The Okamoto-Nolen-Schiffer anomaly without rho-omega mixing
Extracted source text ยท arXiv nucl-th/9602010
\documentstyle[preprint,aps]{revtex}
\tighten
\begin{document}
\draft
\preprint{\vbox{Submitted to Nuclear Physics A
\hfill FSU-SCRI-96-08}}
\title{The Okamoto-Nolen-Schiffer anomaly without
\boldmath{$\!\!\rho$}-\boldmath{$\omega$} mixing}
\author{F. Cardarelli}
\address{Department of Physics and Supercomputer Computations
Research Institute, \\
Florida State University, Tallahassee, FL 32306}
\author{J. Piekarewicz}
\address{Supercomputer Computations Research Institute, \\
Florida State University, Tallahassee, FL 32306}
\date{\today}
\maketitle
\begin{abstract}
We examine the effect of isospin-violating meson-nucleon
coupling constants and of $\pi$-$\eta$ mixing on the
binding-energy differences of mirror nuclei in a model
that possesses no contribution from $\rho$-$\omega$ mixing.
The ${}^{3}$He-${}^{3}$H binding-energy difference is computed
in a nonrelativistic approach using a realistic wave function.
We find the ${}^{3}$He-${}^{3}$H binding-energy difference
very sensitive to the short-distance behavior of the
nucleon-nucleon potential. We conclude that for the typically
hard Bonn form factors such models can not account for the
observed binding-energy difference in the three-nucleon system.
For the medium-mass region (A=15--41) the binding-energy
differences of mirror nuclei are computed using a relativistic
mean-field approximation to the Walecka model. We obtain large
binding-energy differences---of the order of several hundred
keV---arising from the pseudoscalar sector. Two effects are
primarily responsible for this new finding: a) the inclusion
of isospin breaking in the pion-nucleon coupling constant, and
b) the in-medium enhancement of the small components of the
bound-state wave functions. We look for off-shell ambiguities
in these results and find them to be large.
\end{abstract}
\pacs{PACS number(s):~21.10.Dr, 24.10.Jv, 13.75.Gx}
\narrowtext
\section{Introduction}
\label{sec:intro}
Some of our most fundamental ideas about charge-symmetry-breaking
(CSB) phenomena in nuclear physics~\cite{henmil79,miller90} are currently
being revisited. To a large extent this revision has been prompted by
the suggestion of Goldman, Henderson, and Thomas of a suppressed
contribution from $\rho$-$\omega$ mixing to the CSB component of the
nucleon-nucleon (NN) interaction at small momentum transfers~\cite{ght92}.
Since then, many calculations, using a variety of theoretical approaches,
have confirmed this
suppression~\cite{piewil93,krein93,hats94,mitch94,oconn94,malta95}.
Although the issue continues to be
controversial~\cite{miller94,oconn95}, the search for alternate CSB
mechanisms has already started. Indeed, in a recent publication we
have proposed isospin breaking in the meson-nucleon vertices as an
additional source of isospin violation~\cite{ghp95}. We have used
isospin breaking in the pion-nucleon coupling constant~\cite{piekar95}
to account for the large isospin violation in the pion-nucleon
system reported recently by Gibbs, Ai, and Kaufmann~\cite{gak95}.
Moreover, we have studied the impact of isospin violation in the
vector-meson--nucleon coupling constants on the difference between
the neutron and proton analyzing powers ($\Delta A$) measured in
elastic $\vec{n}$-$\vec{p}$ scattering~\cite{knut90}. We have
concluded that the magnitude of the resulting class IV CSB potential
was consistent with that phenomenologically required by the
experiment~\cite{ghp95}.
However, other CSB observables, such as the scattering-length
differences of the $pp$ and $nn$ systems~\cite{slaus89}, and
the binding-energy differences of mirror
nuclei---the Okamoto-Nolen-Schiffer (ONS)
anomaly~\cite{oka64,nolsch69}---are particularly insensitive
to the class IV component of the CSB potential. Rather, these
observables are driven by the class III component.
Indeed, several authors~\cite{bluiqb87,coon87,ishsas90}
have observed that the class III component of the CSB
potential---generated from ``on-shell'' $\rho$-$\omega$
mixing---dominates the binding-energy discrepancy of mirror
nuclei and could explain most of the ONS anomaly~\cite{miller90}.
Motivated by the recent developments, we propose to study
the ONS anomaly in a model which possesses no contribution from
$\rho$-$\omega$ mixing; in our model, the resulting CSB potential
is generated exclusively from $\pi$-$\eta$ mixing and from isospin
violation in the meson-nucleon coupling constants. To compute the
${}^{3}$He-${}^{3}$H binding-energy difference we employ a
realistic three-body wave function obtained from the variational
analysis of Ref.~\cite{pace77} using the Reid-soft core potential.
To study the ONS anomaly in the A=15--41 region we use
a relativistic mean-field approximation to the Walecka
model~\cite{serwal86}. Although the Walecka model has been used
recently to compute binding-energy differences in mirror
nuclei~\cite{nedroo88,bgk95,kkb96}, these studies have been limited
to the vector sector: Coulomb corrections plus on-shell
$\rho$-$\omega$ mixing. However, since these contributions are
dominated by the timelike part of the vector vertex ($\gamma^{0}$)
they are not particularly sensitive to relativistic effects.
In contrast, one expects important relativistic corrections in
the pseudoscalar sector because of the structure of the
($\gamma^{5}$) vertex. Examining the role of relativistic effects
and of isospin violation in the meson-nucleon coupling constants
are two of the major components of the present investigation.
Our paper has been organized as follows. In Sec.~\ref{sec:csbpot}
we introduce the CSB potentials derived from meson mixing and from
isospin violation in the meson-nucleon coupling constants. There,
we review some of the ideas germane to our model.
Sec.~\ref{sec:threebody} is reserved to the analysis of the ONS
anomaly in the three-nucleon system with special emphasis placed
on the sensitivity of our results to the choice of hadronic form
factors. Binding-energy differences of A=15--41 mirror nuclei are
computed in Sec.~\ref{sec:manybody} using various off-shell
extrapolations of the NN potential. Finally, we present our
conclusions in Sec.~\ref{sec:concl}.
\section{Charge-symmetry-breaking Potentials}
\label{sec:csbpot}
\subsection{Vector-meson sector}
The CSB potentials which arise from isospin violation in the
vector-meson--nucleon vertices and from $\rho$-$\omega$ mixing
are given, respectively, by~\cite{henmil79,miller90,ghp95}
%
\begin{mathletters}
\label{vveccsb}
\begin{eqnarray}
\widehat{V}^{\omega}_{\rm CSB} &=&
{V}^{\omega}_{\rm CSB}(Q^{2})
\Big[ \Gamma^{\mu}(1) \gamma_{\mu}(2) \tau_z(1) +
\gamma^{\mu}(1) \Gamma_{\mu}(2) \tau_z(2) \Big]
\;, \label{vomegaa} \\
\widehat{V}^{\rho}_{\rm CSB} &=&
{V}^{\rho}_{\rm CSB}(Q^{2})
\Big[ \Gamma^{\mu}(1) \gamma_{\mu}(2) \tau_z(2) +
\gamma^{\mu}(1) \Gamma_{\mu}(2) \tau_z(1) \Big] +
\label{vrhoa} \\
&\phantom{+}&
{V}^{\prime\rho}_{\rm CSB}(Q^{2})
\left[\Gamma^{\mu}(1) \Gamma_{\mu}(2)
\Big(\tau_z(1)+\tau_z(2)\Big) \right] \;, \nonumber \\
\widehat{V}^{\rho\omega}_{\rm CSB} &=&
{V}^{\rho\omega}_{\rm CSB}(Q^{2})
\left[\gamma^{\mu}(1) \gamma_{\mu}(2)
\Big(\tau_z(1)+\tau_z(2)\Big) \right] +
\label{vrhoomegaa} \\
&\phantom{+}&
{V}^{\prime\rho\omega}_{\rm CSB}(Q^{2})
\Big[\Gamma^{\mu}(1) \gamma_{\mu}(2) \tau_z(1) +
\gamma^{\mu}(1) \Gamma_{\mu}(2) \tau_z(2)\Big] \;. \nonumber
\end{eqnarray}
\end{mathletters}
%
where $Q^{2} \equiv -q_{\mu}^{2}$ is the negative of the
four-momentum transfer, $\Gamma^{\mu} \equiv
i\sigma^{\mu\nu}(p'-p)_{\nu}/2M$, and we have defined
%
%
\begin{mathletters}
\begin{eqnarray}
{V}^{\omega}_{\rm CSB}(Q^{2}) &\equiv&
{g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}^2
\over Q^{2} + m_{\omega}^2} f_{1}^{\omega}
\;, \label{vomegab} \\
{V}^{\rho}_{\rm CSB}(Q^{2}) &\equiv&
{g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}^2
\over Q^{2} + m_{\rho}^2} f_{0}^{\rho}
\;, \label{vrhob} \\
{V}^{\prime\rho}_{\rm CSB}(Q^{2}) &\equiv&
{g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}
f_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}
\over Q^{2} + m_{\rho}^2} f_{0}^{\rho}
\;, \label{vrhopb} \\
{V}^{\rho\omega}_{\rm CSB}(Q^{2}) &\equiv& -
{g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}
g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}
\over (Q^{2} + m_{\rho}^2)(Q^{2} + m_{\omega}^2)}
\langle \rho | H | \omega \rangle
\;, \label{vrhoomegab} \\
{V}^{\prime\rho\omega}_{\rm CSB}(Q^{2}) &\equiv& -
{f_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}
g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}
\over (Q^{2} + m_{\rho}^2)(Q^{2} + m_{\omega}^2)}
\langle \rho | H | \omega \rangle \;.
\label{vrhoomegapb}
\end{eqnarray}
\end{mathletters}
%
Note that $g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}$,
$g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}$, and
$f_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}$, are the
isospin-conserving meson-nucleon coupling constants
extracted from fits to two nucleon data and displayed
in Table~\ref{tableone} using the Bonn B potential
model~\cite{machl89}.
Charge-symmetry breaking in the potentials is
characterized by the appearance of three isospin-violating
parameters: $f_{1}^{\omega}$, $f_{0}^{\rho}$, and
$\langle\rho|H|\omega\rangle$; an isovector-tensor
$NN\omega$ coupling constant, an isoscalar-tensor
$NN\rho$ coupling constant, and the $\rho$-$\omega$
mixing amplitude, respectively.
Our model is inspired by the notion of vector-meson dominance (VMD),
so that the vector mesons couple to the appropriate isospin components
of a conserved electromagnetic current. In models of this kind the
$\rho$-$\omega$ mixing amplitude necessarily vanishes at
$Q^{2}~=~0$~\cite{piewil93,oconn94}. Moreover, our VMD assumption
places important constraints on the form of the isospin-violating
vector-meson--nucleon vertex at $Q^{2}=0$. Specifically, isospin
breaking can occur only in the tensor part of the vertex; the
vector coupling is protected by gauge
invariance~\cite{ghp95,dmitra95,maltb95}. Thus, gauge invariance
precludes the appearance of a class III---vector-vector---component
in the CSB potential generated from vector-meson exchange. For the
analysis of $\Delta A$ this fact is of no consequence, as class III
CSB potentials do not contribute to $np$ observables. Indeed, the
class IV contribution from $\omega$-meson exchange is able to fill
the role demanded by data; recall that this term is identical in
structure and comparable in size to the one generated from
on-shell $\rho$-$\omega$ mixing~\cite{ghp95}. However,
for observables in which the vector-vector component of
$\widehat{V}^{\rho\omega}_{\rm CSB}$ was believed to be
dominant---such as in the case of the ONS
anomaly~\cite{bluiqb87,coon87,ishsas90}---the absence of the
corresponding term from $\widehat{V}^{\omega}_{\rm CSB}$ and
$\widehat{V}^{\rho}_{\rm CSB}$ could have important
phenomenological consequences. This conclusion is unavoidable:
it follows directly from gauge invariance; gauge invariance forces
the isospin-violating vector coupling and the $\rho$-$\omega$ mixing
amplitude to vanish at $Q^{2}=0$.
\subsection{Pseudoscalar-meson sector}
The CSB potential which emerges from isospin violations in the
pseudoscalar-meson sector is given by,
%
\begin{mathletters}
\label{vpscsb}
\begin{eqnarray}
&&
\widehat{V}^{(5)}_{\rm CSB} =
{V}^{(5)}_{\rm CSB}(Q^{2}) \gamma^{5}(1)\gamma^{5}(2)
\Big[\tau_z(1)+\tau_z(2)\Big] \;, \\
&&
{V}^{(5)}_{\rm CSB}(Q^{2}) =
{V}^{\pi}_{\rm CSB}(Q^{2}) +
{V}^{\pi\eta}_{\rm CSB}(Q^{2}) \;,
\end{eqnarray}
\end{mathletters}
%
with ${V}^{(5)}_{\rm CSB}$ made out of contributions arising
from isospin violation in the pion-nucleon vertex and from
$\pi$-$\eta$ mixing~\cite{ghp95}, respectively,
%
\begin{mathletters}
\begin{eqnarray}
{V}^{\pi}_{\rm CSB}(Q^{2}) &\equiv&
{g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}^2
\over Q^{2} + m_{\pi}^2} g_{0}^{\pi}
\;, \label{vpion} \\
{V}^{\pi\eta}_{\rm CSB}(Q^{2}) &\equiv& -
{g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}
g_{\lower 2pt \hbox{$\scriptstyle NN\eta$}}
\over (Q^{2} + m_{\pi}^2)(Q^{2} + m_{\eta}^2)}
\langle \pi | H | \eta \rangle \;.
\end{eqnarray}
\end{mathletters}
%
The isospin-conserving pion-nucleon coupling constant
$g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}$ is given
by the Bonn B potential value listed in Table~\ref{tableone}.
For the $NN\eta$ coupling constant we have decided to use a
value extracted from SU(3)-flavor symmetry~\cite{coon82},
rather than the Bonn-potential value of
$g_{\lower 2pt \hbox{$\scriptstyle NN\eta$}}^{2}/4\pi=2.25$.
Indeed, it is believed that the Bonn potential overestimates
the coupling; a recent analysis based on $\eta$-photoproduction
data suggests values as low as
$g^2_{NN\eta}/4\pi \alt 0.5$~\cite{tiator94} (see also
Ref.~\cite{piekar93}). For the $\eta N$ cutoff parameter
we have simply assumed: $\Lambda_{NN\eta}=\Lambda_{NN\pi}$.
The two parameters driving the CSB potential are the
isospin-violating pion-nucleon coupling constant ($g_{0}^{\pi}$)
and the $\pi$-$\eta$ mixing amplitude ($\langle\pi|H|\eta\rangle$).
CSB potentials derived from $\pi$-$\eta$ mixing have been included
in previous nonrelativistic analyses of the ONS
anomaly~\cite{bluiqb87,ishsas90,coon82}. However, to our knowledge,
CSB potentials arising from isospin violation in the meson-nucleon
coupling constants have never been incorporated into these analyses,
and only recently have they been used to study isospin violations
in low-energy pion-nucleon scattering~\cite{piekar95}.
\section{Binding-energy difference in the A=3 system}
\label{sec:threebody}
In this section we examine the effect of $\pi$-$\eta$ mixing and
of isospin-violating meson-nucleon coupling constants on the
${}^{3}$He-${}^{3}$H binding-energy difference. Experimentally,
this difference is very precisely known and has the value of
764~keV. A nearly model-independent method for estimating the
electromagnetic contribution to the binding-energy difference
has already been developed~\cite{coon87,friar70}; ambiguities remain,
however, on how to separate the meson-exchange-current component
from the experimental charge form factors. These estimates suggest
an electromagnetic contribution to the binding-energy of
$693 \pm 19\pm5$~keV, leaving $71\pm19\pm5$~keV to be explained
by charge-symmetry-breaking effects\cite{miller90,coon87}. Note that
a very similar result has been obtained recently by Ishikawa and
Sasakawa using a different method~\cite{ishsas90}. These authors
have also computed the contribution to the binding-energy
difference arising from (on-shell) $\rho$-$\omega$ mixing and
from $\pi$-$\eta$ mixing, with the former playing the dominant
role. Our model possesses no contribution from $\rho$-$\omega$
mixing. Hence, we would like to investigate the possible role
of alternate CSB mechanisms in accounting for the non-electromagnetic
contribution to the ${}^{3}$He-${}^{3}$H binding-energy difference:
\begin{equation}
\Delta E = \langle {}^{3}He | V_{pp} | {}^{3}He \rangle
- \langle {}^{3}H | V_{nn} | {}^{3}H \rangle
\equiv - \langle {}^{3}He | \Delta V | {}^{3}He \rangle \;.
\end{equation}
We shall compute the ${}^{3}$He-${}^{3}$H binding-energy difference
using the realistic three-body wave function obtained from the
variational analysis of Numberg, Prosperi, and Pace\cite{pace77}.
These authors have written the nuclear Hamiltonian
\begin{equation}
H=\frac{1}{2M}({\bf p}_a^2 + 3{\bf p}_b^2) +
\sum_{i,j=1 \atop (i<j)}^{3} V_{ij} \;,
\end{equation}
in terms of the two intrinsic coordinates
\begin{equation}
{\bf a} = \sqrt{1 \over 2}({\bf r}_1 - {\bf r}_2) \;; \quad
{\bf b} = \sqrt{1 \over 2}(2{\bf r}_3 -{\bf r}_1 -{\bf r}_2) \;,
\end{equation}
and their corresponding conjugate momenta. Note that the
indices 1 and 2 are reserved for the identical particles
in the system; this is particularly convenient for the
perturbative estimate of the ${}^{3}$He-${}^{3}$H
binding-energy shift. The basic dynamical input is contained
in the two-nucleon interaction ($V_{ij}$) which was chosen to
be the Reid soft-core potential\cite{reid68}. Matrix elements of
the Hamiltonian were computed using a truncated basis of 820
harmonic-oscillator states and the ground-state wave function
was obtained from direct matrix diagonalization. This wave
function provides a satisfactory description of the charge form
factors of ${}^{3}$He and ${}^{3}$H in the low-momentum-transfer
region~\cite{pace77,ciofi79}. Moreover, a perturbative estimate of
the Coulomb shift in ${}^{3}$He yields 672~keV---a value
consistent with that reported in Refs.~\cite{coon87,ishsas90}.
For simplicity, the binding energy difference will be evaluated
with the dominant ${}^{1}S_{0}$ component of the trinucleon
wave function. Note that for this case it is sufficient to evaluate
the ${}^{1}S_{0}$ component of the CSB potential:
\begin{equation}
\Delta V \equiv V_{nn}({}^{1}S_{0}) - V_{pp}({}^{1}S_{0}) \;.
\end{equation}
The singlet CSB component of the NN potential is obtained by
performing a nonrelativistic reduction of the various Lorentz
structures given in Eqs.~(\ref{vveccsb},\ref{vpscsb}) and by
evaluating the meson propagators in the static
limit~\cite{henmil79,coon75}. We obtain ($f_1^{\rho}\equiv
f_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}/
g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}$)
\begin{mathletters}
\begin{eqnarray}
\Delta V^{\pi\eta}({\bf q}) &=&
g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}
g_{\lower 2pt \hbox{$\scriptstyle NN\eta$}}
\langle\pi|H|\eta\rangle {{\bf q}^{2}/M^{2} \over
{({\bf q}^2+m_{\pi}^2)}{({\bf q}^2+m_{\eta}^2})} \;, \\
\Delta V^{\pi}({\bf q}) &=& -
g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}^{2}g_0^{\pi}
{{\bf q}^{2}/M^{2} \over {{\bf q}^2+m_{\pi}^2}} \;, \\
\Delta V^{\rho}({\bf q}) &=& -
g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}^{2}
(1+2f_1^{\rho}) f_0^{\rho}
{{\bf q}^{2}/M^{2} \over {{\bf q}^2+m_{\rho}^2}} \;, \\
\Delta V^{\omega}({\bf q}) &=& -
g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}^{2}f_1^{\omega}
{{\bf q}^{2}/M^{2} \over {{\bf q}^2+m_{\omega}^2}} \;.
\end{eqnarray}
\end{mathletters}
Note the appearance of the ``relativistic-correction'' factor
${\bf q}^{2}/M^{2}$. This factor emerges from the allowed
Lorentz structures present in the CSB potentials. Recall
that---in contrast to previous analyses based on on-shell
$\rho$-$\omega$ mixing---no Lorentz vector structure
[$\gamma^{\mu}(1)\gamma_{\mu}(2)$] is allowed in our model.
For the $\pi$-$\eta$ mixing amplitude we have used the accepted
value of $\langle\pi|H|\eta\rangle=-4200$~MeV$^{2}$~\cite{coon82}.
For the isospin-violating meson-nucleon parameter we have adopted
the nonrelativistic quark-model estimate of Refs.~\cite{ghp95}:
\begin{mathletters}
\begin{eqnarray}
g_{0}^{\pi}&=&{3 \over 10}{\Delta m \over m} \approx 0.004 \;, \\
f_{0}^{\rho}&=&{3 \over 2}{\Delta m \over m} \approx 0.020 \;, \\
f_{1}^{\omega}&=&{5 \over 6}{\Delta m \over m} \approx 0.011 \;,
\end{eqnarray}
\end{mathletters}
where $m=313$~MeV is the average constituent quark mass and
$\Delta m=4.1$~MeV is the down-up quark mass difference~\cite{licht89}.
Although these values are model dependent, they are insensitive to the
spatial component of the nucleon wave function; they follow directly
from its spin and flavor content. Moreover, the value of $g_{0}^{\pi}$
is compatible with other estimates available in the
literature~\cite{miller90} and, in particular, with the recent value
reported by Henley and Meissner from QCD sum rules~\cite{henmei96}.
To facilitate the interpretation of our results it is convenient
to transform the above expressions into configuration space:
\begin{mathletters}
\begin{eqnarray}
\Delta V^{\pi\eta}(r) &=&
{g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}
g_{\lower 2pt \hbox{$\scriptstyle NN\eta$}} \over 4\pi}
{\langle\pi|H|\eta\rangle \over m_{\eta}^2-m_{\pi}^2}
\left[
\left({m_{\eta} \over M}\right)^{2}{e^{-m_{\eta}r} \over r} -
\left({m_{\pi} \over M}\right)^{2}{e^{-m_{\pi}r } \over r}
\right] \;, \\
\Delta V^{\pi}(r) &=&
{g_{\lower 2pt \hbox{$\scriptstyle NN\pi$}}^{2}\over 4\pi}
g_0^{\pi}
\left[
\left({m_{\pi} \over M}\right)^{2}{e^{-m_{\pi}r} \over r} -
{\delta(r) \over M^{2}r^{2}}
\right] \;, \\
\Delta V^{\rho}(r) &=&
{g_{\lower 2pt \hbox{$\scriptstyle NN\rho$}}^{2}\over 4\pi}
(1+2f_1^{\rho})f_0^{\rho}
\left[
\left({m_{\rho} \over M}\right)^{2}{e^{-m_{\rho}r} \over r} -
{\delta(r) \over M^{2}r^{2}}
\right] \;, \\
\Delta V^{\omega}(r) &=&
{g_{\lower 2pt \hbox{$\scriptstyle NN\omega$}}^{2}\over 4\pi}
f_1^{\omega}
\left[
\left({m_{\omega} \over M}\right)^{2}{e^{-m_{\omega}r} \over r} -
{\delta(r) \over M^{2}r^{2}}
\right] \;.
\end{eqnarray}
\end{mathletters}
Note that in configuration space the ${}^{3}$He-${}^{3}$H
binding-energy difference emerges from a simple overlap between the
singlet component of the wave function [$R(a)$] and the CSB potential
\begin{equation}
\Delta E = - \int_{0}^{\infty} a^2 da |R(a)|^{2}
\Delta V(\sqrt{2}a) \equiv
\int_{0}^{\infty} I(a)da \;.
\label{iofa}
\end{equation}
We will refer to these estimates as the point-coupling results,
as hadronic form factor have yet to be introduced. We note,
as a consequence of the relativistic factor ${\bf q}^{2}/M^{2}$,
the appearance of a contact term [$\delta(r)$] in all single-meson
exchanges (the contact term is absent from the $\pi$-$\eta$ mixing
potential because of the additional meson propagator). For realistic
wave functions---such as the one used here---the contact term does
not contribute to the overlap as a result of the repulsive short-range
correlations. Thus, we conclude that all single-meson-exchange potentials
generate a repulsive $\Delta V$, namely, a CSB potential that is
attractive(repulsive) in the $pp(nn)$ channel---in contrast to the
experimental observation. Note that this conclusion depends
exclusively on the sign of the isospin-violating meson-nucleon
coupling constants; recall that in our model these constants are all
positive as a result of the up quark being lighter than the down quark.
This conclusion could change dramatically upon the inclusion of
hadronic form factors. With form factors the contact term gets
smeared over a distance of the order of $\Lambda^{-1}$ (with $\Lambda$
being the cutoff parameter) and could contribute to the overlap if this
distance becomes larger than the size of the repulsive core. Most of
our knowledge about hadronic form factor comes from phenomenological
fits to a large body of two-nucleon data~\cite{machl89}. These fits
reveal hadronic form factors that are typically harder
($\Lambda\!\agt\!1.5$~GeV) than the corresponding ones extracted
from electron-nucleon scattering experiments
($\Lambda\!\simeq\!0.8$~GeV). However, the subject of hadronic
form factors is highly controversial. Indeed, (model-dependent) analyses
of deep-inelastic-scattering (DIS) experiments suggest that the
$NN\pi$ form factor is considerably softer than
$\Lambda_{NN\pi}\!\sim\!1.5$~GeV~\cite{fms89,kum91,sst91}.
Moreover, Holinde and Thomas have shown recently that it is possible
to obtain an adequate description of two-nucleon data with the softer
pion-nucleon form factor ($\Lambda_{NN\pi}\!\simeq\!0.8$~GeV) suggested
by the DIS analyses~\cite{holtho90}. These authors have also shown that
the $NN\rho$ form factor can also be made softer---but no softer than
$\Lambda_{NN\rho}\!\simeq\!1.2$~GeV---without compromising the fit.
Hence, we find it instructive to do the calculation of the
${}^{3}$He-${}^{3}$H binding-energy shift using, both, the hard
form factors of Table~\ref{tableone} as well as the softer ones
suggested by the analyses of Refs.~\cite{fms89,kum91,sst91,holtho90}.
In the latter case we fix, for simplicity, both pseudoscalar cutoff
parameters at $\Lambda_{NN\pi(\eta)}=0.8$~GeV and both vector cutoff
parameters at $\Lambda_{NN\rho(\omega)}=1.3$~GeV. We incorporate hadronic
form factors into the calculation by modifying the point meson-nucleon
coupling in the following way~\cite{machl89}:
\begin{mathletters}
\begin{eqnarray}
g_{\lower 2pt \hbox{$\scriptstyle NN\pi(\eta)$}} \rightarrow
g_{\lower 2pt \hbox{$\scriptstyle NN\pi(\eta)$}}({\bf q}^{2})&=&
g_{\lower 2pt \hbox{$\scriptstyle NN\pi(\eta)$}}
(1+{\bf q}^2/\Lambda_{\pi(\eta)}^2)^{-1} \;, \\
g_{\lower 2pt \hbox{$\scriptstyle NN\rho(\omega)$}} \rightarrow
g_{\lower 2pt \hbox{$\scriptstyle NN\rho(\omega)$}}({\bf q}^{2})&=&
g_{\lower 2pt \hbox{$\scriptstyle NN\rho(\omega)$}}
(1+{\bf q}^2/\Lambda_{\rho(\omega)}^2)^{-2} \;.
\end{eqnarray}
\end{mathletters}
In Table~\ref{tabletwo} we report the contribution to the
${}^{3}$He-${}^{3}$H binding-energy difference arising
from $\pi$-$\eta$ mixing, $\pi-$, $\rho-$, and $\omega-$meson
exchange, respectively. For comparison, we have also included
the contribution arising from $\rho$-$\omega$ mixing, with the
mixing amplitude fixed at its on-shell value:
$\langle\rho|H|\omega\rangle=-4520$~MeV$^{2}$~\cite{coon87}.
The first column of numbers displays the
results in the point-coupling limit. As suggested, all single
meson-exchange potentials give rise to a negative binding-energy
shift; only $\pi$-$\eta$ mixing leads to a positive---albeit very
small---contribution. The binding energy-shifts computed with
the Bonn B hadronic form factors are reported in Table~\ref{tabletwo}
under the heading of hard form factors. Evidently, the binding-energy
shift is very sensitive to the short-distance behavior of the
potential; the inclusion of form factors results in a 40\%
reduction relative to the point-coupling estimate. This reduction
is best appreciated by plotting the integrand from which $\Delta E$
is obtained, that is $I(a)$ in Eq.~(\ref{iofa}). In Fig.~\ref{figone}
we display the contribution from one-pion exchange to $I(a)$; its
contribution to $\Delta E$ for each estimate, which is the area
under the curve, appears in parentheses next to its label. The
solid line represents the point-coupling result. In this case the
contact term does not contribute and the integrand is
negative-definite over the whole region. As the (hard) Bonn form
factor is introduced (dashed line) the smeared contact term dominates
the short-distance region and cancels some of the negative contribution
arising from the long-range term. It is then clear that for a
sufficiently soft form factor the node in the potential---arising
from the competition between the short- and long-range pieces---could
be moved to a large enough distance as to make the integral positive;
this is precisely what is observed when the softer form factor is
introduced (dot-dashed line). The complete set of numbers with soft
form factors has been collected on Table~\ref{tabletwo}. For completeness
we have also displayed in Fig.~\ref{figtwo} the corresponding integrand
for the omega meson. The same features are evident.
In the present case, however, the integral is negative in all three
cases because the vector form factor remains fairly hard. Nevertheless,
we should mention that even for the case of (perhaps) unrealistically
soft form factors, namely $\Lambda_{NN\pi(\eta)}=0.6$~GeV and
$\Lambda_{NN\rho(\omega)}=0.85$~GeV, our model can only account
for +40 keV of the ${}^{3}$He-${}^{3}$H binding-energy difference;
a result that is still one standard deviation away from the
experimental value.
In conclusion, we have shown that an analysis based on $\pi$-$\eta$
mixing and on isospin-violating meson-nucleon coupling constants---but
one without $\rho$-$\omega$ mixing---can not describe the
${}^{3}$He-${}^{3}$H binding-energy difference. However, we have
established that this finding is extremely sensitive to the
least understood feature of the NN potential, namely, its short-distance
structure. Thus, it is difficult to draw any definite conclusion until
an accurate determination of the cutoff parameters in the meson-nucleon
form factors is made---a very difficult task indeed.
\section{Binding-energy differences in A=15--41 mirror nuclei}
\label{sec:manybody}
In this section we examine the effect of meson mixing and of
isospin violation in the meson-nucleon coupling constants on
the binding-energy differences of medium-mass mirror nuclei
using a relativistic mean-field approximation to the Walecka
model~\cite{serwal86}. The Walecka model is a renormalizable
strong-coupling field theory of nucleons interacting via the
exchange of scalar and vector mesons. One of the great virtues
of the model is its simplicity. Indeed, already at the mean-field
level the Walecka model rivals some of the best available
nonrelativistic calculations and provides a unified description
of various nuclear properties, such as nuclear saturation, the
spin-orbit force, and the density dependence of the nuclear
interaction. The mean-field approximation is characterized by
the existence of large Lorentz scalar and vector components that
are responsible for a substantial in-medium enhancement of the
small (lower) components of the single-particle wave functions;
the so-called $M^{\star}$ effect.
\subsection{Vector-meson sector}
In our model $\rho$-$\omega$ mixing can not contribute to the
ONS anomaly, at least at $Q^{2}=0$. Yet, a calculation of
binding-energy shifts using the on-shell value for the
$\rho$-$\omega$ mixing amplitude is still useful, as it provides
a baseline against which alternate CSB mechanisms can be tested.
Thus, we start this section by computing the impact of the dominant
class III vector-vector component of
$\widehat{V}^{\rho\omega}_{\rm CSB}$ to the ONS anomaly.
The purpose of this exercise is to establish the size of the
phenomenological gap that will have to be filled by the alternate
CSB mechanisms. Note that we will not report the contribution to
the binding-energy differences arising from the tensor-vector
and tensor-tensor terms in Eq.~\ref{vveccsb}; for spin-saturated
nuclei their numerical impact is small (typically less than 10\%).
We compute the single-particle spectrum in a relativistic
mean-field approximation to the Walecka model~\cite{serwal86} and
incorporate CSB corrections in the Hartree-Fock approximation, i.e.,
%
\begin{equation}
\Delta E_{\alpha}^{(0)} \equiv \Delta E_{\alpha}^{H} +
\Delta E_{\alpha}^{F} \;,
\label{deltaerw}
\end{equation}
%
where
%
\begin{mathletters}
\begin{eqnarray}
\Delta E_{\alpha}^{H} &=& +4\sum_{\beta}^{\rm occ}
\int {d{\bf q} \over (2\pi)^{3}}
V_{\rm CSB}^{\rho\omega}({\bf q}) \;
\rho_{\alpha\alpha}^{(0)\star}({\bf q})
\rho_{\beta\beta}^{(0)}({\bf q}) \;, \\
\Delta E_{\alpha}^{F} &=& -4\sum_{\beta}^{\rm occ}
\int {d{\bf q} \over (2\pi)^{3}}
V_{\rm CSB}^{\rho\omega}({\bf q}) \;
\Big| \rho_{\beta\alpha}^{(0)}({\bf q}) \Big|^{2} \;.
\end{eqnarray}
\label{deltaevec}
\end{mathletters}
%
Note that the factor of four appearing in the above expressions
emerges from the characteristic class III isospin factor
$\Big[\tau_z(1)+\tau_z(2)\Big]$. In Eq.~(\ref{deltaevec}) the
sum runs over occupied orbitals in the Fermi sea, and we have
introduced the Fourier transform of the off-diagonal
timelike-vector density
%
\begin{equation}
\rho_{\beta\alpha}^{(0)}({\bf q}) =
\int d{\bf x} \;
\overline{{\cal U}}_{\beta}({\bf x})
\gamma^{0} e^{-i{\bf q}\cdot{\bf x}} \;
{\cal U}_{\alpha}({\bf x}) \;.
\label{rhovecba}
\end{equation}
%
Several remarks are in order. First, we have ignored the
contribution from the Dirac sea (negative-energy states)
in the evaluation of the single-particle spectrum as well
as in the CSB corrections to the energies. Second, for
spherically symmetric nuclei only the timelike part
[$\gamma^{0}(1)\gamma^{0}(2)$] of the potential contributes
to the Hartree term; for static sources and fields, current
conservation prohibits the appearance of a three-vector
current~\cite{serwal86}. Thus, as far as the Hartree term
is concerned, the CSB potential generated from
$\rho$-$\omega$ mixing behaves as a short-range Coulomb
potential. Third, even though the spacelike-part
[\mbox{\boldmath$\gamma$(1)}$\cdot$\mbox{\boldmath$\gamma$(2)}]
of the potential contributes to the Fock term, it is of little
numerical significance and has been neglected. Finally, we suppress
vector-meson production by ignoring retardation effects in the
meson propagators.
For spherically symmetric nuclei the calculation of the
binding-energy shifts simplifies considerably. In this limit
the eigenstates of the Dirac equation can be classified according
to a generalized angular momentum $\kappa$ and can be written in
a two component representation; i.e.,
%
\begin{equation}
{\cal U}_{\alpha}({\bf x}) \equiv
{\cal U}_{n\kappa m}({\bf x})= {1 \over x}
\left[
\begin{array}{c}
\phantom{i}
g_{a}(x)
{\cal{Y}}_{+\kappa m}(\hat{\bf{x}}) \\
i
f_{a}(x)
{\cal{Y}}_{-\kappa m}(\hat{\bf{x}})
\end{array}
\right] \;; \quad \Big(\alpha \equiv (a;m) = (n,\kappa;m)\Big) \;.
\label{dirwfx}
\end{equation}
%
The upper and lower components are expressed in terms of
spin-spherical harmonics defined by
%
\begin{equation}
{\cal{Y}}_{\kappa m}(\hat{\bf{x}}) \equiv
\langle\hat{\bf{x}}|l{\scriptstyle{1\over 2}}jm\rangle \;; \quad
j=|\kappa|-{1\over 2} \;; \quad
l=\cases{ \kappa\;, & if $\kappa > 0$; \cr
-1-\kappa\;, & if $\kappa < 0$. \cr}
\label{curlyy}
\end{equation}
%
The binding-energy shifts can now be easily evaluated. That is,
\begin{mathletters}
\begin{eqnarray}
\Delta E_{\alpha}^{H} &=& + {2 \over \pi^{2}}
\int q^{2} dq V_{\rm CSB}^{\rho\omega}(q)
{\cal F}_{aH}^{(0)}(q) \;, \\
\Delta E_{\alpha}^{F} &=& - {2 \over \pi^{2}}
\int q^{2} dq V_{\rm CSB}^{\rho\omega}(q)
{\cal F}_{aF}^{(0)}(q) \;,
\label{dehf}
\end{eqnarray}
\end{mathletters}
%
where $q \equiv |{\bf q}|$ and we have defined the following
quantities
%
\begin{mathletters}
\begin{eqnarray}
{\cal F}_{aH}^{(0)}(q) &\equiv& \sum_{b}^{\rm occ}
(2j_{b}+1) \rho_{bb0}^{(0)}(q) \rho_{aa0}^{(0)}(q) \;, \\
{\cal F}_{aF}^{(0)}(q) &\equiv& \sum_{b;L}^{\rm occ}
(2j_{b}+1) \Big[\langle j_{a},-1/2;j_{b},+1/2|L,0\rangle
\rho_{abL}^{(0)}(q)\Big]^{2} \;.
\label{curlyf}
\end{eqnarray}
\end{mathletters}
%
Note that the Fourier transform of the off-diagonal
timelike-vector density has become
%
\begin{equation}
\rho_{abL}^{(0)}(q) = \int_{0}^{\infty} dx
\Big[g_{a}(x)g_{b}(x)+f_{a}(x)f_{b}(x)\Big]j_{L}(qx) \;;
\quad (l_{a}+l_{b}+L={\rm even}) \;.
\label{rhoablvec}
\end{equation}
%
Using the Bonn B potential parameters of
Table~\ref{tableone}~\cite{machl89} we obtained the
binding-energy differences reported in
Table~\ref{tablethree}. These values are in qualitative
agreement with those reported by Blunden and
Iqbal~\cite{bluiqb87} and, more recently, by Barreiro,
Gale\~ao, and Krein~\cite{bgk95}. It is interesting to note
that, in contrast to the long-range Coulomb potential, the
Fock term generates a substantial exchange ``correction''
(of the order of 30-40\%) to the direct Hartree
contribution. Table~\ref{tablethree} also includes the
remaining discrepancy between experiment and three theoretical
calculations of the Coulomb displacement energy~\cite{kkb96,sato76}.
Note that the reported nonrelativistic discrepancies
($\Delta_{\rm DME}$ and $\Delta_{\rm SKII}$)
differ---in some cases by a large fraction---from the recent
relativistic estimates ($\Delta_{\rm REL}$) of Koepf, Krein, and
Barreiro~\cite{kkb96}.
\subsection{Pseudoscalar-meson sector: pseudoscalar $NN\pi$ coupling}
We shall now compute the binding-energy shifts that arise
from the CSB potential given in Eq.~(\ref{vpscsb}) using a
mean-field approximation to the Walecka model. Because of
the assumed pseudoscalar vertex, parity precludes a Hartree
correction to the energy~\cite{serwal86}. Thus, we only need
to evaluate the Fock term. That is,
\begin{equation}
\Delta E_{\alpha}^{({\rm PS})} =
+4\sum_{\beta}^{\rm occ}
\int {d{\bf q} \over (2\pi)^{3}}V_{\rm CSB}^{(5)}({\bf q}) \;
\Big| \rho_{\beta\alpha}^{({\rm PS})}({\bf q}) \Big|^{2} \;;
\label{deltaeps}
\end{equation}
%
where we have introduced the Fourier transform of the
off-diagonal pseudoscalar density
%
\begin{equation}
\rho_{\beta\alpha}^{({\rm PS})}({\bf q}) =
\int d{\bf x} \;
\overline{{\cal U}}_{\beta}({\bf x})
\gamma^{5} e^{-i{\bf q}\cdot{\bf x}} \;
{\cal U}_{\alpha}({\bf x}) \;.
\label{rhopsba}
\end{equation}
%
As in the previous section, the calculation simplifies
considerably in the limit of spherically symmetric nuclei.
In this limit, we obtain a binding-energy shift given by,
%
\begin{equation}
\Delta E_{\alpha}^{({\rm PS})} = + {2 \over \pi^{2}}
\int q^{2} dq V_{\rm CSB}^{(5)}(q)
{\cal F}_{a}^{({\rm PS})}(q) \;,
\label{deps}
\end{equation}
%
where
%
\begin{equation}
{\cal F}_{a}^{({\rm PS})}(q) \equiv \sum_{b;L}^{\rm occ}
(2j_{b}+1) \Big[\langle j_{a},-1/2;j_{b},+1/2|L,0\rangle
\rho_{abL}^{({\rm PS})}(q)\Big]^{2} \;,
\label{curlyfps}
\end{equation}
%
and with
%
\begin{equation}
\rho_{abL}^{({\rm PS})}(q) = \int_{0}^{\infty} dx
\Big[g_{a}(x)f_{b}(x)+f_{a}(x)g_{b}(x)\Big]j_{L}(qx) \;;
\quad (l_{a}+l_{b}+L={\rm odd}) \;.
\label{rhoablps}
\end{equation}
%
Note that the pseudoscalar density is, as expected, linear
in the small components of the wave functions and, thus,
highly sensitive to their in-medium enhancement. Moreover,
the induced energy shifts are positive definite---as
suggested by the ONS anomaly---for all single-particle states.
The pseudoscalar contribution to the binding-energy difference
of mirror nuclei is displayed in Table~\ref{tablethree}. We have
listed separately the contributions arising from isospin violation
in the pion-nucleon coupling constant and from $\pi$-$\eta$
mixing---with the former accounting for about 60\% of the total.
Note that relativistic effects induce an additional enhancement
in the binding-energy differences, relative to the nonrelativistic
estimates of Blunden and Iqbal~\cite{bluiqb87}. From
Table~\ref{tablethree} one observes that the binding-energy
differences from the pseudoscalar sector are comparable---if not
larger---than those computed originally with $\rho$-$\omega$ mixing.
Indeed, the pseudoscalar sector---alone---accounts for about 70-85\%
of the discrepancy between the nonrelativistic theory and experiment.
To gauge the importance of relativity we have also computed the
binding-energy differences by assuming that the lower component
of the wave function is determined from the free-space relation;
the upper component remained unchanged, apart from a small
normalization correction. These ``nonrelativistic'' values appear
in parenthesis in Table~\ref{tablefour} next to the uncorrected
numbers. The relativistic corrections are, indeed, very significant.
This is in sharp contrast to the relativistic enhancements---of
less than 5\%---observed for $\rho$-$\omega$ mixing; recall that
only the dominant vector-vector component has been included,
which is relatively insensitive to the enhancement of the small
components of the wave function. Perhaps the result that best
captures the essence of the present analysis is the
${}^{39}$Ca-${}^{39}$K binding-energy difference (the
$1d_{3/2}^{-1}$ state in Table~\ref{tablefour}). Our baseline
value for this difference is 65~keV; this represents our best
attempt at reproducing the result by Blunden and
Iqbal~\cite{bluiqb87}, namely, a nonrelativistic calculation with
$g_{0}^{\pi}\equiv 0$. However, introducing, both, relativistic
corrections and isospin violation in the pion-nucleon coupling
constant increases the baseline value to 435~keV; this represents
an enhancement of more than a factor of six.
\subsection{Pseudoscalar-meson sector: pseudovector $NN\pi$ coupling}
The above results suggest a large contribution from the pseudoscalar
sector to the ONS anomaly. Yet, a definite statement awaits, as these
results could be sensitive to off-shell extrapolations. Off-shell
ambiguities arise from the fact that different Lorentz structures for
the $NN$--pseudoscalar-meson vertex---which are all equivalent
on-shell---can generate vastly different predictions once the
nucleon is off its mass shell. Indeed, a pseudoscalar ($\gamma^{5}$)
vertex is very sensitive to the in-medium enhancement of the lower
components of the nucleon wave function, while a pseudovector coupling
[$\gamma^{5}\gamma^{\mu}(p'-p)_{\mu}/2M$] is not. In a model in which
chiral symmetry is not explicitly realized---such as the one employed
here---a pseudovector $NN\pi$ coupling seems to be preferred, as it is
known to incorporate the correct low-energy pion-nucleon dynamics without
sensitive cancellations~\cite{matser82}. In contrast, a recent analysis
of eta photoproduction data from the nucleon seems to favor a pseudoscalar
$NN\eta$ coupling~\cite{tiator94}.
In this section we adopt a pseudovector representation for both
($NN\pi$ and $NN\eta$) vertices. The advantage of such a choice
is that it provides us with a lower bound for the binding-energy
differences of mirror nuclei within the pseudoscalar sector. This,
combined with the upper bound set by the pseudoscalar representation,
should adequately reflect the sensitivity of our results to the various
off-shell extrapolations.
For the pseudovector case the angular-momentum algebra becomes
slightly more complicated than before due to the explicit spin
dependence of the operator. Recall that in the pseudoscalar case
the spin dependence is implicit through the upper-to-lower
coupling induced by the pseudoscalar vertex. The binding-energy
shifts computed with a pseudovector $NN\pi$ coupling are now given by
%
\begin{equation}
\Delta E_{\alpha}^{({\rm PV})} = + {2 \over \pi^{2}}
\int q^{2} dq \left({q^2 \over 4M^{2}}\right)
V_{\rm CSB}^{(5)}(q) {\cal F}_{a}^{({\rm PV})}(q) \;,
\label{depv}
\end{equation}
%
where
%
\begin{equation}
{\cal F}_{a}^{({\rm PV})}(q) \equiv \sum_{b;J}^{\rm occ}
(2j_{b}+1) \Big[\langle j_{a},-1/2;j_{b},+1/2|J,0\rangle
\rho_{abJ}^{({\rm PV})}(q)\Big]^{2} \;.
\label{curlyfpv}
\end{equation}
%
The pseudovector density has been computed in the static
[$(p'-p)^{0} \rightarrow 0$] limit and is given by
%
\begin{eqnarray}
\rho_{abJ}^{({\rm PV})}(q) = \sum_{L=J \pm 1}
\Big[
&&C_{JL}(+\kappa_{a},+\kappa_{b})
\int_{0}^{\infty} dx g_{a}(x)g_{b}(x)j_{L}(qx) + \nonumber \\
&&C_{JL}(-\kappa_{a},-\kappa_{b})
\int_{0}^{\infty} dx f_{a}(x)f_{b}(x)j_{L}(qx) \Big] \;;
\quad (l_{a}+l_{b}+L={\rm even}) \;,
\label{rhoabjpv}
\end{eqnarray}
where
%
\begin{equation}
C_{JL}(\kappa_{a},\kappa_{b}) \equiv
{(-1)^{(L+J+1)/2} \over (2J+1)}
\cases{L+(l_{a}-j_{a})(2j_{a}+1)+(l_{b}-j_{b})(2j_{b}+1)\;,
& if $L=J+1 \;;$ \cr
J-(l_{a}-j_{a})(2j_{a}+1)-(l_{b}-j_{b})(2j_{b}+1)\;,
& if $L=J-1 \;.$ \cr}
\label{ccoeff}
\end{equation}
Note that in the static limit a pseudovector vertex does not
induce an off-diagonal coupling. Hence, relativistic effects,
which appear as $(f/g)^{2}$ corrections, are negligible. Moreover,
the extra factor of $q^{2}$ that results from the pseudovector
coupling is expressed in units of the free nucleon mass $M$---not
in units of $M^{\star}$ (see Eq.~\ref{depv}). Thus, the binding-energy
differences computed with a pseudovector vertex could be reproduced
with a pseudoscalar vertex---only in the case in which the lower
component of the wave function be determined from the free-space
relation (see Table~\ref{tablefour}). Thus, the square of the
effective mass in the medium relative to its free space value
(i.e., $M^{\star 2}/M^{2}$) should serve as an indicator of the
sensitivity of the approach to the various off-shell extrapolations.
Note that in a mean-field approximation to the Walecka model
$M^{\star 2}/M^{2}$ is typically of the order of 40\%. Thus,
off-shell ambiguities are expected to be large
(see Table~\ref{tablefive}).
The pseudovector contribution to the binding-energy difference
of mirror nuclei is displayed, alongside the pseudoscalar
results, in Table~\ref{tablefive}. As suggested previously,
these results are very close to those obtained in
Table~\ref{tablefour} using the free upper-to-lower ratio.
More significantly, however, is the very dramatic reduction
in the binding-energy differences relative to the pseudoscalar
results. Indeed, while a calculation with a pseudoscalar vertex
accounts for about 70-85\% of the ONS anomaly, only 30\% of it can
be explained with a pseudovector vertex. It is important to stress
that although some theoretical guidelines do exist, off-shell
ambiguities might be difficult to resolve given that the contribution
from the isospin-violating part of the potential to hadronic
observables is small.
\section{Conclusions}
\label{sec:concl}
Binding-energy differences of mirror nuclei have been computed
in a model that includes two sources of charge-symmetry-breaking
in the NN potential; isoscalar-isovector mixing in the meson
propagator and isospin violation in the meson-nucleon coupling
constants. For the vector-meson sector we have used a VMD-inspired
model. Thus, on the basis of this assumption---but little else---we
concluded that neither $\rho$-$\omega$ mixing nor the dominant
vector-vector component of the vector-meson-exchange potential can
contribute to the binding-energy discrepancy between theory and
experiment. In models of this sort this conclusion seems
unavoidable---as gauge invariance forces, both, the $\rho$-$\omega$
mixing amplitude and the isospin-violating vector coupling to vanish
at $Q^{2}=0$. This is in contrast to previous analyses that employed
the on-shell value of the $\rho$-$\omega$ mixing amplitude to explain
the bulk of the ONS anomaly~\cite{miller90,bluiqb87,coon87,ishsas90}.
Hence, in our model the anomaly must be explained exclusively in
terms of $\pi$-$\eta$ mixing and of isospin violation in the pion-,
rho-, and omega-nucleon coupling constants.
To compute the ${}^{3}$He-${}^{3}$H binding-energy difference a
nonrelativistic estimate of the ${}^{1}S_{0}$ component of the CSB
potential was obtained. In our model all sources of CSB generate
a ${}^{1}S_{0}$ interaction with a long-range part that is
attractive in the $pp$ channel and repulsive in the $nn$ channel.
This result is robust, as it only depends on the sign of the
$\pi$-$\eta$ mixing amplitude and on the sign of the
isospin-violating meson-nucleon coupling constants; recall that
in our model the latter is determined from the up-down quark mass
difference, which we assume to be negative. The ${}^{3}$He-${}^{3}$H
binding-energy difference, however, was seen to be very sensitive
to the short-distance behavior of the potential. Indeed, the
binding-energy difference emerged from a delicate interplay between
the short- and long-range parts of the potential. This interplay
introduces a node in the CSB potential which can be shifted to a
large enough distance as to make the binding-energy difference
almost consistent with experiment. We stress that the (almost)
agreement with experiment could only be achieved at the expense
of introducing unrealistically soft form factors.
For the medium-mass (A=15--41) region we computed the binding-energy
difference of mirror nuclei in a relativistic mean-field approximation
to the Walecka model. Due to the larger size of the system, relative
to ${}^{3}$He, our results were fairly insensitive to the short-distance
structure of the potential. Moreover, explicit spin-dependent effects
are known to be small for these spin-saturated nuclei. As a consequence,
the numerical impact from the vector-meson sector on the ONS anomaly was
seen to be small. We proposed pseudoscalar-meson exchange as an
alternate mechanism to explain the ONS anomaly in the medium-mass region.
We computed binding-energy differences that were comparable---and in most
cases larger---than those obtained from previous estimates using on-shell
$\rho$-$\omega$ mixing. We have concluded that the pseudoscalar
sector---alone---could explain about 70-85\% of the ONS anomaly. Two
effects, ignored until now, were responsible for these findings: a) the
inclusion of isospin violation in the pion-nucleon coupling constant,
and b) relativity, which enhances the small components of the bound-state
wave functions relative to their free-space value. We reached these
conclusions by assuming a pseudoscalar coupling for, both, the $NN\pi$
and $NN\eta$ vertices. Yet other choices---all of them equivalent
on-shell---are possible; such as pseudovector coupling. There is no
definite prescription on how to take the CSB potential off-shell.
Moreover, this issue is complicated further by the fact that chiral
symmetry favors a pseudovector representation for the $NN\pi$ vertex
while a recent analysis of $\eta$-photoproduction data seems to suggest
a pseudoscalar $NN\eta$ coupling. Hence, to gauge the sensitivity of our
results to the various off-shell extrapolations we repeated the calculation
by adopting a pseudovector representation for both vertices. We concluded
that the off-shell ambiguities were large; while 70-85\% of the anomaly
could be explained with a pseudoscalar coupling only 30\% of it could
be accounted for with a pseudovector vertex.
In conclusion, we have used a VMD-inspired model to examine the
effect of isospin-violating meson-nucleon coupling constants and
of $\pi$-$\eta$ mixing on the binding-energy difference of mirror
nuclei. We could account for most of the ${}^{3}$He-${}^{3}$H
binding-energy difference---provided very soft form factors were
assumed. Moreover, we could explain the ONS anomaly in the
medium-mass region---provided a pseudoscalar representation was
adopted. Based on these findings we expect that the
Okamoto-Nolen-Schiffer anomaly will remain---even after more
than a quarter of a century---the source of considerable theoretical
debate.
\acknowledgments
We thank S. Gardner, C.J. Horowitz, D. Robson, and B.D. Serot
for many helpful conversations. We acknowledge the generosity
of E. Pace for providing us with the three-body wave function.
This work was supported by the DOE under Contracts Nos.
DE-FC05-85ER250000, DE-FG05-92ER40750 and DE-FG05-86ER40273.
\begin{references}
\bibitem{henmil79} E.M.~Henley and G.A.~Miller, in {\it Mesons in
Nuclei}, eds. M.~Rho and D.H.~Wilkinson
(North-Holland, Amsterdam 1979).
\bibitem{miller90} G.A.~Miller, B.M.K.~Nefkens, and I.~\v{S}laus,
Phys.~Rep.~{\bf 194} (1990) 1.
\bibitem{ght92} T.~Goldman, J.A.~Henderson, and A.W.~Thomas,
{\it Few-Body Systems} {\bf 12} (1992) 123;
Mod. Phys. Lett. {\bf A7} (1992) 3037.
\bibitem{piewil93} J. Piekarewicz and A.G.~Williams,
Phys.~Rev.~C~{\bf 47} (1993) R2462.
\bibitem{krein93} G. Krein, A.W. Thomas, and A.G. Williams,
Phys. Lett. B {\bf 317} (1993) 293.
\bibitem{hats94} T.~Hatsuda, E.M.~Henley, T.~Meissner, and
G.~Krein, Phys.~Rev.~C~{\bf 49} (1994) 452.
\bibitem{mitch94} K.L.~Mitchell, P.C.~Tandy, C.D.~Roberts,
and R.T.~Cahill,
Phys. Lett. B~{\bf 335} (1994) 282.
\bibitem{oconn94} H.B.~O'Connell, B.C.~Pearce, A.W.~Thomas, and
A.G.~Williams, Phys.~Lett.~B~{\bf 336} (1994) 1.
\bibitem{malta95} K.~Maltman, University of Adelaide preprint
ADP-95-20/T179 and hep-ph/9504237.
\bibitem{miller94} G.A.~Miller, Chin. J. Phys. {\bf 32} (1994) 1075;
G.A.~Miller and W.T.H.~van Oers nucl-th/9409013.
\bibitem{oconn95} H.B.~O'Connell, B.C.~Pearce, A.W.~Thomas, and
A.G.~Williams, University of Adelaide preprint
ADP-95-15/T176 and hep-ph/9503332.
\bibitem{ghp95} S. Gardner, C.J. Horowitz, and J. Piekarewicz,
Phys.~Rev.~Lett.~{\bf 75} (1995) 2462;
Phys.~Rev.~C~{\bf 53} (1996) 1143.
\bibitem{piekar95} J. Piekarewicz,
Phys.~Lett.~B~{\bf 358} (1995) 27.
\bibitem{gak95} W.R. Gibbs, Li Ai, and W.B. Kaufmann,
Phys. Rev. Lett.~{\bf 74} (1995) 3740.
\bibitem{knut90} L.D.~Knutson {\it et al.,}
Nucl.~Phys.~{\bf A508} (1990) 185c;
Phys.~Rev.~Lett.~{\bf 66} (1991) 1410;
S.E.~Vigdor {\it et al.,}
Phys.~Rev.~C~{\bf 46} (1992) 410.
\bibitem{slaus89} I.~{\v S}laus, Y.~Akaishi, and H.~Tanaka, Phys. Rep.
{\bf 173} (1989) 257.
\bibitem{oka64} K. Okamoto,
Phys.~Lett.~{\bf 11} (1964) 150;
K. Okamoto and C. Pask, Ann. of Phys.
{\bf 68} (1971) 18.
\bibitem{nolsch69} J.A. Nolen Jr. and J.P Schiffer,
Ann. Rev. Nucl. Sci.~{19} (1969) 471.
\bibitem{bluiqb87} P.G. Blunden and M.J.~Iqbal, Phys.~Lett.~B~{\bf 198}
(1987) 14.
\bibitem{coon87} S.~A. Coon and R.~C. Barrett, Phys. Rev. C~{\bf 36}
(1987) 2189.
\bibitem{ishsas90} Y.W. Ishikawa, and T. Sasakawa, Phys. Rev. Lett.
{\bf 64} (1990) 1875; Errata Phys. Rev. Lett.
{\bf 66} (1991) 242.
\bibitem{pace77} P. Numberg, D. Prosperi and E. Pace, Nucl. Phys.
{\bf A285} (1977) 58.
\bibitem{serwal86} J.D.~Walecka, Ann. of Phys. {\bf 83} (1974) 491;
B.D. Serot and J.D. Walecka, Adv. in Nucl. Phys.
{\bf 16}, J.W. Negele and E. Vogt, eds.
(Plenum, N.Y. 1986).
\bibitem{nedroo88} Y. Nedjadi and J.R. Rook,
Nucl. Phys. {\bf A484} (1988) 525.
\bibitem{bgk95} L.A. Barreiro, A.P. Gale\~ao, and G. Krein,
Phys.~Lett.~B~{\bf 358} (1995) 7.
\bibitem{kkb96} W. Koepf, G. Krein, and L.A. Barreiro,
nucl-th/9601027.
\bibitem{machl89} R.~Machleidt, in {\it Advances in Nuclear Physics},
v.~19, eds. J.W.~Negele and E.~Vogt (Plenum, New
York, 1989).
\bibitem{dmitra95} V.~Dmitra\v{s}inovi\'c and S.J.~Pollock,
Phys.~Rev.~C~{\bf 52} (1995) 1061.
\bibitem{maltb95} Kim Maltman,
Phys.~Lett.~B~{\bf 362} (1995) 11.
\bibitem{coon82} Sidney A. Coon and Michael D. Scadron,
Phys. Rev. C~{\bf 26} (1982) 562.
\bibitem{tiator94} L. Tiator, C. Bennhold, and S.S. Kamalov,
nucl-th/9404013.
\bibitem{piekar93} J. Piekarewicz, Phys. Rev. C~{\bf 48} (1993) 1555.
\bibitem{friar70} J.L. Friar, Nucl. Phys. {\bf A156} (1970) 43;
M. Fabre de la Ripelle, Fizika {\bf 4} (1972) 1;
R.A. Brandeburg, S.A. Conn, and P.U. Sauer,
Nucl. Phys. {\bf A294} (1978) 305; J.L. Friar
and B.F. Gibson, Phys. Rev. C~{\bf 18} (1978) 908.
\bibitem{reid68} R.V. Reid, Ann. Phys. {\bf 50} (1968) 411.
\bibitem{ciofi79} C. Ciofi degli Atti, E. Pace, and G. Salm\`e, Studies
in High Energy Physics, vol. 2, Horwood Academic
Publishers, 1979, 369; E. Pace, Lett. Nuovo
Cimento~{\bf 26} (1979) 615; C. Ciofi degli Atti,
E. Pace, and G. Salm\`e, Nucl. Phys. {\bf A358}
(1981) 183c.
\bibitem{coon75} Peter C. McNamee, Michael D. Scadron, and Sidney
A. Coon, Nucl. Phys. {\bf A249} (1975) 483.
\bibitem{licht89} D.~B.~Lichtenberg, Phys. Rev. D~{\bf 40} (1989) 3675.
\bibitem{henmei96} E.M. Henley and T. Meissner, nucl-th/9601030.
\bibitem{fms89} L. Frankfurt, L. Mankiewicz, M.I. Strikman,
Z. Phys. {\bf A334} (1989) 343.
\bibitem{kum91} S. Kumano, Phys. Rev. D {\bf 43} (1991) 59.
\bibitem{sst91} A. Signal, A.W. Schreiber, and A.W. Thomas,
Mod. Phys. Lett. {\bf A6} (1991) 271.
\bibitem{holtho90} K. Holinde and A.W. Thomas, Phys. Rev. C {\bf 42}
(1990) R1195.
\bibitem{sato76} Hiroshi Sato,
Nucl. Phys. {\bf A269} (1976) 378.
\bibitem{matser82} T. Matsui and B.D. Serot, Ann. Phys. {\bf 144}
(1982) 107.
\end{references}
%%%%%%%%%%%%%%%%%%%%% Figures %%%%%%%%%%%%%%%%%%%%
\begin{figure}
\caption{The integrand for the ${}^{3}$He-${}^{3}$H
binding-energy difference arising from
one-pion exchange for three different choices
of form factors: $\Lambda_{NN\pi}\rightarrow\infty$
(solid line), $\Lambda_{NN\pi}=1.7$~GeV (dashed line)
$\Lambda_{NN\pi}=0.8$~GeV (dot-dashed line). The
one-pion exchange contribution to $\Delta E$, which
is the area under the appropriate curve, appears in
parentheses next to its label.}
\label{figone}
\end{figure}
\begin{figure}
\caption{The integrand for the ${}^{3}$He-${}^{3}$H
binding-energy difference arising from
one-omega exchange for three different choices
of form factors: $\Lambda_{NN\omega}\rightarrow\infty$
(solid line), $\Lambda_{NN\omega}=1.85$~GeV (dashed line)
$\Lambda_{NN\omega}=1.3$~GeV (dot-dashed line). The
one-omega exchange contribution to $\Delta E$, which
is the area under the appropriate curve, appears in
parentheses next to its label.}
\label{figtwo}
\end{figure}
%%%%%%%%%%%%%%%%%%%%% Tables %%%%%%%%%%%%%%%%%%%%
\mediumtext
\begin{table}
\caption{Meson masses, coupling constants, tensor-to-vector ratio,
and cutoff parameters of the Bonn B potential.}
\begin{tabular}{ccccc}
Meson & Mass (MeV) & $g^{2}/4\pi$ & $f/g$ & $\Lambda$~(MeV) \\
\tableline
$\pi$ & 138 & 14.21 & --- & 1700 \\
$\eta$ & 549 & 0.90 & --- & 1700 \\
$\rho$ & 769 & 0.42 & 6.1 & 1850 \\
$\omega$ & 783 & 11.13 & 0.0 & 1850 \\
\end{tabular}
\label{tableone}
\end{table}
\mediumtext
\begin{table}
\caption{Contribution to the ${}^{3}$He-${}^{3}$H binding-energy
difference (in keV) arising from $\pi$-$\eta$ mixing,
$\pi-$, $\rho-$, and $\omega-$meson exchange for three
different values of the cutoff parameters; see text for
details. The experimental value is $\Delta E =
71\pm19\pm5$~keV. For comparison, we also include the
contribution from on-shell $\rho$-$\omega$ mixing.}
\begin{tabular}{cccc}
& Point coupling & Hard form factors & Soft form factors \\
\tableline
$\pi$-$\eta$ & $ +7.4$ & $+13.6$ & $+22.2$ \\
$\pi$ & $ -27.9$ & $-16.2$ & $+13.0$ \\
$\rho$ & $ -49.4$ & $-37.3$ & $-20.4$ \\
$\omega$ & $ -53.8$ & $-40.9$ & $-22.4$ \\
Total & $-123.7$ & $-80.8$ & $ -7.6$ \\
\tableline
$\rho$-$\omega$ & $ +57.9$ & $+77.9$ & $+87.3$ \\
\end{tabular}
\label{tabletwo}
\end{table}
\mediumtext
\begin{table}
\caption{Contribution to the binding-energy differences of mirror
nuclei (in keV) arising from isospin violation in the
pion-nucleon coupling constant, $\pi$-$\eta$ mixing, their
sum, and on-shell $\rho$-$\omega$ mixing. Also included are
the remaining differences ($\Delta$) between experiment and
the Coulomb energy computed in three different
models\protect{\cite{kkb96,sato76}}.}
\begin{tabular}{cc|ccc|ccc|ccc}
& &
\multicolumn{3}{c|}{Pseudoscalar} &
\multicolumn{3}{c|}{Rho-omega} &
\multicolumn{3}{c} {ONS Anomaly} \\ \hline
A & $\phantom{xx}$State$\phantom{xx}$ &
$\delta g_{\scriptscriptstyle{NN}\pi}$ &
$\pi$-$\eta$ & $\phantom{xx}$Total$\phantom{xx}$ &
Hartree & Fock & $\phantom{xx}$Total$\phantom{xx}$
& $\phantom{x}$$\Delta_{\rm DME}$$\phantom{x}$
& $\Delta_{\rm SKII}$$\phantom{x}$
& $\Delta_{\rm REL}$$\phantom{x}$ \\
\tableline
15 & $1p_{1/2}^{-1}$ & 201 & 145 & 346 & 379
& $-$149 & 230 & 380 & 290
& 95 \\
17 & $1d_{5/2}$ & 83 & 56 & 139 & 246
& $-$91 & 155 & 300 & 190
& 79 \\
27 & $1d_{5/2}^{-1}$ & 223 & 148 & 371 & 445
& $-$176 & 269 & 480 & 490
& 551 \\
29 & $2s_{1/2} $ & 165 & 108 & 273 & 305
& $-$91 & 214 & 290 & 240
& 34 \\
31 & $2s_{1/2}^{-1}$ & 202 & 134 & 336 & 355
& $-$119 & 236 & 540 & 560
& 39 \\
33 & $1d_{3/2}$ & 159 & 113 & 272 & 363
& $-$121 & 242 & 360 & 280
& 213 \\
39 & $1d_{3/2}^{-1}$ & 255 & 180 & 435 & 455
& $-$169 & 286 & 540 & 430
& 404 \\
41 & $1f_{7/2} $ & 133 & 89 & 222 & 355
& $-$129 & 226 & 440 & 350
& 404 \\
\end{tabular}
\label{tablethree}
\end{table}
\mediumtext
\begin{table}
\caption{Contribution to the binding-energy differences of mirror
nuclei (in keV) arising from isospin violation in the
pion-nucleon coupling constant, $\pi$-$\eta$ mixing, their
sum, and on-shell $\rho$-$\omega$ mixing. Included in
parenthesis are the corresponding quantities computed
with lower components generated from the free-space
relation.}
\begin{tabular}{cccccc}
A & State & $\delta g_{\scriptscriptstyle{NN}\pi}$
& $\pi$-$\eta$ & Total & Rho-omega \\
\tableline
39 & $1d_{5/2}^{-1}$ & 205~(103) & 142~(73)
& 347~(176) & 294~(291) \\
& $2s_{1/2}^{-1}$ & 233~(84) & 158~(58)
& 391~(142) & 283~(271) \\
& $1d_{3/2}^{-1}$ & 255~(92) & 180~(65)
& 435~(157) & 286~(275) \\
41 & $1f_{7/2} $ & 133~(78) & 89~(53)
& 222~(131) & 226~(224) \\
\end{tabular}
\label{tablefour}
\end{table}
\mediumtext
\begin{table}
\caption{Contribution to the binding-energy differences of mirror
nuclei (in keV) arising from isospin violation in the
pion-nucleon coupling constant, $\pi$-$\eta$ mixing, and
their sum. The first(second) set of numbers were computed
using a pseudoscalar(pseudovector) $NN\pi$ vertex.}
\begin{tabular}{cc|ccc|ccc}
& &
\multicolumn{3}{c|}{Pseudoscalar} &
\multicolumn{3}{c}{Pseudovector} \\ \hline
A & $\phantom{xx}$State$\phantom{xx}$ &
$\delta g_{\scriptscriptstyle{NN}\pi}$ &
$\pi$-$\eta$ & $\phantom{xx}$Total$\phantom{xx}$ &
$\delta g_{\scriptscriptstyle{NN}\pi}$ &
$\pi$-$\eta$ & $\phantom{xx}$Total$\phantom{xx}$ \\
\tableline
15 & $1p_{1/2}^{-1}$ & 201 & 145 & 346
& 67 & 49 & 116 \\
17 & $1d_{5/2}$ & 83 & 56 & 139
& 52 & 36 & 88 \\
27 & $1d_{5/2}^{-1}$ & 223 & 148 & 371
& 109 & 74 & 183 \\
29 & $2s_{1/2} $ & 165 & 108 & 273
& 54 & 36 & 90 \\
31 & $2s_{1/2}^{-1}$ & 202 & 134 & 336
& 62 & 43 & 105 \\
33 & $1d_{3/2}$ & 159 & 113 & 272
& 49 & 35 & 84 \\
39 & $1d_{3/2}^{-1}$ & 255 & 180 & 435
& 83 & 59 & 142 \\
41 & $1f_{7/2} $ & 133 & 89 & 222
& 74 & 51 & 125 \\
\end{tabular}
\label{tablefive}
\end{table}
\end{document}