curve. The breakdown of each component for different final states for both signal and background events is shown in the legend. The bottom subpanels present the data-to-prediction ratios as well as the statistical and systematic uncertainties. The pink band includes the statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with an addition of detector systematic uncertainty.

Figure 25

Event distribution of FC ${\nu }_e$ CC candidates as a function reconstructed $E_{\nu }$. Same setup as that of Fig. 24 except that the constraints from PC ${\nu }_e$ CC, FC ${\nu }_{\mu }$ CC, PC ${\nu }_{\mu }$ CC, FC $\mathrm{CC}{\pi }^0$, PC $\mathrm{CC}{\pi }^0$, and $\mathrm{NC}{\pi }^0$ are used.

Figure 26

Event distributions of FC ${\nu }_e$ CC candidates as a function of reconstructed $E_{\nu }$. (a) An ${\mathrm{eLEE}}_{x=0}$ hypothesis and (b) an ${\mathrm{eLEE}}_{x=1}$ hypothesis are assumed in MC prediction and uncertainty calculation, respectively. Black points are from data measurements. Red (blue) histograms and error bands are for MC prediction before (after) constraint from the other six channels.

Figure 27

Distributions of simple-vs-simple $\mathrm{\Delta }{\chi }_{\text{simple}}^2$ assuming (a) the ${\mathrm{eLEE}}_{x=0}$ hypothesis is true and (b) the ${\mathrm{eLEE}}_{x=1}$ hypothesis is true. Both distributions are obtained from 4 million pseudoexperiments. Data $\mathrm{\Delta }{\chi }_{\text{simple}}^2$ values are indicated by dashed vertical lines.

Figure 28

The $\mathrm{\Delta }{\chi }_{\text{nested}}^2$ with the nested likelihood ratio test. The best-fit eLEE strength is at $x_{\min }=0$. The Feldman-Cousins 68.3%, 95.5%, and 99.7% confidence intervals are displayed with the red, green, and blue bands, respectively. The 68.3% confidence interval of the eLEE strength extracted from the MiniBooNE 2020 result are also shown for comparison purpose.

Figure 29

Distribution of $\mathrm{\Delta }{\chi }_{\text{nested}}^2$ assuming ${\mathrm{eLEE}}_{x=1}$ is true. The sensitivity of rejecting the ${\mathrm{eLEE}}_{x=1}$ hypothesis is at $2.93\sigma$ following the Feldman-Cousins procedure. Using the same procedure, the data measurement rejects the ${\mathrm{eLEE}}_{x=1}$ hypothesis at the $p$ value of 0.002, which corresponds to $3.14\sigma$. The results are obtained from 10 million pseudoexperiments.

Figure 30

Event distributions of FC ${\nu }_{\mu }$ CC $0p\mathrm{X}\pi$, PC ${\nu }_{\mu }$ CC $0p\mathrm{X}\pi$, FC ${\nu }_{\mu }$ CC $\mathrm{N}p\mathrm{X}\pi$, and PC ${\nu }_{\mu }$ CC $\mathrm{N}p\mathrm{X}\pi$ samples in panels (a)–(d), respectively. The breakdown of each component for different final states for both signal and background events is shown in the legend (see definitions in Sec. 4a). The bottom subpanels present the data-prediction ratios as well as the statistical and systematic uncertainties. The pink band includes the MC statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with an addition of detector systematic uncertainty. No constraint is applied.

Figure 31

Event distributions of FC ${\nu }_e$ CC $0p\mathrm{X}\pi$, PC ${\nu }_e$ CC $0p\mathrm{X}\pi$, FC ${\nu }_e$ CC $\mathrm{N}p\mathrm{X}\pi$, and PC ${\nu }_e$ CC $\mathrm{N}p\mathrm{X}\pi$ samples in panels (a)–(d), respectively. The breakdown of each component for different final states for both signal and background events is shown in the legend (see definitions in Sec. 4a). The bottom subpanels present the data-prediction ratios as well as the statistical and systematic uncertainties. The pink band includes the MC statistical, cross-section, and flux uncertainties. The purple band corresponds to the full uncertainty with an addition of detector systematic uncertainty. No constraint is applied.

Figure 32

$\mathrm{\Delta }{\chi }_{\text{nested}}^2$ value as a function of eLEE strength. “7-ch” (black curve) corresponds to the result from the default 7-channel simultaneous fit which is the same as the one in Fig. 28. “11-ch” (green curve) corresponds to the result from a 11-channel simultaneous fit combining separated $0p\mathrm{X}\pi$ and $\mathrm{N}p\mathrm{X}\pi$ channels for ${\nu }_e$ CC and ${\nu }_{\mu }$ CC FC and PC events together with the dedicated ${\pi }^0$ channels. “$0p\mathrm{X}\pi$” (red curve) corresponds to the result from the $0p\mathrm{X}\pi$ channels for ${\nu }_e$ CC and ${\nu }_{\mu }$ CC FC and PC events as well as the dedicated ${\pi }^0$ channels. “$\mathrm{N}pX\pi$” (blue curve) corresponds to the result from the $\mathrm{N}p\mathrm{X}\pi$ channels for ${\nu }_e$ CC and ${\nu }_{\mu }$ CC FC and PC events as well as the dedicated ${\pi }^0$ channels.