Tests of electron flavor conservation with the Sudbury Neutrino Observatory
Extracted source text ยท arXiv hep-ph/9607433
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% %%
%% Tests of Electron Flavor Conservation %%
%% with the Sudbury Neutrino Observatory %%
%% %%
%% by J.N. Bahcall and E. Lisi %%
%% %%
%% Institute for Advanced Study Report No. 96/33 %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
\documentstyle[prd,aps,preprint,epsfig,tighten]{revtex}
%\documentstyle[prd,preprint,aps]{revtex}
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% F R O N T P A G E
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{document}
\preprint{IASSNS-AST 96/33}
\draft
%\date{\today}
\title{ Tests of Electron Flavor Conservation \\
with the Sudbury Neutrino Observatory }
%
\author{ John N.~Bahcall}
%
\address{ Institute for Advanced Study, Princeton, New Jersey 08540}
%
\author{ Eligio Lisi}
%
\address{ Institute for Advanced Study, Princeton, New Jersey 08540\\
and Istituto Nazionale di Fisica Nucleare, Sezione di Bari, Italy}
%
\maketitle
\begin{abstract}
%...........................................................................
We analyze tests of electron flavor conservation that can be performed at
the Sudbury Neutrino Observatory (SNO). These tests, which utilize $^8$B
solar neutrinos interacting with deuterium, measure: 1) the shape of the
recoil electron spectrum in charged-current (CC) interactions (the CC
spectrum shape); and 2) the ratio of the number of charged current to neutral
current (NC) events (the CC/NC ratio). We determine standard model predictions
for the CC spectral shape and for the CC/NC ratio, together with realistic
estimates of their errors and the correlations between errors. We consider
systematic uncertainties in the standard neutrino spectrum and in the
charged-current and neutral current cross-sections, the SNO energy
resolution and absolute energy scale, and the SNO detection efficiencies.
Assuming that either matter-enhanced or vacuum neutrino oscillations solve
the solar neutrino problems, we calculate the confidence levels with which
electron flavor non-conservation can be detected using either the CC spectrum
shape or the CC/NC ratio, or both. If the SNO detector works as expected,
the neutrino oscillation solutions that best-fit the results of the four
operating solar neutrino experiments can be distinguished
unambiguously from the standard predictions of electron flavor conservation.
%............................................................................
\end{abstract}
\pacs{PACS number(s): 11.30.Fs, 26.65.+t, 13.15.+g, 12.60.$-$i}
%
%.............................................................................
% PACS legenda:
% 11.30.Fs global symmetries, including lepton number
% 26.65.+t solar neutrinos
% 13.15.+g neutrino interactions
% 12.60.-i models beyond the standard model
% 96.60.Jw solar interior
%.............................................................................
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% S E C T I O N I
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Introduction}
\label{sec:intro}
We assess quantitatively the possibility of detecting electron flavor
non-conservation using $^8$B solar neutrino interactions in deuterium at
the Sudbury Neutrino Observatory (SNO) \cite{Sudb}. The separate conservation
of the lepton (electron, muon, tau) flavors is a well known ingredient of
the standard electroweak model \cite{SEW} and of some of its extensions.
Figures~\ref{Summaryfig} and~\ref{Distancefig} (and Table~\ref{Errortab}
and Table~\ref{Deviationtab}) summarize the power of the SNO experiment
to find new physics. We urge the reader to look at these two summary figures
(and tables) before descending into the necessary details, which are analyzed
in this paper.
Solar neutrinos offer a unique possibility to detect electron-flavor
non-conserving processes. In solar neutrino experiments, a pure beam of
electron neutrinos is created in the interior of the sun, passing through
$10^{11}~{\rm gm~cm}^{-2}$ of matter and eventually reaching a terrestrial
detector located at a distance of $10^8$ km from the sun. The tests discussed
in this paper are independent of solar models and are made possible by
the fact that low energy (MeV) nuclear fusion reactions produce only electron
type neutrinos. For neutrino squared mass
differences less than $10^{-4}~{\rm eV^2}$, the solar
neutrino tests are more sensitive than laboratory tests
\cite{PDG} of lepton flavor conservation.
We consider measurements of:
1)\ the energy spectrum of recoil electrons in charged current
(absorption) reactions \cite{Ba91};
2)\ the ratio of the number of charged-to-neutral current events
\cite{Ch85}; and
3)\ the combined measurement of the charged current energy spectrum and
the charged to neutral current ratio.
The shape measurement is sensitive to an energy-dependent depletion of the
created flux of electron flavor neutrinos, and the neutral current to charged
current comparison is sensitive to a non-zero conversion probability to a
different (active) neutrino.
How can we test lepton flavor conservation with solar neutrinos? The energy
spectrum of $^8$B neutrinos is the same in the laboratory and in the sun,
modulo negligible (gravitational redshift) corrections of ${\cal O}(10^{-5})$
\cite{Ba91}. Fortunately, the spectrum, $\lambda(E_\nu)$, can be determined
with relatively small uncertainties from laboratory data on the
$^8$B$(\beta^+){}^8$Be$(2\alpha)$ decay chain \cite{BaLi}. The measurement
of the electron spectrum produced by neutrino absorption
is therefore a test for new physics independent of complications
related to solar physics. The ratio of neutral current to charged current
events is also independent of uncertainties that affect the calculation of
the total flux.%
%---------------------------------
\footnote{The calculated total flux of $^8$B neutrinos at earth
(all flavors) depends on solar physics and on the
extrapolated low-energy cross section for the reaction
${\rm ^7Be} (p,\gamma){\rm ^8B}$.}
%---------------------------------
All of the observed solar neutrinos must be of the electron type unless the
separate conservation of electron flavor is violated.
Neutrino oscillations are used in this paper to illustrate the potential
effects of flavor transitions, but the considerations described here can
be applied to other proposed electron-flavor non-conserving mechanisms,
such as neutrino decay \cite{bah72,ber87}, non-standard electromagnetic
properties \cite{vol86,lim88,akh88}, neutrino violation of the equivalence
principle \cite{gasp88}, and supersymmetric flavor-changing neutral currents
\cite{rou91,guz91}. Many of the key papers and other relevant references
are reprinted in \cite{30yr}.
Will the measurements with SNO be sensitive enough to prove---if it is
present---that new neutrino physics is occurring? Will the uncertainties
(systematic and statistical) be sufficiently small to identify electron
flavor non-conservation if it occurs as previously suggested? The answer
is: ``Yes, if SNO performs as expected'' \cite{Sudb}.
The SNO collaboration is completing the construction of a 1000 ton deuterium
detector in the Creighton Mine (Walden, Canada) \cite{SNOw}. The detector
will measure the rates of the charged (CC) and neutral (NC) current reactions
induced by solar neutrinos in deuterium:
%............................................................................
\begin{equation}
\nu_e + d \rightarrow p+p+e^-\quad({\rm CC\ absorption})\quad,
\label{reactionCC}
\end{equation}
%............................................................................
\begin{equation}
\nu_x + d \rightarrow p+n+\nu_x\quad({\rm NC\ dissociation})\quad,
\label{reactionNC}
\end{equation}
%............................................................................
including the determination of the electron recoil energy in
Eq.~(\ref{reactionCC}). Only the more energetic $^8$B solar neutrinos
will be detected%
%---------------------------
\footnote{ The contribution of {\em hep\/} neutrinos
\cite{Ba89} is negligible
and will be discussed in Sec.~\ref{sec:8bspectrum}.}
%---------------------------
since the expected
SNO threshold for CC events is an electron kinetic energy
of about 5 MeV and the physical threshold for NC dissociation is the
binding energy of the deuteron, $E_b= 2.225$ MeV.
Neglecting all systematic uncertainties, some previous authors
\cite{Sudb,Othe,Fi94,Kw95} have considered how well the tests of flavor
conservation by SNO can discriminate
between new neutrino physics scenarios and
standard model expectations.
The most explicit discussions are given in \cite{Fi94} and \cite{Kw95},
which constitute especially good introductions to the subject.
We evaluate the effects of systematic uncertainties, theoretical and
experimental, on the discriminatory power of the SNO
tests for new physics.
We consider uncertainties related
to the laboratory shape of the neutrino energy spectrum, the calculated
cross sections for charged-current and neutral-current reactions with
deuterium, the energy calibration and resolution, detection efficiencies,
and the CC detection threshold of the SNO detector.
The primary goal of this paper is to
refine the best-estimates and uncertainties for the
theoretical ingredients
that will determine how powerfully SNO will test electron flavor
conservation. In addition,
we carry out a preliminary overall estimate of the
sensitivity of the detector to different types of oscillation phenomena,
including realistic estimates of the experimental characteristics
and their uncertainties.
The various backgrounds \cite{Sudb} will not be known
until after the detector is operating and are not included here.
Our analysis is not a substitute for the detailed Monte
Carlo simulations of the operating detector
which will be performed by the SNO team.
The discriminatory power of the
detector will be established definitively by the
simulations to be performed by the SNO collaboration, which will
include all the theoretical ingredients
discussed here, the detector elements that we high-light, and other
aspects of the detector (such as the backgrounds) that will be
determined during the operation of the experiment. Our calculations can,
however, be a useful guide as to what is likely to be
possible and what uncertainties are most important to try to reduce.
We note that the SNO collaboration has been working for a number of
years to develop the detector and calibration techniques in ways that
will minimize the experimental uncertainties.
We shall show that the recognized systematic uncertainties permit the
observation of new physics at SNO, but the systematic errors
may well dominate the total uncertainties after a relatively short exposure
($\sim1$~year) to solar neutrinos. Our analysis can be
extended easily to include additional sources of uncertainties.
We concentrate here on the most direct tests for new physics, which
involve the
shape of the neutrino spectrum and the charged-current to
neutral-current ratio.
If SNO does find evidence for new physics, the next step will be
to discriminate among competing models of new physics.
Important information will be provided by the
time-dependence of the observed solar neutrino signal (day-night and
seasonal variations) \cite{Sudb}. We do not address questions related
to the time-dependence in this paper.
The SNO collaboration plans an overall test of the detector by measuring
the energy
spectrum of an intense $^8$Li beta-decay source to be placed in the SNO
detector \cite{Cali}. We include in Appendix~A our determination, using
the best-available data, of the $^8$Li$(\beta)$ spectrum,
together with its
estimated uncertainties. We also discuss some possible strategies for the
${\rm ^8Li}$-test.
This paper has the following structure. In Sec.~\ref{sec:nuingredients},
we describe the neutrino-related ingredients of our calculation:
the ${\rm ^8B}$ neutrino energy spectrum, and the charged and neutral
current neutrino cross sections for deuterium.
In Sec.~\ref{sec:detingredients}, we discuss the detector-related
ingredients: the energy resolution, the absolute energy scale, the
detection efficiencies, and the CC energy threshold. We use in
Sec.~\ref{sec:physSNO} the neutrino-related and the detector-related
ingredients to calculate the flavor-conserving expectations for the shape
of the CC electron recoil energy spectrum
and the CC/NC event ratio; we include realistic
estimates of the likely uncertainties and the correlations among the
uncertainties. In Sec.~\ref{sec:newphys}, we calculate the effects upon
measurable quantities of representative neutrino oscillation scenarios,
and assess quantitatively the statistical significance with which new
physics might be observed. We summarize our work in Sec.~\ref{sec:summary}.
Appendix~A presents a calculation and discussion of the
$^8$Li$(\beta)$ spectrum and its use as a test of the overall
performance of the detector. Appendix~B discusses the extent to which
the average value of the electron recoil energy is a good estimator
of possible deviations from the CC shape expected in the absence of flavor
violations.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Neutrino-related Ingredients}
\label{sec:nuingredients}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this Section, we discuss the neutrino-related ingredients of the
analysis that have appreciable, recognized uncertainties.
These ingredients are: the $^8$B neutrino spectrum
(Sec.~\ref{sec:8bspectrum}), the charged-current absorption cross section
(Sec.~\ref{sec:CCcross}), and the neutral-current dissociation cross
section (Sec.~\ref{sec:NCcross}). We discuss the detector-related
ingredients in the following section.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{$^8$B neutrino spectrum}
\label{sec:8bspectrum}
The only component of the solar neutrino spectrum \cite{Ba89} that is
important for the SNO experiment is the $^8$B spectrum, $\lambda(E_\nu)$.
A derivation of the best-estimate ${\rm ^8B}$ spectrum, $\lambda(E_\nu)$,
along with the maximum allowed deviations, $\lambda^\pm (E_\nu)$
($\pm 3$ effective standard deviations away from the best-estimated
spectrum) is presented in reference \cite{BaLi} .
The $^8$B neutrinos are produced in the decay $^8$B$(\beta){}^8$Be followed
by $^8$Be$(2\alpha)$ disintegration. The broad $^8$Be intermediate state
is responsible for important deviations of the neutrino spectrum $\lambda$
from the usual allowed shape. The population of the $^8$Be state is
determined experimentally by measuring the delayed $\alpha$-decay spectrum.
The absolute energy calibration of the measured alpha spectrum is the main
systematic error.
For the calculation of the ${\rm ^8B}$ neutrino spectrum, the experimental and
theoretical uncertainties can be included \cite{BaLi} in a single
effective alpha-energy offset, $b$: $E_\alpha\rightarrow E_\alpha+b$.
The independently-measured positron spectrum in $^8$B$(\beta)$ decay
\cite{Napo} provides a fundamental additional constraint that was used
in \cite{BaLi} to choose a ``best'' reference $\alpha$-spectrum
\cite{Wilk} and to bound its offset: $b=0.025\pm0.104$ MeV
($\pm3\sigma$ uncertainties). An ``infinitely-precise'' measurement of
the $^8$B$(\beta)$ positron spectrum could reduce the effective
$\pm3\sigma$ range of $b$ to $\pm 0.075$ MeV, where 0.075~MeV is the
residual theoretical uncertainty. The uncertainties of the $^8$B neutrino
spectrum would be reduced in the same ratio. Since the uncertainties in
the neutrino spectrum are a significant source of errors for the CC-shape
test with SNO, a reduction of the allowed range of $b$ through more precise
measurements of the $^8$B positron spectrum would be useful.
The {\em hep\/} neutrinos \cite{Ba89} have a maximum energy of 18.8~MeV
and could also in principle contribute to the neutrino spectrum observed
by SNO. The calculated total flux of {\em hep\/} neutrinos is uncertain
by a factor of about six \cite{BP92} because of theoretical difficulties in
estimating the low-energy production cross section. Using the nominal
value $\phi_{hep}= 1 \times 10^{+3}~{\rm cm^{-2}s^{-1}}$ given in
\cite{Ba95}, we estimate that {\em hep\/} neutrinos contribute less than
0.07\% of either the total NC or CC event rates. Therefore, hep neutrinos
can be neglected in calculating the CC/NC ratio. Moreover, we have
verified that the {\em hep\/} contribution to the high-energy tail of
the spectrum is much smaller than the shape uncertainties estimated below
in Sec.~\ref{sec:smpredictsCC}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Charged current $\nu d$ cross section}
\label{sec:CCcross}
The cross section for the charged-current reaction (1) has been calculated
a number of times in the last 30 years, since the original proposal by
Jenkins \cite{Jenk} to use charged-current capture on deuterium to measure
the ${\rm ^8B}$ solar neutrino flux. Kubodera and Nozawa \cite{Ku94} have
recently presented an insightful and thorough summary of the calculations
of both the charged current and the neutral current cross sections. In this
subsection, we assess the reliability of the theoretical calculations of
the total and the differential CC cross section. We establish the
robustness of the calculated cross sections, which is
exemplified by the excellent
agreement between the simple effective range calculations and the more
sophisticated treatments. In addition, we stress the importance for
determining the shape of the electron spectrum of including the final
state interactions among the protons . We discuss the neutral current
cross section in the following subsection.
The kinematics of reaction (1) leads to the following expression \cite{Ko66}
for the neutrino energy, $E_\nu$:
%............................................................................
\begin{equation}
E_\nu = Q + T_e + \frac{P^2}{m_p} + \frac{(\bbox{p}_e-\bbox{p}_\nu)^2}{4m_p}
\quad,
\label{eqkinematic}
\end{equation}
%.............................................................................
where $\bbox{p_\nu}$ is the neutrino 3-momentum, $T_e$ and $\bbox{p_e}$ are
the electron kinetic energy and 3-momentum, $P$ is the relative momentum of
the protons in the proton center-of-mass (c.m.s.) system, and the threshold
$Q=1.442$ MeV. The third term describes the kinetic energies of the two
protons in the proton c.m.s. and is an important contribution to the energy
balance. The fourth (last) term in Eq.~(\ref{eqkinematic}) describes the
small recoil energy of the two-proton center-of-mass system.
The charged-current absorption, reaction (1), is described by the well-known
electroweak Hamiltonian and by less well-known nuclear physics effects that
can be treated at various levels of approximation. A neutrino energy of
$10$~MeV corresponds in natural units to $(20 {\rm\ fm})^{-1}$. Therefore,
we expect that the details of deuteron nuclear physics will play only a
minor role. This expectation is confirmed by comparing the $s$-wave
calculations performed in the 1960's by Kelly and {\"U}berall \cite{Ke66}
and by Ellis and Bahcall \cite{El68} using Bethe's effective range
approximation \cite{Be49} with the recent sophisticated calculations by
Ying, Haxton, and Henley \cite{Yi92} and by Kubodera and collaborators
\cite{Do92,Ku94}. The recent calculations include higher partial waves,
relativistic effects, forbidden matrix elements, and exchange-currents.
Figure~\ref{Totalcross}a shows the excellent agreement between the
effective-range calculations and the more sophisticated treatments. In
the figure, three independent estimates of the total CC cross section are
compared: Ellis-Bahcall (EB) \cite{El68}, Kubodera-Nozawa (KN) \cite{Ku94},
and Ying-Haxton-Henley (YHH) \cite{Yi92}. The main difference between the
calculated cross sections of KN and YHH is an energy-independent
normalization factor (about $\sim 6\%$ uncertainty, as also estimated in
\cite{Ku94}). The EB normalization shows a slight energy dependence,
that amounts to a $\sim4\%$ additional variation over the important
energy range of 5~MeV to 10~MeV. Figure~\ref{Totalcross}b (referring to
the neutral current cross-section) will be discussed in
Sec.~\ref{sec:NCcross}.
When quoting the Ellis-Bahcall cross-section \cite{El68}, we use a
slightly-improved calculation of the differential CC cross section
in which we include the previously neglected $p$-$p$ c.m.s.\
recoil term [the fourth term in Eq.~(\ref{eqkinematic})]. We have also
used the more recent choices of parameters of the effective range
approximation given in reference \cite{Pr75}, in order to obtain
our best estimate of the differential cross section
%......
${d\sigma_{\rm CC}(E_\nu,\,T_e,\,\cos\theta)}/{dT_e\,d\cos\theta}$
%......
for any given value of $E_\nu$, $T_e$ and of the electron scattering
angle $\theta$.
Figure~\ref{Comparecross} shows the results of the improved Ellis-Bahcall
calculation of the normalized differential cross section
$
\sigma^{-1}_{\rm CC}\,d\sigma_{\rm CC}/{dT_e\,d\cos\theta}
$
(dotted line) as a function of the dimensionless variable $T_e/(E_\nu-Q)$,
for representative values of $E_\nu$ and $\theta$.
The Kubodera-Nozawa results appear as a solid line in the same figure.%
%----------------------------
\footnote{ The extensive numerical
tables of the differential charged current cross
section calculated by Kubodera and Nozawa in \cite{Ku94}
are not published.
We thank the SNO collaboration for
providing us with a computer-readable
copy of these tables.}
%----------------------------
The close agreement between the EB and KN normalized recoil electron spectra
is striking. The angular dependence calculated using the effective range
and the Hamiltonian approximations is essentially identical. On the basis
of Fig.~\ref{Comparecross}, we conclude that the uncertainties associated
with the $(T_e,\,\theta)$-shape of the normalized differential
cross-section for the reaction (1) are much smaller than other recognized
uncertainties. In practice, we parametrize the reference cross sections
EB, KN, and YHH in the form:
%.............................................................................
\begin{equation}
\left[
\frac%
{d\sigma_{\rm CC}(E_\nu,\,T_e,\,\cos\theta)}%
{dT_e\,d\cos\theta}
\right]_{\rm X}
=
\sigma_{\rm CC}^{\rm X}(E_\nu)
\left[\frac{1}{\sigma_{\rm CC}(E_\nu)}
\frac{d\sigma_{\rm CC}(E_\nu,\,T_e,\,\cos\theta)}{dT_e\,d\cos\theta}
\right]_{\rm EB}\ ,
\label{eqcross}
\end{equation}
%...........................................................................
for X~=~KN, YHH. The differences between the EB, KN and YHH cross sections
are embedded in a multiplicative factor that depends exclusively on $E_\nu$
and not on the angular distribution.
Contrary to what is sometimes assumed in the literature (see, e.g.,
\cite{Fi94,Bilenky93}),
the electron spectra in Fig.~\ref{Comparecross}, although
peaked, cannot be approximated well by delta functions in the
electron energy. In other words, there is not a one-to-one relation
between the incoming neutrino's energy and the energy of the electron
that is produced. The
final state in the charged current reaction cannot be approximated
by a pure two-body state. Even as early as the seminal Kelly-{\"U}berall
calculation \cite{Ke66}, it was noted that the attractive $^1S$ $p$-$p$
interaction is not sufficient to bind the protons as an effective single
particle, because of the presence of the repulsive Coulomb force.
The two-body approximation would be equivalent to omitting the third and
fourth terms in Eq.~(\ref{eqkinematic}). Removing only the fourth
(the smallest) recoil term in Eq.~(\ref{eqkinematic})
would cause the dotted cross sections in Fig.~\ref{Comparecross}
to be systematically peaked at slightly higher electron energies
(about $+2\%$ at $E_\nu=12$ MeV). The excellent agreement
between our improved Ellis-Bahcall calculation and the
Kubodera-Nozawa differential cross sections would be spoiled by omitting
even this smallest term.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Neutral current $\nu d$ cross section}
\label{sec:NCcross}
We use the recent calculations of the neutral current cross
sections (averaged over final states), $\sigma_{\rm NC}(E_\nu)$,
by Kubodera and Nozawa (KN) \cite{Ku94}, and by Ying, Haxton and Henley
(YHH) \cite{Yi92}. Only the total rate for reaction~(\ref{reactionNC}),
not the differential production rate as a function of energy, will be
measured by SNO. However, the energy dependence of the cross section for
the neutral current reaction is relevant for the SNO experiment, since the
calculated differential cross section will be used in the SNO Monte
Carlo simulations to model the production, and subsequent detection,
of the neutrons produced by the neutral current reaction.
Figure~\ref{Totalcross}b
compares the KN and YHH neutral current cross section as a function
of neutrino energy. The difference is $\sim 6\%$, about the same
magnitude and
in the same direction as for the charged current cross section. There
is a small residual energy-dependence below $E_\nu\simeq 6$ MeV. The
difference shown in Figure~\ref{Totalcross}b between the
theoretical calculations is consistent with the
theoretical error of $\pm
10$\% that was estimated by Bahcall, Kubodera, and Nozawa\cite{bkn}
from various contributions (the
impulse approximation, the nucleon-nucleon potential, meson exchange
currents, and higher partial waves).
The best estimate (and $1\sigma$ uncertainty) for the
neutral current cross section averaged over the $^8$B neutrino
spectrum is
%......................................................................
\begin{equation}
\langle \sigma(^8{\rm B}) \rangle = 0.478 (1 \pm 0.06)
\times10^{-42}~{\rm cm^{2}}.
\label{eqnnccrosssection}
\end{equation}
%......................................................................
We have chosen in Eq.~(\ref{eqnnccrosssection})
the $1\sigma$ error of 6\% to reflect the difference
between the KN and YHH calculations.
The standard solar model\cite{Ba95} prediction for the neutral
current event rate due to $^8$B neutrinos is
%......................................................................
\begin{equation}
\left<\phi\sigma\right> = 3.2_{-0.5}^{+0.6} ~{\rm SNU},
\label{ncbprate}
\end{equation}
%.......................................................................
where the quoted $1\sigma$ error in Eq.~(\ref{ncbprate}) combines
quadratically the uncertainties in the solar model prediction,
the $^8$B neutrino
spectrum, and the neutral current cross section.
The uncertainties in the solar model calculation dominate the error
estimate. The event rate SNU is defined\cite{bahcall69} as $10^{-36}$
interactions per target atom (deuterium atom) per second.
In the calculation of the CC/NC rate,
the (already-small) theoretical cross section
errors largely cancel and therefore do not affect significantly the ratio.
As a default choice, we use the KN neutral current cross section
in our calculations. The YHH cross section is used for comparison and
to evaluate the theoretical uncertainties.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Detector-related ingredients}
\label{sec:detingredients}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this section, we discuss the detector-related ingredients of the
analysis: the energy resolution, the absolute energy scale,
the detector efficiencies, and the energy threshold for detecting CC
events. Accurate determinations of all of these experimental
quantities, and their associated uncertainties, are significant for
the success of the SNO experiment.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Energy Resolution}
\label{sec:energyres}
The measured electron kinetic energy, $T_e$, determined by SNO with the
Cherenkov technique, will be distributed around the {\em true}
energy $T^\prime_e$ with a width established by the photon statistics.
The resolution function $R(T_e^\prime,\,T_e)$
is expected to be well approximated by a normalized Gaussian,
%..........................................................................
\begin{equation}
R(T_e^\prime,\,T_e) = \frac{1}{\sigma(T_e^\prime)\sqrt{2\pi}}
\exp{\left[-\frac{(T_e^\prime-T_e)^2}{2\sigma(T_e^\prime)^2}\right]}
\label{eqgaussian}
\end{equation}
%..........................................................................
with an energy-dependent one-sigma width $\sigma(T^\prime_e)$ given by
\cite{Ba89,Hirata91}
%..........................................................................
\begin{equation}
\sigma(T^\prime_e) =
\sigma_{10}\sqrt{\frac{T^\prime_e}{10 {\rm\ MeV}}}\ ,
\label{eqresolution}
\end{equation}
%..........................................................................
where $\sigma_{10}$ is the resolution width at $T^\prime_e=10$ MeV.
A plausible estimate of the parameter $\sigma_{10}$ is 1.1 MeV
($\sigma_{10} = 1.8$ MeV for Kamiokande, see \cite{Hirata91}), and
$\sigma_{10}$ itself may be uncertain by $10\%$ \cite{Beier96}.
We will use in what follows $\sigma_{10}=1.1 \pm 0.11$ MeV
($1\sigma$ errors) as an illustrative estimate.
In Figure~\ref{normfig}, solid line, we anticipate the results of our
best-estimate of the standard shape of the electron energy spectrum
(see Sec.~\ref{sec:physSNO}). The dotted line in Fig.~\ref{normfig}
represents the same spectrum without the inclusion of the energy
resolution. The area under the curve is normalized to unity in both cases.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{The Absolute Energy Scale}
\label{sec:abenergy}
What is the absolute accuracy of the SNO energy scale? How precisely will
the average SNO energy measurement correspond to the true
electron energy?
The energy resolution function of the previous section describes how the
measured kinetic energy, $T_e$, is distributed around the true energy,
$T'_e$, assuming that the centroid of the distribution, $T_{e, \rm ave}$,
coincides with $ T'_e$.
The calibration of the energy scale will be performed with a series of
$\gamma$-ray sources, the most important of which are monoenergetic.
The primary energy calibration source will be the $6.130$ MeV
$\gamma$-ray from the decay of the first $3^-$ excited
state in $^{16}$O.
We define the systematic error in the
absolute energy calibration, $\delta$, by the relation
%...........................................................................
\begin{equation}
\delta\equiv T_{e, \rm ave} - T^\prime_e\ .
\label{eqabsolute}
\end{equation}
%...........................................................................
A reasonable $1\sigma$ estimate \cite{Beier96} of $\delta$ is:
%...........................................................................
\begin{equation}
\delta=\pm100{\rm \ keV}\,\left(\frac{T_e^\prime}{\rm 10\ MeV}\right)^\alpha
\quad,\quad 0\leq\alpha\leq1\quad,
\label{eqdeltaalpha}
\end{equation}
%...........................................................................
which corresponds to a $\pm 1\%$ error at 10 MeV. For comparison, the
Kamiokande collaboration achieved \cite{Hirata91} a $\pm 3\%$ energy scale
error.
The case $\alpha=0$ ($\alpha=1$) would correspond to an energy-independent
scale shift (scale factor). The intermediate case $\alpha=\frac{1}{2}$
would apply to a scale uncertainty dominated by the width
of calibration lines (error $\propto \sqrt{T_e^\prime}$).
In general, the phenomenological parameter $\alpha$ will depend both on the
physical sources of the scale uncertainties and on the calibration technique.
It may even remain an unknown parameter after calibration. However, as we
will see, the energy-scale induced uncertainties of the SNO observables
depend only mildly on $\alpha$. The worst case appears to be $\alpha=0$
(i.e., a uniform energy bias), which we adopt for a conservative estimate
of the errors.
In practice, we introduce the energy scale shift $\delta$ by modifying the
energy resolution function [Eq.~(\ref{eqgaussian})] with the replacement:
%...........................................................................
\begin{equation}
R(T_e^\prime,\,T_e) \longrightarrow R(T_e^\prime+\delta,\,T_e)
\label{eqTTprime}
\end{equation}
%...........................................................................
and $\delta$ given by Eq.~(\ref{eqdeltaalpha}). The reader can verify
that the transformation given in Eq.~(\ref{eqTTprime}) is an
appropriate representation of the energy-calibration uncertainty by
writing the total rate for the process under consideration as a
triple integral over the neutrino energy $(E_\nu)$, the true electron
recoil energy $(T^\prime_e)$, and the measured electron recoil energy
($T_e$, between a specified minimum and maximum value).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Detection efficiencies}
\label{sec:detection}
The detection efficiencies, $\epsilon_{\rm CC}$ and $\epsilon_{\rm NC}$,
for detecting charged and neutral current events, will be measured with
calibration experiments at SNO along with their uncertainties,
$\sigma(\epsilon_{\rm CC})$ and $\sigma(\epsilon_{\rm NC})$. The calibration
experiments will also measure possible variations of the CC efficiency
with the (true) electron recoil energy,
$\epsilon_{\rm CC}=\epsilon_{\rm CC}(T_e^\prime)$, which can in principle
affect the CC-shape measurements. If the experiments work as expected
\cite{Rob96}, then $\sigma(\epsilon_{\rm CC}) < \sigma(\epsilon_{\rm NC})$ and
%...........................................................................
\begin{equation}
\sigma(\epsilon_{\rm NC}) \simeq 2\%\;\;(1\sigma)\quad.
\label{eqNCeff}
\end{equation}
%............................................................................
In Sec.~\ref{sec:physSNO}, we shall include the efficiencies
$\epsilon_{\rm CC}(T_e^\prime)$ and $\epsilon_{\rm NC}$ in the general
expressions for the predicted quantities. As default values, we will assume
that $\epsilon_{\rm CC}(T_e)$ is constant and equal to 1, and that
$\epsilon_{\rm NC} = 0.50 \pm 0.01$ ($1\sigma$ error). It will be shown in
Sec.~\ref{sec:errors} that plausible energy variations of $\epsilon_{\rm CC}$
induce deviations in the CC spectrum that are much smaller than
other sources of error.
In any event, after $\epsilon_{\rm CC}(T_e^\prime)$ and $\epsilon_{\rm NC}$
are measured in the SNO detector, their effects on the predictions, and
their uncertainties, can be included easily using the formalism given here.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Threshold Energy}
\label{sec:efficiency}
The measured kinetic energy threshold, $T_{\rm min}$,
for counting charged current events is expected to be fixed around 5~MeV
\cite{Sudb}. Below $\sim5$~MeV, the signal-to-background ratio is expected
to decrease very rapidly. In principle, one would like to have the
threshold as low as possible in order to increase the number of events
that are detected for a given exposure and in order to observe more of
the curvature of the spectrum at
lower energies (cf.~Fig.~\ref{normfig}).
The actual background level in the operating SNO detector will determine
how low the energy threshold may be set. In Sec.~\ref{sec:dependences},
we will show that the discriminatory power of the experiment is not very
sensitive to the actual threshold level as long as the threshold is in
the vicinity ($\pm 1$ MeV) of the nominal value, $T_{\rm min}=5$~MeV.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Standard neutrino physics at SNO}
\label{sec:physSNO}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this section, we calculate the standard predictions, and their associated
uncertainties, for the shape of the CC electron recoil energy spectrum
and the ratio of total number of CC to NC events. We assume standard neutrino
properties (lepton flavor conservation and zero neutrino masses) and
the ingredients discussed in the previous sections
(Secs.~\ref{sec:nuingredients} and~\ref{sec:detingredients}).
We adopt the Kubodera-Nozawa (KN) neutrino interaction cross sections
as ``standard,'' since they are the most recent and complete for both the
charged and the neutral current reactions. In particular, we evaluate the
standard CC differential cross-section as indicated by Eq.~(\ref{eqcross})
with X~=~KN, i.e., with absolute normalizations given by the
Kubodera-Nozawa \cite{Ku94} calculations and relative differential
cross sections given by the Ellis-Bahcall \cite{El68} calculation.
The shape of the recoil electron energy spectrum is given by the
normalized distribution of charged current events $(N_{\rm CC})$
as a function of the measured kinetic energy, $T_e$:
%.............................................................................
\begin{equation}
\frac{1}{N_{\rm CC}}\frac{dN_{\rm CC}}{dT_e}=
\frac
{\displaystyle \int dE_\nu\lambda(E_\nu)\int dT'_e
\frac{d\sigma_{\rm CC}}{dT'_e} R(T'_e,\,T_e) \epsilon_{\rm CC}(T^\prime_e)}
{\displaystyle \int_{T_{\rm min}} dT_e
\int dE_\nu\lambda(E_\nu)\int dT'_e
\frac{d\sigma_{\rm CC}}{dT'_e} R(T'_e,\,T_e)\epsilon_{\rm CC}(T_e^\prime)
}\quad,
\label{eqCCshape}
\end{equation}
%.............................................................................
where $T_{\rm min}$ is threshold for the measured electron kinetic
energy and $\epsilon_{\rm CC}(T_e^\prime)$ is the efficiency for detecting
an electron of true energy $T_e^\prime$. The energy resolution function,
$ R(T'_e,\,T_e)$, is given by Eq.~(\ref{eqgaussian}), with an allowance
for an energy scale shift [Eqs.~(\ref{eqdeltaalpha}, \ref{eqTTprime})].
The CC differential cross section, $\frac{d\sigma_{\rm CC}}{dT'_e}$,
is implicitly integrated over the entire solid angle since events will be
detected for all electron recoil angles.
The ratio of charged to neutral current events may be written:
%...........................................................................
\begin{equation}
\frac{N_{\rm CC}}{N_{\rm NC}}=
\frac{
\displaystyle \int_{T_{\rm min}} dT_e
\int dE_\nu\lambda(E_\nu)\int dT'_e
\frac{d\sigma_{\rm CC}}{dT'_e} R(T'_e,\,T_e)\epsilon_{\rm CC}(T_e^\prime)
}
{\displaystyle\epsilon_{\rm NC}\int dE_\nu \lambda(E_\nu) \sigma_{\rm
NC}(E_\nu)}\ ,
\label{eqNCrate}
\end{equation}
%...........................................................................
where $\epsilon_{\rm NC}$ is the overall efficiency of neutral
current event detection.
It is necessary to adopt specific values for the efficiencies
$\epsilon_{\rm CC}$ and $\epsilon_{\rm NC}$ in order to evaluate the
relative {\em number\/} of CC and NC events and thus the statistical
errors. We adopt plausible default values, $\epsilon_{\rm CC}=1$
and $\epsilon_{\rm NC} = 0.50$. However, after calculating the statistical
and efficiency errors, we prefer to convert the results to, and to quote,
an essentially efficiency-independent CC/NC ratio,
$R_{\rm CC}/R_{\rm NC}$, defined as
%..........................................................................
\begin{equation}
\frac{R_{\rm CC}}{R_{\rm NC}}=\frac{\epsilon_{\rm NC}}{\epsilon_{\rm CC}}
\frac{N_{\rm CC}}{N_{\rm NC}}\quad.
\label{eqefficindependent}
\end{equation}
In writing Eq.~(\ref{eqefficindependent}), we have assumed that
$\epsilon_{\rm CC}$ is equal to a constant, which turns out to be a
good approximation.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Standard model predictions for the CC spectrum shape}
\label{sec:smpredictsCC}
What effects do the different uncertainties have on the shape of the
electron energy spectrum? Figure~\ref{Threesigfig} answers this question
by showing the standard spectrum (solid line) and the effective 3$\sigma$
shape errors (dashed lines) due to neutrino-related and to detector-related
uncertainties. We postpone the discussion of the sensitivity of the
results to the assumed CC threshold energy to Section~\ref{sec:dependences}
(see especially Tables~\ref{sigmabest} and \ref{sigmaTh}).
The dominant neutrino-related uncertainties are due to the $^8$B neutrino
energy spectrum $\lambda(E_\nu)$ (Fig.~\ref{Threesigfig}a). The theoretical
CC cross section uncertainties are very small; Fig~\ref{Threesigfig}b shows
the ``greatest'' deviation, induced by the use of the EB instead of the KN
cross-sections.
The detector related uncertainties are due to statistics
(Fig.~\ref{Threesigfig}c), energy resolution (Fig.~\ref{Threesigfig}d),
and energy scale (Figs.~\ref{Threesigfig}e,f). The statistical
errors bars in Fig.~\ref{Threesigfig}c refer to a hypothetical sample of 5000
CC events collected above threshold and divided in 10 bins.
In Fig.~\ref{Threesigfig}d, the dotted curves have been obtained by using
energy resolution widths $\sigma_{10}=1.1\pm0.33$ MeV
($\pm3\sigma$ errors, see Sec.~\ref{sec:energyres}). The last two subplots
of Fig.~\ref{Threesigfig} show the effect of the absolute scale uncertainty,
with $\alpha=0$ or $\alpha=1$ (see Sec.~\ref{sec:abenergy}). The two cases
are almost indistinguishable. We adopt in the following the value $\alpha=0$,
since it gives slightly more conservative error estimates.
Figure~\ref{Threesigfig} shows that the systematic errors due to the
adopted uncertainties in the energy scale, in the resolution width and in
the $^8$B neutrino spectrum are at least as important as the statistical
errors, with the additional complication that the non-statistical errors are
correlated point-by-point. The correlations of the uncertainties imply that
the analysis of a realistic spectrum divided in $N$ bins
could become rather
cumbersome, requiring consideration of an $N\times N$ covariance matrix
with large off-diagonal elements for the usual $\chi^2$ statistics. Moreover,
the $\chi^2$ test is not powerful when bins are affected by significant
systematic errors, since the additional information embedded in sequences
of equal-sign deviations (typically all positive or all negative in one
half of the spectrum, as in Fig.~\ref{Threesigfig}) is lost.
The simplest quantity that characterizes a generic electron spectrum, while
avoiding the use of bins, is the average value of the measured recoil energy,
$\langle T_{e}\rangle$ (see also \cite{Fi94,Kw95}).%
%-----------------
\footnote{For a discussion of the extent to which the average electron
recoil energy is a good estimator of the deviations from the expected
shape in the absence of electron flavor violation, see Appendix~B
and Ref.~\protect\cite{newpaper}.}
%-----------------
The basic question
then becomes: ``Can SNO detect significant deviations of
$\langle T_{e}\rangle_{\rm measured}$ from
$\langle T_{e}\rangle_{\rm standard}$?''
We evaluate $\langle T_e\rangle$ with the aid of Eq.~(\ref{eqCCshape}) by
using the definition
%..........................................................................
\begin{equation}
\langle T_e\rangle \equiv \int_{T_{\rm min}} dT_e \, T_e \frac{1}{N_{\rm
CC}} \frac{dN_{\rm CC}}{dT_e}\ .
\label{eqdefTe}
\end{equation}
%..........................................................................
We determine the effects of uncertainties in different ingredients by
carrying out the integration indicated in Eq.~(\ref{eqdefTe}) with different
assumptions. For example, we estimate the 1$\sigma$ uncertainty associated
with the neutrino spectrum by evaluating $\langle T_e\rangle$ using the
neutrino spectra $\lambda^+(E_\nu),\ \lambda^-(E_\nu)$, which are
$\pm 3\sigma$ away from the best-estimate neutrino spectrum \cite{BaLi}.
Then the 1$\sigma$ difference is $\sigma(\langle T_e \rangle) =
\frac{1}{6}[\langle T_e(\lambda^+)\rangle - \langle T_e(\lambda^-)\rangle]$.
Analogously, the $1\sigma$ errors due to the energy resolution uncertainties
were estimated by recalculating the spectra, and thus $\langle T_e \rangle$,
with $\sigma_{10}=1.1\pm0.33$ MeV ($\pm3\sigma$), and dividing the total
shift by six. A similar procedure was adopted for determining the energy
scale error. For the CC cross section uncertainty, we have attached
a $1\sigma$ significance to the deviation obtained when the Ellis-Bahcall
CC cross sections were used instead of the Kubodera-Nozawa CC cross sections,
i.e., $\sigma(\langle T_e \rangle) =
\langle T_e({\rm EB})\rangle - \langle T_e({\rm KN})\rangle$.
Our standard estimate, for $T_{\rm min}=5$ MeV, is then:
%..........................................................................
\begin{eqnarray}
\langle T_{e}\rangle&=&7.658
\pm0.025^a
\pm0.011^b
\pm0.029^c
\pm0.024^d
\pm0.052^e
\rm{\ MeV}\ , \nonumber \\
&=&7.658\cdot(1\pm0.009)
\rm{\ MeV}\ ,
\label{eqTav}
\end{eqnarray}
%...........................................................................
where the errors $(\pm 1\sigma)$ are due to:
a) statistics of 5000 CC events;
b) difference between EB and KN cross sections;
c) uncertainties in the neutrino spectrum;
d) energy resolution; and
e) the absolute energy calibration.
The statistical error of the average is calculated, according to the central
limit theorem, as $\sigma_{\rm stat}=\sqrt{{\rm Var}/N_{\rm CC}}$, where
Var is the variance of the standard distribution above threshold,
${\rm Var}=(1.74~{\rm MeV})^2$.
The assumed uncertainty in the absolute energy scale,
Eq.~(\ref{eqdeltaalpha}), dominates the total error.
In the best-estimate calculation, the detection efficiency $\epsilon_{\rm CC}$
has been taken constant. A linear dependence of $\epsilon_{\rm CC}$ on
$T_e$ (neglecting temporarily the distinction between measured and true
energies, $T_e$ and $T_e^\prime$) would modify the CC spectrum as:
%...........................................................................
\begin{equation}
\frac{1}{N_{\rm CC}}\frac{dN_{\rm CC}}{dT_e}
\longrightarrow
\frac{1}{N_{\rm CC}}\frac{dN_{\rm CC}}{dT_e}
\left( 1+\beta\frac{T_e-\langle T_{e}\rangle}{\langle T_{e}\rangle} \right)
\label{eqCCmodify}
\end{equation}
%............................................................................
and the average value as
%.............................................................................
\begin{equation}
\langle T_{e}\rangle\longrightarrow \langle T_{e}\rangle\left(1+\beta
\frac{{\rm Var}}{\langle T_{e}\rangle^2} \right)
\label{eqmodifyTav}
\end{equation}
%.............................................................................
where $\beta$ is the slope of $\epsilon_{\rm CC}=\epsilon_{\rm CC}(T_e)$.
A plausible variation (or uncertainty) of $\epsilon_{\rm CC}$ over the
interval 5--15 MeV is a few percent, say $3\%$ for definiteness. This
would correspond to $\beta =0.023$ and to a $0.12\%$ variation of
$\langle T_{e}\rangle$, comparable to the smallest error shown in
Eq.~(\ref{eqTav}). We conclude that uncertainties in the CC efficiency will
not be an important source of error, if SNO performs as expected.
The characterization, Eq.~(\ref{eqdefTe}), of the spectrum given
in Eq.~(\ref{eqCCshape}) is the first (non-trivial) step in a complete
description by a series of moments: the zeroth moment (the area), the
first moment (the average), the second moment (the variance), and
higher-order moments. In the present case, the zeroth moment is equal to
unity by definition, and small variations in the shape of the spectrum
affect primarily the average value. More sophisticated
unbinned tests, such as the Kolmogorov-Smirnov (K-S) test, may be useful to
apply after the SNO collaboration has estimated by empirical calibrations
the systematic errors in the experimental input quantities. One could then
determine by Monte Carlo simulations the distribution function for the K-S
statistic with an inferred model for the systematic errors.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Standard model predictions for the CC/NC ratio}
\label{sec:smprediictsCCNC}
In Eq.~(\ref{eqefficindependent}), we have defined a charged to neutral
current ratio, ${R_{\rm CC}}/{R_{\rm NC}}$, which is independent of the
average absolute values of the efficiences but incorporates the
efficiency-induced errors.
The calculation of the standard value and $\pm1\sigma$ errors
for ${R_{\rm CC}}/{R_{\rm NC}}$ is done with the same logic as
for $\langle T_e \rangle$. The final result is:
%.........................................................................
\begin{eqnarray}
\frac{R_{\rm CC}}{R_{\rm NC}}&=&1.882
\pm0.058^a
\pm0.010^b
\pm0.008^c
\pm0.009^d
\pm0.034^e
\pm0.038^f
\nonumber \\
&=&1.882(1 \pm 0.042)\quad.
\label{eqNCratio}
\end{eqnarray}
%.........................................................................
The individual contributions result from:
a) statistics of 5000 CC events and 1354 NC events
($\epsilon_{\rm NC}=0.5$);
b) difference between YHH and KN cross sections;
c) neutrino spectrum;
d) energy resolution;
e) energy scale;
f) NC efficiency.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Correlation of errors}
\label{sec:errors}
Some of the systematic errors affecting $\langle T_{e}\rangle$ and
$R_{\rm CC}/R_{\rm NC}$ have the same origin and are correlated.
In particular, a variation of the neutrino spectrum $\lambda(E_\nu)$
causes both $\langle T_e\rangle$ and $R_{\rm CC}/R_{\rm NC}$ to increase
or decrease at the same time. A variation of $\sigma_{10}$, the energy
resolution width, produces instead opposite effects (anti-correlation).
Energy scale errors are positively correlated.
In Table~\ref{Errortab}, we summarize the separate error components and
their correlations. The correlation of the total errors, $\rho=0.32$, is
small because of accidental cancellations, but it is not entirely
negligible. When the actual error budget for SNO is determined
experimentally, the approximate cancellation of the correlations may not
be as strong.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Could SNO prove the occurrence of new physics?}
\label{sec:newphys}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In the previous section, we calculated the standard predictions
for the CC-shape and the CC/NC ratio, $\langle T_e \rangle$ and
$R_{\rm CC}/R_{\rm NC}$, along with a realistic estimate of these
uncertainties. In this section, we show that the anticipated uncertainties
are sufficiently small to allow SNO to prove the occurrence of new physics
with a high degree of confidence.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Neutrino oscillations}
\label{sec:oscillations}
We examine the implications of three representative scenarios in which the
solar neutrino problem is solved by neutrino oscillations:
the two (best-fit) Mikheyev-Smirnov-Wolfenstein (MSW) \cite{MSW}
solutions at small and at large mixing angle (SMA and LMA), and the purely
vacuum (VAC) oscillation \cite{Pont} solution (see \cite{BaKr} and refs.
therein).%
%-------------------------
\footnote{The best-fit mass and mixing values $(\Delta m^2,\,
\sin^2 2\theta)$ are taken from \cite{BaKr}: ($5.4 \times
10^{-6}~{\rm eV^2},\ 7.9 \times 10^{-3}$) for small angle MSW,
($1.7 \times 10^{-5}~{\rm eV^2}$, 0.69) for large angle MSW, and
($6.0 \times 10^{-11}~{\rm eV^2}$, 0.96) for vacuum oscillations.
Numerical tables of the oscillation probabilities for these best-fit
scenarios were prepared by P.~Krastev \cite{BaKr} and are
available at the following URL:
http://www.sns.ias.edu/$^\sim$jnb.}
%----------------------------
Figure~\ref{Scenariosfig} illustrates some of the principal differences
among the three oscillation scenarios and also shows that all of the
oscillation solutions differ significantly from the standard model
expectations (STD). The survival probabilities for electron-type neutrinos
vary greatly among the oscillation scenarios, as can be seen clearly in
Fig.~\ref{Scenariosfig}a. The energy spectrum of electron-type neutrinos at
earth, shown in Fig.~\ref{Scenariosfig}b, is affected strongly by
oscillations. Figure~\ref{Scenariosfig}c represents the different CC
electron recoil spectra that are predicted for SNO, with the normalization:
Area~=~1. Until the neutral current is measured, the different recoil
spectra must be compared with the same normalization, as in
Fig.~\ref{Scenariosfig}c, because we do not know {\it a priori\/} the total
number of electron-type neutrinos that are created in the solar interior.
Once the neutral current is measured, we can compare the different
oscillation and no-oscillation scenarios in a more informative way.
Figure~\ref{Scenariosfig}d makes use of assumed measurements of the total
neutral and charged current rates and shows the standard electron spectra
normalized to { {\rm Area}\/}~=~$N_{\rm CC}/N_{\rm NC}$.%
%-----------------
\footnote{If we had normalized the area to the efficiency-independent
ratio, { {\rm Area}\/}~=~$R_{\rm CC}/R_{\rm NC}$, then the scale of
the ordinate in Fig.~5d would be decreased by a factor of 2, but the
relative shapes would remain the same.}
%-----------------
A comparison of Figs.~\ref{Scenariosfig}c and Fig.~\ref{Scenariosfig}d
makes clear the importance of the neutral current measurement for
interpreting the shape of the electron recoil spectrum.
Figure~\ref{Spectraratios} shows the ratio of the normalized electron spectra
after oscillation (as displayed in Fig.~\ref{Scenariosfig}c) to the standard
spectrum. These ratios of spectra are approximately linear in
energy. Therefore, the main effect of oscillations on the normalized CC
spectrum is to change the first moment of the energy distribution, i.e.,
the mean value $\langle T_e\rangle$. In Ref.~\cite{Kw95}, it is shown that the
approximate linearity of the ratios of recoil energy spectra is a general
feature of the resonant MSW effect. The representation of the spectral
information by one parameter, $\langle T_e\rangle$, is efficient because
the ratios of spectral shapes are approximately linear (i.e., are determined
rather well by just one parameter).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Statistical analysis}
\label{sec:analysis}
For the three neutrino oscillation scenarios considered in the previous
section, we calculate the observables $\langle T_e\rangle$,
$R_{\rm CC}/R_{\rm NC}$, and their distance---in units of standard
deviations---from the standard predictions, Eqs.~(\ref{eqTav}) and (18).
For each test, the distance is defined simply as:
%.............................................................................
\begin{equation}
{\cal N}(\sigma)=(X-X_{\rm standard})/\sigma_{X,{\rm standard}}\ ,
\label{eqdistance}
\end{equation}
%.............................................................................
where $X=\langle T_e\rangle,\, R_{\rm CC}/R_{\rm NC}$. In the combined tests,
we have calculated the $\chi^2$ including the correlation of the
total errors (see Table~\ref{Errortab}), and defined
${\cal N}(\sigma)=\sqrt{\chi^2}$ \cite{PDG}.
When ${\cal N}(\sigma) \gg 3$, the physical interpretation
is that the statistical probability of the result under consideration
is negligibly
small; the normal distribution presumably
does not describe the extreme tails of
the probability distribution.
The expected deviations for different oscillation scenarios are shown in
Table~\ref{Deviationtab}.
The measurement of the $R_{\rm CC}/R_{\rm NC}$ ratios is a powerful test
for occurrence of new physics; the three oscillation cases are each
separated from the standard expectations by a distance that is formally
more than $15\sigma$ (cf. comment above regarding
${\cal N}(\sigma) \gg 3$ ).
The CC-shape test is less powerful. This is of course
expected for the large-angle MSW case, but is somewhat surprising for
the small-angle MSW case, that was generally expected to be separated from
the standard model expectations at a high confidence level \cite{Fi94,Kw95}.
The reason that the significance level found here for the measurement of the
CC shape is much less than previously calculated is that we have included
estimates for the systematic uncertainties. The systematic errors
in measuring the CC-shape may be twice as large as typical statistical errors
(5000 CC events in our case), as evidenced in Table~\ref{Errortab}.
The statistical power of the combined tests (CC-shape and NC/CC) is dominated
by the measurement of the CC/NC ratio. The effect of the correlation of the
errors is small but not entirely negligible.
Figures~\ref{Summaryfig} and \ref{Distancefig} display graphically the
information contained in Tables~\ref{Errortab} and \ref{Deviationtab}.
In Fig.~\ref{Summaryfig} we show the standard predictions for $\langle
T_e\rangle$ (upper panel) and $R_{\rm CC}/R_{\rm NC}$ (lower
panel), together with the separate and combined $3\sigma$ errors. The
values of $\langle T_e\rangle$ and $R_{\rm CC}/R_{\rm NC}$
for the different oscillation channels are also displayed.
In Fig.~7a, the efficiency error (labeled by a question mark) should
be negligible if SNO works as expected.
In Fig.~\ref{Distancefig} we show the results of the combined tests
(correlations included) in terms of iso-sigma contours in the plane
$(\langle T_e\rangle,\,R_{\rm CC}/R_{\rm NC})$, where ${\cal
N}(\sigma)=\sqrt{\chi^2}$. The three oscillation scenarios can be well
separated from the standard case, but the vertical separation
($R_{\rm CC}/R_{\rm NC}$) is larger and dominating\ with respect to
the horizontal separation ($\langle T_e\rangle$).
The error bars on the SMA point in Fig.~\ref{Summaryfig} and
Fig.~\ref{Distancefig} represent the range of values allowed at
$95\%$ C.L.\ by a fit of the oscillation predictions to the four
operating solar neutrino experiments \cite{BaKr}; they are intended to
indicate the effect of the likely range of
the allowed oscillation parameters.
The choice of $\langle T_e\rangle$ as a characterization of the CC-shape
is not unique. $\langle T_e\rangle$ has been chosen because it is a single
and well-defined number (the first moment of the electron distribution),
whose systematic uncertainties can be determined independent of the event
binning. If the measured electron distribution at SNO has significant
deviations of the second moment (the variance), or higher moments, from
the standard expectations, then $\langle T_e\rangle$ may not be the optimal
statistical estimator. If a more sophisticated statistical test is used to
test for non-standard curvature in the spectrum, then the assessment of
statistical significance will require a full Monte Carlo simulation of
SNO detector, with systematic effects calculated by brute force.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Threshold Dependences}
\label{sec:dependences}
All of the previous calculations were carried out assuming an recoil energy
threshold, $T_{\rm min}$, of 5 MeV. The actual value of $T_{\rm min}$ that
will be used will depend upon the observed or estimated backgrounds in the
operating SNO detector.
Table~\ref{sigmabest} summarizes the dependence of $\langle T_e\rangle$ and
$R_{\rm CC}/R_{\rm CC}$ on the adopted energy threshold. We give values for
the standard model and for the three exemplary oscillation scenarios.
In Table~\ref{sigmaTh} we give the distances of the oscillation scenarios
from the standard predictions, in units of standard deviations.
We see from Tables \ref{sigmabest} and \ref{sigmaTh} that the differences
resulting from changing the threshold by $\pm 1$ MeV are not expected to be
decisive for the SNO discovery potential. However, the diagnostic power of
the measurement of the shape of the electron recoil energy spectrum would
be significantly enhanced if the energy threshold were lowered. The
small mixing angle MSW solution is $3.8 \sigma$ away from the standard
model prediction if the threshold is 4~MeV but is only $2.4 \sigma$
away if the threshold is 6~MeV.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Summary and conclusions}
\label{sec:summary}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The Sudbury Neutrino Observatory has the potential to reveal new phenomena
with a high level of confidence, but the detector must work well
in order to discriminate among different physics options.
The accurate calibration of the absolute energy scale, the
energy resolution width for electron detection, and a high
sensitivity for neutral current detection, are especially important.
We have determined both the the best-available estimates and the
uncertainties of three important neutrino-related input quantities
(available at the
following URL: http://www.sns.ias.edu/$^{\sim}$jnb)
that will be needed in the analysis of the SNO data: the laboratory shape
of the $^8$B neutrino energy spectrum, the charged current neutrino
absorption cross section, and the neutral current dissociation cross
section.
We have also estimated the effects on the tests of electron flavor violation
of five detector-related aspects: the energy resolution, the
absolute energy scale, the energy threshold, and the detection
efficiencies of the charged current events and of the neutral current events.
The principal uncertainties that affect the predictions are shown in
Table~\ref{Errortab} and in Fig~\ref{Summaryfig}. For the measurement
of the shape of the electron recoil energy spectrum,
the largest estimated error is contributed by the uncertainty
in the absolute energy scale, with significant additional errors arising
from the energy resolution and from the shape of the $^8$B neutrino
energy spectrum.
Unfortunately, the systematic uncertainties reduce the power of SNO to
detect new physics via the measurement of the shape of the recoil
electron spectrum. For example, a previous analysis\cite{Kw95},
which considered only statistical errors, indicated that a $3\sigma$
distinction between the standard model prediction and the small angle
MSW solution would be possible with only $1800$ CC events observed with
the SNO observatory. This same statistical-only analysis suggests that
5000 CC events would give more than an $8\sigma$ distinction.
With our adopted estimates of the systematic uncertainties and 5000 CC
events, we find that, instead of $8\sigma$,
the standard model prediction and the
best-fit small angle MSW solution differ by only $3.1\sigma$,
as judged by $\langle T_e \rangle$.
Even with zero statistical error, the difference between the standard
model prediction and the small angle MSW solution would be only $3.3\sigma$.
Indeed, the systematic uncertainties begin to dominate the statistical
uncertainties for this case after (less than) a year of operation.
The statistical significance can be improved by $\sim 1\sigma$ for
the SMA solution if the variance of the spectrum as well as
$\langle T_e \rangle$ is measured \cite{newpaper}.
Fortunately, the shape of the recoil spectrum predicted by vacuum
neutrino oscillations is distinctive and the difference between
standard model physics and vacuum oscillations represents, with our
adopted uncertainties, a $10 \sigma$ distinction in the SNO detector.
The main information content of the measured shape of the electron
recoil spectrum can be summarized by evaluating the average recoil
electron energy. For standard model physics, we find that the average
electron kinetic energy is $\langle T_e\rangle = 7.658(1\pm 0.009) $ MeV,
$1 \sigma$ total errors.
Figure~\ref{Scenariosfig}c compares the normalized electron recoil
spectra computed for the standard model, the LMA, the SMA, and
the vacuum neutrino oscillation scenarios.
The differences in the positions of the peaks of the spectra shown in
Fig.~\ref{Scenariosfig}c are larger than the differences in $\langle
T_e\rangle$ given in Table~\ref{Deviationtab} and
Table~\ref{sigmabest}.
Although previous
authors have
shown similar figures with error bars
due only to the statistical
fluctuations assigned to individual bins, we refrain
from showing error bars in Fig.~\ref{Scenariosfig}c
since the systematic errors
will likely dominate the statistical uncertainties and since
systematic errors are correlated from bin to bin.
There is not a one-to-one relation between the incoming neutrino
energy in the charged current reaction and the energy of the electron
that is produced. Figure~2 shows, in fact, that there is a
significant spread in electron recoil energies for a specified
neutrino energy. This result, unfortunately, contradicts the
assumption of a one-to-one energy relation used by a number of
authors \cite{Fi94,Bilenky93} in describing potential applications of SNO
measurements.
The neutral current to charged current event ratio is a sensitive
probe of lepton flavor violation. For standard model physics, we find
a charged-to-neutral current ratio
$R_{\rm CC}/R_{\rm NC} = 1.882 (1 \pm 0.042)$, $1\sigma$ total error.
The measurement of the absolute neutral current rate
will test directly the solar model
prediction\cite{Ba95} of $3.2_{-0.5}^{+0.6}$ SNU, $1\sigma$ total error,
for the $^8$B neutrino flux. This test of solar models is independent
of uncertainties in the fundamental physics
related to oscillations into active neutrinos.
The predictions of the three favored oscillation solutions
considered here (small mixing and large mixing
MSW solutions, and vacuum oscillations) are all separated by more than
$16\sigma$ from the predictions of the standard model with no lepton
flavor violations, as shown in Table~\ref{Deviationtab} and
Fig.~\ref{Distancefig}. The combined test, shape of the electron
recoil spectrum and ratio of neutral current events to charged events,
is, in our simulations, only slightly more powerful than the neutral
current to charged current ratio alone.
An accurate measurement of the neutral current rate
is essential in order to exploit the full
potential for new physics of the Sudbury Neutrino Observatory.
Are the conclusions about the statistical significance of
the flavor tests
robust with respect to the SNO charged-current energy threshold? This
question is answered in Table~\ref{sigmabest} and Table~\ref{sigmaTh}.
These two tables show that the discriminatory ability of the SNO
detector does not depend critically upon the detection threshold for
the CC reaction. However, lowering the threshold to 4~MeV would
separate by the shape measurement alone the small mixing
angle MSW solution by $3.8 \sigma$ from the
standard prediction, instead of the $3.1 \sigma$ that applies for a
threshold of 5~MeV.
The relative insensitivity of the diagnostic power of SNO to the CC
energy threshold suggests one possible strategy for dealing, especially
in the initial stages of the experiment, with the most troubling
backgrounds.
Without
seriously affecting the discriminatory power of the CC to NC ratio
(see Table~\ref{sigmabest} and Table~\ref{sigmaTh}),
the CC energy threshold for events being analyzed can be set
sufficiently high, at 6 MeV or perhaps even at 7 MeV, that no
significant background contamination is plausible.
Are the absolute event rates for the CC events sensitive to the assumed CC
threshold? Table~\ref{cceventrates} gives the expected event
rates for different assumed thresholds and neutrino oscillation
scenarios. The range of expected event rates is about a factor of two
for CC thresholds from $4$ MeV to $7$ MeV for the standard model and
for the SMA and LMA MSW solutions; the variation is about $40$\% for
the vacuum oscillations.
We have also varied the assumed value of the energy
resolution width at 10 MeV,
$\sigma_{10}$. We find that a one-third worsening of the
energy resolution $\sigma_{10}$, which is defined by
Eq.~(\ref{eqresolution}), from the current best-guess of
1.1 MeV to 1.5 MeV decreases the difference between the
standard model value for $\langle T_e \rangle$ and the small mixing
angle (MSW) value from $3.1\sigma$ to
2.8$\sigma$. Thus it is important that the energy resolution
width is kept as low as possible.
The expected background level in SNO decreases strongly with
increasing energy \cite{Sudb}.
Increasing the electron energy threshold for the charged current
reaction, can decrease the fractional contribution of background events.
The principal lesson of
Table~\ref{sigmabest} and Table~\ref{sigmaTh} is that the SNO experiment will
provide a powerful diagnostic for new physics even if the observed
backgrounds are somewhat higher than expected.
We have varied the mass and mixing parameters
of the small mixing angle (MSW) case within the 95\% C.L.\
limits of the fit to the four operating experiments \cite{BaKr}. The
difference between the standard value of $\langle T_e \rangle$ and the
value calculated for the small mixing angle solution varies
between $3.1\pm0.6$ standard deviations, depending on which values
one adopts within the allowed MSW region.
(Somewhat more powerful discrimination can be achieved if both the
dispersion and the mean recoil energy are calculated \cite{newpaper}.)
The formal difference between
the standard and the MSW value of $R_{\rm CC}/R_{\rm NC}$ is always
within $15.7\pm 1.6$ standard deviations at 95\% C.L. Thus the MSW solution
at small mixing angles (SMA) can be tested with a high level of confidence
with the SNO experiment, although the information coming from
the CC-shape might not be sufficient by itself.
We discuss in Appendix~A the information that can be gained from
an overall test of the SNO detector using an intense
$^8$Li($\beta^-$) source. We calculate a theoretical
$^8$Li($\beta^-$) spectrum and compare with the available data. We
conclude that the existing data for the $^8$Li($\beta^-$) spectrum
are not sufficient to permit an
accurate test of the SNO detector and that a new, laboratory experiment
is required.
.
Appendix $B$ discusses the extent to which $\langle T_e \rangle$
is a good statistical estimator of possible deviations in the
CC electron spectrum.
Finally, we must ask: How general are the conclusions given in this paper?
The method of analysis and the discussion of the principal ingredients and
their uncertainties will be of use in considering how well SNO can
test potential
new physics scenarios. We have evaluated the sensitivity of the
likely operation of the Sudbury Neutrino Observatory to the MSW and vacuum
neutrino oscillations that best-fit (see \cite{BaKr}) the results of the
four pioneering
solar neutrino experiments. The correct physical explanation
may differ from the oscillation solutions considered here. The ``true''
solution of the solar neutrino problems may involve, for example, a
distortion of the CC electron recoil spectrum that is much more drastic than
is implied by the oscillation solutions considered here.
In this case, the shape of the electron recoil spectrum from
the CC reaction might indicate new physics that is not
apparent by comparing the rates
of the CC and the NC reactions \cite{Bilenky93}.
The analysis presented in this paper is
illustrative of the power of the SNO detector, but specific,
quantitative inferences depend upon what, if any, new physics exists
in the accessible domain.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\acknowledgments
We are grateful to the SNO collaboration for providing information on
the best-estimates and uncertainties for the experimental quantities.
In addition, we are especially grateful to
F. Ajzenberg-Selove,
H. A. Bethe,
F.~Calaprice,
W.~C.~Haxton,
B.~R.~Holstein,
E.~Kolbe,
P.~I.~Krastev,
K.~Kubodera,
Y.~Nir,
W.~Press,
A.~Yu.~Smirnov, and
J.~Sromicki
for valuable advice, discussions, and suggestions. For important
comments on an early draft of this paper, we are grateful to
E.~W.~Beier,
W.~Frati, and
R.~G.~H. Robertson.
The work of JNB is supported in part by NSF grant No. PHY95-13835.
The work of EL was supported in part by the Institute for Advanced
Study through a Hansmann fellowship, and in part by INFN. The
research of EL was also performed under the auspices of the
Theoretical Astroparticle Network, under contract No. CHRX-CT93-0120
of the Direction General XII of the E.E.C.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\appendix
\section{Testing SNO by Measuring {\it in~situ\/}
$^8$L{\lowercase{i}} beta-decay}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The SNO collaboration plans to perform an overall test of the
experiment by
measuring the $\beta$-decay spectrum of an intense $^8$Li source
that will be placed in different locations in the detector.
The measurement by SNO of the $^8$Li($\beta^-$) spectrum
will be used as a demonstration that
the results obtained for a known
beta-decay spectrum are consistent with those measured in the laboratory.
Figure~\ref{compareLiB} compares our calculated (see below)
$^8$Li spectrum with
the standard model electron spectrum from $^8$B solar neutrino
CC absorption [reaction~(\ref{reactionCC})].
The test is based upon the fact that, despite the different physical
processes
that are involved in the two cases, the electron spectra
from $^8$Li $\beta$-decay and from $^8$B solar neutrino absorption
on deuterium have somewhat similar shapes and
cover essentially the same energy range (from 0 to $\sim13$ MeV).
The spectra displayed in Fig.~\ref{compareLiB} are separately
normalized to unity above
the standard SNO threshold of 5 MeV and do not include broadening due
to the finite energy resolution in the detector (or other signatures
of the SNO detector).
We discuss in this appendix some of the things that can be
learned from the ${\rm ^8Li}$ test. For background information, we
first summarize in subsection~\ref{App:relations} the relations
between the $^8$Li and the $^8$B electron spectra.
We then describe in subsection~\ref{App:snoresponse} how a future
precision
laboratory measurement of the $^8$Li spectrum could be used, in
conjunction with a SNO measurement of the $^8$Li spectrum, to help
determine characteristics of the SNO detector.
\subsection{Relations Between the $^8$Li and $^8$B Spectra}
\label{App:relations}
The $^8$Li electron spectrum is produced by the
beta decay
%.......................................................................
\begin{equation}
\label{App:lithium}
{}^8{\rm Li}\to {}^8{\rm Be}+e^-+\bar\nu_e\ .
\end{equation}
%........................................................................
The CC electron spectrum, whose measurement is one of the primary
goals of SNO, is produced by
a two-step reaction (beta decay followed by neutrino capture):
%.......................................................................
\begin{mathletters}
\label{App:boron}
\begin{equation}
{}^8{\rm B}\to {}^8{\rm Be}+e^+ +\nu_e\ ,
\label{App:b8}
\end{equation}
\begin{equation}
\nu_e+d\to p + p + e^- \ .
\label{App:nuabs}
\end{equation}
\end{mathletters}
%........................................................................
All three of these reactions are different, but
the intermediate $^8$Be states are the same in both $^8$B
(reaction~\ref{App:b8})
and $^8$Li (reaction~\ref{App:lithium})
decay. The $^8$Be excited states are unstable and break
up into two alpha particles.
As a consequence, the shape of the $^8$Li $\beta^-$
spectrum [Eq.~(\ref{App:b8})] and of the $^8$B $\beta^+$ spectrum
[Eq.~(\ref{App:nuabs})]
deviate significantly from the standard allowed shape. The shape
of the $^8$B neutrino spectrum and its uncertainties are
affected as well (see \cite{BaLi} and references therein).
Measurements of the delayed $\alpha$ spectrum allow one to determine the
profile of the intermediate ${\rm ^8Be}$ state and thus to calculate
the deviations of the $^8$Li electron, $^8$B positron, and $^8$B neutrino
spectra from their allowed
shapes. The best-estimated $^8$B positron and neutrino spectra
and their uncertainties were discussed extensively
in \cite{BaLi}.
In Fig.~\ref{Lifig}a we show our
best-estimate for the ${\rm ^8Li} (\beta^-)$ electron
spectrum, together with its $\pm 3 \sigma$ uncertainties.
The calculation that leads to Fig.~\ref{Lifig}a
is similar to the calculation
of the ${\rm ^8B}(\beta^+)$ spectrum performed in \cite{BaLi}, modulo
different values for the radiative corrections, forbidden corrections,
and Coulomb effects.
The form factors entering the forbidden corrections are, by isospin
symmetry,
the same as used for ${\rm ^8B}(\beta^+)$ \cite{BaLi}.
As discussed in Sec.~II~A,
the $3\sigma$ uncertainties shown in Fig.~\ref{Lifig}a are related to
a possible offset, $b$, in the energy of the $\alpha$ particles from
$^8$Be break-up: $E_\alpha\to E_\alpha+b$. The value of this
offset affects, in particular, the calculated
peaks of the $\beta^+$, $\beta^-$,
and $\nu$ spectra in Eqs.~(A1) and (A2).
The effective
$3\sigma$ uncertainty of the offset $b$ is estimated to be
$\pm 0.104$ MeV \cite{BaLi}; this estimate includes
theoretical errors.
There are at least three complementary experiments
that could help to reduce the offset uncertainty, $b$, and thereby
make the prediction of the $^8$B solar neutrino spectrum more precise.
The potential experiments are:
1) a high-precision measurement of the $\alpha$ spectrum from $^8$Be
break-up; 2) a high-precision measurement of the $\beta^+$ spectrum
from $^8$B decay; and 3) a high-precision measurement of the
$\beta^-$ spectrum from $^8$Li decay. In all three
cases, dedicated laboratory
experiments with a carefully calibrated spectrograph
would be needed.
The SNO detector is expected to have an uncertainty in the
absolute energy calibration of $\sim
100$~keV ($1 \sigma$), and therefore probably cannot be used as a
$^8$Li spectrometer at the
level of precision needed to further constrain $b$.
\subsection{The SNO Response Function: $S_{\rm SNO}(T^\prime_e,\,T_e)$}
\label{App:snoresponse}
In addition to an overall demonstration that the detector is working
as expected,
can one learn more about the characteristics of
SNO by studying
the ${\rm Li}(\beta^-)$ spectrum? The answer is: ``Yes,
provided that a precision measurement of the ${\rm
^8Li} (\beta^-)$ spectrum is made with a laboratory spectrograph.''
If future laboratory experiments determine accurately
the $^8$Li beta decay spectrum, then the measurement with SNO of
this same spectrum can be used to constrain possible
systematic effects that apply
in the energy range that is also relevant for $^8$B neutrino absorption
%----------------------------------------
\footnote{
The theoretical $^8$Li spectrum in Fig.~\ref{Lifig}a
is affected by significant shape uncertainties (dotted lines) and is
not a valid substitute for a high-precision laboratory measurement.}.
%------------------------------------------
Specifically, one could use the following strategy. Let
$\lambda^{\rm Li}(T_e^\prime)$ be the {\em true\/} lithium spectrum
as a function of the {\em true\/} electron energy, $T_e^\prime$.
Suppose, as a first approximation, that this spectrum
is known with ``infinite'' precision as a result of an error-free
laboratory
spectrographic measurement:
%...........................................................................
\begin{equation}
\lambda^{\rm Li}_{\rm lab}\simeq\lambda^{\rm Li} \ .
\label{App:errorfree}
\end{equation}
%...........................................................................
Then the lithium spectrum measured in the SNO detector,
$\lambda^{\rm Li}_{\rm SNO}$, will be given by
a convolution of $\lambda^{\rm Li}(T_e^\prime)$ with the SNO response
function, $S_{\rm SNO}(T^\prime_e,\,T_e)$, where $T_e$ is the
{\em measured\/} electron kinetic energy and $T_e^\prime$ is the true
electron energy. Thus
%...........................................................................
\begin{equation}
\lambda^{\rm Li}_{\rm SNO}(T_e)=\int\,\lambda^{\rm Li}(T_e^\prime)\cdot
S_{\rm SNO}(T^\prime_e,\,T_e)\; dT_e^\prime\ .
\label{App:liconvolve}
\end{equation}
%...........................................................................
In our notation [see Eq.~(13)], the SNO response function can be
written as
the product of the energy resolution and the CC detection
efficiency:
%...........................................................................
\begin{equation}
S_{\rm SNO}(T^\prime_e,\,T_e) = R(T^\prime_e,\,T_e)\cdot
\epsilon_{\rm CC}(T_e^\prime)\ .
\end{equation}
%...........................................................................
Therefore $S_{\rm SNO}(T^\prime_e,\,T_e)$
depends on three parameters: the energy
resolution width $\sigma_{10}$ [Eq.~(8)], the absolute energy scale error
$\delta$ [Eq.~(9)], and the slope of the CC efficiency function
$\beta$ [Eq.~(18)].
(Of course, other free parameters could be
eventually introduced to model $S_{\rm SNO}$ more accurately.)
Once $\lambda^{\rm Li}$ and $\lambda^{\rm Li}_{\rm SNO}$ are
{\em experimentally\/} determined, one can use Eq.~(A4) to fit
the parameters of the SNO response function, $S_{\rm SNO}$, which
describes the
spectral distorsion effects induced by the detector. Since
the SNO detector will also be calibrated with more
traditional techniques (e.g., with $\gamma$ rays of known energy),
the determination of the free parameters in $S_{\rm SNO}$
will be overconstrained.
The experimental overdetermination of $S_{\rm SNO}$ will limit the
effects of unknown systematic errors.
Moreover,
the comparison of the fitted parameter values
with those estimated by Monte~Carlo simulations
will provide further consistency checks.
In practice, one has to take account of uncertainties in
the spectrum that is measured in the laboratory,
$\lambda^{\rm Li}_{\rm lab}$,
in order to infer the true spectrum $\lambda^{\rm Li}$. The corrections
will depend on the response function of the laboratory
spectrograph,
$S_{\rm lab}$:
%...........................................................................
\begin{equation}
\lambda^{\rm Li}_{\rm lab}(T_e)=\int\,\lambda^{\rm Li}(T_e^\prime)\cdot
S_{\rm lab}(T^\prime_e,\,T_e) \;dT_e^\prime\ .
\end{equation}
%...........................................................................
The response function $S_{\rm lab}$ must be known with a precision
higher than what one hopes to achieve for the overall SNO response
function, $S_{\rm SNO}$.
Any uncertainty, $\delta S_{\rm lab}$, in the laboratory response function,
will be propagated to $S_{\rm SNO}$.
Are the available laboratory data on the Li($\beta^-$) decay
sufficiently good that their uncertainties would not introduce large
errors in the SNO response function if determined via
Eq.~(\ref{App:liconvolve})? Unfortunately, the answer is ``No.''
In Fig.~\ref{Lifig}b we show the published data from \cite{Horn}
and from \cite{Alle}, which are superimposed on the theoretical spectrum.
(We have taken the data at face value and made no attempt to deconvolve
resolution effects.) The agreement of the data with themselves and with
the theoretical spectrum is unsatisfactory.
In Fig.~\ref{Lifig}c,
the $\beta$ kinetic energies of each
data set are linearly transformed, $T_\beta \to A T_\beta + B$, so that
the two (renormalized) experimental spectra match each other and fit the
theoretical spectrum. A good fit-by-eye (Fig.~\ref{Lifig}c) is obtained with
parameters: $A = 0.94$, $B = 0.45$ MeV for the data of reference
\cite{Horn}
and $A = 1.06$, $B = -0.2$ MeV for the data of reference
\cite{Alle}. The results of these transformations suggest
that true spectrum, $\lambda^{\rm Li}$, cannot be inferred
from the available data with a precision better than a few hundred keV
in the energy scale.
A new, precision laboratory measurement of the ${\rm ^8Li} (\beta^-)$
spectrum is needed.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% A P P E N D I X B
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Statistics of linear deviations from the standard spectrum}
\label{app:statistics}
We state in the text that the average value
of the electron kinetic energy, $\langle T \rangle$, is a good
statistical estimator of {\em linear\/}
deformations of the recoil spectrum.
What we mean by this claim is that $\langle T \rangle$ contains most
of the information about spectral deformations, especially if the
deformations are small.
In this Appendix we show heuristically
the basic correctness of the above statement in two
simple but representative cases,
purely statistical errors and a single dominant systematic error.
More precisely, we show in Eq.~(\ref{eq:proofstat}) (stat.\ error) and
Eq.~(\ref{eq:proofsyst}) (syst.\ error)
that, for a linear spectral deformation,
the $\chi^2$ associated with
deviations of $\langle T \rangle$ is approximately equal to the $\chi^2$
obtained by binning the observed spectrum in a histogram.
However,
if there are several comparable systematic errors, or if the
deviation is nonlinear, significant additional information
may be obtained from the higher moments \cite{newpaper}.
The arguments given below are in the spirit of a ``physicist's proof''
rather than a mathematical theorem.
We note that for probability distributions with long tails the
analysis in terms of moments may not be appropriate.
Fortunately, the spectrum of
electron recoil energies does not have pathologically long tails so
this last remark does not apply in the case we are considering.
Let $\rho(T)$ be the {\em expected\/} normalized electron recoil
spectrum $[\int\!dT\,\rho(T)=1]$, with average kinetic energy
$\langle T \rangle$ and variance $\sigma^2$. Let $\rho'(T)$ be the
{\em observed\/} normalized spectrum, with average energy
$\langle T \rangle'$. In the hypothesis of a perfectly linear spectral
deformation, one can always write:
%.............................................................................
\begin{equation}\label{eq:betaslope}
\frac{\rho'(T)}{\rho(T)}=1+\beta\,
\frac{T-\langle T\rangle}{\langle T\rangle}\ ,
\end{equation}
%.............................................................................
where $\beta$ is a slope parameter. Then the shift in the average energy
is given by
%.............................................................................
\begin{equation}\label{eq:DeltaT}
\Delta \langle T\rangle \equiv \langle T\rangle'-\langle T\rangle=
\beta\,\frac{\sigma^2}{ \langle T\rangle}\ ,
\end{equation}
%.............................................................................
and its $\chi^2$ statistic simply reads
%.............................................................................
\begin{equation}\label{eq:chiT}
\chi^2_{\langle T\rangle}=\left(
\frac{\Delta \langle T\rangle}{\sigma_{\langle T\rangle}} \right)^2\ ,
\end{equation}
%.............................................................................
where $\sigma_{\langle T\rangle}$ is the total error affecting
$\langle T\rangle$.\
%-----------------
\footnote{
Notice from Eq.~(\ref{eq:DeltaT}) that
$\Delta\langle T\rangle$
and $\beta$ are in one-to-one correspondence, so that a determination of the
shift in the average energy $\langle T\rangle$ is equivalent to a
determination of the slope of $\rho'(T)/\rho(T)$ with the same
fractional accuracy, and vice~versa. The observables $\Delta\langle T\rangle$
and $\beta$ are interchangeable for a linear spectral distortion. We
prefer $\Delta\langle T\rangle$
because the average kinetic energy is well-defined
also in the case of a non-linear distortion, while $\beta$ is not.}
%----------------------
For purely statistical
errors, $\sigma^2_{\langle T\rangle}=\sigma^2/N$ by the
central limit theorem, where $N$ is the total number of observed
electrons. From Eqs.~(\ref{eq:DeltaT}) and (\ref{eq:chiT}) one has:
%.............................................................................
\begin{equation}\label{eq:chistat}
\chi^2_{\langle T\rangle}
=\beta^2\frac{\sigma^2}{\langle T\rangle^2} N\quad
{(\rm stat.\ dominated)}\ .
\end{equation}
%.............................................................................
A single small, purely systematic error (such as the uncertainty
in the $^8$B neutrino spectrum shape or in the absolute
energy calibration) also produces, in first order,
a linear deformation of the expected recoil spectrum.
The spectral distortion can thus be represented as
an uncertainty $\sigma_\beta$ of the slope parameter $\beta$. Then,
from Eq.~(\ref{eq:DeltaT}),
the propagated error on $\langle T \rangle$ is
$\sigma_{\langle T \rangle}=\sigma_\beta\sigma^2/{\langle T \rangle}$, and
one has from Eqs.~(\ref{eq:DeltaT}) and (\ref{eq:chiT}):
%.............................................................................
\begin{equation}\label{eq:chisyst}
\chi^2_{\langle T\rangle}=
\frac{\beta^2}{\sigma^2_\beta}\quad
{(\rm syst.\ dominated)}\ ,
\end{equation}
%.............................................................................
as would be expected intuitively.
Let us divide now the spectra in $n$ bins of width $\Delta T_i$:
$\rho\equiv\{\rho_i,\,\Delta T_i\}_{i=1,\dots, n}$ and
$\rho'\equiv\{\rho'_i,\,\Delta T_i\}_{i=1,\dots, n}$, with
$\sum_i\rho_i\Delta T_i=\sum_i\rho'_i\Delta T_i=1$.
Then the shift in the height
of the $i$-th bin associated to the linear deformation in
Eq.~(\ref{eq:betaslope}) is:
%.............................................................................
\begin{equation}\label{eq:Deltarho}
\Delta\rho_i\equiv\rho'_i-\rho_i=\rho_i\,\beta\,
\frac{T_i - \langle T \rangle}{\langle T \rangle}\ ,
\end{equation}
%.............................................................................
where $T_i$ is the average value of $T$ in the $i$-th bin.
If counting statistics dominates the errors, the fractional
uncertainty $\sigma_i/\rho_i$ of
the $i$-th bin height is $\sigma_i/\rho_i=1/\sqrt{N\rho_i\Delta T_i}$,
and the total $\chi^2$
of the histogram differences,
$\chi^2_{\rm hist}=\sum_i(\Delta\rho_i/\sigma_i)^2$, is easily derived:
%.............................................................................
\begin{equation}\label{eq:chihist}
\chi^2_{\rm hist}=\beta^2\,\frac{\hat{\sigma}^2}{\langle T\rangle^2}\,N\ ,
\end{equation}
%.............................................................................
where $\hat{\sigma}^2=\sum_i \rho_i \Delta T_i(T_i- \langle T\rangle)^2$
is just a discretized estimate of the variance $\sigma^2$, and therefore
$\hat{\sigma}^2\simeq\sigma^2$. One gets the desired proof by comparing
Eqs.~(\ref{eq:chistat}) and (\ref{eq:chihist}):
%.............................................................................
\begin{equation}\label{eq:proofstat}
\chi^2_{\rm hist}\simeq\chi^2_{\langle T\rangle}
\quad{\rm (stat.\ dominated)}\ .
\end{equation}
%.............................................................................
If systematic errors dominate, the corresponding analysis of a binned
spectrum is somewhat trickier.
Let the error be represented by
an overall uncertainty $\sigma_\beta$ of the slope parameter
$\beta$. From Eq.~(\ref{eq:Deltarho}), this uncertainty propagates to an error
$\sigma_i=\sigma_\beta\rho_i(T_i-\langle T \rangle)/\langle T \rangle$
of the $i$-th bin.
The formula $\chi^2_{\rm hist}=\sum_i(\Delta\rho_i/\sigma_i)^2$ naively
(and incorrectly) applied to this case would give
$\chi^2_{\rm hist}=n\beta^2/\sigma_\beta^2=n\chi^2_{\langle T\rangle}$,
with $\chi^2_{\langle T\rangle}$ given by Eq.~(\ref{eq:chisyst}).
However, a systematic shift in the slope
$\beta$ produces completely correlated bin errors: ${\rm corr}(i,\,j)=1$.
Therefore, only one out the $n$ bin residuals is independent, and
the ``effective'' number of bins to be considered in the $\chi^2$ is 1
%----------------------
\footnote{The reader more experienced in
statistical analyses may have noticed that, in this
case, the square error matrix including the error correlations would
have rank 1 and not $n$, signaling that only one bin error is independent.},
%---------------------
so that $\chi^2_{\rm hist}=\beta^2/\sigma_\beta^2$ and:
%.............................................................................
\begin{equation}\label{eq:proofsyst}
\chi^2_{\rm histo}=\chi^2_{\langle T\rangle}
\quad{\rm (syst.\ dominated)}\ .
\end{equation}
%.............................................................................
Equations (\ref{eq:proofstat}) and (\ref{eq:proofsyst})
show that, if the errors are dominated by statistics or by
a single systematic uncertainty, the use of the integrated variable
$\langle T \rangle$ is as informative as a spectrum
histogram, provided that the spectral deformations are linear in $T$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%% R E F E R E N C E S
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{thebibliography}{99}
\bibitem{Sudb} SNO Collaboration, G.~T.~Ewan {\em et al.\/},
``Sudbury Neutrino Observatory Proposal,''
Report No.\ SNO-87-12, 1987 (unpublished);
``Scientific and Technical Description of
the Mark~II SNO Detector,'' edited by
E.~W.~Beier and D.~Sinclair, Report No.\
SNO-89-15, 1989 (unpublished). A.\ B.\ McDonald,
Proceedings of the 9th Lake Louise Winter Institute,
edited by A.\ Astbury {\em et al.\/} (World
Scientific, Singapore, 1994), p. 1.
\bibitem{SEW}S. L. Glashow, Nucl. Phys. {\bf 22}, 579 (1961);
S. Weinberg, Phys. Rev. Lett. {\bf 19}, 1264 (1967);
A. Salam, in {\it Elementary Particle Theory}, edited
by N. Svartholm (Almqvist and Wiksells, Stockholm, 1968),
p. 367.
\bibitem{PDG} Particle Data Group, L.~Montanet {\em et al.\/},
Phys.\ Rev. \ D {\bf 50}, 1173 (1994).
\bibitem{Ba91} J.~N.~Bahcall, Phys.\ Rev.\ D {\bf 44}, 1644 (1991).
\bibitem{Ch85} H.~H.~Chen, Phys.\ Rev.\ Lett.\ {\bf 55}, 1534 (1985).
\bibitem{BaLi} J.~N.~Bahcall, E.~Lisi, D.~E.~Alburger, L.~De Braeckeleer,
S.~J.~Freedman, and J.~Napolitano,
Institute for Advanced Study Report No.\
IASSNS-AST 95/47, e-print archive No.\ nucl-th/9601044,
to appear in Phys.\ Rev.\ C.; see also:
J.\ N.\ Bahcall and B.\ R.\ Holstein,
Phys.\ Rev.\ C {\bf 33}, 2121 (1986).
\bibitem{bah72} J.~N.~Bahcall, N.~Cabibbo, and A.~Yahil, Phys.\ Rev.\
Lett.\ {\bf 28}, 316 (1972).
\bibitem{ber87} Z.~G.~Berezhiani and M. I. Vysotsky, Phys.\ Lett.\ B
{\bf 199}, 281 (1987).
\bibitem{vol86} M.~B.~Voloshin, M.~I.~Vysotskii, and L.~B.~Okun,
Soviet Phys.\ JETP {\bf 64}, 446 (1986);
{\bf 65}, 209(E) (1987).
\bibitem{lim88} C.-S.~Lim and W.~J.~Marciano, Phys.\ Rev.\ D {\bf 37},
1368 (1988).
\bibitem{akh88} E.~K.~Akhmedov, Phys.\ Lett.\ B {\bf 213}, 64 (1988).
\bibitem{gasp88}M.~Gasperini, Phys.\ Rev.\ D {\bf 38}, 2635 (1988).
\bibitem{rou91} E.~Roulet, Phys.\ Rev.\ D {\bf 44}, R935 (1991).
\bibitem{guz91} M.~M.~Guzzo, A.~Masiero, and S.~T.~Petcov, Phys.\ Lett.\ B
{\bf 260}, 154 (1991).
\bibitem{30yr} J.~N.~Bahcall, R.~Davis, Jr., P.~Parker, A.~Smirnov,
and R.~Ulrich, editors: {\em Solar Neutrinos, The First Thirty
Years\/}, Frontiers in Physics, Vol.~92
(Addison-Wesley, 1994).
\bibitem{SNOw} Updated information on the status of the SNO experiment
can be retrieved from the following URL:
http://snodaq.phy.queensu.ca/SNO/sno.html.
\bibitem{Ba89} J.~N.~Bahcall, {\em Neutrino Astrophysics\/} (Cambridge
University Press, Cambridge, England, 1989).
\bibitem{Othe} P.~I.~Krastev and A.~Yu.~Smirnov, Phys.\ Lett.\ B
{\bf 338}, 282 (1994); N.~Hata and P.~Langacker,
Phys.\ Rev.\ D {\bf 50}, 632 (1994); P.~I.~Krastev
and S.~T.~Petcov, Nucl.\ Phys.\ {\bf B449}, 605 (1995).
\bibitem{Fi94} G.~Fiorentini, M.~Lissia, G.~Mezzorani, M.~Moretti, and
D.~Vignaud, Phys.\ Rev.\ D {\bf 49}, 6298 (1994).
\bibitem{Kw95} W.~Kwong and S.~P.~Rosen, Phys.\ Rev.\ D {\bf 51},
6159 (1995); see also
Phys.\ Rev.\ Lett.\ {\bf 68}, 748 (1992).
\bibitem{Cali} SNO collaboration, Internal Report (unpublished).
\bibitem{Napo} J.~Napolitano, S.~J.~Freedman, and J.~Camp,
Phys.\ Rev.\ C {\bf 36}, 298 (1987).
\bibitem{Wilk} D.~H.~Wilkinson and D.~E.~Alburger, Phys.\ Rev.\ Lett.\
{\bf 26}, 1127 (1971); numerical data tables are reported
in E.~K.~Warburton, Phys.\ Rev.\ C {\bf 33}, 303 (1986).
\bibitem{BP92} J.~N.~Bahcall and M.~H.~Pinsonneault, Rev.\ Mod.\ Phys.
{\bf 64}, 885 (1992).
\bibitem{Ba95} J.~N.~Bahcall and M.~H.~Pinsonneault,
Rev.\ Mod.\ Phys.\ {\bf 67}, (1995) 781.
\bibitem{Jenk} T.~L.~Jenkins, ``A Proposed Experiment for the Detection of
Solar Neutrinos,'' Case Western Reserve Univ.\ Report
No.\ C00-818-62, 1962 (unpublished).
\bibitem{Ku94} K.~Kubodera and S.~Nozawa,
Int.\ J.\ Mod.\ Phys.\ E {\bf 3}, 101 (1994).
\bibitem{Ko66} Yu.~S.~Kopysov and V.~A.~Kuz'min,
Yad.\ Fiz.\ {\bf 4}, 1031 (1966)
[Sov.\ J.\ Nucl.\ Phys.\ {\bf 4}, 740 (1967)].
\bibitem{Ke66} F.~J.~Kelly and H.~{\"U}berall,
Phys.\ Rev.\ Lett.\ {\bf 16}, 145 (1966).
\bibitem{El68} S.~D.~Ellis and J.~N.~Bahcall,
Nucl.\ Phys.\ {\bf A114}, 636 (1968).
\bibitem{Be49} H.~A.~Bethe,
Phys.\ Rev. {\bf 76}, 38 (1949);
H.~A.~Bethe and C.~Longmire,
Phys.\ Rev. {\bf 77}, 647 (1950).
\bibitem{Yi92} S.~Ying, W.~C.~Haxton, and E.~M.~Henley,
Phys.\ Rev.\ C {\bf 45}, 1982 (1992).
\bibitem{Do92} M.~Doi and K.~Kubodera,
Phys.\ Rev.\ C {\bf 45}, 1988 (1992).
\bibitem{Pr75} M.~A.~Preston and R.~K.~Bhaduri, {\em Structure of
the Nucleus\/} (Addison-Wesley, 1975, Reading,
Massachusetts).
\bibitem{Bilenky93} S.~M.~Bilenky and C.~Giunti, Phys.\ Lett.\ B
{\bf 311}, 179 (1993).
\bibitem{bkn} J. N. Bahcall, K. Kubodera, and S. Nozawa, Phys.\ Rev.\
D {\bf 38}, 1030 (1988).
\bibitem{bahcall69} J. N. Bahcall, Phys.\ Rev.\ Lett.\ {\bf 23}, 251
(1969).
\bibitem{Hirata91}Kamiokande collaboration, K.~S.~Hirata {\em et al.\/},
Phys.\ Rev.\ D {\bf 44}, 2241 (1991).
\bibitem{Beier96}G.~Beier, private communication (1996); A.~Hallin,
private communication (1996).
\bibitem{Rob96} R.~G.~H.~Robertson, private communication (1996).
\bibitem{newpaper} J.~N.~Bahcall, P.~I.~Krastev, and E.~Lisi,
in preparation.
\bibitem{MSW} L.~Wolfenstein, Phys.\ Rev.\ D {\bf 17}, 2369 (1978);
S.~P.~Mikheyev and A.~Yu.~Smirnov,
Yad.\ Fiz.\ {\bf 42}, 1441 (1985)
[Sov.\ J.\ Nucl.\ Phys.\ {\bf 42},913 (1985)].
\bibitem{Pont} V.~N.~Gribov and B.~M~Pontecorvo, Phys.\ Lett.\ B {\bf 28},
493 (1969);
J.~N.~Bahcall and S.~C.~Frautschi, Phys.\ Lett.\ B {\bf 29},
623 (1969);
S.~L.~Glashow and L.~M.~Krauss, Phys.\ Lett.\ B {\bf
190}, 199 (1987).
\bibitem{BaKr} J.~N.~Bahcall and P.~I.~Krastev,
Phys.\ Rev.\ D {\bf 53}, 4211 (1996).
\bibitem{Horn} W.~F.~Hornyak and T.~Lauritsen, Phys.\ Rev.\ {\bf 77},
160 (1950).
\bibitem{Alle} M.~Allet {\em et al.\/}, Phys.\ Rev.\ Lett. {\bf 68}, 572;
see also J.~Sromicki {\em et al.\/}, Phys.\ Rev.\ C
{\bf 53}, 932 (1996).
\end{thebibliography}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% T A B L E I
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{table}
\caption{ The percentage $1\sigma$ errors from different ingredients
that affect the standard predictions,
$\langle T_e\rangle = 7.658$ MeV and
$R_{\rm CC}/R_{\rm NC} = 1.882$.
The numbers given are for $N_{\rm CC} = 5000$ events above
threshold ($T_{\rm min}=5$ MeV), and $\epsilon_{\rm CC}=1$,
$\epsilon_{\rm NC} = 0.5$. Uncertainties due to the
backgrounds are neglected. The approximate
cancellation of the correlation of total the errors may not
be as strong for the actual SNO error budget.}
\begin{tabular}{lccc}
Error component &
$\sigma (\langle T_e\rangle)\;\,\%$ &
$\sigma (R_{\rm CC}/R_{\rm NC})\;\,\%$ &
Correlation \\
\hline
Neutrino spectrum & 0.38 & 0.43 & $+1$ \\
Cross section & 0.14 & 0.53 & $\sim0$ \\
Statistics & 0.33 & 3.09 & 0 \\
Energy resolution & 0.31 & 0.47 & $-1$ \\
Energy scale & 0.68 & 1.81 & $+1$ \\
Efficiency & $\sim0$ & 2.00 & $\sim0$ \\
\hline%---------------------------------------------------------------------
TOTAL & 0.91 & 4.18 & 0.32 \\
\end{tabular}
\label{Errortab}
\end{table}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% T A B L E II
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{table}
\caption{ Deviations of $\langle T_e\rangle$ and
$R_{\rm CC}/R_{\rm NC}$ from the standard model
(electron flavor conserved) predictions.
The results are shown for representative neutrino
oscillation scenarios in units of standard deviations
($\sigma$). In the last two columns, the
combined $\chi^2$ for the CC-shape and CC/NC ratio tests
is calculated with and without the correlation of the total
errors ($\rho = 0.32$), and the deviation is given as:
Dev.~$(\sigma)=\protect\sqrt{\chi^2}$. Uncertainties
due to the backgrounds are neglected.}
\begin{tabular}{lccccccc}
&&
\multicolumn{2}{c}{CC-shape test}& \multicolumn{2}{c}{CC/NC test} &
\multicolumn{2}{c}{Combined tests} \\
%
&&
\multicolumn{2}{c}{$\langle T_e \rangle$ (MeV)} &
\multicolumn{2}{c}{$R_{\rm CC}/R_{\rm NC}$} &
$\rho=0.22$ & $\rho=0$ \\
%
Scenario & Acronym &
Value & Dev.\ ($\sigma$) &
Value & Dev.\ ($\sigma$) &
Dev.\ ($\sigma$) & Dev.\ ($\sigma$) \\
\hline
Standard & STD & 7.658 & --- & 1.882 & --- & --- & --- \\
Small Mixing Angle (MSW) & SMA & 7.875 & 3.1 & 0.639 & 15.7 & 17.9 & 16.0 \\
Large Mixing Angle (MSW) & LMA & 7.654 & 0.0 & 0.422 & 18.5 & 19.5 & 18.5 \\
Vacuum Oscillations & VAC & 8.361 &10.0 & 0.411 & 18.6 & 25.1 & 21.1 \\
\end{tabular}
\label{Deviationtab}
\end{table}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% T A B L E III
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{table}
\caption{ The dependence of $\langle T_e\rangle$,
$R_{\rm CC}/R_{\rm NC}$, and of their errors, on the
threshold energy, $T_{\rm min}$. The correlation of errors
is given in the last column. The results shown
assume 5000 CC events collected above
threshold. Uncertainties due to the backgrounds are
neglected. The acronyms for the scenarios (STD, SMA, LMA,
VAC) are the same as in Table~\protect\ref{Deviationtab}.}
\begin{tabular}{ccccccccccc}
$T_{\rm min}$ &
\multicolumn{4}{c}{Average energy $\langle T_e \rangle$ (MeV)} &&
\multicolumn{4}{c}{CC/NC ratio, $R_{\rm CC}/R_{\rm NC}$} &
Correlation \\
%
(MeV) &
STD $\pm$ $1\sigma$ & SMA & LMA & VAC &&
STD $\pm$ $1\sigma$ & SMA & LMA & VAC &
$\rho$ \\
\hline
4.0&$7.234\pm0.079$&7.533&7.228&8.101&&$2.177\pm0.088$&0.712&0.489&0.440&0.23\\
5.0&$7.658\pm0.070$&7.875&7.654&8.361&&$1.882\pm0.079$&0.639&0.422&0.411&0.32\\
6.0&$8.187\pm0.063$&8.337&8.184&8.678&&$1.509\pm0.067$&0.534&0.338&0.369&0.43\\
7.0&$8.798\pm0.058$&8.897&8.796&9.110&&$1.107\pm0.056$&0.409&0.248&0.307&0.52\\
\end{tabular}
\label{sigmabest}
\end{table}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% T A B L E IV
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{table}
\caption{ The energy threshold $(T_{\rm min})$ dependence
of the deviations, $\xi$, of the
predictions with neutrino oscillations from the
standard predictions. The entries
in the table, $\xi (\langle T_e\rangle)$ and
$ \xi (R_{\rm CC}/R_{\rm NC})$, are in units of
standard deviations. The acronyms for the
oscillation scenarios
(SMA, LMA, VAC) are the same as in
Table~\protect\ref{Deviationtab}. Uncertainties due to
the backgrounds are neglected.}
\begin{tabular}{cccccccccccccc}
$T_{\rm min}$ &&
\multicolumn{3}{c}{$\xi (\langle T_e\rangle)$} &&
\multicolumn{3}{c}{$\xi (R_{\rm CC}/R_{\rm NC})$} &&
\multicolumn{3}{c}{Combined, $\xi=\protect\sqrt{\chi^2}$} \\
%
(MeV)&& SMA & LMA & VAC && SMA & LMA & VAC && SMA & LMA & VAC \\
\hline
4.0 && 3.8 & 0.1 & 11.0 && 16.6 & 19.2 & 19.7 && 18.3 & 19.8 & 25.4 \\
5.0 && 3.1 & 0.0 & 10.0 && 15.7 & 18.5 & 18.6 && 17.5 & 19.5 & 25.1 \\
6.0 && 2.4 & 0.0 & 7.8 && 14.6 & 17.4 & 17.0 && 17.4 & 19.2 & 23.9 \\
7.0 && 1.7 & 0.0 & 5.4 && 12.5 & 15.3 & 14.3 && 15.8 & 17.9 & 20.7 \\
\end{tabular}
\label{sigmaTh}
\end{table}
\begin{table}
\caption[]{The CC event rates as a function of the CC energy
threshold and the neutrino oscillation solution.}
\begin{tabular}{ccccc}
$T_{\min}$&STD&SMA&LMA&VAC\\
(MeV)&(SNU)&(SNU)&(SNU)&(SNU)\\
\hline
\noalign{\smallskip}
4&6.9&2.3&1.6&1.4\\
5&6.0&2.0&1.3&1.3\\
6&4.8&1.7&1.1&1.2\\
7&3.5&1.3&0.8&1.0\\
\end{tabular}
\label{cceventrates}
\end{table}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% F I G U R E C A P T I O N S
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%.......................................................................... 1
\begin{figure}
\caption{ (a) Total CC cross section as calculated by Kubodera and
Nozawa (KN), Ying, Haxton and Henley (YHH), and Ellis and
Bahcall (EB), slightly improved. (b) Total NC cross section
as calculated by Kubodera and Nozawa (KN), and Ying, Haxton
and Henley (YHH).}
\label{Totalcross}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 2
\begin{figure}
\caption{ Comparison of the CC differential cross-section at various
energies and scattering angles. Solid: Kubodera and Nozawa
\protect\cite{Ku94}. Dotted: Ellis and Bahcall
\protect\cite{El68}, slightly improved.}
\label{Comparecross}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 3
\begin{figure}
\caption{ The normalized electron spectrum, with and without inclusion
of the energy resolution function.}
\label{normfig}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 4
\begin{figure}
\caption{ Three standard deviation departures from the standard electron
spectrum (solid line) due to neutrino-related and
detector-related errors.}
\label{Threesigfig}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 5
\begin{figure}
\caption{ Neutrino Oscillation Scenarios:
(a) survival probabilities for oscillation test cases;
(b) effect of neutrino oscillations on neutrino spectrum
at earth;
(c) effect of neutrino oscillations on normalized electron
spectrum at SNO. Area under curves = 1;
(d) effect of neutrino oscillations on electron spectrum
at SNO.
Area under curves = $N_{\rm CC}/N_{\rm NC}$.
Labels: STD~=~standard (no oscillation); SMA~=~small
mixing angle (MSW); LMA~=~large mixing angle (MSW);
VAC~=~vacuum oscillation. See the text for details.}
\label{Scenariosfig}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 6
\begin{figure}
\caption{ Ratios of the normalized neutrino spectra for different
oscillation scenarios. The normalized spectra are
displayed in Fig.~\protect\ref{Scenariosfig}c. Labels as
in Fig.~5.}
\label{Spectraratios}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 7
\begin{figure}
\caption{ Values of the characteristic CC-shape variable, the average
energy $\langle T_e \rangle$, and of the
CC/NC ratio, $R_{\rm CC}/R_{\rm NC}$,
together with $3\sigma$
error bars. Uncertainties due to the backgrounds are
neglected. Labels as in Fig.~5.}
\label{Summaryfig}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 8
\begin{figure}
\caption{ Iso-sigma contours $(\sigma=\protect\sqrt{\chi^2})$
for the combined CC-shape and CC/NC test, for the
representative oscillation cases shown in Fig.~5
and discussed in the text.
Uncertainties due to the backgrounds are neglected.
For values of the iso-sigma distance
${\cal N}(\sigma) \gg 3$,
the number of standard deviations is only a formal
characterization; the tail of the
probability distribution is not expected to be Gaussian
for very large values of ${\cal N}(\sigma)$.
Labels as in Fig.~5. }
\label{Distancefig}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 9
\begin{figure}
\caption{ A comparison of the $^8$Li beta-decay spectrum and the
standard electron spectrum
from $^8$B neutrino absorption, as a function of the
electron
kinetic energy above the standard SNO threshold (5 MeV).
The spectra shown are both theoretical: the effects of
finite energy resolution
are not included. The area under the curves is
normalized to unity.}
\label{compareLiB}
\end{figure}
%\vspace*{-2mm}
%.......................................................................... 10
\begin{figure}
\caption{ (a) Theoretical $^8$Li spectrum and its $3\sigma$
uncertainties.
(b) Experimental determinations of the $^8$Li spectrum.
(c) Experimental data with an allowance for a linear
recalibration of the energy. See the text for details.}
\label{Lifig}
\end{figure}
%\end{document}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%% E N D %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%% P O S T S C R I P T F I G U R E S
%%%%%%% memo: to include them add epsfig in the \documentstyle
%%%%%%% and move this part befor \end{document}.
%%%%%%% Include the following \newcommand:
%%----------------------------------------------------------------------------
\newcommand{\InsertFigure}[2]{\newpage\begin{center}\mbox{%
\epsfig{bbllx=1.4truecm,bblly=1.3truecm,bburx=19.5truecm,bbury=26.5truecm,%
height=21.4truecm,figure=#1}}\end{center}\vspace*{-1.8truecm}%
\parbox[t]{\hsize}{\small\baselineskip=0.5truecm\hspace*{0.5truecm} #2}}
%----------------------------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%..............................................................................
\InsertFigure{sno_01.ps}%
{ FIG.~1. (a) Total CC cross section as calculated by Kubodera and
Nozawa (KN), Ying, Haxton and Henley (YHH), and Ellis and
Bahcall (EB), slightly improved. (b) Total NC cross section
as calculated by Kubodera and Nozawa (KN), and Ying, Haxton
and Henley (YHH).}
%..............................................................................
\InsertFigure{sno_02.ps}%
{ FIG.~2. Comparison of the CC differential cross-section at various
energies and scattering angles. Solid: Kubodera and Nozawa
\protect\cite{Ku94}. Dotted: Ellis and Bahcall
\protect\cite{El68}, slightly improved.}
%..............................................................................
\InsertFigure{sno_03.ps}%
{ FIG.~3. The normalized electron spectrum, with and without inclusion
of the energy resolution function.}
%..............................................................................
\InsertFigure{sno_04.ps}%
{ FIG.~4. Three standard deviation departures from the standard electron
spectrum (solid line) due to neutrino-related and
detector-related errors.}
%..............................................................................
\InsertFigure{sno_05.ps}%
{ FIG.~5. Neutrino Oscillation Scenarios:
(a) survival probabilities for oscillation test cases;
(b) effect of neutrino oscillations on neutrino spectrum
at earth;
(c) effect of neutrino oscillations on normalized electron
spectrum at SNO. Area under curves = 1;
(d) effect of neutrino oscillations on electron spectrum
at SNO.
Area under curves = $N_{\rm CC}/N_{\rm NC}$.
Labels: STD~=~standard (no oscillation); SMA~=~small
mixing angle (MSW); LMA~=~large mixing angle (MSW);
VAC~=~vacuum oscillation. See the text for details.\hfill}
%..............................................................................
\InsertFigure{sno_06.ps}%
{ FIG.~6. Ratios of the normalized neutrino spectra for different
oscillation scenarios. The normalized spectra are
displayed in Fig.~\protect\ref{Scenariosfig}c. Labels as
in Fig.~5.}
%..............................................................................
\InsertFigure{sno_07.ps}%
{ FIG.~7. Values of the characteristic CC-shape variable, the average
energy $\langle T_e \rangle$, and of the
CC/NC ratio, $R_{\rm CC}/R_{\rm NC}$,
together with $3\sigma$
error bars. Uncertainties due to the backgrounds are
neglected. Labels as in Fig.~5.}
%..............................................................................
\InsertFigure{sno_08.ps}%
{ FIG.~8. Iso-sigma contours $(\sigma=\protect\sqrt{\chi^2})$
for the combined CC-shape and CC/NC test, for the
representative oscillation cases shown in Fig.~5
and discussed in the text.
Uncertainties due to the backgrounds are neglected.
For values of the iso-sigma distance
${\cal N}(\sigma) \gg 3$,
the number of standard deviations is only a formal
characterization; the tail of the
probability distribution is not expected to be Gaussian
for very large values of ${\cal N}(\sigma)$.
Labels as in Fig.~5. }
\InsertFigure{sno_09.ps}%
{ FIG.~9. A comparison of the $^8$Li beta-decay spectrum and the
standard electron spectrum
from $^8$B neutrino absorption, as a function of the
electron
kinetic energy above the standard SNO threshold (5 MeV).
The spectra shown are both theoretical: the effects of
finite energy resolution
are not included. The area under the curves is
normalized to unity.}
%..............................................................................
\InsertFigure{sno_10.ps}%
{ FIG.~10. (a) Theoretical $^8$Li spectrum and its $3\sigma$
uncertainties.
(b) Experimental determinations of the $^8$Li spectrum.
(c) Experimental data with an allowance for a linear
recalibration of the energy. See the text for details.}
%..............................................................................
%..............................................................................
\end{document}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% E N D
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Edited:
% John June 25, 1996 (2:30 PM)
% John June 25, 1996 (1 PM)
% Eligio,June 21, 1996 (3 P.M.)
% Eligio,June 20, 1996 (4 P.M.)
% ----------------------------------> submission to PRD (rcvd May 17)
% John May 16, 1996 (9 A.M.)
% John May 15, 1996 (9 A.M.)
% John May 14, 1996 (7 P.M.)
% John May 14, 1996 (2 P.M.)
% John May 14, 1996 (10 A.M.)
% John May 1, 1996 (4 PM)
% Eligio, May 1, 1996 (11:00 am)
% John, May 1, 1996 (morning)
% Eligio, April 30, 1996 (evening)
% John, April 30, 1996 (afternoon)
% Eligio, April 30, 1996 (morning)