Final results from the Palo Verde Neutrino Oscillation Experiment

TL;DR

This work reports the final Palo Verde reactor-neutrino oscillation results from 350 days of data, finding no evidence for $\bar{\nu}_e$ disappearance. Using two complementary analyses (reactor-power and swap), the study employs a Gd-loaded scintillator detector to detect inverse beta-decay events and places 90% CL limits on $\Delta m^2$ and $\sin^2 2\theta$ across the two-flavor parameter space, with the swap method yielding the strongest constraints. The measured rate ratio is $R_{\rm obs}/R_{\rm calc}=1.01\pm0.024\,\text{(stat)}\pm0.053\,\text{(syst)}$, indicating consistency with no oscillations and highlighting systematic uncertainties as the dominant error. When combined with results from Chooz and Super–Kamiokande, the findings disfavor $\nu_\mu-\nu_e$ mixing as the source of the atmospheric neutrino anomaly, supporting $\nu_\mu-\nu_\tau$-driven oscillations instead. The analysis showcases rigorous MC validation, detailed calibration, and robust background handling, contributing important constraints to the neutrino-oscillation landscape at small $Δm^2$.

Abstract

The analysis and results are presented from the complete data set recorded at Palo Verde between September 1998 and July 2000. In the experiment, the $\nuebar$ interaction rate has been measured at a distance of 750 and 890 m from the reactors of the Palo Verde Nuclear Generating Station for a total of 350 days, including 108 days with one of the three reactors off for refueling. Backgrounds were determined by (a) the $swap$ technique based on the difference between signal and background under reversal of the positron and neutron parts of the correlated event and (b) making use of the conventional reactor-on and reactor-off cycles. There is no evidence for neutrino oscillation and the mode $\nuebar\to\barν_x$ was excluded at 90% CL for $\dm>1.1\times10^{-3}$ eV$^2$ at full mixing, and $\sinq>0.17$ at large $\dm$.

Figures (8)

• Figure 1: The calculated $\bar{\nu}_{\rm e}$ interaction rate in the detector target for the case of no oscillations. The four long periods of reduced flux from reactor refuelings were used for background subtraction. The decreasing rate during the full power operation is a result of the changing core composition as the reactor fuel is burned.
• Figure 2: The upper plots show the simulated and measured trigger efficiency for low and high thresholds as a function of energy deposited in the center of one cell. Dots represent data while the solid line shows the simulated efficiency. The lower plots show the energy corresponding to a trigger efficiency of 50% for each cell. The spread between data and Monte Carlo has been improved by a factor of about two compared to Boehm:1999gl.
• Figure 3: Comparison of data (points) and Monte Carlo (histograms) for detection efficiency for $^{22}$Na and Am-Be source runs at various locations. For positions of the radioactive source near the border of the central detector we measure lower efficiencies in good agreement with the simulation (see locations 3, 5, 22, ... for $^{22}$Na or locations 11, 12, 18 ... for Am-Be).
• Figure 4: Comparison of data (points) and Monte Carlo simulation (histograms) for the spectra of total energy and first, second, and third most energetic hit ($E_{\rm total}$, $E_1$, $E_2$, and $E_3$) for capture cosmic muon induced neutrons. The sharp feature at 3.5 MeV total energy is related to the requirement that at least one sub-event has total energy above this value (see text).
• Figure 5: The event rates $R_{{\rm exp}}$ for different data taking periods, corrected for deadtime and neutrino detection efficiency, plotted versus the expected neutrino interaction rate $R_{{\rm calc}}$ for no oscillations. Errors are statistical only. Points corresponding to data taking periods with same reactor power conditions should lie on top of each other. Also shown is the result, discussed in the text, of a linear fit to the data.