HAL Id: jpa-00209058
https://hal.archives-ouvertes.fr/jpa-00209058
Submitted on 1 Jan 1981
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Improved Debye Hückel theory for one-and multicomponent plasmas
P. Vieillefosse
To cite this version:
P. Vieillefosse. Improved Debye Hückel theory for one-and multicomponent plasmas. Journal de
Physique, 1981, 42 (5), pp.723-733. �10.1051/jphys:01981004205072300�. �jpa-00209058�
Improved Debye Hückel theory for one-and multicomponent plasmas
P. Vieillefosse
Laboratoire de Physique Théorique des Liquides (*), Université P. et M. Curie, 4, place Jussieu, 75230 Paris cedex 05, France
(Rep le 25 novembre 1980, accepté le 20 janvier 1981)
Résumé.
2014La théorie bien
connuede Debye Hückel appliquée
auxplasmas classiques neutralisés par
unfond continu n’est valide que lorsque le couplage est suffisamment faible. La raison essentielle de ceci est que l’approxi-
mation de champ moyen
surlaquelle est fondée la théorie de Debye Hückel néglige les corrélations entre les ions du nuage d’écrantage. Nous montrons ici comment il est possible de tenir compte de la contribution de ces corré- lations à l’énergie interne
enmodifiant de façon adéquate la formule donnant l’énergie
enfonction du champ
moyen. L’énergie ainsi obtenue
abien les deux bons comportements limites en couplages faible et fort. De plus,
en
couplage fort, la linéarité dans les concentrations de l’énergie interne des plasmas à plusieurs composants est montrée et
sonlien
avecla condition d’écrantage parfait est établi. Enfin nous présentons
unmodèle simple de
démixtion sous forte pression des mélanges binaires
enprésence d’électrons.
Abstract.
2014When applied to classical plasmas which
areneutralized by
auniform background, the well known Debye Hückel theory is only valid in the weak coupling limit. The main
reasonfor this limitation is that the
meanfield approximation, upon which the Debye Hückel theory is based, neglects correlations between ions of the
screening cloud. Here
weshow how it is possible to take into account the correlation contributions to the internal energy by properly modifying the formula which determines the internal energy
as afunction of the mean field.
The internal energy
weobtain correctly exhibits the two asymptotic behaviours in the weak and strong coupling
limits. Moreover, at strong coupling,
weget the well known linear law for the internal energy of multicomponent plasmas
as afunction of the concentrations, and its relationship to the perfect screening condition is proved.
Finally,
weshow
asimple model for phase separation of binary mixtures neutralized by electrons under high
pressure.
Classification Physics Abstracts
05.20
-52.25
Introduction. - Dense fully ionized matter has
been extensively studied for several years. A simple
and accurate model is the classical one-component
plasma (OCP). The point ions interact via purely
coulombic forces and a uniform and rigid background,
which represents the degenerate electron gas, ensures overall charge neutrality. The strength of the coupling
is defined by the dimensionless parameter r which is essentially the ratio of the potential energy of two ions separated by the characteristic mean distance to their mean kinetic energy.
In the weak coupling limit, thermodynamical
functions can be analytically obtained by two theories (i) The Debye Hfckel one [1] (DH1 which is based on a mean field approximation and gives accurate results
for r lower than 0.01 when the Boltzmann factor is linearized and up to F - 1 without linearization [2] ; (ii) the exact Abe expansion [3], which is rapidly diverg-
(*) Equipe associ6e
auC.N.R.S.
ing for r larger than 0.3 because of its asymptotic
character.
When the coupling is stronger, more powerful
methods must be used. First, there are purely numeri-
cal ones like Monte Carlo [4] (MC) and molecular dynamics (MD) computer simulations [5] which yield,
of course, exact results. In addition to their own
interest, these results allow one to test theories the accuracy of which cannot otherwise be directly evaluate
ed. Among the theories which give results close to the exact ones, the hypernetted chain equations [6] (HNC)
and the mean spherical approximation [7] (MSA)
must be mentioned.
Unfortunately, all these theories require lengthy
numerical calculations. Moreover, going from one,
to several components leads to serious complications.
Here we present an extension of the DH theory, called improved DH theory (IDH) in the following, which
is also a mean field approximation. The equation
to solve is identical to the nonlinearized DH one. On the other hand, the expression for the internal energy
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01981004205072300
is modified in order to take into account the correla- tions between ions of the screening cloud which are neglected by the mean field approximation. The result-
ing expressions correctly exhibits the leading terms
-
(J3/2) r3/2 and - (9/10) r in the weak and strong coupling limits, respectively. The maximum deviation
from exact results is of the order of 5 % for T close to
one.
Going from one to several (two or more) compo- nents presents no difficulty because the mean poten- tials around ions of different species are not correlat-
ed. Moreover, it is indeed found that the excess energy is remarkably linear in the concentrations for suffi-
ciently strong coupling [2] and the relationship of this linearity with the perfect screening condition is clearly
shown.
In the first part, the basis of this theory is discussed
for the OCP case; some exact results and expansions
are given and results obtained in the one-component
case are extended to the multicomponent case. We also present a simple model of phase separation of binary
mixtures neutralized by a degenerate electron gas in the strong coupling limit; in this calculation, the linear
law for the ion energy is used.
Some partial results have been published else-
where [8].
1. One-component plasma.
-Here we consider the simplest system of particles interacting via Coulomb forces, i.e. a set of N ions of charge ze, of mass m in a
box of volume V ; the numerical density is p
=N/V.
A uniform and rigid background neutralizes the
charge of the ions so that the thermodynamic limit
is well defined. The ions are supposed classical (non quantum) and the interactions are purely coulombic.
The Hamiltonian is the sum of three parts,
which are respectively the kinetic energy, the potential
energy of the ions and the potential energy of the
background.
As the potential is a homogeneous function of the
coordinates of the ions, the excess free energy takes the form
where k is the Boltzmann constant, T the temperature and T(F) is some function of the coupling parameter
For r « 1, the gas is nearly perfect. On the other hand,
when the ions are strongly coupled, the system beco-
mes solid [9] (F - 155).
The various thermodynamic excess quantities are easily derived from IF(F) :
Sex is the entropy per particle, eex the energy per parti- cle, Pex the pressure and Mex the chemical potential.
The knowledge of the excess internal energy is
sufficient to determine all thermodynamic quantities
and we shall be essentially interested in this quantity
because approximations can be more easily made for
it.
Fixing an ion at the origin of coordinates, we obtain the excess internal energy per particle in the form :
where p(r) is the exact mean density of particles
around the central ion Of course, p(r) depends only
on the distance r II and tends to p as 11 r II goes to
infinity, 7¡)*(r) is the exact mean potential due to all charges except the central ion; the total potential 7¡)(r) is equal to the sum 7¡)*(r) + ze/r.
1.1 MEAN FIELD APPROXIMATION.
-The Debye
Hfckel theory [1], which is principally known under
its linearized form, amounts to assuming that each
ion of the cloud around the central ion interacts with the others only by a mean potential cp(r) which is produced by the equilibrium distribution p(r) of the
ions assumed to form a perfect gas in the external field cp(r) ; i.e. p(r) is given by the Boltzmann distribu- tion
whereas the potential cp(r) is determined by
where the background is taken into account.
Such an approximation clearly neglects the corre-
lations between the ions of the cloud, and the distri- bution p(r) as well as the potential g(r) are approximate
values of p(r) and (j)(r) (p(r) is not equal to
p exp {
-zT(r) 1).
Some exact results following from equations (6)
and (7) will be shown later. Here we only mention
three properties, the first one of which is well known :
By taking 1/2 ze(p*(O) for the excess energy we do not make a good approximation, except in the case of small F values where the correlations between the ions of the cloud are weak and give terms of higher
order in F. Moreover, we know that Jieex is essentially given by - (9/10) F when F is large [4]. The ions sphere model [10] gives this value correctly and
shows us how the excess energy must be calculated.
In the strong-coupling limit, the cloud is likened to a sphere of radius A
=rp, in which the only ion is
the central one and outside of which the ions have a
uniform density p. The excess energy is taken as the total electrostatic energy of such a distribution with the energies of the central ion and of the background
inside the sphere giving contributions to fleex equal
to - (3/4) T and - (3/20) F respectively. As we shall
see later, the equations (6) and (7) exhibit, when r
is large enough, the same distribution of ions
(where Y is the Heaviside function). These remarks suggest to us that the potential energy of the charge
distribution ze { p(r) - p I must be added to the
central ion energy 1/2 zecp*(O). However, it must be carefully noted that this supplementary term is to be
divided amongst the n ions inside the screening
cloud. The number n is not well defined; it can be
estimated as p(4/3) nA 3
=(Alrp)3, where A is the
screening length, i.e. the length at which (p tends to zero. Thus, we are led to the expression
At large values of r, A is equal to rp and
At small values of r, A is equal to rpl.,,13-F and the
second term in (9) gives a negative correction, pro-
portional to F’, which is qualitatively correct.
However, the expression (9) is not satisfactory
because of the poor definition of n (n is defined within
a numerical factor, except in the strong-coupling limit). Another way to divide the cloud energy amongst the ions is to take for eex the work spent to polarize
the cloud, i.e. the work done when the charge of the
central ion is switched on from zero to ze with the other ions keeping a constant charge ze. If the central
charge is Aze (0 K h K 1), the potential cp ;.(r) and the density pi(r) satisfy the equations :
The excess energy which we write as,
takes the following form by taking (10) and (11) into account
where (p(r) and p(r) are solutions of (6) and (7) (1).
Such an expression correctly gives the two limiting
behaviours of peex : - (/2) r3/2 as r goes to zero
and - (9/10) F when F is large compared to one
Another expression of fleex can be given by taking (7)
into account :
IPeex’ as a function of p(rx is minimum for
The expression (12) of eex must be compared to
that one which gives the excess chemical poten- tial [11] :
where, it is recalled that 7P is the exact mean potential (calculated with the exact distribution p(r)). In the
weak coupling limit, cp and 7P are close; as cp). and 7P).
(1) This is obtained by integrating by parts. The calculation is
very similar to that
oneof ppex in the HNC approximation [12].
are linear in À. then, peex is equal to pJlex. This result
can easily be deduced from (4) with
On the other side, (p and 7P are different in the strong coupling limit; fie,, - - (9/10) F leads to
As an example it can be noticed that :
1
Another comparison can also be made with the HNC equations. The mean density around the central ion is written in the form
where the mean forces potential 0 is the solution of the equations :
The internal energy takes the form :
which looks very much like formula (13). The che-
mical potential, given by (16) is obtained with the help of the HNC equations [12] as :
As r tends to zero, v(r) is equal to 1/r except in a small region around the origin; 0(r) is then close to p(r). Moreover, the following equation can be
written :
When r is large compared to one, the region where v(r) is different from 1/r is no longer negligible and
(p becomes different from cpo Keeping only the leading
terms in (21) and (22), we get :
from which we deduce
These results have to be compared to (17) and (8).
The proof of the validity of expression (12) for the
internal energy ultimately lies in the good numerical
accuracy of the results when compared to those
obtained by Monte Carlo calculations [4] or by the
solution of HNC equations [6]. The difference is in the worst case (r - 1), of the order of 5 %.
1.2 EXACT RESULTS FOR THE DEBYE HUCKEL EQUA- TIONS.
-We introduce the following dimensionless
quantities :
The screening factor satisfies the following equation :
the solution of which can be written
The derivative at the origin §’(0) must be chosen so
that §(x) tends to zero when x goes to infinity.
According to (27), 0" and 0 have the same sign
so §’(0) must be negative; in the opposite case,
§ would indefinitely increase from one. Hence,
§(x) decreases initially; if 0’(x) went to zero before §(x),
then beyond this point t/J(x) would indefinitely increase
while if t/J(x) went to zero before §’(x), then beyond
this point O(x) would indefinitely decrease. There-
fore, 0 and 41’ must be zero at the same point (x = oo,
in fact) and the function 0 monotonically decreases
from one at x
=0 to zero at x
=oo. g(x) is mono-
tically increasing and does not exhibit oscillations.
The integral in the equation (28) is positive so the,
function
must be positive; y(x) has a minimum (at
which has also to be positive. We deduce from this the following inequality :
For very large r, the integral in (28) is zero as long as 0 is not of the order of 1/T ; i.e. § is equal to
y as long as y is not of the order of 1/r. Thus, the
minimum value of y(x) is of the order of 11F and
§’(0) tends to its lower bound with a difference of the order of 11F when r becomes large.
The excess energy given by (13) takes the form :
We show in appendices A and B how the solution of (27) can be obtained in the two limits of weak and strong coupling; the following expansions are found
where y is the Euler constant, and,
The weak coupling expansion must be compared
to the exact expansion of Abe [3]
Numerical comparison (Table I) shows that the
expansion (32) is more accurate than the exact one, due to the poor convergence of the latter.
The strong coupling expansion correctly contains
the leading term, linear in r, which corresponds to
the electrostatic part of the energy. On the other hand, the second term, which corresponds to the
thermal part of the energy, does not increase with r
as the results from the Monte Carlo calculations [4]
or from HNC [6] and Mean Spherical Approxima-
tion [7] indicate.
Table I.
-Comparison of - peex(r) for the OCP,
calculated by various methods :
a) IDH expansions (32) and (33) in the weak and strong coupling limits, respectively.
b) Abe expansion (34).
c) IDH (31) numerical calculation.
d) HNC Springer et al. for F 50 and Ng for T > 100.
e) MC Hansen.
1. 3 NUMERICAL RESULTS.
-The numerical inte-
gration of equation (27) by formula (28) is very easy and can be done on a pocket calculator. In table I,
we compare these results to those of Monte Carlo and HNC calculations. The maximum relative devia- tion is obtained for r close to one where it reaches 5 %.
2. Ionic mixtures.
-Here we look at mixtures of several species of ions which have charges of the
same sign. Each species i is characterized by its
number of ions Ni, the charge zi e and the mass mi of the ions. The numerical density pi and p, the charge density po and the concentrations ci and yi are defined by the equations :
where
and V is the volume of the system. The charge of
the ions is neutralized by a uniform background of charge density - po e. Interactions are assumed to be purely coulombic.
Two length units,
and two coupling parameters
are introduced; the notation f indicates the mean
value of a quantity f over the concentrations :
The excess free energy is written here in the form :
The entropy per particle Sex
=SE.IN, the energy per
particle eex
=Eex/N, the pressure Pex and the chemical potentials Pie. are given by the formulae
2.1 DH EQUATION FOR MIXTURES.
-We apply to
the ionic mixtures the formalism that has been
developed for the OCP in the first part.
The density of a species j around an ion i is written :
where x
=r/rpo. The screening factor §;(x) is solu-
tion of the differential equation
Here two remarks must be made :
-
The functions ql, which correspond to different species, each satisfy a separate differential equation;
going from one to several components does not make difficulties and leads only to complications of writing
at the utmost.
-
Generally, gij will be different from gji; this is related to the previously mentioned fact that /(r)
does not stand for the exact mean density of ions j
around an ion i (except in the weak coupling limit).
The internal excess energy per particle is obtained in the same way as in the OCP case
The first term is the usual definition of the energy while the second occurs as the result of the correlations bet-
ween the ions in the screening cloud.
The functions are always monotically decreasing and their derivatives at the origin satisfy
and tend to their lower bound when To is large compared to one.
The weak and strong coupling expansions of the OCP are easily generalized
where y is the Euler constant.
An important point is that the excess energy per particle becomes a linear function of the concentrations,
at constant To, when Fo is large enough compared to one. The leading term in (46) can be obtained by a simple
calculation which shows the reason for the linearity.
It is easily seen that the leading term comes from the two following contributions :
Integrating by parts, we obtain
At large Fo, zi t/J;’lx tends to a step function 3 Y(xo - x),
the length xo of which is determined by the perfect screening condition which can be written
A straight forward calculation of (48) yields
for f3eex. Therefore, the linearity of fle,,. in the concen-
trations is closely related to the perfect screening
condition.
Numerical integration of the equations confirms
the linearity of f3eex as a function of the concentra-
tions at constant ro :
The approximation is all the more valid when To is larger and the charges are not too different. For a
binary mixture Z1
=1, Z2
=2, the maximum relative deviation is 4 x 10-2 at To = 0.01, 5 x 10-3 at to = 1.0, and 5 x 10-6 at Fo
=100. At Fo
=10.0,
this deviation increases from 2 x 10-4 for Z2 = 2,
to 3 x 10-4 for z2
=3 and to 6 x 10-4 for Z2
=10.
Some results are given in tables II and III. Finally, f3eex(r 0’ { ci }) is always found (for binary mixtures)
to be larger than when calculated from the linear law (50).
Comparison with the HNC and MC results [6, 4]
is satisfactory, the deviation being of the same order
of magnitude as for the OCP case. It must be noticed that each calculation leads to the linear law with a
very high accuracy; this agrees with the fact that the linear law is related to the perfect screening condi-
tion.
2.2 APPLICATION TO THE MISCIBILITY PROBLEM.
-Here we apply the previous results to the problem
of the miscibility of a binary ionic mixture in the presence of electrons. We limit ourselves to high
pressures (i.e. large densities) so that the coupling
Table II.
-Comparison of - peex(r) for a mixture
Z1
=1, Z2
=2, calculated by HNC, IDH and IDH expansions ((45) for r
=0.1 and (46) for r
=10).
,r’B...
Table III.
-Excess internal energy - peex and excess
internal energy of mixing
calculated by IDH for To
=0.1 and three values of Z2
(z 1
=1).
between ions and electrons would be negligible. A
similar calculation has already been done by Ste-
venson [13], but with a perfect gas term unfortunately
wrong.
For sufficiently high pressures, the free energy of the ionic mixture in the presence of electrons is the
sum of the perfect gas free energy of the ions, of the
excess free energy of the ions and of the zero tempe-
rature free energy of the electrons :
The excess free energy is taken in the form
The free energy of electrons is given by the well known
expansion :
Iwhere ao is the electron Bohr radius, rso
=rpolao and
the constants 7, 6 and e are equal to 1.105, - 0.458
and 0.0311, respectively.
To the lowest order in the variable
where P is the pressure, the Gibbs free energy of
mixing is obtained as :
This term does not depend on the pressure, i.e. the
phase diagram T(cl) becomes independent of P
when the pressure is high enough. An exact calcula-
tion with the expression (55) shows that the mixture is unstable and separates into two phases below the critical temperature T c
which of course tends to zero as Z1 tends to z2.
The critical temperatures are 2.7 x 103 K for Z1 = 1, z2 =2 and 1.1 x 104 K for Z1
=1, z2
=3.
The critical concentration is given by
Calculation of the Gibbs free energy of mixing at
the next order in r,,o yields a correction to the critical temperature (56) which is positive and proportional
to rSo (i.e. P - 1/5) and a correction to the critical concentration (57) which has the same sign as Zl -Z2 and is also proportional to r,, ,o :
Comparison with exact results [2, 14] shows that
the agreement is very good for the critical concen-
tration (better than 1 %). The critical temperature is of the right order of magnitude but is lower than the
«