% Chapter on Jupiter's interior for Jupiter book
% Submitted version: 25 April 2002

\documentclass{jupiter}
\bibliographystyle{apalike}
\usepackage{graphicx}
%\usepackage{wasysym}
%\include{psfig}

% If your system has the AMS fonts version 2.0 installed, jupiter.cls can be
% made to use them by uncommenting the line: \AMSfontstrue
%
% By doing this, you will be able to obtain upright Greek characters.
% e.g. \umu, \upi etc.  See the section on "Upright Greek characters" in
% this guide for further information.
%
% If you are using AMS 2.0 fonts, bold math letters/symbols are available

\newif\ifAMSfonts
\AMSfontstrue %uncomment if you have AMS fonts

%%%%% AUTHORS - PLACE YOUR OWN MACROS HERE %%%%%

\def\beq{\begin{equation}}
\def\eeq{\end{equation}}

\def\eref#1{(\ref{eq:#1})}
\def\mea{\mbox{M}_\oplus}
\def\rm{\mathrm}
\def\bf{\mathbf}
\def\itt{\textit}
\def\etal{{et al.}}
\def\ie{{i.e.}}
\def\eg{{e.g.}}
\def\HH{H$_2$}
\def\hho{H$_2$O}
\def\meth{CH$_4$}
\def\amm{NH$_3$}
\def\mic{\,$\mu$m}
\def\bar{\,\rm{bar}}
\def\K{{\,\mbox{K}}}
\def\cm{{\,\rm{cm}}}
\def\gcc{\,\rm{g\,cm^{-3}}}
\def\rgas{{R}}
\def\ti#1{$^{{#1}}$}
\def\cm{\,\mbox{cm}}
\def\sec{\,\mbox{s}}
\def\m{\,\mbox{m}}
\def\bfnabla{\mathbf{\nabla}}
\def\hp{H_P}
\def\req{R_\rm{eq}}
\def\rme{\rm{e}}
\def\zmet{Z_\rm{met}}
\def\zmol{Z_\rm{mol}}
\def\mcore{M_\rm{core}}
\def\nabt{{\nabla_T}}
\def\nabmu{\nabla_\mu}
\def\nabe{{\nabla_\rm{e}}}
\def\nabad{{\nabla_\rm{ad}}}
\def\nabrad{{\nabla_\rm{rad}}}
\def\dnab{(\nabla_T-\nabla_\rm{e})}
\def\teff{T_\rm{eff}}
\def\kcrit{\kappa_\rm{crit}}
\def\hp{H_P}
\def\mea{\mbox{M}_\oplus}
%\def\rjup{\rm{R}_{\rm{J}}
\def\eref#1{(\ref{eq:#1})}
\def\rfig#1{fig.~\ref{fig:#1}}
\def\rfigure#1{Figure~\ref{fig:#1}}
\def\rjup{{\small${\overline\rm{R}}_\rm{J}$}}
\def\wig#1{\mathrel{\hbox{\hbox to 0pt{%
          \lower.6ex\hbox{$\sim$}\hss}\raise.4ex\hbox{$#1$}}}}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Journals
%%%%%%%%%%%%%%%%%%%
\def\aj{AJ}
\def\apj{ApJ}
\def\apjs{ApJS}
\def\apjl{ApJL}
\def\jgr{JGR}
\def\aap{A\&A}
\def\aaps{A\&AS}
\def\araa{Ann. Rev. Astron. Astrophys.}
\def\grl{Geophys. Res. Let.}
\def\jcp{Jour. Chem. Phys.}
\def\mnras{MNRAS}
\def\nat{Nature}
\def\pra{Phys. Rev. A}
\def\prb{Phys. Rev. B}
\def\pre{Phys. Rev. E}
\def\prl{Phys. Rev. Let.}
\def\planss{Plan. Space Sci.}
%%%%%%%%%%%%%%%%%%%%%

  \providecommand\boldsymbol[1]{\mbox{\boldmath $#1$}}
  \def\textbfit{\protect\txtbfit}
  \def\textbfss{\protect\txtbfss}
  \long\def\txtbfit#1{{\fontfamily{cmr}\fontseries{bx}\fontshape{it}%
    \selectfont #1}}
  \long\def\txtbfss#1{{\fontfamily{cmss}\fontseries{bx}\fontshape{n}%
    \selectfont #1}}
  %
  \DeclareMathAlphabet{\mathbfit}{OT1}{cmr}{bx}{it}
  \DeclareMathAlphabet{\mathbfss}{OT1}{cmss}{bx}{n}
  %
  \ifAMSfonts
    \DeclareSymbolFont{UPM}{U}{eur}{m}{n}
    \DeclareSymbolFont{AMSa}{U}{msa}{m}{n}
    \DeclareMathSymbol{\upi}{0}{UPM}{"19}
    \DeclareMathSymbol{\umu}{0}{UPM}{"16}
    \DeclareMathSymbol{\upartial}{0}{UPM}{"40}
    \DeclareMathSymbol{\leqslant}{3}{AMSa}{"36}
    \DeclareMathSymbol{\geqslant}{3}{AMSa}{"3E}
    %
    \let\oldle=\le     \let\oldleq=\leq
    \let\oldge=\ge     \let\oldgeq=\geq
    \let\leq=\leqslant \let\le=\leqslant
    \let\geq=\geqslant \let\ge=\geqslant
    %
  \else % if you don't have AMS fonts
    \def\upi{\pi}
    \def\umu{\mu}
    \def\upartial{\partial}
  \fi

% for guide only
\long\def\boxit#1{\noindent\ignorespaces
  \framebox[\hsize][l]{\hbox{\vbox{\raggedright #1\par}}}\par
  \medskip\noindent\ignorespaces
} 
\makeatletter
% define \thebiblio (same as thebibliography, but
% without the section heading)
\def\thebiblio#1{%
 \list{}{\usecounter{dummy}%
         \labelwidth\z@
         \leftmargin 1.5em
         \itemsep \z@
         \itemindent-\leftmargin}
 \reset@font\small
 \parindent\z@
 \parskip\z@ plus .1pt\relax
 \def\newblock{\hskip .11em plus .33em minus .07em}
 \sloppy\clubpenalty4000\widowpenalty4000
 \sfcode`\.=1000\relax
}
\let\endthebiblio=\endlist
\makeatother
% end of guide macros


\chapter{3}% insert your chapter number here
\title[Jupiter's interior]
  {The interior of Jupiter}
\author[T. Guillot et al.]
  {Tristan Guillot\\ 
  Observatoire de la C\^ote d'Azur, CNRS UMR 6529
  \and
  David J. Stevenson\\
  California Institute of Technology
  \and
  William B. Hubbard\\
  Lunar and Planetary Laboratory, University of Arizona
  \and
  Didier Saumon\\
  Los Alamos National Laboratory
}

\begin{document}
\maketitle

\section{Introduction}

Jupiter, owing to its large mass and rapid formation, played a crucial
role in shaping the solar system as we know it today. Jupiter mostly
contains hydrogen and helium (more than 87\% by mass), and as such
bears a close resemblance to the Sun. However, the Sun has only 2\% of
its mass in elements other than hydrogen and helium (the \itt{heavy
elements}), whereas Jupiter has between 3 and 13\%. The exact amount
of these heavy elements in the planet and their distribution are keys
to understand how the solar system formed. 

Yet, it would seem that since the first Jupiter book was published, more
than twenty five years ago, there has been little qualitative
change to our vision of the interior of Jupiter, as a planet with a
central dense core and a
surrounding hydrogen and helium envelope (Stevenson \& Salpeter 1976;
Hubbard \& Slattery 1976). Fortunately, several factors have
led to significant quantitative improvements to that picture. 
Jupiter's gravity field has been measured with a better accuracy by
the Voyager flybys in 1979, thereby yielding stronger constraints on
the interior models. Our understanding of its atmosphere has been
steadily improved, in particular by the \itt{in situ}
measurements of the Galileo probe in 1995, but also by the Galileo and
Cassini missions, and by more accurate ground-based observations. On
the experimental side, hydrogen (actually deuterium)
has been successfully compressed to pressures up to several
Mbar. Although the latest experiments remain controversial,
this has generally led to the calculation of improved equations of
state, a crucial ingredient for the calculation of interior models of
the giant planet. Last but not least, the discovery of giant
planets in orbit around other stars and of the related
\itt{brown dwarfs} has motivated more detailed studies of the
evolution of substellar objects, with direct applications to Jupiter.

To first order, Jupiter's interior can be described by simple
arguments. Jupiter is a hydrogen-helium planet in hydrostatic
equilibrium. Its interior is warm ($\sim 20000\,$K) because it formed
from an extended gas cloud whose gravitational energy was converted
into heat upon contraction.  (It is still contracting at the rate of
$\sim 3$\,cm per year while its interior cools by $\sim 1$\,K per
million year.) This has several important consequences: The relatively
warm conditions imply that Jupiter's interior is \itt{fluid}, not
solid. The cooling and contraction yield a significant intrinsic
energy flux (revealed by the fact that Jupiter emits more energy than
it receives from the Sun) that drives convection in most parts of the
interior. Convection ensures the planet's homogeneity and generates
the observed magnetic field through a dynamo mechanism.

Were the above description entirely true, one would be able to derive
the planet's composition directly from the determination of the
atmospheric abundances.  However, several factors contribute to a more
complex picture of Jupiter's interior. As discussed in
Section~\ref{sec:data}, the observation of the planet's atmosphere
indicates that several major chemical species (such as helium, neon
and water) are partly sequestered into the interior. In the interior,
the degenerate nature of the electrons and the Coulomb interactions
between ions can be responsible for phase transitions and/or phase
separations, synonymous of chemical inhomogeneities
(Section~\ref{sec:eos}). Energy transport is complicated by the
possibility of radiative transport of the intrinsic heat flux in some
regions, while convection itself is complicated by the presence of
molecular weight gradients and by intricate coupling with rotation and
magnetic fields (Section~\ref{sec:heat}). Finally, interior models
based on the measurements of the planet's gravity field generally (but
not always) require the presence of a central, dense core of uncertain
mass and composition (Section~\ref{sec:interior}). As shown in
Section~\ref{sec:evolution}, this has major consequences for the
planet's evolution and our understanding of its formation. Answering
the most fundamental questions concerning Jupiter's origin (and by
extension, the origin of the Solar System) requires a renewed
exploration of this planet.  

%\hfill
%% The observational data used for interior modeling are examined in
%% section~\ref{sec:data}, with particular emphasis on results of the
%% Galileo mission. We then describe the latest high-pressure experiments
%% and related equations of state issues and, in Section~\ref{sec:heat},
%% opacities and heat transport in Jupiter's interior. Interior models are
%% presented in section~\ref{sec:interior} and evolution scenarios in
%% section~\ref{sec:evolution}. The last section is devoted to future
%% prospects, which we feel should be bright for Jupiter's exploration.


\section{Available data}
\label{sec:data}

\subsection{Gravity field and global properties}

Jupiter is a rapid rotator: its rotation period derived from
its magnetic field is 9h 55m 27.2s.
A fluid body with an internal flow field that is purely
zonal and symmetric about the equatorial plane will have a
particularly simple gravity field of the form:
%Although the consequences of rotation on
%its very structure, and in particular the presence and strength of
%convection in its interior, are difficult to assess (Stevenson and
%Salpeter 1977;
%see also Busse 1976, Zhang \& Schubert 1996), this property is a
%chance because it allows us
%to constrain its interior density profiles, by measuring the
%departures from sphericity of its gravitational potential:
%
\begin{equation}
U(r,\theta)={GM\over
r}\left\{1-\sum_{i=1}^{\infty}\left({\req\over r}\right)^{2i}
J_{2i} P_{2i}(\cos \theta)\right\},
\end{equation}
%
where $G$ is the gravitational constant, $M$ the mass of the planet,
$\req$ its equatorial radius, $r$ the distance to the planetary
center, $\theta$ the colatitude and 
$P_{2i}$ are Legendre polynomials. 
Current gravity field data show no detectable deviation from
this assumed form. The $J_{2i}$ are the 
gravitational moments. Their observed values (inferred mostly from the
trajectories of the Pioneer and Voyager spacecrafts) are given in
Table~\ref{tab:moments}. The gravitational moments can also be related to
the internal density
profile $\rho(r)$ (and thus to theoretical models) by the following
relation (e.g., Zharkov \& Trubitsyn 1974): 
\begin{equation}
J_{2i}=-{1\over M \req^{2i}}\int \rho(r)r^{2i}P_{2i}(\cos
\theta)d\tau, 
\label{eq:j2i}
\end{equation}
in which the integral is calculated over the total volume $\tau$ of the
planet. A common method for the solution of Eq.~\ref{eq:j2i} is the
so-called theory of figures, which will not be developed here (see
Zharkov \& Trubitsyn 1978). 

\begin{table}
\begin{center}
\caption{Characteristics of the gravity fields and radii}
\label{tab:moments}
\begin{tabular}{l r@{.}l r} \hline 
Mass (total)& 1&8986112(15)$\times 10^{30}$\,g & (a) \\
Equatorial radius (1 bar) &  7&1492(4)$\times 10^9$\,cm & (b) \\
Polar radius (1 bar) & 6&6854(10)$\times 10^9$\,cm & (b) \\
$P_{\omega}$ (magnetic)& 3&57297(41)$\times 10^4$\,s & (c) \\
$J_2$ (1 bar)& 1&4697(1)$\times 10^{-2}$ & (a) \\
$J_4$ (1 bar)& $-5$&84(5)$\times 10^{-4}$ & (a) \\
$J_6$ (1 bar)& 0&31(20)$\times 10^{-4}$ & (a) \\
\hline
\multicolumn{4}{p{8cm}}{%
The numbers in parentheses are the uncertainty in the last digits 
of the given value. The value of the gravitational constant used to
calculate the mass of Jupiter is $G=6.67259\times
10^{-8}\,\mbox{ dyn\,cm}^2\,\mbox{g}^{-1}$ (Cohen \& Taylor 1987).}\\
\multicolumn{4}{l}{(a) Campbell \& Synott (1985)}\\
\multicolumn{4}{l}{(b) Lindal \etal\ (1981)}\\
\multicolumn{4}{l}{(c) Davies \etal\ (1986)}\\
%\multicolumn{4}{l}{(d) Adjusted from Hubbard (1982)}\\
\end{tabular}
\end{center}
\end{table}

A special and important case for which equation (1) is exactly true
arises when the flow field depends at most on distance from the
rotation axis. (The case of rigid rotation is a trivial example of
this). Deviations from rigid rotation will be small in a region of
substantial electrical conductivity (Hide \& Stannard 1976), and
deviations from rigid rotation are likely to take the form of
geostrophic flow (rotation on cylinders) elsewhere (Busse
1976). Symmetry-breaking flows such as convection are expected to be
tiny by comparison (Stevenson and Salpeter 1977b). Most analyses of
Jupiter gravity assume rigid body rotation throughout.

Differential rotation is suggested by the observed
atmospheric motions (e.g. Gierasch and Conrath 1993) as well as
theoretical considerations (Busse 1976).
%A complication arises from the fact that the equations derived from
%that theory generally assume the planet to be rotating as a solid body. 
%Observations of the atmospheric winds show significant variations with
%latitude, however (e.g., Gierasch \& Conrath 1993). 
The question of the
depth to which these differential
rotation patterns extend is still open. Hubbard (1982) has proposed
a solution to the planetary figure problem in the case of a deep rotation
field that possesses cylindrical symmetry. It is thus possible to
derive, from interior models assuming solid rotation, the value of the
gravitational moments that the planet would have if its surface
rotation pattern extended deep into its interior. It is \itt{a priori}
impossible to prefer one model to the other.
The correction due to differential rotation is of order $-1.5\times
10^{-5}$ for $J_2$, $+4\times 10^{-6}$ for $J_4$ and $-10^{-6}$ for
$J_6$. Due to the unknown rotation of the planet at deep levels, this
correction has to be treated as an uncertainty (its value can be added
to the measured moments given by Table~\ref{tab:moments} when
theoretical models are calculated with equations valid for solid body
rotation). (For further discussion of the influence of differential
rotation on the external gravity field, see Section~\ref{sec:gravity}).


\subsection{Magnetic field}

The external magnetic field ${\bf B}$ is generally expressed as an expansion
in spherical harmonics of the scalar potential $W$,
such that ${\bf B}=-\bfnabla W$:
\begin{eqnarray}
W&=&\req\sum_{n=1}^{\infty} \left(\frac{\req}{r} \right)^{n+1}
\sum_{m=0}^n \left\{g_n^m \cos(m\phi)+\right. \\ \nonumber 
&& \left. h_n^m \sin(m\phi)\right\} P_n^m(\cos\theta),
\label{eq:W}
\end{eqnarray}
where $\phi$ is the longitude and the $P_n^m$ are the associated
Legendre polynomials. The coefficients $g_n^m$ and $h_n^m$ are the
magnetic moments that characterize the field. They are expressed in
magnetic field units (\eg\ Gauss). 

\begin{table}
\caption{Magnetic moments (in Gauss) for Jupiter and Saturn}
\label{tab:W_jup}
\begin{center}
\begin{tabular}{cc r@{.}lr@{.}l c r@{.}lr@{.}l} \hline
 & & \multicolumn{4}{c}{Jupiter$^a$} && \multicolumn{4}{c}{Saturn$^b$}
\\ \cline{3-6} \cline{8-11}
$n$ &\ $m$\ & \multicolumn{2}{c}{$g_n^m$} & \multicolumn{2}{c}{$h_n^m$}
&& \multicolumn{2}{c}{$g_n^m$} & \multicolumn{2}{c}{$h_n^m$} \\ \hline
1 & 0 &  4&208 & 0& && 0&2144 & 0& \\
1 & 1 & $-$0&660 & 0&261 && $-$0&0014 & 0&0014 \\
2 & 0 & $-$0&034 & 0& && 0&0188 & 0& \\
2 & 1 & $-$0&759 & $-$0&294 && $-$0&0052 & $-$0&0043 \\
2 & 2 &  0&483 & 0&107 && 0&0050 & $-$0&0004 \\
3 & 0 & \multicolumn{2}{c}{\dots} & 0& && \multicolumn{2}{c}{\dots} &
0& \\
3 & 1 & \multicolumn{2}{c}{\dots} & \multicolumn{2}{c}{\dots} &&
\multicolumn{2}{c}{\dots} & \multicolumn{2}{c}{\dots} \\
3 & 2 & 0&263 & 0&695 &&
\multicolumn{2}{c}{\dots} & \multicolumn{2}{c}{\dots} \\
3 & 3 & $-$0&069 & $-$0&247 && \multicolumn{2}{c}{\dots} &
\multicolumn{2}{c}{\dots} \\ \hline
\multicolumn{11}{l}{$^a$ from Connerney \etal\ (1982)} \\
\multicolumn{11}{l}{$^b$ from Acu\~na \etal\ (1983)} 
\end{tabular}
\end{center}
\end{table}


One can show that the first coefficients in Eq.~\eref{W} (for
$n=0$ and $n=1$) correspond to the potential of a magnetic dipole such
that $W={\bf{M\cdot r}}/r^3$ where $M$ is the moment.
%\beq
%M=\req^3 \left\{\left(g_1^0\right)^2 + \left(g_1^1\right)^2 +
%\left(h_1^1\right)^2\right\}^{1/2}.
%\label{eq:M}
%\eeq
As seen from table~\ref{tab:W_jup}, Jupiter and Saturn have magnetic
fields of essentially dipolar nature, with a magnetic axis close to
the rotation axis ($g_1^0$ is much larger than the 
other harmonics). Note that Uranus and Neptune have magnetic fields
that are intrinsically much more complex. 
The magnetic dipolar moments are
$M=4.27\,\rm{Gauss\,R_J^3}$ for Jupiter, and $M=0.21\,\rm{Gauss\,R_S^3}$ for
Saturn. The dipole tilts (angle between the magnetic and rotation
axis) are about $9.6^\circ$ for Jupiter and less than $1^\circ$ for
Saturn (Connerney \etal\ 1982; Acu\~na \etal\ 1983).

        It is thought likely that Jupiter's field, like Earth's
field, is generated by a dynamo process, arising through convective
motions in an electrically conducting deep interior. Although
Jupiter's field is considerably larger than Earth's field, almost all
other known aspects of the field are remarkably like those of Earth
(Stevenson 1983). One obvious similarity is the dipole tilt. However,
the similarity between the two fields runs much deeper. To
appreciate this, it is important to ``downward continue'' the measured
field to the level that one thinks represents the surface of the
field generating region. If one chooses this to be at 0.75 to 0.8
Jupiter radii, and uses the known correct location for Earth (0.55
Earth radii, the top of Earth's core, seismologically determined)
then the ratio of quadrupole to dipole and the ratio of octupole to
dipole are the same for Jupiter and for Earth at the core ``surface''.
In both cases, the quadrupole is suppressed, and this is
characteristic of dipole family solutions to the dynamo generation of
magnetic fields (Roberts \& Glatzmeier 2000). The physical
significance of the ``core surface'' for Jupiter is discussed in
section 3.5.3.

Comparison of spacecraft data 
over twenty-five years suggests that Jupiter's field may undergo 
secular variation with a time scale similar to that observed for 
Earth. However, the accuracy of earlier field models is sufficiently 
uncertain that no variation has been definitively observed (Russell 
et al. 2001). The old ideas of Westward drift are no longer accepted 
in the geomagnetic community as a sufficient explanation for the 
variation of Earth's field (Dormy et al. 2000) but it is believed 
that these field variations are indicative of the convective motions 
in the uppermost region of the planet in which the field is strongly 
coupled to the flow. For the largest variations suggested by Russell 
et al (2001), motions in Jupiter could be as much as an order of 
magnitude larger than those for Earth (merely because Jupiter's 
radius is an order of magnitude larger). This suggests motions of up 
to 0.1 to $1\cm\sec^{-1}$ and the significance of this is discussed further 
in sections 3.4 and 3.5. Note, however, that this has no direct 
bearing on the important question of whether there are large, uniform 
zonal flows since they can only produce unobserved toroidal field.

%         Comparison of spacecraft data over twenty -five years suggest
% that Jupiter's field exhibits secular variation on time scales
% similar to Earth's field variations (Russell et al. 2002?[This is a
% bit controversial. I need to check on the status of this work]) The
% old ideas of Westward drift are no longer accepted in the geomagnetic
% community as an explanation for Earth's field (Bloxham...,Courtillot
% et al. 2001) but it is believed that these variations are indicative
% of the convective motions in the uppermost region where the field is
% coupled to the flow. This would suggest motions of about an order of
% magnitude larger than for Earth (because Jupiter has a core radius
% that is an order of magnitude larger value), or ~0.1 -1 cm/sec. This
% is discussed further when we talk about convective transport. Note,
% however, that this has no bearing on the important question of
% whether there are large zonal flows, which can affect both gravity
% interpretation and the magnetic field (independent of whether there
% is secular change of the field).

        Saturn's field is different from Jupiter's field in having no
detectable dipole tilt. However, it has been suggested that this is
an effect external to the region where the magnetic field is produced
so that the deep-seated field of Saturn 
might be very similar to Jupiter (Stevenson 1980, 1982a). By contrast the
fields of Uranus and Neptune are profoundly different. They have
large dipole tilts and large quadrupole moments, which is suggestive
of quadrupole family solutions of the dynamo equations.



\subsection{Atmospheric composition}
\label{sec:composition}

The most important components of the atmosphere of Jupiter (as well as
Saturn, Uranus and Neptune) 
are also among the most difficult to detect: H$_2$ and He have no
electric dipolar moments, and mostly absorb during collisions. 
However, an accurate helium to hydrogen ratio is now available thanks
to the \itt{in situ} measurement by the Galileo probe (von Zahn \etal\
1998). The helium mole fraction (\ie\ number of 
helium atoms over the total number of species in a given volume) is
$q_\rm{He}=0.1359\pm 0.0027$. The helium mass mixing ratio $Y$ (\ie\
mass of helium atoms over total mass) is obtained from 
$Y/(X+Y)=0.238\pm 0.05$, where $X$ is the hydrogen mass mixing
ratio. This value of $Y$ is by coincidence very close to 
that of the Sun's atmosphere, but because of helium
sedimentation in the Sun's radiative zone, $Y$ was larger in the
protosolar nebula: $Y_\rm{proto}=0.275\pm 0.01$ and
$(X+Y)_\rm{proto}\approx 0.98$ (\eg\ Bahcall \& Pinsonneault
1995). Less helium is
therefore found in the atmosphere of Jupiter than inferred to be
present when the planet 
formed. The most natural explanation is a phase separation of helium
in hydrogen and the subsequent formation of heavy helium-rich droplets
falling to deeper levels under the action of gravity (see \eg\
Stevenson \& Salpeter 1977b for a detailed discussion). 
This explanation is also favored by the fact that neon which tends to
dissolve into helium-rich droplets (Roulston \& Stevenson 1995) is found
to be significantly depleted in the atmosphere compared to the solar
abundance (Niemann \etal\ 1998). 

If the interpretation for helium is 
correct then there is a tendency for the atmosphere to become top 
heavy over time (i.e. the atmospheric material is more dense than 
the material located just \itt{above} the helium separation region when
compared at the same T and P). This will 
tend to promote mixing and homogenization to great depth, even if 
there is an intervening radiative zone (as discussed in 3.4). 
Consequently, the atmosphere is expected to provide a useful guide to 
the composition of a large fraction of Jupiter's volume.

The situation is more complex in the case of other elements because
chemical and meteorological effects are sources of considerable
heterogeneity in the atmosphere (see chapters by Taylor \etal\ and
Ingersoll \etal). Rather than trying to describe the many subtleties
of Jupiter's chemistry, we will focus globally on all the elements
other than hydrogen and hydrogen, i.e. the \itt{heavy elements}:  A
surprising result of the Galileo probe is that within the error bars,
carbon, sulfur, phosphorus, krypton, xenon and argon are all enriched
relative to hydrogen by a factor $\sim 3$ compared to the Sun's
atmosphere (Mahaffy \etal\ 2000; Owen \etal\ 1999). The case
of nitrogen is yet unclear but it may be consistent with this
figure. Last but not least, oxygen is the most abundant atom in the
Universe after hydrogen and helium, but its abundance in Jupiter's
deep atmosphere is still unknown, due to meteorological effects
affecting the abundance of water. Interpretation of the water features
in 5 micron spectra 
%TG:
(which probe deeper levels than other spectral regions)
%%
indicate that the bulk abundance of oxygen is probably at least solar.

Assuming that heavy elements are uniformly enriched by a factor 3 over
solar in Jupiter's interior, their total mass would amount to $18\mea$. 
However, in a solar composition mixture, oxygen contributes up to
half of the mass of heavy elements. The determination of its
abundance should therefore be a critical goal of future space missions
to Jupiter.



\subsection{Energy balance and atmospheric temperatures}

As known since the infrared observations of Low (1966), Jupiter emits more
energy than it receives from the Sun. This is the natural result of the
cooling and contraction of the planet, as it slowly releases its
primordial potential gravitational energy (see Hubbard 1968, 1977). 
Table~\ref{tab:flux} shows that about 40\% of Jupiter's \itt{infrared}
flux is due to its intrinsic flux, the rest being due to the
\itt{absorbed} solar flux (the directly reflected part is not included
in the energy balance). 


\begin{table}
\begin{center}
\caption{Energy balance and atmospheric temperatures\ti{a}.}
\label{tab:flux}
\begin{tabular}{l l} \hline
Absorbed power & 5.014(248)$\times 10^{24}$\ erg\,s$^{-1}$ \\
Emitted power  & 8.365(84)$\times 10^{24}$\ erg\,s$^{-1}$ \\
Intrinsic power & 3.350(260)$\times 10^{24}$\ erg\,s$^{-1}$ \\
Intrinsic flux & 5440(430)\ erg\,s$^{-1}$\,cm$^{-2}$ \\
Bond albedo & 0.343(32) \\
Effective temperature & 124.4(3)\ K \\
1-bar temperature & 170.4(50)\ K$\quad$ radio-occultation\ti{b} \\
 & 166.1(4)\ K$\quad$ hot spot\ti{c} \\
22-bar temperature & 427.7(12)\ K$\quad$ hot spot\ti{c} \\
\hline 
\multicolumn{2}{l}{\ti{a} Hanel \etal\ (1981); see also Pearl \&
Conrath (1991)} \\ 
\multicolumn{2}{l}{\ti{b} Lindal \etal\ (1981), adjusted (see text)} \\
\multicolumn{2}{l}{\ti{c} Seiff \etal\ (1998)} 
\end{tabular}
\end{center}
\end{table}

A crucial consequence of the presence of an intrinsic heat flux is
that it requires high internal temperatures ($\sim 10,000\K$ or more),
and that consequently Jupiter is \itt{fluid} (not solid)
(Hubbard 1968; see also Hubbard \etal\ 1995). Another consequence is
that its interior is essentially
convective with a temperature profile that is close to
an \itt{adiabat}. We will come back to this in more detail in
section~\ref{sec:heat}. 

Jupiter's deep atmosphere (more accurately troposphere)
is indeed observed to be close to an adiabat, a result first
obtained by Trafton (1967), but verified by radio-occultation
experiments by the Voyager spacecrafts (Lindal \etal\ 1981), and by
the \itt{in situ} measurement from the Galileo probe (Seiff \etal\
1998).

The atmospheric temperatures are important because they define the
outer boundary condition of interior models. Such a boundary condition
is often defined at the 1 bar level. The temperature retrieved from the
Voyager data (radio-occultation and infrared spectra) led to a value
of $165\pm 5$\,K (Lindal \etal\ 1981). However, this was using an incorrect
helium to hydrogen ratio. Using the He/H value from Galileo (von Zahn
\etal\ 1998) one finds that a reanalysis of the radio occultation data
leads to a 1 bar temperature of $170.4\pm 5$\,K. On the other hand,
the Galileo probe found a 1 bar temperature of
$166.1\pm0.4\K$ (Seiff \etal\ 1998). Horizontal temperature variations
can only be 
%TG:
measured down to
%%
pressures smaller than a fraction of a bar, and are of order
$\sim 5\,$K (Conrath \etal\ 1989). 

%TG:
The Galileo probe measurements indicate a temperature profile close to
a \itt{dry} adiabat down to 22\,bar. However, the probe fell into a
\itt{hot spot}, a relatively rare, dry region of
Jupiter's atmosphere. It is not clear that the measurements are also
relevant outside hot spots. For
example, the mean temperature profile could be closer to a \itt{moist}
adiabat (a more uniform temperature profile due to the release of
latent heat by condensation), as for the Earth's atmosphere. In that
case the planet's 
interior temperature would be colder by a factor that would
depend almost linearly on the bulk abundance of water, \ie\ about
0.5\% for a solar abundance, 1\% for a factor 2 enrichment in
water...etc. Furthermore, because water is heavier than hydrogen, the
reverse can also be true, \ie\ a \itt{superadiabatic} mean temperature
profile, especially for large enrichments (see Guillot 1995). In that
case, the interior temperature would be larger than expected. 
This can potentially affect the amount of heavy elements obtained from
interior models: at fixed density profile, a higher temperature will
be compensated by a larger amount of heavy elements. However, for most
enrichments the effect is relatively small. 


\subsection{Tidal response}

The tidal distortion of Jupiter arising from the Galilean satellites
has not yet been detected in the gravity field, because the largest
(Io) tide would correspond to a gravity coefficient $C_{22}$ of order
$10^{-7}$, even smaller than the measured $J_6 \sim 10^{-5}$.  The
corresponding real part of the Love number would tell us
(independently of $J_2$) about the linear response of Jupiter to
degree-two disturbances and may be readily detected in a future
orbiter mission. However ,we have a long-standing constraint on the
imaginary part of this Love number, directly related to the more
commonly referenced tidal $Q$. The tidal $Q$ is an inverse measure of
the tidal dissipation, so that very high $Q$ implies very low
dissipation. From the requirement that the orbit of Io expands and the
resonant heating of Io persists, we require $Q\sim 10^5$ (Goldreich
and Soter 1966; Peale 1999). Unfortunately, there is no consensus
yet on what this value tells us about the nature of Jupiter's interior.

A similar value appears to be needed to explain the tendency of
extrasolar planets to have low eccentricity orbits at small semi-major
orbital distances ($a\wig{<}0.07\,$AU) from their central stars
(e.g. Marcy et al. 1997).  This could suggest that the cause of this
$Q$ is generally present in giant planets and perhaps not related to
some ``special'' mechanism such as helium raindrops (Stevenson 1983b).
On the other hand, contrary to Jupiter, extrasolar planets with short
orbital periods develop a thick external radiative region (Guillot et
al. 1996; Guillot \& Showman 2002) in which tidal waves can be
efficiently dissipated (Lubow et al. 1997). The discovery of a
statistically significant number of extrasolar planets and their
characterization (e.g. by transit detections) should shed light on
the problem of tidal dissipation in giant planets.  (For further
discussion of Jupiter's tidal response to Galilean
satellites, see Section~\ref{sec:gravity}). 




\subsection{Oscillations}

The case for the existence of free oscillations of
the giant planets is still unresolved. 
Such a discovery would lead to
great leaps in our knowledge of the interior of these planets, as can
be seen from the level of accuracy reached by solar interior models
since the discovery of its oscillations. 
At present, there is no clearly identified mechanism capable of
exciting Jupiter's vibrations to an observationally detectable
level. Turbulent convection, thought to be responsible for the
Sun's oscillations (Goldreich \& Kumar 1988) yields negligible
amplitudes when applied to Jupiter (Deming et al. 1989). Other
possibilities generally invoke the presence of stable regions (e.g. at
the transition between molecular and metallic hydrogen) but are
largely speculative. On the other hand, observations aimed at
detecting modes of Jupiter have shown promising results (Schmider et
al. 1991), but have thus far been limited by instrumental and
windowing effects, and by the complex nature of the atmosphere
(Lederer \etal\ 1995). Recent work by Mosser \etal\ (2000) puts an upper
limit to the amplitude of the modes at $0.6\m\sec^{-1}$, and shows 
increased power in the Fourier spectrum in the expected range of
frequencies. Observations from space or from an Earth-based network of
telescopes should be pursued in order to verify these results. 


\section{Equations of state}
\label{sec:eos}

The equation of state is the ensemble of equilibrium thermodynamic
properties of a material, here taken to be a fluid mixture
predominantly composed of hydrogen, helium, and a small amount of
heavier elements.  At the high pressures and moderate temperatures
characteristic of Jupiter's interior, it is extremely challenging to
study the equation of state either experimentally or
theoretically. Yet, it is one of the most important factors in
modeling Jupiter's structure and inferring its internal composition
(\S 3.5.2).  At low pressures and temperatures, in the outer regions
of the planet, hydrogen is a molecular fluid. In the deep interior,
where pressures of several tens of Mbar are achieved, the average
distance between two neighboring nuclei is smaller than a Bohr radius
and the electrons can no longer be bound to nuclei and the fluid is
ionized.  This warm, dense plasma is characterized by strong electron
degeneracy and strong Coulomb interactions between the ions.  These
physical conditions are similar to those found in liquid metals and
this plasma is often referred to as ``liquid metallic hydrogen.'' The
bulk of Jupiter is made of this material.  The transitional regime
where hydrogen goes from a mostly molecular fluid to the fluid
metallic state is known as pressure ionization, a process of great
significance to the interior of Jupiter but that remains poorly
understood.

Equations of state used in modeling the interior of the planet are
based on theoretical calculations involving various approximations
that until recently had been largely untested in the appropriate
regimes of pressure and temperature.  Several recent experiments on
hydrogen and deuterium now provide data in regimes of interest for
modeling Jupiter and can constrain modeled equations of state.  This
renewed focus on hydrogen at Mbar pressures as so far spurred more
controversy than resolved important questions.

\subsection{Hydrogen: Summary of high-pressure experiments}
\label{sec:high-P exp}

The high pressures and high temperatures typical of the interiors of
giant planets can be achieved in the laboratory by shock-compression
of a small sample of material.  The shock is typically generated by a
hypervelocity impactor or by a powerful laser.  Measuring the
thermodynamic properties of the compressed sample is quite difficult
since such dynamical experiments last only 5 - 100$\,$ns and the
sample can be very small (0.4 -- 500$\,$mm$^3$).  For a given initial
state of the sample, the family of shocked states that can be achieved
follows a curve in the $(P,\rho,T)$ phase diagram known as a Hugoniot.
The Hugoniot is one of the Rankine-Hugoniot relations that result from
the conservation of energy, momentum, and matter flux across the shock
front.  Nearly all dynamical experiments on hydrogen and deuterium
performed share the same cryogenic initial state and therefore
measurements from different experiments can be directly compared.  By
reflection of the shock wave on a back plate made of a material
stiffer than the sample, a double-shocked state can be achieved that
reaches even higher pressures with a modest increase in temperature.
Multiple shock reflections, known as shock reverberation, lead to a
succession of compressed states that approach adiabatic compression.

Since 1995, deuterium has been the subject of intense experimental
study using several independent techniques.\footnote{Due to its higher
density, deuterium is experimentally more advantageous than hydrogen
because higher shock pressures can be achieved for a given impactor
speed.} Measurements of the pressure, density, temperature,
reflectivity, electrical conductivity, and sound speed have been
performed along the single-shock Hugoniot and, in some cases, along
double-shock Hugoniots.

The most reliable experimental results come from experiments where the
impactor is accelerated with a gas gun.  This technique allows for
larger samples ($\sim 500\,$mm$^3$) and longer lasting ($\sim
100\,$ns) experiments but is generally limited to pressures below
1$\,$Mbar.  Pressures and densities have been measured along the
single-shock Hugoniot up to 0.2$\,$Mbar and along the double-shock
Hugoniot up to 0.8$\,$Mbar (Nellis et al. 1983).  The reshocked states
reproduce the $(P,T)$ conditions of the molecular hydrogen envelope of
Jupiter and provide a direct probe of the thermodynamics of hydrogen.

%The electrical conductivity of deuterium compressed by shock reverberation 
%increases by almost 4 orders of magnitude between 0.9 and 1.4$\,$Mbar (Weir,
%Mitchell \& Nellis 1996). It then 
%remains approximately constant up to the largest pressure obtained (1.8$\,$Mbar).
%The maximum conductivity obtained of $\sim 2000\,(\Omega\,{\rm cm})^{-1}$ is
%comparable to that of liquid alkali metals but remains
%well below the calculated value for the fully dissociated, fully ionized state
%that we defined as ``fluid metallic hydrogen'' above.
%Under these experimental conditions ($T \wig< 4000\,$K), hydrogen remains 
%mainly molecular.
%The different nature of this unexpected conducting hydrogen fluid has
%been modeled following several approaches (Ross 1996; Lenosky et al. 1997;
%Nellis, Weir \& Mitchell 1999).  
%This discovery is important when considering the 
%magnetic field in the envelope of Jupiter (Nellis, Weir \& Mitchell 1996) 
%but has little consequence on the equation of state, however.

Under conditions where the dissociation of molecules becomes
significant, the temperature becomes a sensitive test of the EOS.
Processes that can absorb substantial amounts of energy like
dissociation and ionization result in relatively cool temperatures and
higher degrees of compression for a given pressure along the
Hugoniot. In the absence of such processes, the energy of the shock is
expended mostly in the kinetic degrees of freedom with a corresponding
increase in temperature.  The temperature of double-shocked deuterium
(Holmes, Ross \& Nellis 1995) was found to be lower than all EOS
predictions by about 30-40\%, indicating that dissociation plays a
more important role than predicted by contemporaneous models.

Finally, the sound speed has been measured along the Hugoniot in gas
gun experiments up to 0.28$\,$Mbar (N. C. Holmes, priv. comm.).  Since
it is a derivative of the pressure, the sound speed is a sensitive
test of EOS models with the advantage of being measurable very
reliably.

\begin{figure}
%\vspace*{6cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f1.eps}}
\caption{Comparison of experimental data and theoretical Hugoniot for
deuterium (densities are twice larger than expected for hydrogen at
any given pressure). Empty ellipses correspond to data points obtained
from laser compression (Collins \etal\ 1998). Filled ellipses were
obtained by magnetic compression (Knudson \etal\ 2001). Theoretical
calculations are represented by lines. They are respectively: the
``PPT'' (solid) and ``interpolated'' (dashed) Saumon-Chabrier
equations of state (Saumon, Chabrier \& Van Horn 1995), and a Path
Integral Monte Carlo EOS (Militzer \& Ceperley 2000). The solid line
to the left shows the T=0 equation of state for D$_2$ as determined by
an exp-6 potential fit to diamond-anvil cell measurements (Hemley et
al. 1990). The temperatures along the Hugoniot have been calculated
using the PPT-EOS.  }
\label{fig:hugoniot}
\end{figure}

With powerful lasers, deuterium can be shocked to much higher
pressures than with gas guns but the small sample size and the very
short duration of the experiments make accurate diagnostics very
challenging.  The $(P,\rho,T)$ single shock Hugoniot has been measured
recently up to 3.5$\,$Mbar with the NOVA Laser Facility (Da Silva et
al. 1997; Collins et al. 1998; Collins et al. 2001), reaching a
maximum density of $\sim 1\,$g\,cm$^{-3}$ at $\sim 1\,$Mbar
(\rfigure{hugoniot}).  Such a high compressibility was not anticipated
by most EOS models and this work sparked the current interest in the
thermodynamics of warm dense hydrogen as well as controversy, both on
the theoretical and experimental fronts.  The reflectivity of shocked
deuterium reaches about 60\% for pressures above 0.5$\,$Mbar along the
Hugoniot (Celliers et al.  2000), a value indicative of a large
density of free electrons and of a high electric conductivity
characteristic of fluid metallic hydrogen.
%Under the most extreme pressure achieved in this experiment 
%($P \sim 5\,$Mbar and $T \sim 80000\,$K), hydrogen is unlikely to be fully ionized
%but is close to the state commonly referred to as fluid metallic hydrogen.
Second-shock compression up to 6$\,$Mbar with the Nike laser give
results in agreement with the NOVA $(P,\rho)$ data (Mostovych et
al. 2000).  On the other hand, Knudson \etal\ (2001) used a magnetic
Z-accelerator to accelerate impactors to very high velocities.  Their
single-shock Hugoniot agrees well with the NOVA data for $P \wig<
0.4\,$Mbar but it is not as compressible at higher pressures, reaching
a density of only $\sim 0.7\,$g\,cm$^{-3}$ at 0.7$\,$Mbar
(\rfigure{hugoniot}).

\subsection{Hydrogen: EOS calculations}
\label{sec:high-P theory}

While the temperatures obtained along the single-shock Hugoniot
rapidly become much higher than those inside Jupiter at the same
pressure (\rfigure{phase_diag}), these measurements provide very
important, and heretofore unavailable tests of equations of state in
the 0.5 to 5 Mbar range where pressure ionization of hydrogen occurs.
Conversely, EOS models can be used to compute the various physical
quantities measured in the lab and to interpret the experimental
results.

Theoretical single-shock Hugoniots computed from a wide variety of EOS
models basically fall into two groups. First principle calculations
(\itt{e.g.}  Militzer \& Ceperley 2000; Lenosky et al. 1997, 2000;
Galli et al. 2000) all predict a rather stiff Hugoniot that is in
general agreement with the Z-accelerator data of Knudson et
al. (2001).  This is illustrated in \rfigure{hugoniot} by the Path
Integral Monte Carlo calculation of Militzer \& Ceperley (2000). On
the other hand, models that are partly calibrated with experimental
data (Saumon, Chabrier \& Van Horn 1995; Ross 1998; Saumon et
al. 2000, Ross \& Yang 2001), obtain a generally good agreement with
the NOVA data (\rfigure{hugoniot}; Collins et al. 1998).
Interestingly, the standard SESAME EOS of deuterium (Kerley 1972)
predicts a Hugoniot that generally agrees with the much more
sophisticated \itt{ab initio} calculations.

Our study of a number of theoretical Hugoniots shows that EOS that
have been fitted to the gas gun single- and double-shock $(P,\rho,T)$
data of Nellis et al. (1983) and Holmes et al. (1995) -- all taken
below 0.8$\,$Mbar and 5300$\,$K -- reproduce the high compression of
the NOVA data (Collins et al. 1998) and the sound speed data along the
single shock (N. C. Holmes, priv. comm.).  On the other hand, the
first-principle calculations generally agree with the stiffer Hugoniot
of Knudson et al. (2001) and cannot reproduce the high compression of
the NOVA data. They also fail to reproduce the double-shock
temperatures and the sound speed measurements.  Some of the \itt{ab
initio} calculations disagree with the low-pressure gas gun data
(\itt{e.g.} Lenosky et al. 2000).  On the one hand, the Knudson et al.
(2001) data and nearly all first-principle EOS calculations are in
good agreement with each other.  On the other hand, more heuristic EOS
models clearly show that four independent EOS experiments
(second-shock temperature, sound speed, the NOVA single shock and the
Nike double shock) are fully consistent with each other but neither
with first principle calculations nor the Knudson et al.  (2001) data.
Both the high compressibility of the NOVA Hugoniot and the low gas-gun
reshock temperatures can be explained by the absorption of the shock
energy resulting from molecular dissociation.

This polarization of EOS calculations along different data sets has
created a lively debate and is stimulating much additional (and
challenging) experimental and theoretical work.  The EOS of hydrogen
in the 0.5 to 5 Mbar regime, where it is transformed from an
insulating molecular fluid to a conducting liquid metal remains
uncertain to a level that is significant for modeling the interior of
Jupiter.  The recent progress in this area as been very beneficial,
however, as it appears that the current data and models bracket the
actual EOS of hydrogen.

In order to model Jupiter's interior with confidence, a careful study
of the uncertainties arising from the EOS would be required.  This is
not presently available, but fig.~\ref{fig:hugoniot} shows that this
can be crudely approximated by using the ``interpolated'' and the
``PPT'' equations of state of Saumon, Chabrier \& Van Horn (1995) even
though they do not fit the experimental data well. A comparison of
\ref{fig:hugoniot} and \ref{fig:densities} shows that large
uncertainties in density along the Hugoniot at 1 Mbar ($\sim$ 30\%)
result in much smaller differences along the Jupiter adiabat ($\sim$
8\%).  The effects on the inferred core mass and the mass of heavy
elements in Jupiter are discussed in \S3.5.2.

\subsubsection{A Plasma Phase Transition?}

We have seen that hydrogen undergoes a transition from a low-pressure
molecular insulating fluid to a high-pressure conductive fluid.  Is
the transition continuous, as is the case for temperature ionization,
or rather a first order phase transition (the so-called \itt{Plasma
Phase Transition}, or PPT) with discontinuities in density and entropy
across the coexistence curve?  Such a first-order transition was first
suggested by Wigner \& Huntington (1935) on the basis of the different
nature of the interaction potentials in metals (a weakly repulsive,
screened Coulomb potential) and in insulators (a strongly repulsive
``hard-sphere'' potential).

The PPT has not been observed experimentally in hydrogen (\ie\ there
is no evidence for the expected discontinuities), but it can be argued
that the gas-gun experiments have not reached high enough pressures,
and that laser-shocks may be supercritical. Note for example that
using the new data, the critical point for the PPT computed by Saumon
\etal\ (2000) is lower ($T\approx 14600\K$; $P\approx 0.73$\,Mbar)
than shown in fig.~\ref{fig:phase_diag}. The PPT is predicted by some
of the more heuristic ``chemical picture'' EOS models (Saumon \etal\
1995 and references therein) and Beule \etal\ (1999). On the other
hand, none of the first-principle EOS calculations show evidence for a
first order phase transition in warm dense hydrogen.

\begin{figure*}
%\vspace*{6cm}
\resizebox{14cm}{!}{\includegraphics[angle=0]{chap3_f2.eps}}
\caption{Hydrogen phase diagram, with interior profiles of present-day
Jupiter and Saturn overlaid, and with some experimental data shown.
The boundary between liquid H$_2$ and solid H$_2$ is somewhat
uncertain in the Mbar pressure range (2 estimates are shown), but is
not relevant to Jupiter.  The laser shock measurements of Collins et
al. (2001) and the gas-gun measurements of Holmes et al. (1995) are
shown as triangles and filled circles in the upper left-hand corner,
respectively.  Single- and double-shock hydrogen Hugoniots calculated
by Saumon et al. (2000) are shown as dot-dashed lines in the same
region of the plot. The solid line labeled ``50\%'' shows where 50\%
of molecular dissociation is obtained in the model of Ross (1998).}
\label{fig:phase_diag}
\end{figure*}

If present, the PPT would have significant consequences for the
structure of Jupiter, Saturn, and low-entropy extrasolar giant
planets.  Its main effect would be to create an impenetrable barrier
for convection between the molecular and metallic hydrogen parts of
the envelope, affecting the mixing of chemical species (Stevenson \&
Salpeter 1977b). The thermodynamic conditions of phase equilibrium
imply that the chemical composition across the PPT must be
discontinuous (Landau \& Lifschitz 1969), with the consequence that
atmospheric abundances of all elements would no longer be indicative
of their bulk abundance in the planet.  In addition, as the planet
cools, a fraction of the mass of the envelope is converted from one
phase to the other with an associated latent heat release (or
absorption).  The effect on the evolution is not very pronounced for a
latent heat of $\sim 0.5k_{\scriptscriptstyle B}$ per proton (Saumon
\etal\ 1992).

\subsection{Helium: Emphasis on the H-He phase diagram}
\label{sec:h-he}

In comparison with hydrogen, the EOS of helium under the conditions of
interest for Jupiter has been little studied.  Single-shock Hugoniot
data is available up to 0.56 Mbar (Nellis et al. 1984) only.  The
current controversy surrounding the hydrogen EOS suggests that
untested EOS calculations of warm dense helium should be viewed with
caution.  A much more difficult problem is that of the phase diagram
of H/He mixtures.  Nevertheless, its importance in giant planets is
such that this topic deserves a detailed discussion.

As first proposed for Jupiter and Saturn by Smoluchowski (1967) and
Salpeter (1973), hydrogen and helium mixtures can undergo a \itt{phase separation}:
at low temperatures, helium (or other elements) can become
insoluble and form helium-rich droplets. Under the action of gravity, these
droplets will tend to fall toward the central regions of the planet. 

Physically, a phase separation arises in a binary mixture of
concentration $x$ when the second derivative of the Gibbs free energy
$\partial^2 G/\partial x^2<0$. The two concentrations $x_1$ and $x_2$
that correspond to phase equilibrium between the droplets and their
surroundings are given by the equality of the chemical potentials:
$${\partial G \over \partial x}\Big|_{x_1}={\partial G \over \partial
x}\Big|_{x_2}.$$ The lower the temperature, the closer $x_1$ and $x_2$
are to 0 and 1, respectively.

The calculation of the miscibility of hydrogen and helium requires a
knowledge of the (non-ideal) EOS of each element and, of course, of
the interactions between hydrogen and helium species.  Given the
difficulty in modeling the EOS for hydrogen alone, it may not be so
surprising that the question of the helium phase separation in the
giant planets is still unsolved.


One approach has been to calculate the hydrogen-helium phase diagram
assuming complete ionization. In that case, critical temperatures of
order $8000\K$ at 2\,Mbar can be calculated (Stevenson 1982b). Even
more importantly, this leads to a critical temperature that
\itt{decreases} with increasing pressure. The consequence is that (i)
this would imply that a phase separation has occurred in Jupiter, and
earlier in Saturn, as suggested by the abundance of helium measured in
the atmosphere (see section~\ref{sec:composition}); (ii) helium would
be most insoluble near the molecular/metallic transition.  Results
from that approach (Hubbard \& DeWitt 1985) are illustrated in
\rfig{phase_diag}\ by two lines labelled ($Y=0.21$ and
$Y=0.27$). These lines represent, in the liquid metallic-hydrogen
phase, the limiting He mass fraction Y for a single-phase H-He
solution.

Other calculations have been attempted in the local density
approximation. Earlier work (Klepeis \etal\ 1991) suggested an
unrealistically high critical temperature ($40000\K$ at
10.5\,Mbar). However, a more careful study by Pfaffenzeller \etal\
(1995) with the same basic technique led to a lower critical
temperature (less than $5000\K$ at 4\,Mbar). This value would imply no
demixing of helium in Jupiter and Saturn. More importantly, the work
of Pfaffenzeller \etal\ implies a critical temperature that
\itt{increases} with pressure. This can be explained if hydrogen is
still not fully ionized at the pressures considered (4 to 24\,Mbar),
which seems difficult to reconcile with the more standard hydrogen
EOSs. Another problem of the work of Pfaffenzeller \etal\ is that it
does not recover the fully ionized limit. If the critical temperature
increases with pressure, this would open the possibility that helium
separates from hydrogen over an extended fraction of the planetary
radius, with significant consequences for the interior and evolution
models.

An alternative approach is to constrain the phase diagram by requiring
that evolution models for both Jupiter and Saturn yield a cooling time
that agrees with the age of the Solar System, 4.55 Ga.  Recent
calculations (Fortney \& Hubbard, in preparation) indicate that the
phase diagram shown on \rfig{phase_diag}\ implies only a modest
differentiation of helium in Saturn, and a cooling time that is short
of the required one by $\sim 1$ Ga. However, Fortney \& Hubbard
propose that a slight modification of the phase diagram is sufficient
to obtain the required cooling time: The solubility of He in an
electron gas goes approximately as $x\sim\exp(-\Delta G/kT)$, where
$x$ is the concentration of He and $\Delta G$ is the Gibbs energy of
immersion of the helium atom (Hubbard \& Stevenson 1984).  The results
of Hubbard and DeWitt (1985) can be well represented by $\Delta G\sim
1.4\,$eV at the pressures in Jupiter or Saturn near the PPT
coexistence curve (Fig. 3.2).  Increasing $\Delta G$ by a factor $\sim
2$ yields the correct age for Saturn, and at the same time has no
effect on Jupiter, for which interior temperatures lie above the
solar-composition line on Fig. 3.2. However, this approach has
limitations inherent to the uncertainties of the evolution
calculations (see section~\ref{sec:evolution}).

Clearly, physical and astrophysical constraints on the hydrogen/helium
phase diagram are consistent with a scenario in which the phase
separation has begun recently in Jupiter, and a few billion years ago
in Saturn due to its colder interior. However, the uncertainties on
both constraints are significant, and a more detailed test of
this scenario must await (i) an accurate understanding of the pressure
ionization of hydrogen alone and (ii) calculations and experiments
predicting the behavior of hydrogen-helium mixtures at Mbar pressures. 




\subsection{Heavy elements}

Due to their small molecular abundances, an approximate treatment of
the equations of state of heavy elements is generally sufficient for
the calculation of Jupiter's internal structure. However, the
derivation of the amount of heavy elements present in the planet
depends on the accuracy of these equations of state. Heavy elements
also have indirect but crucial contributions, such as in calculating
radiative opacities, and the possible existence of molecular weight
gradients.


\begin{figure*}
%\vspace*{6cm}
\resizebox{14cm}{!}{\includegraphics{chap3_f3.eps}}
\caption{% 
Comparison of density profiles: \itt{Solid line} Jupiter
model with a pure hydrogen-helium envelope (X=0.70; Y=0.30) and a
central ice/rock core, using the Saumon-Chabrier ``interpolated EOS''
(Saumon et al. 1995). \itt{Dotted line}: \itt{Idem} with the ``PPT''
SC EOS for hydrogen (see text). \itt{Dashed lines}: Theoretical
zero-temperature EOSs for ice VII (Zharkov and Trubitsyn 1978) and
olivine (ANEOS). \itt{Dash-dotted lines}: Analytic fits of Hubbard and
Marley (1989) to experimental results of adiabatic compression of
``synthetic ices'' (a mixture of water, methane and ammonia) and
``synthetic rocks'' (a mixture of SiO$_2$, MgO, FeS and
FeO). \itt{Asterisks}: Shock-reverberation measurements for H$_2$O
(Chau \etal\ 2001). \itt{Squares}: Single- and double-shock
measurements for ``synthetic ices'' (Nellis et al. 1997).}
\label{fig:densities}
\end{figure*}


\rfigure{densities} shows adiabats and isotherms for various elements
and mixtures. The EOSs for the simplest elements, hydrogen and helium,
yield density profiles with an inherent uncertainty of about $10\%$
(the uncertainty due to the temperature profile being only a few
percent). Other elements are roughly categorized into ``ices''
(elements which condensed in the protosolar nebula at temperatures
$\sim 150$\,K, including water, ammonia, methane), and ``rocks'' (a
mixture of more refractory materials). Polynomial fits to the
pressure-density profiles for both kinds of mixtures have been
calculated by Hubbard \& Marley (1989) on the basis of experimental
measurements (Nellis et al. 1997 and references therein). 
These are shown in \rfig{densities}. 
Note that these relations are independant of temperature. Thermal
perturbations can be included by adding a thermal
pressure component, which can be approximated by $P_\rm{therm}\approx
3 \gamma \rho kT/m$, where $k$ is Boltzmann's constant, $m$ is the
molecular mass and $\gamma$ the Gruneisen parameter (e.g. Zharkov
1986). For $T\approx 10^4\K$, $m\approx 20$\,a.m.u., $\rho\approx 2\gcc$,
$\gamma\approx 1$, one finds $P_\rm{therm}\approx 0.3$\,Mbar. Thermal
effects can be neglected in the metallic region only. The most
important source of uncertainty is the unknown composition of the
heavy elements.

Water is especially important because it should contribute to about
half of the mass of heavy elements if the ratios of elemental abundances
to that of oxygen (e.g. C/O, N/O, Fe/O ...etc.)  remain
approximatively solar.  Experimental results on H$_2$O, as indicated
by asterisks in \rfig{densities} (Chau et al. 2001) are in good
agreement with theoretical zero-temperature estimates by Zharkov \&
Trubitsyn (1978), but rules out the commonly used ANEOS H$_2$O
equation of state which predicts densities that are systematically
smaller than
the laboratory measurements by $\sim 8\%$.  The relatively good
agreement between experiments and one theoretical prediction should
however not occult two important unknowns: (i) when does water
metallize? (ii) how soluble is it in hydrogen? The first is important
because water could provide a significant amount of electrons if its
metallization takes place at low pressures. The second could limit the
amount of water in the external envelope and create an inner
``water-rich'' region (this is possible for large enrichments of water
over the solar value).

Globally, heavy elements are thought to be uniformly mixed throughout
the hydrogen-helium envelope because of convection (next section). A
phase separation can be important for mixtures with large enough
abundances (helium and possibly water)\footnote{%
This is due to the fact that, as discussed in the case of helium,
given $x$ the concentration of a minor 
species in hydrogen, a phase separation
occurs when $T\wig{<}-\Delta G/(k\ln x)$. For chemical species with
a small abundance $x$, the Gibbs energy of immersion $\Delta G$ would
need to be unrealistically large for demixing to occur.}.
Other factors can also affect
the distribution of elements in Jupiter's interior in a more limited
way. The dissolution of neon in helium raindrops (Roulston \&
Stevenson 1995) is one example, but other elements could also dissolve
in liquid silicate raindrops, thought to form clouds at temperatures
$T \sim 2000\K$ and $P\sim 10\,$kbar (a region where non-ideal effects
are important). This partitioning could affect atmospheric abundances.
Another crucial problem is the delivery of planetesimals at
depths that depend on their composition (Podolak et al. 1988) and
their subsequent mixing. Depending on the formation mechanisms and
efficiency of convection, Jupiter's envelope could have either a solar
ice to rock ratio, or be depleted in refractory material.




\section{Transporting Jupiter's intrinsic heat}
\label{sec:heat}

\subsection{Radiation \& Conduction}

The fact that Jupiter emits more energy than it receives from the Sun
implies that internal heat is transported from the planet's central
region to space. The structure of the interior and the planet's
evolution are intimately related to the mechanisms that transport this
heat. Radiation and conduction can be modeled with the same
approach. In the diffusion approximation (which assumes that the
radiation field is isotropic, a very good hypothesis at the
deep levels considered here), one defines a radiative gradient:
\beq
\nabrad=3{\kappa P \over g}\left(\teff\over T\right)^4,
\label{eq:nabrad}
\eeq
where $P$ is the pressure, $\kappa$ the Rosseland opacity (including
contributions due both to radiation and conduction), $g$ is the gravity
$\teff$ the effective temperature at the level considered
($\sigma\teff^4$ is the intrinsic energy flux), and $T$ the
temperature. 

The \itt{critical opacity} $\kcrit$ is defined as the (Rosseland)
opacity above
which the medium is too opaque/resistant to transport the totality of
the heat flux by radiation/conduction, and becomes convective. In a
homogeneous inviscid mixture, this occurs when $\nabrad>\nabad$, where
$\nabad\equiv (\partial\ln T/\partial\ln P)_S$ is the adiabatic
gradient. 
Assuming furthermore that $g$, $\teff$ and the temperature
gradient $\nabt\equiv d\ln T/d\ln P$ are independant of depth, one
finds that
\beq
\kcrit \approx {\nabad\over 2} \kappa_0 \left({P\over
P_0}\right)^{4\nabt-1},
\label{eq:kcrit}
\eeq
where we have introduced an equivalent opacity
$\kappa_0=(2/3)(g/P_0)$, and $P_0$ is for example the pressure at the
tropopause. 

Since $\nabt\approx 0.3$ in Jupiter, $\kcrit$ is only weakly dependant
on the level considered. However the opacity itself (in cm$^2\,\rm
g^{-1}$) is a strong function of $P$ and $T$. At pressures of a bar or
more and relatively low temperatures (less than 1000\,K), the three
dominant sources of opacities are water, methane and collision-induced
absorption by hydrogen molecules. Except possibly in limited regions
($P\wig{<}10$ bar) where meteorology due to water condensation can
affect chemical abundances and/or yield a stable stratification (see
discussion hereafter), \rfig{opacities} shows that convection is
ensured by the rapid rise of opacity with increasing pressure. (Note
that the critical opacity shown by a dashed line also slowly increases
in regions where $\nabt>0.25$).

However, in the intermediate temperature range between $\sim 1200$ and
$1500\K$, the Rosseland opacity due to the hydrogen and helium
absorption behaves differently: the absorption at any given
wavelength increases with density, but because the temperature
also rises, the photons are emitted at shorter wavelengths, where the
monochromatic absorption is smaller. As a consequence, the opacity can
decrease. This was shown by Guillot et al. (1994a,b) 
to potentially lead to the presence of a deep radiative zone in the
interiors of the giant planets. 

\begin{figure}
%\vspace*{6cm}
%\centerline{\hbox{\psfig{file=opacities.eps,width=8cm}}}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f4.eps}}
\caption{Comparison of Rosseland mean opacities (solid lines) along a
model of Jupiter to the critical opacity (dashed line) above which the
medium is convectively unstable (Schwarzschild's criterion). The
opacity calculations assume a three times solar enrichment in heavy
elements. Depending on the presence of alkali metals, Jupiter is found
to be fully convective or to possess a small, deep radiative zone.}
\label{fig:opacities}
\end{figure}

This problem must however be reanalyzed in the light of recent
observations and analyses of brown dwarfs. Their spectra show 
unexpectedly wide sodium and potassium absorption lines (see
Burrows, Marley \& Sharp 2000), in spectral regions where
hydrogen, helium, water, methane and ammonia are relatively
transparent. It thus appears that the added contribution of these
elements (if they are indeed present inside Jupiter) would wipe out
any radiative region at these levels (Guillot, Freedman, Lodders \&
Marley, in preparation). Figure~\ref{fig:opacities}
shows that opacities calculated without the contribution of
these alkali metals present a marked dip whereas those
including them in the mixture do not in which case the planet is
expected to be fully 
convective. However the presence of alkali metals in Jupiter depends
on the history of the formation of the planet. Although these elements
are probably present, one could imagine that they can be retained in
the central core of the planet and be underabundant in the
envelope. 

At temperatures above $1500\sim 2000\K$ two important sources of opacity
appear: (i) the rising number of electrons greatly enhances the
absorption of H$_2^-$ and H$^-$; (ii) TiO, a very strong absorber at
visible wavelengths is freed by the vaporization of CaTiO$_3$. Again,
the opacity rises rapidly which ensures a 
convective transport of the heat. Still deeper, conduction by free
electrons becomes more efficient, but the densities are found not to
be high enough for this process to be significant, except perhaps near
the central core (see Hubbard 1968; Stevenson \& Salpeter 1977a). 


\subsection{Convection}

\begin{table*}
\begin{center}
\caption{Properties of convection in Jupiter (order of
  magnitude estimates from the mixing length theory)}
\vspace*{-0.5cm}
\label{tab:propconv}
\begin{tabular}{rccccccccccc}
\hline\hline
                 & $P$            & $T$   & $\hp$ & $\nabt-\nabad$    & $v_\rm{conv}$ & $\tau_\rm{conv}$ & $\nu$    & $Pr={\nu\over\kappa}$ & $Re={vd\over\nu}$ & $Ro={v\over\omega d}$ & $E={\nu\over\omega d^2}$\\
                 & [bar]          & [K]   & [km]  &                   & [m\,s$^{-1}$] & [yrs]            & [$\rm{m}^{2}\,\rm{s}^{-1}$] &        &                   &                 &            \\ \hline
Atmosphere       & 1              & 165   & 20    & $10^{-5}$         & 1             & $10^{-3}$        & $10^{-5}$      & $10^{-4}$             & $10^9$            & 0.4             & $10^{-10}$ \\ 
Molecular region & $5\times 10^5$ & 5100  & 3000  & $2\times 10^{-9}$ & 0.1           & 1              & $10^{-6}$       & $10^{-1}$             & $10^{11}$         & $10^{-4}$       & $10^{-15}$ \\
Metallic region  & $4\times 10^6$ & 8400  & 8000  & $4\times 10^{-10}$& 0.1           & 2                & $3\times 10^{-7}$& $10^{-2}$             & $10^{12}$         & $10^{-4}$       & $10^{-16}$ \\
Core interface   & $4\times 10^7$ & 20000 & 30000 & $5\times 10^{-11}$& 0.03          & 20               & $10^{-6}$       & $10^{-2}$             & $10^{12}$         & $10^{-5}$       & $10^{-17}$ \\
\hline
\end{tabular}
\end{center}
\end{table*}

Modeling convection in the presence of rapid rotation, of a powerful
magnetic field and of compositional gradients, is of extreme
complexity. Fortunately, we are often mostly interested in
order-of-magnitude estimates of two quantities: the superadiabatic
temperature
gradient necessary to transport the intrinsic heat flux, and the
mean convective velocity. 
Rotation and magnetic field can significantly affect the type of
convective  
style and fluid velocities, but it is useful to look first at the 
commonly used crude approximation provided by mixing length theory:
\begin{eqnarray}
\nabt-\nabad&\sim&\left[{4\sqrt{2}\over \alpha^2\delta^{1/2}}
{F_\rm{conv}\over c_P T(\rho P)^{1/2}}\right]^{2/3},
\label{eq:superad}\\
v_\rm{conv} &\sim&\left[{\alpha\delta\over 4}{P\over \rho c_P T}
{F_\rm{conv}\over\rho}\right]^{1/3},
\label{eq:vconv}
\end{eqnarray}
where $\alpha\sim 1$ is the mixing length parameter,
$\delta=-(\partial\ln\rho/\partial\ln T)_P$, and $c_P$ is the specific
heat at constant pressure. The convective flux is
$F_\rm{conv}=F(1-\nabt/\nabrad)$, 
where $F$ is the total flux. In our case, the opacities become rapidly
large, therefore $\nabrad\gg\nabt\approx\nabad$ and
$F_\rm{conv}\approx F$.  
Physically, the superadiabatic gradient $\nabt-\nabad$ depends on
the ratio of the energy per unit mass $F/\sqrt{\rho P}$ to
be transported to that of a given layer, $c_P T$.  The convective
velocity is essentially proportional to $(F/\rho)^{1/3}$: since $F$ is
a slowly varying function, $v_\rm{conv}$ should be expected to be larger near
the surface, where $\rho$ is smaller. This corresponds to the fact
that in a low density material, transporting the same energy requires
higher velocities.

Values typical for Jupiter's interior yield the following quantitative
expressions for the superadiabatic gradient and convective velocity:
\begin{eqnarray*}
\nabt-\nabad &\sim& 2\times 10^{-9} 
\left(\frac{H_p}{{10^8\cm}}\right)^{-1/3}
\left(\frac{\rho}{{1\,\rm{g\,cm^{-3}}}}\right)^{-2/3}\times\\ \nonumber
&&\left(\frac{T}{{10^3\K}}\right)^{-2/3}
\left(\frac{{F}_\rm{conv}}{{10^3\,\rm{erg\,cm^{-2}\,s^{-1}}}}\right)^{2/3}\\
v_\rm{conv} &\approx& 20 \left(\frac{\nabt-\nabad}{10^{-8}}\right)^{1/2}
\left(\frac{H_p}{10^8\cm}\right)^{1/2}\:\rm{cm\,s^{-1}}.
\end{eqnarray*}
As shown in table~\ref{tab:propconv} the temperature gradient in
convective zone is thus almost indistinguishable from the adiabatic
gradient. 
The velocities derived are relatively small (for example, they
allow helium droplets to grow and fall before they can be transported
efficiently; Stevenson \& Salpeter 1977b), yet the convective
timescale is always much shorter than the evolution timescale. 

A few dimensionless numbers characterize convection itself and are
given in table~\ref{tab:propconv}, based on values of viscosities and
conductivities from Stevenson \& Salpeter (1977a), and radiative
diffusivities as discussed in the previous section.  These values are
only order of magnitude estimates. As shown by the relatively small
values of the \itt{Prandtl} number, radiation (near the upper layers)
and conduction (in the metallic interior) generally dominate over the
microscopic diffusion of heat. However, the large \itt{Reynolds}
number shows the overwhelming importance of macroscopic diffusion, and
the inherently turbulent nature of convection inside the planet. The
low \itt{Rossby} number indicates that rotation will significantly
affect convective motions, implying that in the planet's interior,
convective motions will be mostly confined to a plane perpendicular to
the axis of rotation. (Note that \itt{horizontal} motions in the
atmosphere are characterized by $Ro\ll 1$ because the characteristic
length is then $d\gg H_P$). Finally, the \itt{Ekman} number describes
the relative magnitude of viscous and Coriolis forces. It is extremely
low, indicating that frictionnal forces are largely negligible for
large scale motions in the planet.


\subsection{Factors inhibiting convection}

The simple expectations offered by mixing length theory are 
modified, sometimes dramatically, by the following phenomena:

\begin{enumerate}
\item \itt{Rotation}: 
In the relevant limit of very small Rossby 
numbers, the convective motions must conform to the Taylor-Proudman 
theorem which enforces a columnar or plate-like pattern in which at 
least one of the characteristic length scales perpendicular to the 
rotation axis is much smaller than the large (planet scale) length of 
variation parallel to the rotation axis. It is likely (but not 
certain) that this requires a significantly larger superadiabaticity 
than mixing length theory predicts (for the same specified heat flux) 
but a convective velocity that is relatively unchanged (Stevenson
1979). However, there remains a major unresolved issue: Does 
convection in the presence of rotation excite large differential 
rotation (e.g. Aurnou and Olson 2001)? This issue is 
still beyond current computational capability because of the very low 
Ekman number of these flows. (In the actual planet, the 
viscosity is negligible but in numerical calculations it is chosen to 
be quite large since the flow would otherwise be imperfectly resolved 
by the numerical grid that is used).

\item \itt{Magnetic field}: In an electrical conductor, fluid motions
modify the field to the extent that there is flow across field
lines. The resulting induced electrical current together with the
field leads to a Lorentz force whose influence on the motion can be
large. In dynamos, numerical and theoretical analysis suggests that
the Lorentz force, Coriolis force and buoyancy force are all
comparable. The highly anisotropic nature of the rotation-dominated
convection is then reduced %(the motions become more equidimensional)
and the convection requires less superadiabaticity than in the absence
of the field. This so-called MAC state (cf. Hide and Stannard 1976;
Jones 2000) leads to a different scaling of convective velocity on
heat flux. If $\Delta T$ is the temperature anomaly ($\sim$ superadiabatic
excess over a scale height) then the buoyancy force $g \delta \Delta T/T
\sim 2 \omega v_\rm{conv}$ (the Coriolis force), where $\omega$ is the planetary
angular velocity. Together with $F \sim \rho c_P \Delta T v_\rm{conv}$ this
leads to
\begin{equation}
	v_\rm{conv} \sim \left(g \delta F\over \rho c_P T \omega\right)^{1/2}
\end{equation}
which predicts motions of order $0.1-1\,$cm\,s$^{-1}$ and fractional
superadiabaticities of order $10^{-6}$ or less. In this regime where both
rotation and magnetic field are important, the convective motions are
one to two orders of magnitude smaller than mixing length theory predicts and
compatible with the previously mentioned constraints on magnetic
secular variation (Russell et al. 2001). We discuss the dynamo and its
relation to convection further in 3.5.3.

\item \itt{Compositional gradients}: The presence of compositional
gradients ($\nabmu\equiv d\ln\mu/d\ln P > 0$, where $\mu$ is the mean
molecular weight) can lead to the inhibition of convection. The
problem becomes complex because this gradient is \itt{a priori}
unknown: on one hand, convection tends to homogenize layers, leading
to $\nabmu\rightarrow 0$; on the other hand, sharp, diffusive
interfaces can form for which $\nabmu\rightarrow \infty$. This is
indeed observed in the Earth oceans, where salt and heat have opposite
effects. Note, however, that even in the deep oceans of Earth,
vertical transport and mixing is much faster than simple diffusion
would predict, because of the presence of small scale but efficient
double diffusive or salt finger instabilities. These would also occur
in Jupiter, in any region where there is overall stability but the
tendencies of the thermal and compositional gradients are opposed
(e.g. thermal stratification that overwhelms compositional
instability, or where there is compositional stratification that
overwhelms thermal buoyancy).

\item \itt{Condensation}: Phase changes of minor species, such as
water can strongly modify convection. First, the latent heat released
favors updrafts, as observed in Earth's cumulus clouds. In Jupiter's
atmosphere, this leads to convective updrafts of tens of m/s. However,
in hydrogen-helium atmospheres, another effect can be potentially
important: in this case, because any condensing species is heavier
than the surrounding environment, condensation tends to yield a stable
compositional gradient ($\nabmu>0$)\footnote{This is unlike the
Earth's atmosphere in which the condensing molecule, water ($\mu=18$),
is lighter than air ($\mu=29$).}. Guillot (1995) shows that convection
is locally inhibited when the abundance of the condensing species is
larger than a certain critical value. This value is of the order of 5,
15 and 40 times the solar values for \hho, \meth\ and \amm,
respectively. The temperature profiles of Uranus and Neptune retrieved
from radio occultation of Voyager~2 indeed show a strong
superadiabaticity in the region of methane condensation, implying that
convection is probably inhibited by this mechanism.
\end{enumerate}



\section{Internal structure}
\label{sec:interior}

\subsection{The three-layer model}

A division of Jupiter's interior into a minimum of three layers is a
natural consequence of the previously discussed observational facts
and theoretical inferences. These regions are: (i) a helium-poor molecular
hydrogen envelope which includes the atmosphere; (ii) deeper, a
helium-rich metallic hydrogen envelope; (iii) a central dense core of
uncertain composition.  
The necessity of the helium-poor and helium-rich regions arises from
the low atmospheric helium abundance and its explanation by a phase
separation of helium into metallic hydrogen. The presence of a dense
core is generally required to fit the gravitational moments. 
Because the planet is mostly convective, the helium-poor and
helium-rich regions should be relatively homogeneous in terms of
composition. The structure of the core itself is unknown. 

\begin{figure}
%\vspace*{6cm}
\hspace*{-1.5cm}\resizebox{12cm}{!}{\includegraphics[angle=0]{chap3_f5.eps}}
\caption{A schematic view of Jupiter's interior, together with
some of the most important unanswered questions.}
\label{fig:jupint}
\end{figure}

This division in three regions is a good working hypothesis but
it should be stressed that Jupiter may be more complex, as
indicated by the questions raised in fig.~\ref{fig:jupint}. In order
of significance, the first uncertainty is on the location and
extent of the inhomogeneous region in which helium demixing
occurs. The actual representation assumes that this region is narrow,
and occurs in the low-pressure metallic hydrogen (see
section~\ref{sec:h-he}). It is possible that helium separates from
hydrogen at larger pressures and/or on a wider pressure interval,
hence requiring more complex models which are presently not
available. 
The second unknown is the existence of a first order phase
transition between molecular and metallic hydrogen (the PPT). The
equilibrium of the chemical potential would then inevitably yield a
different chemical composition in the two phases. This is taken into
account in the three layer framework only if the helium phase
separation region occurs close to the PPT. 
If not, another layer (between the PPT and
the helium demixing zone)  would have to be added to the models. 
Thirdly, the structure of the central core is unknown. If it is
diluted, its mass would tend to be underestimated, part of it being
added to the metallic region. 
Finally, the possible existence of stable regions at low pressures
(either because of water condensation or because of a radiative zone
at kbar levels) would affect mostly the temperatures at deeper
levels. 

In the framework of the three-layer model, one is conveniently left
with only three main parameters to seek: The mass of the core,
$\mcore$, and the mass mixing ratio of heavy elements in the molecular
and metallic envelopes, $\zmol$ and $\zmet$, respectively. (The helium
mixing ratio in the molecular envelope is set equal to the atmospheric
value; That in the metallic region is constrained by the fact that the
total helium/hydrogen ratio should be equal to the protosolar value).
Although this is clearly a simplification of the problem, the extent
of the ensemble of solutions found within this framework appears to be
a useful representation of the possible structure of Jupiter's
interior. This is due to the fact that because we are interested in
globally averaged quantities significant departures from the results
are possible only through very different initial hypothesis on the
qualitative structure of the planet.
%Note finally that this discussion would apply to Saturn as well. 




\subsection{Constraints on Jupiter's composition from interior models}

Constraints on the compositions of Jupiter and also Saturn can be
obtained from numerical models solving the differential equations of
hydrostatic equilibrium of a rotating fluid object. The system of
equations is very similar to that describing stellar structure (\eg\
Kippenhahn \& Weigert 1990; for its application to giant planets
see DeMarcus 1958; Stevenson \& Salpeter 1976; Zharkov
\& Trubitsyn 1978; Hubbard 1984; for the description of a numerical
method Guillot \& Morel 1995; for the extension to extrasolar giant
planets and brown dwarfs Saumon \etal\ 1996; Guillot 1999b; Burrows et
al. 2001).

The interiors of Jupiter and Saturn share many similarities. While
focussing this review on Jupiter, we will also mention
Saturn when results obtained for this planet can shed light on
Jupiter's internal structure. 

A relatively large number of models of the two giant planets have been
calculated since the pioneering work of Jeffreys (1924). We will not
attempt to review them exhaustively (see Hubbard \& Smoluchovski
1973; Stevenson 1982b; Zharkov 1986 for reviews with a
historical perspective). 
Focussing instead on the models calculated in the past twelve years,
it is interesting to notice a relative convergence of results:
Hubbard \& Marley (1989) allowed the density
profile to vary continuously in the molecular/metallic hydrogen
transition region and found cores of the order of 8
to 14 M$_\oplus$ for Jupiter, and 9 to 20 M$_\oplus$ for Saturn,
depending on the interpolation profiles between molecular and metallic
hydrogen. The
total masses of heavy elements that can be inferred from that work is
about 30 M$_\oplus$ and $20-30$ M$_\oplus$ for Jupiter and Saturn,
respectively. 

In the beginning of the 90s', a new astrophysical EOS for
high-pressure hydrogen and helium became available (Saumon \etal\
1995), and has been used in all interior models of Jupiter calculated
since then. With that EOS, Chabrier \etal\ (1992) found core masses of
4 to 8 M$_\oplus$ and 1 to 20 M$_\oplus$, and total masses of heavy
elements of 10 to 16 M$_\oplus$ and 24 to 30 M$_\oplus$ for Jupiter
and Saturn, respectively. Guillot \etal\ (1994a,b) studied the effect
of a radiative region at the kbar level, and obtained similar results,
but with a bigger core mass (by $1-2\mea$) and a smaller
total amount of heavy elements (by $\sim 10-20\%$). 

In 1995, after a long journey, the Galileo probe successfully measured
the composition and structure of Jupiter's atmosphere. The new value
for the helium mass mixing ratio prompted a reexamination of previous
models: because the new value of $Y$ was higher than the Voyager
measurement, the models matching the gravitational moments would
necessarily yield a smaller quantity of heavy elements in the
molecular region. 
Here, we will focus on results obtained since then by extensive
numerical calculation in which uncertainties on observations (\eg\
gravitational 
moments) and data (\eg\ EOSs) are taken into account (Guillot, Gautier
\& Hubbard 1997; Guillot 1999a). These models are specifically
calculated in the three-layer framework.

Before discussing these results, it is interesting to examine an
alternative approach by Zharkov \& Gudkova (1991) and Gudkova \&
Zharkov (1999). These authors calculated Jupiter's and Saturn's
structure assuming a five layer structure consisting of a helium-poor
external molecular region, a helium-rich internal region, a metallic
region, and a two-layer helium and ice/rock core. This division is
\itt{ad hoc} and it is difficult to imagine formation and/or
high-pressure processes that would have led to this particular
structure. However, it is instructive to examine their results:
ice/rock core masses 5 M$_\oplus$ for Jupiter and 7 M$_\oplus$ for
Saturn, these planets containing about 50 M$_\oplus$ and 25 M$_\oplus$
of heavy elements, respectively (Zharkov \& Gudkova 1991). With the
Galileo data and the Saumon \etal\ (1995) EOS, Gudkova \& Zharkov
(1999) found generally smaller quantities (about $3\mea$ for Jupiter's
core, and $\sim 24\mea$ for the total mass of heavy elements). Even if
the resulting global structure for Jupiter is extremely different,
their results more or less agree with those presented below.

The interior three-layer models of Jupiter from Guillot \etal\ (1997)
and Guillot (1999a) matching all
available observations are shown in fig.~\ref{fig:zel_jup}. 
Hundreds of models
have been calculated, but the solution is represented as a filled area
instead of dots for an easier interpretation of the figure. 
A striking result obtained from Jupiter's modeling is the large
uncertainty due to our relatively poor knowledge of the behavior of
hydrogen at Mbar pressures. As a consequence, two kinds of solutions
are found assuming (i) the Saumon-Chabrier PPT EOS, or (ii) the one
that is smoothly interpolated between the molecular and the
metallic fluids. The uncertainties in the solutions are \itt{not} due
to the qualitative difference at the molecular/metallic 
transition but instead by the quantitatively different density
profiles, as seen in fig.~\ref{fig:densities}. Any solution between the
two regions in fig.~\ref{fig:zel_jup} would be valid, provided the
``true'' EOS for hydrogen lies between the PPT and interpolated EOSs. 

\begin{figure*}
%\vspace*{6cm}
%\centerline{\hbox{\psfig{file=zel_jup.ps,width=14cm}}}
\resizebox{14cm}{!}{\includegraphics[angle=0]{chap3_f6.ps}}
\caption{Constraints on Jupiter's interior structure. The upper panel
shows values of the core mass ($M_\rm{core}$) and total mass of heavy
elements ($M_{Z\rm tot}$) of models matching all available
observational constraints. The lower panel shows the mass mixing
ratio of heavy elements of the molecular ($Z_\rm{mol}$) and metallic
($Z_\rm{met}$) regions, in solar units ($Z_\odot=0.0192$). The two
different regions correspond to different EOSs for hydrogen (see
text). Arrows indicate the direction and magnitude of the 
assumed uncertainties, if $J_4$ or $Y_\rm{proto}$ are increased by
$1\sigma$, rotation is assumed to be solid (``$\Omega$''), the core is
assumed to be composed of ices only (``$f_\rm{ice}$'') and if Jupiter's
interior becomes fully adiabatic (``$\nabla_T$''). 
The dashed line in the lower panel indicates a homogeneous abundance
of heavy elements ($Z_\rm{mol}=Z_\rm{met}$). 
%The hashed regions correspond to models which fit all static
%constraints but yield theoretical ages that disagree with the age of
%the Solar System by more than 10\%. 
The letters A to E label sample models (see
Table~\ref{tab:int_models}). Model F is indicated by a diamond.
[Adapted from Guillot 1999a].}
\label{fig:zel_jup}
\end{figure*}

More quantitatively, fig.~\ref{fig:zel_jup} shows that an upper limit
to Jupiter's core mass is rather small, \ie\ about $10\mea$ only. This
is significantly smaller than found $\sim20$ years ago, the main
difference being due to an (hopefully) improved EOS. The lower limit
on the core 
mass is found to be zero: in this case, Jupiter could have no core, or
a very small one. This corresponds however to rather extreme models,
assuming a hydrogen EOS close to the interpolated one, and a large
$J_4$ value. The lower panel of fig.~\ref{fig:zel_jup} also indicates
that this corresponds to a planet that is enhanced in heavy elements
by 4 to 6 times over the solar value (assuming
$Z_\rm{mol}=Z_\rm{met}$, a consequence of the presence of no physical
discontinuity of
the EOS). Generally, it is found that Jupiter's molecular region is
enriched in heavy elements by 1.5 to 6.5 times the solar value, in
agreement with the observations that indicate a $\sim 3$ times solar
enrichment for C, N, S. 

The results obtained with the PPT EOS indicate that if this EOS is
valid, Jupiter would contain less heavy elements (10 to $20\mea$) than
Saturn (20 to $30\mea$), which is unlikely. 
%The latest magnetic
%compression experiments are in better agreement with EOSs that predict
%a smaller compression of hydrogen, which includes ab-initio models,
%but also the interpolated EOS. 
It appears that therefore Jupiter's
total mass of heavy elements should be in the high end of the
[10,42]$\mea$ range. Note that this does not necessarily imply that the mass
of the core is extremely small because it could be diluted into the
bottom of the metallic envelope. Also, the uncertainty of the EOSs at
pressures of tens of Mbars is difficult to estimate and could be larger
than assumed here. 

It is also interesting to notice that in that case, Jupiter's
molecular region could be consistent with a global enrichment in heavy
elements of three times the solar value, and a larger abundance of
heavy elements in the metallic envelope. Globally, a three times
solar enrichment of heavy elements uniformly over Jupiter's envelope
corresponds to a total mass of $18\mea$. 
However, it has been proposed (Gautier \etal\ 2001) that the heavy
elements brought to Jupiter do not necessarily need to have solar
abundances, and that water could be significantly enriched. In this
case, because oxygen is the third most abundant element in the
universe, one can derive a useful (but high) upper limit to the
abundance of water in the planet: about 10 times the solar abundance. 

\begin{table}
\begin{center}
\caption{Sample models of Jupiter's interior}
\label{tab:int_models}
\begin{tabular}{cccccc}
\hline\hline
Model & H EOS & Radiative? & $T$(1bar) & $M_{\rm
  core}$ & $M_{Z}$ \\ 
 & & & [K] & [M$_\oplus$] & [M$_\oplus$] \\
\hline
A & Int. & no & 165 & 4.2 & 33.1 \\
B & Int. & no & 170 & 0 & 35.3 \\
C & Int. & yes & 165 & 1.5 & 30.1 \\
D & PPT & no & 170 & 10.0 & 17.5 \\
E & PPT & no & 165 & 4.3 & 19.4 \\
F & PPT & yes & 165 & 5.8 & 15.9 \\
\hline
\end{tabular}
\end{center}
\end{table}

For reference, Table~\ref{tab:int_models} lists a few models calculated
with either the interpolated or PPT hydrogen EOS, assuming a fully
convective interior or including the presence of a deep (kbar)
radiative zone, and assuming a 1 bar temperature of either 165 or
170\,K. These models are included as ascii files in the additional
material appended to this chapter. 


\subsection{Location and nature of magnetic field generation}

We could test the interior models if we knew exactly what the 
magnetic field says about the radius below which there is sufficient 
conductivity to generate a dynamo. Alternatively, we could use 
interior models and experimental high pressure data to test our 
theories of dynamos and magnetic field behavior. In reality, we are 
in a situation where neither the interior models nor the theory of 
the field is so well understood that we can treat either as the basis 
for a good test of the other. The best we can do is assess 
consistency between the different considerations. There is indeed a 
fairly good consistency, based on four factors: The magnetic Reynolds 
number (which assesses whether a dynamo is possible), the spectrum of 
the field (which determines the outer radius of dynamo generation), 
the Elsasser number (which determines the expected magnitude of the 
field) and the observational constraint on secular variation (which 
limits the amount of electromagnetic shielding that occurs exterior 
to the dynamo region). In addition to these four consistency checks, 
we have the unanswered questions concerning differential rotation and 
large zonal flows. We also discuss this constraint, though no data 
are yet available to test it.

\itt{(i) The magnetic Reynolds number}. Dynamo theory (Roberts and 
Glatzmaier 2000; Jones 2000) tells us that a dynamo is not possible 
except when the magnetic Reynolds number is sufficiently large:
\begin{equation}
	R_\rm{m} = v_\rm{conv} L/\lambda \wig{>} 10 \mbox{\ to\ } 100
\end{equation}
where $v_\rm{conv}$ is the convective velocity, $L$ is some characteristic
length scale ($\sim$ size of the conducting core) and $\lambda$ is the
magnetic diffusivity. In SI units, $\lambda = 1/(\mu_0\sigma)$ where
$\sigma$ is the electrical conductivity and $\mu_0$ is the
permeability of free space. If we choose $v_\rm{conv} \sim 0.1$ to
$1\cm\sec^{-1}$, as previously discussed, and $L\sim 10^9\cm$ (15\% of
Jupiter's radius) then the largest acceptable value for $\lambda$ is
$\sim 10^7\cm^2\sec^{-1}$. For comparison, the Earth's core magnetic
diffusivity is around $10^4\cm^2\sec^{-1}$, and for idealized monotonic
metallic hydrogen is $10^2$ to $10^3\cm^2\sec^{-1}$ (Stevenson and
Ashcroft 1974). The highest conductivity observed in the pioneering
metallization experiments (Weir et al. 1996; Nellis et al. 1999)
corresponds to $\lambda \sim \mbox{few}\times 10^5\cm^2\sec^{-1}$. The
implication is that the dynamo may operate out to a pressure that is
even lower than the metallization pressure of those experiments,
perhaps into a regime where hydrogen is best thought of as a
semiconductor. In this regime, $\sigma \sim \exp(-\delta E/2kT)$ where
$\delta E$ is the pressure-dependent band gap in molecular hydrogen
and ``metallization'' is interpreted as the location ($\sim1.6$\,Mbar)
where $\delta E$ goes to zero. If the outermost dynamo
generating region is at the pressure corresponding to $\lambda\sim
10^7\cm^2\sec^{-1}$, then this pressure is $\sim 1.3$ Mbar,
corresponding to a radius $\sim 85$\% of Jupiter's radius. This is a
slightly larger radius than previous estimates, but the rapid drop-off
of conductivity with radius means that it may correspond to a
significantly different (and much lower) conductivity than the usual
definition of Jupiter's "metallic core". The problem with this
estimate is that we have no appropriate theoretical models for dynamos
functioning in large conductivity gradients.

\itt{(ii) The magnetic spectrum}. As discussed in 3.2.2, the downward
continuation of the field to 0.8 or 0.85 of Jupiter's radius leads to 
a spectrum (field energy as a function of harmonic degree) that looks 
similar in shape (not magnitude) to the spectrum of Earth's field 
evaluated at the known surface of Earth's core. At this level, the 
spectrum is rather ``flat'' (i.e not much less energy in the octupole 
than in the dipole). This indicates that the argument in (i) is 
consistent with the observed field, the interior models, and our 
knowledge of how dynamos operate. However, it must be emphasized that 
there is a fundamental difference between Earth and Jupiter: The 
core-mantle boundary in Earth is very precisely defined by multiple 
data sets because it is a profound jump in many material properties. 
The surface of the dynamo in Jupiter by contrast appears to be 
ill-defined, e.g. probably not at the location of a phase transition 
or material change. It is therefore unclear to what extent we should 
use a direct comparison between Earth and Jupiter.

\itt{(iii) The Elsasser number}. Dynamo theory tells us that we should
expect a field magnitude corresponding to $\Lambda \sim 1$, where
$\Lambda$ is called the Elsasser number and defined:
\begin{equation}
		\Lambda = {B^2\over 8 \pi\rho\lambda\omega}
\end{equation}
(cgs units with $B$ in Gauss). If we use the low conductivity of the 
outermost region of Jupiter's dynamo and allow for the likely larger 
toroidal field (several tens to 100 Gauss) then $\Lambda\sim 1$ is 
satisfied.

\itt{(iv) Secular variations}. It is also necessary that the low but
non-zero conductivity outside the dynamo allow outward penetration of
any observed time variation of the field. (In geophysics, this is
directly analogous to the constraint people place on the conductivity
of the lower mantle by analyzing geomagnetic ``jerks''). If the outer
edge of dynamo action is placed at $\sim 0.85$ Jupiter radius, then
Liu and Stevenson (in preparation, 2002) find that there is not much
electromagnetic attenuation of variations with characteristic times
similar to Earth or longer, consistent with the upper bound on
observed secular variation (Russell et al. 2001). However, if the
dynamo ceases at a much higher conductivity, then the external field
would be forced to have much less time variation than Earth's field.

\itt{(v) Coupling to the zonal flow}. Last but not least, we should
ask about the coupling of large zonal winds to the field. For flows
like those observed in the atmosphere ($\sim 10^4\cm\sec^{-1}$), one
could get a magnetic Reynold's number exceeding unity even for a
conductivity that is six orders of magnitude less than that usually
associated with a metal (e.g. a conductivity like ocean water on
Earth). This could potentially happen as far out as 90-95\% of
Jupiter's radius (Kirk and Stevenson 1987). Persistence of such large
flows, even in a low conductivity region, would create large but not
directly observable toroidal fields that would act back on the flow,
reducing its strength or creating torsional oscillations (Moffatt
1978). This is poorly understood, but it suggests that there should be
a connection between the structure of the magnetic field and the
postulated gravity signature of differential rotation, discussed in
3.2.1. This is a challenge for future exploration.

\subsection{Relationship to external gravity field}
\label{sec:gravity}

Accurate measurement of Jupiter's external gravitational potential to
very high degree and order would provide data about interior structure
at a level of detail well beyond that currently available.  The
measurements that could be made with a low-periapse orbiter, optimized
to measure subtle gravitational anomalies ($\sim$ 1 mgal), could
potentially answer a number of important questions, such as the depth
of the large-scale zonal flows and the mass of a possible high-density
core.  

\begin{figure}
%\vspace*{6cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f7.eps}}
\caption{Predicted spectrum of Jupiter zonal harmonics for solid-body rotation
in all layers contributing to the external gravitational potential.}
\label{fig:solid_body_rot}
\end{figure}

If there were no interior dynamics in Jupiter and the planet conformed
to hydrostatic equilibrium in the presence of global solid-body
rotation at the rotation rate of the magnetic field, the external
gravitational potential would have a simple form.  All odd
gravitational harmonic coefficients $J_3$, $J_5$, $J_7$, etc. would be
identically zero by symmetry.  The coefficient $J_2$ would be directly
excited by the quadrupole component of the rotational potential.  All
higher finite gravity coefficients are not directly excited by the
solid-body rotational potential, but instead result from the planet's
nonlinear response to that potential.  In this case, the zonal
harmonic spectrum takes the form of a geometric progression with
alternating signs:
\begin{equation}
J_{2n+2} \sim -J_2 J_{2n}.
\end{equation}
\rfigure{solid_body_rot} shows such a zonal-harmonic spectrum for Jupiter
(Hubbard 1999).  The figure shows the predicted absolute values of
zonal harmonics out to degree 20, calculated for a simple polytropic
equation of state.  The open diamonds show the available observed
values for Jupiter (on this plot, an error bar is only discernable for
$J_6$).  Also shown are estimated sensitivity limits for detection of
the $J_n$ with a low-periapse orbiter (J. D. Anderson and E. Lau,
private communication, 2001). 

The observed zonal flows on Jupiter persist to an as-yet unknown
depth.  Schubert and Zhang (2000) have considered possible convection
modes in giant planets, and they predict that for planets such as
Jupiter, intrinsic heat flow will cause large-scale convective rolls
parallel to the rotation axis to be established.  The zonal flows in
the observable atmosphere could thus be the surface expression of such
rolls, as was initially suggested by Busse (1976).  Although such
convective rolls may not penetrate far into the liquid
metallic-hydrogen interior at depths below $\sim 8000$ km due to MHD
stresses, they may persist deep enough to involve significant mass,
even at high latitudes.

\begin{figure}
%\vspace*{6cm}\hspace*{-1.5cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f8.eps}}
\caption{Estimated spectrum of Jupiter zonal harmonics for cylindrical rotation
in all layers contributing to the external gravitational potential. Equivalent
tidal response to Galilean satellites is also shown.}
\label{fig:cylindrical_rot}
\end{figure}

Hubbard (1999) carried out an analysis of the implications for the
high-order Jupiter gravity potential if observed atmospheric zonal
flows are mapped onto concentric cylinders within the jovian interior.
The steady-state structure of the interior is then given by the
two-dimensional solution to the steady-state inviscid fluid equation,
\begin{equation}
{\bf v} \cdot \nabla {\bf v}=-{1 \over \rho}\nabla P + \nabla V,
\end{equation}
where $V$ is the gravitational potential.  Here the velocity field
${\bf v}$ is assumed to be time-independent and strictly zonal.  Since
the actual zonal flow pattern in Jupiter is not exactly north-south
symmetric, rotation on cylinders cannot be consistently applied to
both hemispheres simultaneously.  Hubbard (1999) carries out separate
solutions for Jupiter's northern and southern hemispheres.
\rfigure{cylindrical_rot} shows a guess for the predicted harmonic
spectrum for Jupiter, obtained by fitting a global gravitational
potential to the northern and southern potentials simultaneously.
This figure does not ``predict'' the magnitude of the odd zonal
harmonics; such a calculation would require a detailed solution of
Eq. (1) with north-south variation of the zonal rotation rates and
baroclinic temperature profiles.  However, the figure does suggest
that low-order odd harmonics may be eventually detectable from a
spacecraft with moderately improved sensitivity to the jovian gravity
field, perhaps even with a single orbit.

According to Hubbard's (1999) calculations, if Jupiter's zonal flows
persist to a depth of about 1000 km below the cloud layers, i.e. to
pressures exceeding $\sim$ 10 kbar, they will involve enough mass to
produce gravity anomalies detectable by a low-periapse orbiter.
\rfigure{gravity_anom} shows a diagram illustrating the correlation
between gravity anomalies and possible jovian ``topography''.  The
topography, actually the shape of isobars under the influence of zonal
rotation, is shown exaggerated by about a factor of 1000.  The vectors
to the right of the figure show gravity anomalies calculated with
respect to a planet rotating as a solid body.  The maximum values of
these gravity anomalies are about 30 mgal.  As can be seen, the rapid
zonal rotation at low latitudes causes a low-latitude height increase
of the isobars, yielding extra gravity.  However, the reference
solid-body rotation gravity field is not observationally accessible.
Instead, one will need to reference to a ``smooth'' gravity field,
such as one generated by means of the lower zonal harmonics, say out
to $J_{10}$.  The vectors on the left-hand side of \rfig{gravity_anom}
show such anomalies, this time with a maximum amplitude $\sim$ 1 mgal.

\begin{figure}
%\vspace*{6cm}\hspace*{-1.5cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f9.eps}}
\caption{Schematic figure showing possible correlation between Jupiter
``topography'' produced by cylindrical zonal flows and gravity
  anomalies (arrows). The vectors to the right represent gravity
  anomalies calculated with respect to a planet rotating as a solid
  body. The vectors to the left represent gravity anomalies with
  respect to a ``smooth'' gravity field (calculated only from the
  first few gravitational moments) (see text).}
\label{fig:gravity_anom}
\end{figure}

Although the dynamics-excited components of Jupiter's gravity field
will be very informative about the depth of penetration of zonal
flows, they are a noise source for attempts to infer the mass of a
possible dense core by combining gravity data with high-pressure
equations of state (Guillot, Gautier, and Hubbard 1997).  An
independent measurement of Jupiter's response to a perturbation field
would help to remove such ambiguity.  The tidal response of Jupiter to
the gravitational perturbations from the Galilean satellites appears
to be measurable by a low-periapse orbiter, and may
provide such constraints. \rfigure{cylindrical_rot} shows the equivalent
components of Jupiter's gravity potential (expressed as zonal-harmonic
coefficients) in response to the tidal perturbations of the four
Galilean satellites.  These components were calculated for a simple
polytropic equation of state, using an approach similar to that of
Gavrilov and Zharkov (1977).  Although Jupiter's tidal response is
$\sim$ 5 to 6 orders of magnitude smaller than its response to
rotation, the tidal response potential has a different spatial
dependence than the rotation response potential, and, using multiple
orbits, could be synchronously detected.  Importantly, Jupiter tides
represent a linear response to a well-characterized perturbation.
Because they are time-dependent, the possibility of resonance with
free-oscillation modes must be investigated.  Their detection from a
Jupiter orbiter would open up rich possibilities for detecting
interior structure, including a dense core.


\section{Evolution}
\label{sec:evolution}

\subsection{Jupiter's contraction}

Another constraint to the problem of Jupiter's interior comes from the
age of the planet as inferred from that of the solar system itself (4.55
Gyr). However, that constraint is difficult to use because it requires
a good knowledge of Jupiter's atmosphere and interior not only
presently but in the past as well. Modern evolution models calculated
assuming a homogeneous structure (i.e. neglecting the presence of a
hydrogen-helium phase separation) predict cooling times of 4.7 to
5.1 Gyr (Saumon \etal\ 1992; Guillot \etal\ 1995; Guillot 1999a). In
the absence of alkali metals the presence of a radiative zone
leads to a faster loss of the internal entropy of the planet and can
lead to a minimum age of 3.6 Gyr. However, as discussed previously,
this fast cooling is unlikely because alkali metals are expected to be
present in Jupiter's atmosphere. Jupiter's models appears therefore to
be cooling too slowly. 

The problem is even more acute when one considers the phase
separation of helium. In that case the falling of helium-rich droplets
provides an additional source of energy which slows the cooling of the
planet by several hundreds of millions of years (Hubbard \etal\ 1999;
Guillot 1999a). 

Several solutions around that problem exist. A first one is
the fact that the atmospheric boundary condition still used in these
evolution models is extremely coarse: basically, it is fitted to the
planet's present 1 bar temperature and extrapolated to higher
luminosities and atmospheric temperature using a simple relation
between atmospheric temperature, effective temperature and gravity
(Hubbard 1977; see also Saumon et al. 1996). 
This is expected to introduce an uncertainty on the calculation of
about 10\%. 
Another possible factor is the fact that Jupiter was not necessarily
formed from an initially infinitely extended, isotropic gas cloud. In
the core-accretion formation scenario, the planet can have an extended
phase of several Myr in which its mass is of order $20\mea$. In the
extreme case that this protoplanetary core is supposed to lose its
entropy efficiently during that period, the subsequent evolution to
Jupiter's present luminosity and radius is hastened by as much as 500
Myr. 
Even in the case of the gravitational collapse, the formation of an
accretion disk can slow the accretion onto the central planet,
allowing it to lose its entropy more efficiently than is assumed for
isotropic contraction (radiation can escape freely in all directions
except in the plane of rotation of the disk). 
Another possible 
explanation would be the presence of a compositional gradient deep 
within the planet. This can arise through the accretion but incomplete
mixing of large planetesimals (up to $\sim$Earth mass in size). In this 
situation, a significant portion of the planet's heat reservoir is 
unavailable for the total luminosity. The time required to cool to 
the present observed temperature is then reduced. A scenario like 
this has been proposed for Uranus and Neptune (Podolak et al. 1991; 
Hubbard et al. 1995) where there is a much more striking discrepancy 
between the observed heat flow and the predictions of a fully 
adiabatic cooling model.
Finally, a perhaps even more promising explanation is that Jupiter may
have formed from an initially more 
massive dense core and is progressively eroding it (\S 3.6.3). The
massive core  
would have yielded an initially high intrinsic luminosity, and more
energy would thus have been lost from the planet in, say, the first few
millions of years. The erosion would then imply a transformation of
internal energy into gravitational energy, thereby further quickening
its cooling. 


\begin{figure}
%\vspace*{6cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f10.ps}}
\caption{Evolution of the radius (corresponding to the 10 bar level)
of a $318\mea$ Jupiter. The dashed line corresponds to an idealized
scenario in which the planet is assumed to have formed at $t=0$ with
an arbitrarily 
high initial radius. In the second case (solid line), an inner $\sim
30\mea$ is assumed to formed and cooled in 3\,Myr, epoch at which the
planet suddenly accreted most of its present gaseous envelope. The
dotted line indicates in this case the locus of the 100\,mbar radius
level. As indicated by the hashed region, the inital evolution of
Jupiter is expected to have occurred when the protosolar nebula was
still present. The present positions of the corotation radius and
semi-major axis of Metis, Jupiter's innermost moon are indicated by
dotted horizontal lines.} 
\label{fig:evjup}
\end{figure}

Figure~\ref{fig:evjup} shows the expected variation in radius of a
fully-formed Jupiter as a function of time. A first, important
fact is that the planet's radius has decreased over time only by a
modest factor. The evolution curve is reasonnably well
understood for the period that extends from about 10\,Myr
after Jupiter's formation onward. However, before that epoch, several
factors prevent an accurate representation of the planet's evolution. 
As inferred from observations of young stars, it is expected
that the protosolar nebula was present for about 10\,Myr
(this precise value being itself quite uncertain). Jupiter could then
have accreted enough material from the nebula during that time to
alter its evolution. Alternatively, it could have led to the end of
the nebula phase by accreting most of its gas rapidly. Depending on
the duration of the formation phase, Jupiter's inner regions could
have had the time to cool or not, affecting the planet's energy
budget. More detailed studies of that problem would require modeling
the formation of a planet with a circumplanetary disk, which itself
led to the formation of the regular moons. 



\subsection{On the origin of the heavy element enrichment}

Jupiter and Saturn both contain a significant mass of heavy
elements. Were these elements acquired early during the first phases
of formation of the planets or later by impacts similar to the
Shoemaker-Levy 9 event? This question is relatively new, because it is
only since the in situ measurements by the Galileo probe that it has
been proven that Jupiter's atmosphere is enriched over the solar value
in several elements.

A first possible explanation may be that Jupiter and Saturn, after
their formation, cleared the nebula of its planetesimals and acquired
a fraction of them. Dynamical simulations (Guillot \& Gladman 2001;
Hahn \& Malhotra 1999) show however that as soon as Jupiter grows to
its present mass, the accretion efficiency of all giant planets
becomes small. Quantitatively, if they were suddenly plunged into a
$100\mea$ disk of planetesimals distributed between 4 and 35\,AU,
Jupiter and Saturn (with their present masses, radii and orbital
parameters) would accrete about $4.4\mea$ and $0.9\mea$, respectively,
the rest of the planetesimals being ejected by Jupiter or sent to the
Oort cloud (Guillot \& Gladman 2001). This result is for a disk
surface density $\sigma\propto r^{-3/2}$ ($r$ being the heliocentric
distance), but changing that distribution has only modest
consequences. The result should also hold for the same fully-formed
planets 4.5 Gyr ago, because of the modest increase in their radii
(fig.~\ref{fig:evjup}). For comparison, Pollack et al. (1996) require
a disk mass of $220\mea$ in solids (and its complement in gas) between
4 and 35\,AU to form Jupiter, Saturn and Uranus. Given the fact that
most of the mass that is initially close to a giant planet and has the
largest impact probability is used to form a core during the runaway
growth phase, it appears that it is difficult to account for the
observed atmospheric enrichment of Jupiter and Saturn in the framework
of a late accretion of planetesimals scenario (after Jupiter's mass
increased to $\sim 300\mea$).

An alternative scenario could be that Jupiter and Saturn formed
rapidly in a still massive nebula (Boss 2000 and references
therein). However, the problem may be even more difficult because
a mechanism then needs to be found to prevent gas accretion by the giant
planets, while allowing the capture of solid bodies.

A more plausible explanation is hence that the giant planets acquired
their heavy elements through the rapid growth of their cores (the
runaway growth phase -see Safronov 1972; Lissauer 1993), and in a
phase in which they had grown to about $\sim 20\mea$ and in which they
were in quasi-static equilibrium with the surrounding nebula (Pollack
\etal\ 1996). In that later phase, the accretion efficiency is
increased significantly (Guillot \& Gladman 2001)
due (i) to a smaller probability of ejection of planetesimals out of
the solar system (Jupiter being much less massive than it is today)
and (ii) to a larger effective capture radius by the very extended
gaseous envelopes (\eg\ Podolak \etal\ 1988). 

Because the planetesimals are almost always dissolved very close
to the central regions (Podolak \etal\ 1988; Pollack \etal\ 1996), the
rapid capture of the gas of the nebula is expected to create a giant
planet consisting mostly of a massive core and a low-metallicity
gaseous envelope. This scenario implies that part of this
primordial core had to be eroded to enrich the gasous envelope in
heavy elements. 


\subsection{Core erosion}

To first order, the redistribution of a small core of mass $\Delta
m_\rm{core}$ in a planet of total mass $M$ and radius $R$  ``costs''
in terms of energy:
\beq
\Delta E_\rm{grav}=-\varpi{GM\over R}\Delta m_\rm{core},
\eeq 
where $\varpi$ is a factor of order unity, which is equal to $3/10$ in
the limit of a very small core of finite density, and a fully
incompressible planet (a sufficient approximation for our purposes).
The erosion of the core can be thought to proceed through the
penetration of downward convective plumes. In that case, the
erosion mass flux is:
\beq
\dot{m}_\rm{core}(t)=-{\chi \over \varpi}{R L_1(t)\over GM},
\label{eq:mdot}
\eeq
where $L_1(t)$ is the planet's intrinsic luminosity in the first
convective cell (after one pressure scale height), and $\chi$ is an
efficiency factor. As discussed by Stevenson (1982c), the erosion is
limited by diffusion processes at the core/envelope interface, and one
can expect that $\chi \approx (D/K)^{1/2}$ where $D$ and $K$ are the
molecular and thermal diffusivity, respectively. For metallic
hydrogen, one expects $D\sim
10^{-3}-10^{-4}\cm^2\sec^{-1}$ and $K\sim
10^{-1}\cm^2\sec^{-1}$ (Stevenson \& Salpeter 1977a; Stevenson 1982c),
implying, within an order of magnitude, $\chi \sim 0.1$.

\begin{figure}
%\vspace*{6cm}
\resizebox{\hsize}{!}{\includegraphics[angle=0]{chap3_f11.eps}}
\caption{Mass of heavy elements initially embedded in a central core
and progressively redistributed, according to a speculative scenario
in which 10\% 
($\chi=0.1$) of the intrinsic energy flux in the first pressure scale
height is used to mix the heavy elements upward [see
eq.~(\ref{eq:mdot})]. 
Jupiter and Saturn were assumed to consist of a $30\mea$ core and a
hydrogen-helium envelope. The decrease of the core mass in time and
the non-zero compressibility of core material were neglected in the
calculation of $\varpi$.}
\label{fig:erosion}
\end{figure}

Figure~\ref{fig:erosion} shows the evolution of $\int_0^t
\dot{m}_\rm{core} dt$ for Jupiter and Saturn. In the figure, the $t=0$
origin corresponds to the time at which the planets accrete their
gaseous envelopes. In the simulations, the two planets were assumed to
possess a $30\mea$ core. The value of $\varpi$ was calculated for
an incompressible planet (both for the core and the envelope). The
parameter $\chi$ was arbitrarily set to 0.1. Apart from the
quantitative results which are quite uncertain (both because of
uncertainties of $\chi$ and of our assumption of downward overshoot of
convective plumes in the first pressure scale height which represent
a significant $\sim 10\%$ fraction of the planet), it is interesting to
notice that the redistribution
of the core mass would have been much more pronounced in Jupiter than
in Saturn. This is in agreement with interior models that generally
predict that Jupiter has a smaller core than Saturn (\eg\ Guillot
1999a). With simple hypotheses (quasi-homogeneous planets and
a solar C/O ratio), the observed atmospheric abundances appear to
require the redistribution of $\sim 18\mea$ in Jupiter and $\sim
4\mea$ in Saturn. The difference between Jupiter and Saturn is more
pronounced in fig.~\ref{fig:erosion}, but it may be advocated (i) that
with a slightly larger $\chi$ factor, Jupiter's core may have been
fully eroded and (ii) that while we have assumed $\varpi$ of order
unity, this parameter will be smaller for material with a large
compressibility. A full modeling of the evolution of the cores of the
two planets including consistent high-pressure equations of state
would be required in order to further test this hypothesis. 

The possible erosion of the cores of Jupiter and Saturn is important for
formation models in the core accretion scenario. As shown by Pollack
\etal\ (1996) (see also the chapter by Lunine \etal), the formation
of a small ($\sim 10\mea$) core requires
a relatively low surface density of solids in the protosolar nebula
and a consequently relatively long formation timescale ($\sim
8$\,Myr). However, if 
Jupiter's primordial core is allowed to be relatively large, this
constraint is relaxed, and the planet could thus form rapidly. 


% \footnote{
% Another process that would lead to a redistribution of the core mass
% is the likely presence of a strong shear between the core and the
% gaseous envelope (due to the fact that core accretion is a stochastic
% process whereas envelope capture is an orderly process which maximizes
% rotation rates). In that case the energy gained by mixing a
% non-rotating core of mass $dm_\rm{core}$ with an envelope with a
% rotation rate $\omega$ is the difference in kinetic energy of the
% initial and final states:
% \beq 
% \Delta E_\rm{kin}={k^2\over 2}R^2 \omega^2 \Delta m_\rm{core},
% \label{eq:ekin}
% \eeq
% where $k$ is the dimensionless radius of gyration ($k^2=2/5$ for
% a planet of uniform density). This shear instability would thus be
% favored for $\Delta E_\rm{kin}>\Delta E_\rm{grav}$. To first order,
% both terms depend linearly on $\Delta m_\rm{core}$: this process can
% occur only for large values of the initial rotation rate $\omega$ (of
% order the breakup velocity), or low values of $\varpi$. Because for
% present values of $\omega$ $\Delta E_\rm{kin}$ is always two orders of
% magnitude smaller than $\Delta E_\rm{grav}$ the process can be
% neglected. 
% }


\section{Future prospects}
\label{sec:future}

Despite numerous space missions that have flown past
Jupiter, the planet has kept many of its secrets: we do not know what
quantities of heavy elements it contains, we do not know if it
possesses a central core, and we still have to guess how and where its
magnetic field is generated.  Progress concerning these key
questions will be partly addressed by better experimental results on
hydrogen compression to ultra-high pressures. However, improvement in
our knowledge of Jupiter's interior will eventually require three key
measurements: (i) a determination of the bulk abundance of water; (ii)
mapping the planet's gravity field with high accuracy and spatial
resolution; (iii) mapping the planet's magnetic field with high
accuracy and spatial resolution.

These measurements could be accomplished in the near future by
sending a spacecraft within $\sim 3000$\,km
of Jupiter's cloud tops, much closer than any previous spacecraft
(except the Galileo Probe).
The abundance of water at deep levels ($\sim 100\,$bars and more)
could then be retrieved from the brightness temperature at millimeter
wavelengths, the low altitude preventing confusion from
synchrotron emission (Bolton et al. 2001). With a low-periapse
orbiter, the gravity field can be mapped with very high accuracy, so that
gravitational components arising from non-hydrostatic effects can be
measured (Hubbard 1999). The magnetic field can be characterized at
the same time in all its complexity.

In parallel, further progress in understanding Jupiter's interior will
come from theoretical studies of: (i) improved equations of state and
solubities of hydrogen, helium and other elements; (ii) specific
atmospheric models that account for the stellar irradiation, observed
non-solar abundances of chemical elements and presence of moist
convection; (iii) models of convection in the presence of rotation and
magnetic fields; (iv) a description of the core/envelope interface
from the point of view of fluid dynamics (including diffusion,
conduction and convection in the presence of molecular weight
gradients); (v) a theory of rotating fluid figures of (non-idealized)
giant planets in the presence of complex internal rotation
patterns. These improvements will pave the way for more realistic
interior models fitting the gravitational field in all its complexity,
and for detailed evolution models that can account for Jupiter's age
of $\sim 4.55\,$Gyr.

Jupiter's exploration must be pursued because it has far-reaching
astrophysical consequences.  While the Galileo probe has precisely
measured the abundance of helium in Jupiter's atmosphere, the
determination of the abundance of water in Jupiter's interior would
constitute a great leap in completing the inventory of oxygen in the
Solar System.  Oxygen being the third most abundant element in the
universe, this is an important clue to understanding how our Solar
System was formed.  Similarly, measuring Jupiter's deep winds has
implications not only on the planetary structure but on meteorology in
general. Jupiter also has the second most powerful magnetic dynamo in
the Solar System (after the Sun's), and understanding it is crucial
for magnetohydrodynamics in general. Finally, Jupiter appears to be a
nearby example of a new class of astronomical objects: giant planets
and substellar objects (brown dwarfs).  With more than a hundred
members (see e.g. Mayor \& Queloz 1995; Marcy et al. 2000 and the
discoverers' web pages), this new class must characterized by a dual
approach: (i) observations of extrasolar planetary systems, for which
statistically significant information can be sought and (ii) a precise
exploration of the giant planets in the Solar System, especially
Jupiter, for which much more elaborate data can be gathered and
physical processes can be studied in detail.


\section*{Acknowledgements}

This work was supported in part by the \itt{Programme National de
Plan\'etologie}, by an \itt{Action Concert\'ee Incitative} of the
French \itt{Minist\`ere de la Recherche}, by NASA grants NAG5-7211,
NAG5-7499, NAG5-10629 (Origins Program), NAG5-4214 (Planetary
Astronomy), NAG5-7073, NAG5-10760 (Astrophysics Theory), NAG5-8906
(Planetary Geology and Geophysics Program), and by the US Department
of Energy under contract W-7405-ENG-36.



\section*{Material appended to this chapter}
15 ascii files (representing evolution and interior models of Jupiter)
are joined to this chapter and can also be found at the following site:\\
ftp://ftp.obs-nice.fr/pub/guillot/jupbook/


\begin{thebibliography}{}

% Most of the references were compiled from ADS using the following
% custom format:
% \\bibitem%[\2N]%{R} %\10L (%\Y). %\t.{\\em %\j}, %\V:%\p%-\P\.\n 

\bibitem[Acu\~na et al.(1983)]{1983JGR....88.8771A} Acu\~na, M.~H., Connerney, 
J.~E.~P., and Ness, N.~F.\ (1983). The Z3 zonal harmonic model of Saturn's 
magnetic field Analyses and implications.\ {\em \jgr}, 88: 8771-8778. 

\bibitem[Aurnou and Olson(2001)]{2001GeoRL..28.2557A} Aurnou, J.~M.~and 
Olson, P.~L.\ (2001). Strong zonal winds from thermal convection in a 
rotating spherical shell.\ {\em \grl}, 28: 2557. 

\bibitem[Bahcall and Pinsonneault (1995)]{1995Bahcall} Bahcall J.N.,
Pinsonneault M.H. (1995). Solar models with helium and heavy elements
diffusion. {\em Rev. Mod. Phys.} 67: 781.

\bibitem[Beule et al.(1999)]{1999PhRvB..5914177B} Beule, D., Ebeling, W., 
F{\" o}rster, A., Juranek, H., Nagel, S., Redmer, R., \& R{\" o}pke, G.\ 
(1999). Equation of state for hydrogen below 10000 K: From the fluid to
the plasma. {\em Phys. Rev. B}, 59: 14177-14181.

\bibitem[Bolton et al.(2001)]{2001iaop.work.4038B} Bolton, S.~J., Owen, T., 
Gautier, D., Gulkis, S., Janssen, M., Atreya, S., Guillot, T., Anderson, 
J., Allison, M., and Lunine, J.\ (2001). Jupiter: Atmospheric Sounding and 
Sensing of the Interior (JASSI).\ {\em Forum on Innovative Approaches to 
Outer Planetary Exploration 2001-2020}, 4038. 

\bibitem[Boss(2000)]{2000ApJ...536L.101B} 
Boss, A.~P. (2000). Possible Rapid Gas Giant Planet Formation in the
Solar Nebula and Other Protoplanetary Disks. {\em \apjl}, 536: L101-104. 

\bibitem[Burrows et al.(2000)]{2000ApJ...531..438B} Burrows, A., Marley, 
M.~S., and Sharp, C.~M.\ (2000). The Near-Infrared and Optical Spectra of 
Methane Dwarfs and Brown Dwarfs.\ {\em \apj}, 531: 438-446. 

\bibitem[Burrows et al.(2001)]{2001RvMP...73..719B} Burrows, A., Hubbard, 
W.~B., Lunine, J.~I., and Liebert, J.\ (2001). The theory of brown dwarfs 
and extrasolar giant planets.\ {\em Reviews of Modern Physics}, 73: 
719-765. 

\bibitem[Busse(1976)]{1976Icar...29..255B} Busse, F.~H.\ (1976). A simple 
model of convection in the Jovian atmosphere.\ {\em Icarus}, 29: 255-260. 

\bibitem[Campbell and Synnott(1985)]{1985AJ.....90..364C} Campbell, 
J.~K.~and Synnott, S.~P.\ (1985). Gravity field of the Jovian system from 
Pioneer and Voyager tracking data.\ {\em \aj}, 90: 364-372. 

\bibitem[Celliers et al.(2000)]{ } Celliers, P.~M., et al. (2000). Shock-induced    
transformation of liquid deuterium into a metallic fluid.\ 
{\em Physical Review Letters}, 84: 5564-5567.

\bibitem[Chabrier et al.(1992)]{1992ApJ...391..817C} Chabrier, G., Saumon, 
D., Hubbard, W.~B., and Lunine, J.~I.\ (1992). The molecular-metallic 
transition of hydrogen and the structure of Jupiter and Saturn.\ {\em 
\apj}, 391: 817-826. 

\bibitem[Chau et al.(2001)]{2001JChPh.114.1361C} Chau, R., Mitchell, A.~C., 
Minich, R.~W., and Nellis, W.~J.\ (2001). Electrical conductivity of water 
compressed dynamically to pressures of 70-180 GPa (0.7-1.8 Mbar).\ {\em 
\jcp}, 114: 1361-1365. 

\bibitem[Cohen and Taylor(1987)]{1987RvMP...59.1121C} Cohen, E.~R.~and 
Taylor, B.~N.\ (1987). The 1986 adjustment of the fundamental physical 
constants.\ {\em Reviews of Modern Physics}, 59: 1121-1148. 

\bibitem[Collins et al.(1998)]{1998Collins} Collins, G.W., Da Silva,
L.B., Celliers, P., Gold, D.M., Foord, M.E., Wallace, R.J., Ng, A.,
Weber, S.V., Budil, K.S., and Cauble, R. (1998). Measurements of the
equation of state of deuterium at the fluid insulator-metal
transition. {\em Science}, 281: 1178-1181.

\bibitem[Collins et al.(2001)]{2001Collins}
Collins, G. W., Celliers, P. M., da Silva, L. B., Cauble, R., Gold,
D. M., Foord, M. E., Holmes, N. C., Hammel, B. A., Wallace, R. J., and
Ng, A. (2001). Temperature Measurements of Shock Compressed Liquid
Deuterium up to 230 GPa. {\em Phys. Rev. Let.}, 87: 165504-+4.

\bibitem[Connerney et al.(1982)]{1982JGR....87.3623C} Connerney, J.~E.~P., 
Acu\~na, M.~H., and Ness, N.~F.\ (1982). Voyager 1 assessment of Jupiter's 
planetary magnetic field.\ {\em \jgr}, 87: 3623-3627. 

\bibitem[Conrath et al.(1989)]{1989oeps.book..513C} Conrath, B.~J., Hanel, 
R.~A., and Samuelson, R.~E.\ (1989). Thermal structure and heat balance of 
the outer planets.\ {\em Origin and Evolution of Planetary and Satellite 
Atmospheres}, 513-538. 

\bibitem[Da Silva et al.(1997)]{ } Da Silva, L.~B., et al. (1997). Absolute equation
of state measurements on shocked liquid deuterium up to 200 GPa (2 Mbar).\ 
{\em Physical Review Letters}, 78: 483-486.  

\bibitem[Davies et al.(1986)]{1986CeMec..39..102D} Davies, M.~E., Abalakin, 
V.~K., Bursa, M., Lederle, T., Lieske, J.~H., Rapp, R.~H., Seidelman, 
P.~K., Singclair, A.~T., Tejfel, V.~G., and Tjuflin, Y.~S.\ (1986). Report 
of the IAUIAG COSPAR working group on cartographic coordinates and 
rotational elements of the planets and satellites: 1985.\ {\em Celestial 
Mechanics}, 39: 102-113. 

\bibitem[DeMarcus(1958)]{1958AJ.....63....2D} DeMarcus, W.~C.\ (1958). The 
constitution of Jupiter and Saturn.\ {\em \aj}, 63: 2. 

\bibitem[Deming et al.(1989)]{1989ApJ...343..456D} Deming, D., Mumma, 
M.~J., Espenak, F., Jennings, D.~E., Kostiuk, T., Wiedemann, G., 
Loewenstein, R., and Piscitelli, J.\ (1989). A search for p-mode 
oscillations of Jupiter - Serendipitous observations of nonacoustic thermal 
wave structure.\ {\em \apj}, 343: 456-467. 

\bibitem[Dormy et al. 2000]{2000Dormy} Dormy, E., Valet, J. P., and
Courtillot, V. (2000) Numerical models of the geodynamo and
observational constraints. G-cubed. 1: 62

\bibitem[Galli et al.(2000)]{2000PhRvB..61..909G} Galli, G., Hood, R.~Q., 
Hazi, A.~U., and Gygi, F. (2000). Ab initio simulations of compressed 
liquid deuterium.\ {\em \prb}, 61: 909-912. 

\bibitem[Gautier et al.(2001)]{2001ApJ...559L.183G} Gautier, D., Hersant, 
F., Mousis, O., and Lunine, J.~I.\ (2001). Erratum: Enrichments in 
Volatiles in Jupiter: A New Interpretation of the Galileo Measurements.\ 
{\em \apjl}, 559: L183-L183. 

\bibitem[Gavrilov and Zharkov(1977)]{1977Icar...32..443G} Gavrilov, 
S.~V.~and Zharkov, V.~N.\ (1977). Love numbers of the giant planets.\ {\em 
Icarus}, 32: 443-449. 

\bibitem[Gierasch and Conrath(1993)]{1993JGR....98.5459G} Gierasch, 
P.~J.~and Conrath, B.~J.\ (1993). Dynamics of the atmospheres of the outer 
planets - Post-Voyager measurement objectives.\ {\em \jgr}, 98: 5459-5469. 

\bibitem[Goldreich and Kumar(1988)]{1988ApJ...326..462G} Goldreich, P.~and 
Kumar, P.\ (1988). The interaction of acoustic radiation with turbulence.\ 
{\em \apj}, 326: 462-478. 

\bibitem[Goldreich and Soter(1966)]{1966Icar....5..375G} Goldreich, P.~and 
Soter, S.\ (1966). Q in the Solar System.\ {\em Icarus}, 5: 375-389. 

\bibitem[Gudkova and Zharkov(1999)]{1999P&SS...47.1201G} Gudkova, T.~V.~and 
Zharkov, V.~N.\ (1999). Models of Jupiter and Saturn after Galileo 
mission.\ {\em \planss}, 47: 1201-1210. 

\bibitem[Guillot(1995)]{1995Sci...269.1697G} Guillot, T.\ (1995). 
Condensation of Methane Ammonia and Water in the Inhibition of Convection 
in Giant Planets.\ {\em Science}, 269: 1697. 

\bibitem[Guillot(1999a)]{1999P&SS...47.1183G} Guillot, T.\ (1999a). A 
comparison of the interiors of Jupiter and Saturn.\ {\em \planss}, 47: 
1183-1200. 

\bibitem[Guillot(1999b)]{1999Sci...286...72G} Guillot, T.\ (1999b). Interior 
of Giant Planets Inside and Outside the Solar System.\ {\em Science}, 286: 
72-77. 

\bibitem[Guillot et al.(1996)]{1996ApJ...459L..35G} Guillot, T., Burrows, 
A., Hubbard, W.~B., Lunine, J.~I., and Saumon, D.\ (1996). Giant Planets at 
Small Orbital Distances.\ {\em \apjl}, 459: L35. 

\bibitem[Guillot et al.(1995)]{1995ApJ...450..463G} Guillot, T., Chabrier, 
G., Gautier, D., and Morel, P.\ (1995). Effect of Radiative Transport on 
the Evolution of Jupiter and Saturn.\ {\em \apj}, 450: 463. 

\bibitem[Guillot et al.(1994a)]{1994Icar..112..337G} Guillot, T., Gautier, 
D., Chabrier, G., and Mosser, B.\ (1994a). Are the giant planets fully 
convective?\ {\em Icarus}, 112: 337-353. 

\bibitem[Guillot et al.(1994b)]{1994Icar..112..354G} Guillot, T., Chabrier, 
G., Morel, P., and Gautier, D.\ (1994b). Nonadiabatic models of Jupiter and 
Saturn.\ {\em Icarus}, 112: 354-367. 

\bibitem[Guillot et al.(1997)]{1997Icar..130..534G} Guillot, T., Gautier, 
D., and Hubbard, W.~B.\ (1997). NOTE: New Constraints on the Composition of 
Jupiter from Galileo Measurements and Interior Models.\ {\em Icarus}, 130: 
534-539. 

\bibitem[Guillot and Morel(1995)]{1995A&AS..109..109G} Guillot, T.~and 
Morel, P.\ (1995). CEPAM: a code for modeling the interiors of giant 
planets..\ {\em \aaps}, 109: 109-123. 

\bibitem[Guillot and Gladman(2001)]{2001Guillot} Guillot T. and
Gladman B. (2000). Late Planetesimal Delivery and the Composition of
Giant Planets, {\em Proceedings of the Disks, Planetesimals and
Planets Conference, ASP Conference Series vol. 219, F. Garzon et
al. eds.}, 475-485.

\bibitem[Guillot and Showman(2002)]{2002A&A...385..156G} Guillot, T.~and 
Showman, A.~P.\ (2002). Evolution of ``51 Pegasus b-like'' planets.\ {\em 
\aap}, 385: 156-165. 

\bibitem[Hahn and Malhotra(1999)]{1999AJ....117.3041H} Hahn, J.~M.~and 
Malhotra, R.\ (1999). Orbital Evolution of Planets Embedded in a 
Planetesimal Disk.\ {\em \aj}, 117: 3041-3053. 

\bibitem[Hanel et al.(1981)]{1981JGR....86.8705H} Hanel, R., Conrath, B., 
Herath, L., Kunde, V., \& Pirraglia, J. (1981). Albedo, internal heat,
and energy balance of Jupiter - Preliminary results of the Voyager
infrared investigation. {\em \jgr}, 86: 8705-8712. 

\bibitem[Hemley et al.(1990)]{1990PhRvB..42.6458H} Hemley, R.~J., Mao, 
H.~K., Finger, L.~W., Jephcoat, A.~P., Hazen, R.~M., and Zha, C.~S.\ 
(1990). Equation of state of solid hydrogen and deuterium from 
single-crystal x-ray diffraction to 26.5 GPa.\ {\em \prb}, 42: 6458-6470. 

\bibitem[Hide, R. and Stannard, D. (1976)]{1976Hide} Hide, R. and
Stannard, D. (1976) Jupiter's magnetism: Observations and theory. In
{\em Jupiter}, ed. T. Gehrels, Univ. Arizona Press, Tucson,  
pp 767-787.

\bibitem[Holmes et al.(1995)]{ } Holmes, N.~C., Ross, M. and Nellis, W.~J. (1995). 
Temperature measurements and dissociation of shock-compressed liquid deuterium
and hydrogen.\ {\em Physical Review B}, 52: 15835-15845.

\bibitem[Hubbard(1968)]{1968ApJ...152..745H} Hubbard, W.~B.\ (1968). 
Thermal structure of Jupiter.\ {\em \apj}, 152: 745-754. 

\bibitem[Hubbard(1977)]{1977Icar...30..305H} Hubbard, W.~B.\ (1977). The 
Jovian surface condition and cooling rate.\ {\em Icarus}, 30: 305-310. 

\bibitem[Hubbard(1982)]{1982Icar...52..509H} Hubbard, W.~B.\ (1982). 
Effects of differential rotation on the gravitational figures of Jupiter 
and Saturn.\ {\em Icarus}, 52: 509-515. 

\bibitem[Hubbard(1984)]{1984QB603.I53H83...} Hubbard, W.~B.\ (1984). 
Planetary interiors.\ {\em New York, Van Nostrand Reinhold Co.}

\bibitem[Hubbard(1999)]{1999Icar..137..357H} Hubbard, W.~B.\ (1999). NOTE: 
Gravitational Signature of Jupiter's Deep Zonal Flows.\ {\em Icarus}, 137: 
357-359. 

\bibitem[Hubbard and Smoluchowski(1973)]{1973SSRv...14..599H} Hubbard, 
W.~B.~and Smoluchowski, R.\ (1973). Structure of Jupiter and Saturn.\ {\em 
Space Science Reviews}, 14: 599. 

\bibitem[Hubbard and Slattery(1976)]{1976jsia.coll..176H} Hubbard, 
W.~B.~and Slattery, W.~L.\ (1976). Interior structure of Jupiter - Theory 
of gravity sounding.\ {\em Jupiter: Studies of the 
Interior, Atmosphere, Magnetosphere and Satellites, Univ. of Arizona
Space Science Series}, 176-194. 

\bibitem[Hubbard and Stevenson(1984)]{1984satn.book...47H} Hubbard, 
W.~B.~and Stevenson D.~J.~(1984).\ Interior structure of Saturn.\ In
{\em Saturn, Univ. Arizona Space Science Series}, 47-87. 

\bibitem[Hubbard and Dewitt(1985)]{1985ApJ...290..388H} Hubbard, W.~B.~and 
Dewitt, H.~E.\ (1985). Statistical mechanics of light elements at high 
pressure. VII - A perturbative free energy for arbitrary mixtures of H and 
He.\ {\em \apj}, 290: 388-393. 

\bibitem[Hubbard and Marley(1989)]{1989Icar...78..102H} Hubbard, W.~B.~and 
Marley, M.~S.\ (1989). Optimized Jupiter, Saturn, and Uranus interior 
models.\ {\em Icarus}, 78: 102-118. 

\bibitem[Hubbard et al.(1995)]{1995nep.book...109P} Hubbard, W. B.,
Podolak, M. and Stevenson, D. J. (1995). The interior of Neptune. In
{\em Neptune and Triton, Univ. Arizona Space Science Series}, 109-138.

%\bibitem[Hubbard et al.(1997)]{1997PhPl....4.2011H} Hubbard, W.~B., 
%Guillot, T., Lunine, J.~I., Burrows, A., Saumon, D., Marley, M.~S., and 
%Freedman, R.~S.\ (1997). Liquid metallic hydrogen and the structure of 
%brown dwarfs and giant planets.\ {\em Physics of Plasmas}, 4: 2011-2015. 

\bibitem[Hubbard et al.(1999)]{1999P&SS...47.1175H} Hubbard, W.~B., 
Guillot, T., Marley, M.~S., Burrows, A., Lunine, J.~I., and Saumon, D.~S.\ 
(1999). Comparative evolution of Jupiter and Saturn.\ {\em \planss}, 47: 
1175-1182. 

\bibitem[Jeffreys(1924)]{1924MNRAS..84..534J} Jeffreys, H.\ (1924). On the 
internal constitution of Jupiter and Saturn.\ {\em \mnras}, 84: 534. 

\bibitem[Jones (2000)]{2000Jones} Jones,
C.A. (2000). Convection-driven geodynamo models. {\em
Philos. Trans. Roy. Soc. A} 358: 873-897.

\bibitem[Kerley(1972)] { } Kerley, G.~I. (1972). Equation of state and phase diagram 
of dense hydrogen.\ {\em Phys. Earth Planet. Interiors}, 6:78-82.

\bibitem[Kippenhahn and Weigert(1990)]{1990sse..book.....K} Kippenhahn, 
R.~and Weigert, A.\ (1990). Stellar Structure and Evolution.\ {\em Stellar 
Structure and Evolution, Springer-Verlag Berlin}. 

\bibitem[Kirk and Stevenson(1987)]{1987ApJ...316..836K} Kirk, R.~L.~and 
Stevenson, D.~J.\ (1987). Hydromagnetic constraints on deep zonal flow in 
the giant planets.\ {\em \apj}, 316: 836-846. 

\bibitem[Klepeis et al.(1991)]{1991Sci...254..986K} Klepeis, J.~E., 
Schafer, K.~J., Barbee, T.~W., and Ross, M.\ (1991). Hydrogen-helium 
mixtures at megabar pressures - Implications for Jupiter and Saturn.\ {\em 
Science}, 254: 986-989. 

\bibitem[Knudson et al.(2001)]{2001PhRvL..87v5501K} Knudson, M.~D., Hanson, 
D.~L., Bailey, J.~E., Hall, C.~A., Asay, J.~R., and Anderson, W.~W.\ 
(2001). Equation of State Measurements in Liquid Deuterium to 70 GPa.\ {\em 
Physical Review Letters}, 87: 5501. 

\bibitem[Landau and Lifshitz(1969)]{1969stph.book.....L} Landau, L.~D.~and 
Lifshitz, E.~M.\ (1969). Statistical physics.\ {\em Course of 
theoretical physics, Oxford: Pergamon Press, and 
Reading: Addison-Wesley,  |c1969, 2nd rev.~- enlarg.ed.}. 

\bibitem[Lederer et al.(1995)]{ } Lederer, S.~M., Marley, M.~S.,
  Mosser, B., Maillard, J.~P., Chanover, N.~J.~and Beebe, R.~F.\
  (1995). Albedo features and Jovian seismology, Icarus, 114: 269.

\bibitem[Lenosky et al.(1997)]{ } Lenosky, T.~J., Kress, J.~D. and
Collins, L.~A. (1997). Molecular -dynamics modeling of the Hugoniot of shocked 
liquid deuterium.\ {\em Physical Review B}, 56: 5164-5169.

\bibitem[Lenosky et al.(2000)]{ } Lenosky , T.~J., Bickham, S.~R., Kress, J.~D. and
Collins, L.~A. (2000). Density-functional calculation of the Hugoniot of shocked 
liquid deuterium.\ {\em Physical Review B}, 61: 1-4.

\bibitem[Lindal et al.(1981)]{1981JGR....86.8721L} Lindal, G.~F.~and 11 
colleagues (1981). The atmosphere of Jupiter - an analysis of the Voyager 
radio occultation measurements.\ {\em \jgr}, 86: 8721-8727. 

\bibitem[Lissauer(1993)]{1993ARA&A..31..129L} Lissauer, J.~J.\ (1993). 
Planet formation.\ {\em \araa}, 31: 129-174. 

\bibitem[Low(1966)]{1966AJ.....71R.391L} Low, F.~J.\ (1966). Observations 
of Venus, Jupiter, and Saturn at $\lambda=20 \mu$m.\ {\em \aj}, 71: 391. 

\bibitem[Lubow et al.(1997)]{1997ApJ...484..866L} Lubow, S.~H., Tout, 
C.~A., and Livio, M.\ (1997). Resonant Tides in Close Orbiting Planets.\ 
{\em \apj}, 484: 866. 

\bibitem[Mahaffy et al.(2000)]{2000JGR...10515061M} Mahaffy, P.~R., 
Niemann, H.~B., Alpert, A., Atreya, S.~K., Demick, J., Donahue, T.~M., 
Harpold, D.~N., and Owen, T.~C.\ (2000). Noble gas abundance and isotope 
ratios in the atmosphere of Jupiter from the Galileo Probe Mass 
Spectrometer.\ {\em \jgr}, 105: 115061-15072. 

\bibitem[Marcy et al.(1997)]{1997ApJ...481..926M} Marcy, G.~W., Butler, 
R.~P., Williams, E., Bildsten, L., Graham, J.~R., Ghez, A.~M., and 
Jernigan, J.~G.\ (1997). The Planet around 51 Pegasi.\ {\em \apj}, 481: 
926. 

\bibitem[Marcy et al.(2000)]{2000prpl.conf.1285M} Marcy, G.~W., Cochran, 
W.~D., and Mayor, M.\ (2000). Extrasolar Planets around Main-Sequence 
Stars.\ {\em Protostars and Planets IV}, 1285. 

\bibitem[Mayor and Queloz(1995)]{1995Natur.378..355M} Mayor, M.~and Queloz, 
D.\ (1995). A Jupiter-Mass Companion to a Solar-Type Star.\ {\em \nat}, 
378: 355. 

\bibitem[Militzer and Ceperley(2000)]{2000PhRvL..85.1890M} Militzer, B.~and 
Ceperley, D.~M.\ (2000). Path Integral Monte Carlo Calculation of the 
Deuterium Hugoniot.\ {\em Physical Review Letters}, 85: 1890-1893. 

\bibitem[Militzer and Ceperley(2001)]{2001PhRvE..63f6404M} Militzer, B.~and 
Ceperley, D.~M.\ (2001). Path integral Monte Carlo simulation of the 
low-density hydrogen plasma.\ {\em \pre}, 63: 6404. 

\bibitem[Moffatt(1978)]{1978mfge.book.....M} Moffatt, H.~K.\ (1978). 
Magnetic field generation in electrically conducting fluids.\ {\em 
Cambridge, England, Cambridge University Press}, Chap. 12. 

\bibitem[Mosser et al.(2000)]{2000Icar..144..104M} Mosser, B., Maillard, 
J.~P., and M{\' e}karnia, D.\ (2000). New Attempt at Detecting the Jovian 
Oscillations.\ {\em Icarus}, 144: 104-113. 

\bibitem[Mostovych et al.(2000)]{ } Mostovych , A.~N., Chan, Y., Lehecha, T., Schmitt, A.
and Sethian, J.~D. (2000). Reflected shock  
experiments on the equation-of-state properties of liquid deuterium at 100-600 GPa 
(1-6 Mbar).\ {\em Physical Review Letters}, 85: 3870-3873.

\bibitem[Nellis et al.(1997)]{1997JChPh.107.9096N} Nellis, W.~J., Holmes, 
N.~C., Mitchell, A.~C., Hamilton, D.~C., and Nicol, M.\ (1997). Equation of 
state and electrical conductivity of "synthetic Uranus," a mixture of 
water, ammonia, and isopropanol, at shock pressure up to 200 GPa (2 Mbar).\ 
{\em \jcp}, 107: 9096-9100. 

\bibitem[Nellis et al.(1984)]{ } Nellis, W.~J., Holmes, N.~C., Mitchell, A.~C.,
Trainor, R.~J., Governo, G.~K., Ross, M. and Young, D.~A. (1984).  Shock compression of 
liquid helium to 56 GPa (560 kbar).\ {\em Physical Review Letters}, 53: 1248-1251.

\bibitem[Nellis et al.(1983)]{ } Nellis, W.~J., Mitchell, A.~C., Van Thiel, M.,
Devine, G.~J., Trainor, R.~J., Brown, N. (1983).  Equation-of-state data for
molecular hydrogen and deuterium at shock pressures in the range 2-76 GPa (20-760 kbar).
\ {\em Journal Chemical Physics}, 79: 1480-1486.

\bibitem[Nellis et al.(1999)]{1999PhRvB..59.3434N} Nellis, W.~J., Weir, 
S.~T., and Mitchell, A.~C.\ (1999). Minimum metallic conductivity of fluid 
hydrogen at 140 GPa (1.4 Mbar).\ {\em \prb}, 59: 3434-3449. 

\bibitem[Niemann et al.(1998)]{1998JGR...10322831N} Niemann, H.~B.~and 11 
colleagues (1998). The composition of the Jovian atmosphere as determined 
by the Galileo probe mass spectrometer.\ {\em \jgr}, 103: 222831-22846. 

\bibitem[Owen et al.(1999)]{1999Natur.402..269O} Owen, T., Mahaffy, P., 
Niemann, H.~B., Atreya, S., Donahue, T., Bar-Nun, A., and de Pater, I.\ 
(1999). A low-temperature origin for the planetesimals that formed 
Jupiter.\ {\em \nat}, 402: 269-270. 

\bibitem[Peale(1999)]{1999ARA&A..37..533P} Peale, S.~J.\ (1999). Origin and 
Evolution of the Natural Satellites.\ {\em \araa}, 37: 533-602. 

\bibitem[Pearl and Conrath(1991)]{1991JGR....9618921P} Pearl, J.~C.~and 
Conrath, B.~J. (1991).\ The albedo, effective temperature, and energy balance 
of Neptune, as determined from Voyager data.\ {\em \jgr} 96: 18921-18930. 

\bibitem[Pfaffenzeller et al.(1995)]{1995PhRvL..74.2599P} Pfaffenzeller, 
O., Hohl, D., and Ballone, P.\ (1995). Miscibility of hydrogen and helium 
under astrophysical conditions.\ {\em Physical Review Letters}, 74: 
2599-2602. 

\bibitem[Podolak et al.(1988)]{1988Icar...73..163P} Podolak, M., Pollack, 
J.~B., and Reynolds, R.~T.\ (1988). Interactions of planetesimals with 
protoplanetary atmospheres.\ {\em Icarus}, 73: 163-179. 

\bibitem[Podolak et al.(1991)]{1991uran.book...29P} Podolak, M., Hubbard, 
W.~B., and Stevenson, D.~J.\ (1991). Model of Uranus' interior and magnetic 
field.\ In {\em Uranus, Univ. Arizona Space Science Series}, 29-61. 

\bibitem[Pollack et al.(1996)]{1996Icar..124...62P} Pollack, J.~B., 
Hubickyj, O., Bodenheimer, P., Lissauer, J.~J., Podolak, M., and 
Greenzweig, Y.\ (1996). Formation of the Giant Planets by Concurrent 
Accretion of Solids and Gas.\ {\em Icarus}, 124: 62-85. 

\bibitem[Roberts and Glatzmaier(2000)]{2000RvMP...72.1081R} Roberts, 
P.~H.~and Glatzmaier, G.~A.\ (2000). Geodynamo theory and simulations.\ 
{\em Reviews of Modern Physics}, 72: 1081-1123. 

\bibitem[Rogers(2000)]{2000PhPl....7...51R} Rogers, F.~J.\ (2000). 
Ionization equilibrium and equation of state in strongly coupled plasmas.\ 
{\em Physics of Plasmas}, 7: 51-58. 

\bibitem[Ross(1998)]{ } Ross, M. (1998). Linear-mixing model for shock-compressed 
liquid deuterium.\ {\em Physical Review B}, 58: 669-677.

\bibitem[Ross and Yang(2001)]{ } Ross, M. and Yang, L.~H. (2001). Effect of chainlike
structures on shock-compressed liquid deuterium.\ {\em Physical Review B}, 64:
134210. 

\bibitem[Roulston and Stevenson (1995)]{1995Roulston} Roulston, M.S.,
Stevenson, D.J. (1995). Prediction of neon depletion in Jupiter's atmosphere
{\em EOS} 76: 343 [abstract].

\bibitem[Russell et al.(2001)]{2001Russell} Russell, C. T., Yu,
Z. J., Khurana, K. K., et al. (2001). Magnetic field changes in the
inner magnetosphere of Jupiter. {\em Adv. Space Res.} 28:897-902.

\bibitem[Safronov (1972)]{1972Safronov} Safronov,
  V.S. (1972). Evolution of the protoplanetary cloud and formation of
  the Earth and planets. {\em Israel Program for Scientific
  Translations, Jerusalem}.

\bibitem[Salpeter(1973)]{1973ApJ...181L..83S} Salpeter, E.~E.\ (1973). On 
Convection and Gravitational Layering in Jupiter and in Stars of Low Mass.\ 
{\em \apjl}, 181: L83. 

\bibitem[Saumon et al.(1992)]{1992ApJ...391..827S} Saumon, D., Hubbard, 
W.~B., Chabrier, G., and Van Horn, H.~M.\ (1992). The role of the 
molecular-metallic transition of hydrogen in the evolution of Jupiter, 
Saturn, and brown dwarfs.\ {\em \apj}, 391: 827-831. 

\bibitem[Saumon et al.(1995)]{1995ApJS...99..713S} Saumon, D., Chabrier, 
G., and Van Horn, H.~M.\ (1995). An Equation of State for Low-Mass Stars 
and Giant Planets.\ {\em \apjs}, 99: 713. 

\bibitem[Saumon et al.(1996)]{1996ApJ...460..993S} Saumon, D., Hubbard, 
W.~B., Burrows, A., Guillot, T., Lunine, J.~I., and Chabrier, G.\ (1996). A 
Theory of Extrasolar Giant Planets.\ {\em \apj}, 460: 993. 

\bibitem[Saumon et al.(2000)]{ } Saumon, D., Chabrier, G., Wagner, D.~J. and Xie, X. (2000).
Modeling pressure-ionization of hydrogen in the context of astrophysics.\ {\em High Pressure
Research}, 16:331-343.

\bibitem[Schmider et al.(1991)]{1991A&A...248..281S} Schmider, F.-X., 
Fossat, E., and Mosser, B.\ (1991). Possible detection of Jovian global 
oscillations.\ {\em \aap}, 248: 281-291. 

\bibitem[Schubert and Zhang(2000)]{2000fgpc.conf..210S} Schubert, G.~and 
Zhang, K.\ (2000). Dynamics of Giant Planet Interiors.\ {\em ASPC 
Conf.~Ser.~212: From Giant Planets to Cool Stars}, 210. 

\bibitem[Seiff et al.(1998)]{1998JGR...10322857S} Seiff, A., Kirk, D.~B., 
Knight, T.~C.~D., Young, R.~E., Mihalov, J.~D., Young, L.~A., Milos, F.~S., 
Schubert, G., Blanchard, R.~C., and Atkinson, D.\ (1998). Thermal structure 
of Jupiter's atmosphere near the edge of a 5-\&micro;m hot spot in the north 
equatorial belt.\ {\em \jgr}, 103: 222857-22890. 

\bibitem[Smoluchowski(1967)]{1967Smol} Smoluchowski,
  R. (1967). Internal structure and energy emission of Jupiter. {\em
Nature}, 215: 691-695.

\bibitem[Stevenson(1979)]{1979GApFD..12..139S} Stevenson, D.~J.\ (1979).\ 
Turbulent thermal convection in the presence of rotation and a magnetic 
field - A heuristic theory.\ {\em Geophys. and Astrophys. Fluid Dyn.},
2: 139-169. 

\bibitem[Stevenson(1980)]{1980Sci...208..746S} Stevenson, D.~J.\ (1980). 
Saturn's luminosity and magnetism.\ {\em Science}, 208: 746-748. 

\bibitem[Stevenson(1982a)]{1982GApFD..21..113S} Stevenson, D.~J.\ (1982a). 
Reducing the non-axisymmetry of a planetary dynamo and an application to 
Saturn.\ {\em Geophysical and Astrophysical Fluid Dynamics}, 21: 113-127. 

\bibitem[Stevenson(1982b)]{1982AREPS..10..257S} Stevenson, D.~J.\ (1982b). 
Interiors of the giant planets.\ {\em Annual Review of Earth and Planetary 
Sciences}, 10: 257-295. 

\bibitem[Stevenson(1982c)]{1982P&SS...30..755S} Stevenson, D.~J.\ (1982). 
Formation of the giant planets.\ {\em \planss}, 30: 755-764. 

\bibitem[Stevenson(1983a)]{1983RPPh...46..555S} Stevenson, D.~J.\ (1983a). 
Planetary magnetic fields.\ {\em Reports of Progress in Physics}, 46: 
555-620. 

\bibitem[Stevenson(1983b)]{1983JGR....88.2445S} Stevenson, D.~J.\ (1983b). 
Anomalous bulk viscosity of two-phase fluids and implications for planetary 
interiors.\ {\em \jgr}, 88: 2445-2455. 

\bibitem[Stevenson and Ashcroft(1974)]{1974PhRvA...9..782S} Stevenson, 
D.~J.~and Ashcroft, N.~W.\ (1974). Conduction in fully ionized liquid 
metals.\ {\em \pra}, 9: 782-789. 

\bibitem[Stevenson and Salpeter(1976)]{1976jsia.coll...85S} Stevenson, 
D.~J.~and Salpeter, E.~E.\ (1976). Interior models of Jupiter.\ {\em
Jupiter: Studies of the Interior, Atmosphere, Magnetosphere and
Satellites, Univ. of Arizona Space Science Series}, 85-112. 

\bibitem[Stevenson and Salpeter(1977a)]{1977ApJS...35..221S} Stevenson, 
D.~J.~and Salpeter, E.~E.\ (1977a). The phase diagram and transport 
properties for hydrogen-helium fluid planets.\ {\em \apjs}, 35:
221-237. 

\bibitem[Stevenson and Salpeter(1977b)]{1977ApJS...35..239S} Stevenson, 
D.~J.~and Salpeter, E.~E.\ (1977b). The dynamics and helium distribution in 
hydrogen-helium fluid planets.\ {\em \apjs}, 35: 239-261. 

\bibitem[Trafton(1967)]{1967ApJ...147..765T} Trafton, L.~M.\ (1967). Model 
atmospheres of the major planets.\ {\em \apj}, 147: 765-781. 

\bibitem[Weir et al.(1996)]{1996PhRvL..76.1860W} Weir, S.~T., Mitchell, 
A.~C., and Nellis, W.~J.\ (1996). Metallization of Fluid Molecular Hydrogen 
at 140 GPa (1.4 Mbar).\ {\em Physical Review Letters}, 76: 1860-1863. 

\bibitem[Wigner and Huntington(1935)]{1935Wigner} Wigner, E. and
Huntington, H.~B.\ (1935). On the possibility of a metallic
modification of hydrogen.\ {\em \jcp}, 3: 764-770.

\bibitem[von Zahn et al.(1998)]{1998JGR...10322815V} von Zahn, U., Hunten, 
D.~M., and Lehmacher, G.\ (1998). Helium in Jupiter's atmosphere: Results 
from the Galileo probe helium interferometer experiment.\ {\em \jgr}, 103: 
222815-22830. 

\bibitem[Zharkov(1986)]{1986harw.book.....Z} Zharkov, V.~N.\ (1986). 
Interior structure of the earth and planets (Revised and enlarged 
edition).\ {\em Chur, Switzerland and New York, Harwood Academic 
Publishers}. 

\bibitem[Zharkov and Trubitsyn(1974)]{1974Icar...21..152Z} Zharkov, 
V.~N.~and Trubitsyn, V.~P.\ (1974). Determination of the equation of state 
of the molecular envelopes of Jupiter and Saturn from their gravitational 
moments.\ {\em Icarus}, 21: 152-156. 

\bibitem[Zharkov and Trubitsyn(1978)]{1978ppi..book.....Z} Zharkov, 
V.~N.~and Trubitsyn, V.~P.\ (1978). Physics of planetary interiors.\ {\em 
Astronomy and Astrophysics Series, Tucson: Pachart, 1978}, . 

\bibitem[Zharkov and Gudkova(1991)]{1991AnGeo...9..357Z} Zharkov, V.~N.~and 
Gudkova, T.~V.\ (1991). Models of giant planets with a variable ratio of 
ice to rock.\ {\em Annales Geophysicae}, 9: 357-366. 



\end{thebibliography}

\end{document}








