• Aucun résultat trouvé

Semi-empirical correlation function for one and two-ionic component plasmas

N/A
N/A
Protected

Academic year: 2021

Partager "Semi-empirical correlation function for one and two-ionic component plasmas"

Copied!
11
0
0

Texte intégral

(1)

HAL Id: jpa-00210223

https://hal.archives-ouvertes.fr/jpa-00210223

Submitted on 1 Jan 1986

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.

Semi-empirical correlation function for one and two-ionic component plasmas

B. Held, P. Pignolet

To cite this version:

B. Held, P. Pignolet. Semi-empirical correlation function for one and two-ionic component plasmas.

Journal de Physique, 1986, 47 (3), pp.437-446. �10.1051/jphys:01986004703043700�. �jpa-00210223�

(2)

437

Semi-empirical correlation function for one and two-ionic

component plasmas

B. Held and P. Pignolet

Laboratoire d’Electronique, Avenue de l’Université, Université de Pau, 64000 Pau, France (Reçu le 12 juillet 1985, revise le 28 octobre 1985, accepte le 18 novembre 1985)

Résumé.

2014

Dans cet article une expression générale semi-empirique de la fonction de corrélation g(y) est proposée

dans le cas des plasmas à une composante ainsi que des plasmas à deux composantes ioniques. Les résultats, en bon accord avec les calculs de Monte Carlo, sont déduits d’une expression analytique. Cette formulation permet d’entreprendre des calculs pour des mélanges d’ions avec des temps de calcul non prohibitifs.

Abstract

2014

This paper proposes a general semi-empirical expression for the correlation function g(y) in the one- component plasma and the two-ionic component plasma cases. The results, in good agreement with Monte Carlo simulations, are deduced from an analytic expression. This formulation permits extensive calculations for ionic mixtures without prohibitive calculation times.

J. Physique 47 (1986) 437-446 MARS 1986,

Classification Physics Abstracts

52.25

1. Introduction.

During the last few years, the Stark broadening diagnostic of stripped ions immersed in ionic per- turbers have been used with success to determine the density-temperature conditions encountered in

plasmas of inertial-confinement fusion produced by

laser [1-4].

With the development of laser compression experi-

ments on exploding pusher targets comes the interest-

ing problem of ionic mixtures in high density-high temperature conditions.

For the diagnostics, it is necessary to be sure of the

experimental and theoretical profiles, the most accu-

rate results being given by the best fit.

In practice, the Stark broadening calculations

require knowledge of the ionic microfield. The theo- retical profile is then dependent on the microfield,

and this has widely motivated many works during the

last few years [5] (and references given there).

The production of ion mixtures at high density

and high temperature, by laser compression experi-

ments, introduces the difficult problem of correlated

multicomponent plasmas.

To take these effects into account in the micro- field theory, we have to consider the inter-ionic correlations in dense, high-temperature plasmas.

The pair correlation function can be obtained in the weak coupling limit by two theories

-

the Debye-

Hiickel [6, 7] and the Abe-Mayer cluster expan-

sion [8 to 12]

-

while in the strong coupling limit,

other methods must be used : Monte Carlo [13 to 17]

or hypemetted chain equations [18 to 24] for instance.

Because, it is necessary to know the ionic micro-

field, for the Stark broadening application, the pair

correlation function for ions will be at the center of

our preoccupation [25 to 31].

In recent works, particular attention has been

paid to one or two-ionic component plasmas in view

of applications to plasmas of inertial-confinement fusion produced by laser [32 to 36].

Unfortunately, all these theories and computer simulations require lengthy numerical calculations.

That is why this paper proposes a semi-empirical approach. The ionic pair correlation function will be derived from physical considerations and numerical results; the general law must be valid in two limits : weak coupling and strong coupling. Finally, in this general form, the ionic correlation function must agree with the empirical approach encountered in

some particular cases [23, 26, 33, 35, 38].

In section 2, the general correlation expression is given for a one-component plasma.

Section 3 is a generalization of section 2. The semi-

empirical correlation function is obtained under a

general form allowing applications for a plasma parameter going from zero to one hundred.

The results are presented in section 4 and compared

with other theories and computer simulations.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01986004703043700

(3)

Finally, section 5 gives general conclusions and the field of application for these results.

2. One-component plasma.

2 .1 CORRELATION FUNCTION - We consider a plasma

with ions of charge Ze, density ni and equilibrium temperature T. The ions are assumed to be classical,

and a uniform and rigid background neutralizes the ionic charges.

For a one-ionic-component plasma, the plasma parameter r can be written (Appendix A)

where F, is the electronic plasma parameter.

It is then useful to define the reduced length :

r; being the inter-ionic length.

For single-charged ions, the correlation function

can be expressed by [33, 35]

where H(y) is the screening potential.

The empirical form

verifies the conditions [25, 27, 35]

and respects the asymptotic limit of g(y)

with

I --

The unknown F(y) function must be an oscillating

function of y for large values off satisfying the condi-

tions

2.2 PHYSICAL AND NUMERICAL CONSTRAINTS. - The purpose of this section is to introduce a realistic and physical expression for H(o) and F(y) giving

accurate results for g(y).

Firstly, it is interesting to notice [35] that a cha-

racteristic reduced length d can be defined with the condition

Furthermore, a constraint at the origin is given

with the exact result [29, 33]

Starting from the one-component plasma screening length d in the hypemetted chain scheme computed by Tanaka and Ichimaru [35], a good approximation

for d as a function of r can be given by (Appendix B)

with the limits

General expressions have also been given for H(o) in the low-density and the high-density

limits [25, 27, 35]. An approach of H(o) valid for small, large and intermediate values of T can be obtained (Appendix B) for single-charged ions :

with the limits

An analysis of the results computed by Monte Carlo

simulations [ 13 to 17] shows a periodicity d for g(y)

from the first maximum when r > 3 (the first maxi-

mum corresponding to y

=

d).

Because of this periodicity, the correlation function

g(y) is unity for several values of y ; consequently,

the unknown function F(y) is zero for these particular

values of y : F(y) is therefore the oscillating term for H( y) with a periodicity d An approximation of F(y)

can be given by the empirical expression proposed by Hubbard [23, 26, 37] for y > 1.8 :

These different constraints permit one to assume

that F(y) can be written in the form

where u is a useful variable for the periodic functions

For increasing values of y, the first term vanishes

(4)

439

and the developments of H( y) and g(y) give, in first approximation, expression (17).

The coefficients A, B, and C are evaluated by a simple criterion at three particular points, these

results being confirmed by an accurate numerical

method (Appendix C).

It is of interest to notice that, in the weak coupling

limit (F, 1), equation (15) approaches the Debye-

Hfckel value for an electron-ion plasma with Z

=

1.

Under these conditions the electron shielding cannot

be ignored, and these calculations must be convenient-

ly modified by multiplying the pure Coulomb potential by an electronic screening function in the expression

of g(y) [35].

2. 3 SEMI-EMPIRICAL EXPRESSION.

-

To sum up, we propose a fairly simple and accurate expression for

the correlation function g(y) in the case of single- charged ions :

with

and

with

d, H(0), A, B, C being defined by :

3. Two-component plasma.

We now consider a plasma with ions of charges Z. e

and Zb e, densities na and nb, and equilibrium ionic temperature T;. The ions are assumed to be classical,

and a uniform and rigid background neutralizes the ionic charge.

For a two-ionic component plasma, the plasma parameter T is given by (Appendix A)

with

where RT is the temperature ratio

and p the proportion of ions « b »

C. and Cb being the concentrations.

Noticing that a distinction must be made between interaction « a - a » and « a - b » (the central charge being Z. e), we assume that the correlation function is

given by

with

and

The screening potentials at the origin H_(0) and Hab(o) are given by (Appendix B)

with b1(i,j), b2(i, j), b3(i, j) and b4(i,j) being defined in Appendix B (B. 11) to (B. 14) and (B. 20) to (B. 23).

The generalization of the correlation function is obtained assuming that the characteristic reduced

length d(r) and the coefficients A(r), B(r), and C(F)

are functions dependent on the generalized plasma parameter r for a two-ionic component plasma.

Expressions (18) to (22) are unchanged with this assumption, the results presented in section 4 being in good agreement with those obtained by Monte Carlo

simulation [38].

(5)

4. Applications.

Numerical results for the correlation function g(y) are presented in tables I and II in the case of one-compo- nent plasmas for 0.05 F , 100. In figures 1 and 2,

these results are compared with Monte Carlo calcula- tions [ 13 to 17].

For 0.05 , F 3, the precision varies from a few per cent to better than one per cent. For increasing

values off, the uncertainties can be about 10 % in the

first dip.

To sum up, the semi-empirical expression (3) used

for g(y) is a good approximation for 0 F , 100

and all values of y.

In figure 3, the results presented are in good agree- ment with Monte Carlo simulations [38] for a one-

ionic component plasma (charge Za

=

9).

Finally, because of the Stark broadening diagnostics

of highly stripped heavy ions immersed in dense and hot plasmas of inertial-confined fusion produced by

laser, figure 4 (gaa(y) and gab(y)) is devoted to the case

of an ionic mixture with the conditions

wherep is the proportion of ions b » and v the correla- tion parameter (Appendix A).

5. Conclusion.

In this paper, general semi-empirical expressions for

the correlation function g(y) have been introduced with a view toward applications to microfield calcula- tions for the Stark broadening diagnostics of highly stripped ions.

For the inertial-confinement fusion applications, the

plasma parameter is generally less than twenty

(6)

441

Table II.

-

Correlation function g( y) for a one-component plasma (Z

=

1).

Fig. 1.

-

Correlation function g(y) for a one-component

plasma (Z

=

1) compared with HNC [19] when r

=

5

and 10.

Fig. 2.

-

Correlation function g(y) for a one-component plasma (Z

=

1) compared with Monte Carlo [14, 15]

when r

=

20, 50 and 100.

(7)

Fig 3.

-

Correlation function g(y) for a one-component plasma (Z

=

9) compared with Monte Carlo [38] when

r

=

0.52, 2.08 and 8.33.

(because of the current range of variation for the para- meters Z., Zb, p, v), and the results are given with good precision.

Finally, this analytic formulation should permit

extensive application to ionic mixtures with very short calculation time compared to Monte Carlo simu-

lations.

Appendix A.

PARAMETERS. - We consider a two-ionic component

plasma with a neutralizing background

For a one-electronic component plasma, the plasma parameter F. is defined by

where re is the inter-electronic length

Introducing the electronic Debye length

the electronic plasma parameter becomes

This result can be generalized to a two-ionic compo- nent system, writing the two-ionic plasma parameter

where ni is the ionic density

and Â.Di the ionic Debye length [5]

with

RT being the temperature ratio

p the proportion of ions « b »

C., and Cb being the concentration of ions.

The two-ionic plasma parameter r can then be written

If Ce is the concentration of electrons, we have

with the conditions of neutrality

and it is easy to obtain

Using equations (A. 7), (A. 12), and (A. 15), the two-

ionic plasma parameter can be written

For a one-ionic component plasma with tempe-

rature equilibrium conditions (Te

=

Ti), the plasma parameter becomes

Finally, the electronic plasma parameter Te can be expressed using the correlation parameter v :

with

(8)

443

and F. is connected to the plasma parameter Ae [5]

by the equation

Appendix B.

EXPRESSIONS FOR d AND H(o).

-

The characteristic reduced length d can be expressed in a general form

as a function of r, using the numerical results given by Tanaka and Ichimaru [35] and the limits

In particular, writing d in the general form

it is easy to see that the asymptotic limits corres- pond to

For two particular values of r (10-2 and 1), the

curve d(r) [35] gives two supplementary relation- ships

and

Four coefficients are obtained by resolving this system :

General expressions for the screening potential H(y) at the origin have also been proposed in papers

relating to electron screening and thermonuclear reactions [25, 27].

These results, obtained in two limits, can be expressed as

noticing that, for a two-ionic component plasma, a

distinction must be made between interactions « a-a »

and « a-b », assuming that « a » is a central ion, « a »

or « b » being the perturbers.

Under these conditions

in the approximation of a classical system, i and j being the notations for ions « a » or « b ».

Assuming that HiJ(0) can be generalized by [35]

the asymptotic limits give

Resolving this system, we find

For a one ionic component plasma (Z.

=

Zb

=

Z)

in equilibrium conditions (RT

=

1), these general expressions are reduced to

and the numerical results are in good agreement

with those of Tanaka and Ichimaru [35] for all values

of r.

(9)

Appendix C

DETERMINATION OF A, B, C.

-

The determination of A, Band C can be made using Monte Carlo nume-

rical results [13 to 17] for one-component plasmas.

The choice of three particular points (i.e., the first

maximum and the first two values of y corresponding

to the condition g(y)

=

1 when r > 3) is the most simple way. This approach gives a three equation system. A comparison between these results and the accurate numerical method presented in the second part of this Appendix (Figs. 5 to 7) shows the existence

of a general law for A(r), B(r), and C(r).

Fig. 4.

-

Correlation function gaa(y) and gab(y) for a two-

ionic component plasma (Z.

=

9, Zb

=

1, p

=

0.5, v = 0.8).

Fig. 5.

-

Parameter A as a function of r. The solid line represents the analytical fit for A(r). The height of squared

dots corresponds to the approximation bare of numerical

values.

Fig. 6.

-

Parameter B as a function of r. The solid line represents the analytical fit for B(r). The height of squared

dots corresponds to the approximation bare of numerical values.

Fig. 7.

-

Parameter C as a function of r. The solid line represents the analytical fit for B(r). The height of squared

dots corresponds to the approximation bare of numerical values.

The numerical method used for the calculation of the coefficients A, B, and C is based on both, an iterative method for numerical solution of non-

linear equations and on a least square method.

It follows from equation (18) that the function

F(y) depends explicitly on the coefficients A, B, and

C. Thus we can write

(10)

445

Let E(A, B, C, Yi) be a discrete function defined by

i indicates the i-th value of the reduce length y, and

f is the corresponding value of the function F(y)

deduced from numerical data of the correlation

function gi (i

=

1,

...,

M) :

where T and H(O) have been previously defined in

section 2, equations (4) and (5).

For each r, estimated values of the three coeffi- cients can be obtained by resolution of the following equation :

At the first order of the Taylor expansion, the last equation becomes

In a simple form, this can be written as follows :

where Xj is associated, for j = 1, 2, and 3, to the coeffi-

cients A, B, and C, and i corresponds to the values

of the derivatives DEij(Xt, X 2, X 3 ) or the function Ei(X(10), X2(0), x1°») for yi.

Thus an iterative solution can be obtained as follows [39] :

with the iteration XiN + 1)

=

x!N) + AXj and the

set of initial values X 1 (0), X 2 (0) and X 3 (0) j *

However, for each step of the iterative approxima-

tion equation (C. 3) must be verified for each y;, and equation (C. 6) becomes an overdetermined system of linear equations. A classical least square method is required to get an optimal solution.

The solution is given by the following linear system :

Thus the coefficients A, B, and C or X 1 (N), X 2 (N),

and X3(N) can be computed for each r with a relatively good approximation of the order of 10-2, and the

iterative method converges with a small number of iterations (about 5 to 15 iterations).

Finally, the semi-empirical analytic dependence

of the coefficients A, B, and C on r is obtained by

using a classical least square fit method.

(11)

References

[1] WHITNEY, K. G., KEPPLE, P. C., J. Quant. Spectrosc.

Radiat. Transfer, 27 (1982) 281.

[2] LEE, R. W., MATTHEWS, D. L., LANE, S. M., CAMPBELL,

E. M., KOPPEL, L., SCOFIELD, J., AUERBACH, J., LIM, T., J. Ph ys. B 15 (1982) L 317.

[3] RosE, S. J., Spectral Line Shapes, (W. de Gruyter,

Berlin-New York) 1985, vol. 3, p. 147.

[4] D’ETAT, B., NGUYEN, H., Spectral Line Shapes, (W. de Gruyter, Berlin-New York) 1985, vol. 3,

p. 209.

[5] HELD, ’3., J. Physique, 45 (1984) 1731.

[6] DERY, P., HÜCKEL, E., Z. Phys. 24 (1923) 185.

[7] HANSEN, J. P., TORRIE, G. M., VIEILLEFOSSE, P., Phys.

Rev. A 16 (1977) 2153.

[8] ABE, R., Prog. Theor. Phys. 22 (1959) 213.

[9] DE WITT, H. E., Phys. Rev. 140A (1965) 466.

[10] HIRT, C. W., Phys. Fluids 8 (1965) 693.

[11] HIRT, C. W., Phys. Fluids 10 (1967) 565.

[12] DEL RIO, F., DE WITT, H. E., Phys. Fluids 12 (1969) 791.

[13] BRUSH, S. G., SAHLIN, H. L., TELLER, E., Bull. Am.

Phys. Soc. 10 (1965) 232.

[14] BRUSH, S. G., SAHLIN, H. L., TELLER, E., J. Chem.

Phys. 45 (1966) 2102.

[15] HANSEN, J. P., Phys. Rev. A, 8 (1973) 3096.

[16] SLATTERY, W. L., DOOLEN, G. D., DE WITT, H. E., Phys. Rev. A 21 (1980) 2087.

[17] SLATTERY, W. L., DOOLEN, G. D., DE WITT, H. E., Phys. Rev. A 26 (1982) 2255.

[18] CARLEY, D. D., J. Chem. Phys. 43 (1965) 3489.

[19] SPRINGER, J. F., POKRANT, M. A., STEVENS, F. A.,

J. Chem. Phys. 58 (1973) 4863.

[20] IYETOMI, H., Prog. Theor. Phys. 71 (1984) 427 ; COOPER, M. S., Phys. Rev. A 7 (1973) 1.

[21] NG, K. C., J. Chem. Phys. 61 (1974) 2680.

[22] DEUTSCH, C., FURUTANI, Y., GOMBERT, M. M., Phys.

Rev. A 13 (1976) 2244.

[23] DEUTSCH, C., FURUTANI, Y., GOMBERT, M. M., Phys.

Rep. 69, (1981) 85.

[24] BERNU, B., HANSEN, J. P., HAZIGHI, R., Phys. Lett.

100A (1984) 28.

[25] SALPETER, E. E., Austr. J. Phys. 7 (1954) 373.

[26] CANUTO, V., Astrophys. J. 159 (1970) 641.

[27] DE WITT, H. E., GRABOSKE, H. C., COOPER, M. S., Astrophys. J. 181 (1973) 439.

[28] ITOH, N., TOTSUJI, H., ICHIMARU, S., Astrophys. J.

218 (1977) 477.

[29] JANCOVICI, B., J. Stat. Phys. 17 (1977) 357.

[30] ALASTUEY, A., JANCOVICI, B., Astrophys. J. 226 (1978)

1034.

[31] ITOH, N., TOTSUJI, H., ICHIMARU, S., DE WITT, H. E., Astrophys. J. 234 (1979) 1079.

[32] VIEILLEFOSSE, P., J. Physique 42 (1981) 723.

[33] MCGOWAN, D., J. Phys. A 15 (1982) 1007.

[34] KALMAN, G., Phys. Rev. A 30 (1984) 627.

[35] TANAKA, S., ICHIMARU, S., J. Phys. Soc. Japan 53 (1984)

2039.

[36] ICHIMARU, S., TANAKA, S., IYETOMI, H., Phys. Rev. A

29 (1984) 2033.

[37] HUBBARD, W. B., Astrophys. J. 146 (1966) 858.

[38] ANGELIE, A., GILLES, D., Private communication.

[39] PELLETIER, P., Techniques numériques appliquées au

calcul scientifique (Masson) 1979, p. 96.

Références

Documents relatifs

Although crossover phenomenon can be general upon ap- proach of the Ising-like critical point, such a modeling, in which G is a tunable parameter, is essential to check carefully

Large-momentum behavior of the momentum distribu- tion, short-distance behavior of the pair distribution function and of the one-body density matrix, derivative of the energy

We started our presentation with explaining the discrepancies between two widely used formulas for the interdiffusion coefficient in binary liquids, the Hartley-Crank

A n explicit evaluation of Rye /3/ is done by perturbation method with the acouraay up to terms proportio- nal to the square of the wave vector, Moreover the

Klibanov, Global uniqueness of a multidimensionnal inverse problem for a non linear parabolic equation by a Carleman estimate, Inverse Problems, 20, 1003-1032, (2004).

Furthermore, unwanted and uncontrolled visceral metastases, particularly to the lung, can significantly decrease mouse survival span, limiting the time frame of skeletal metastases

Despite a lower level of plcR mRNA in ΔyvfTU, the transcriptomic and proteomic analysis both revealed that the expression of the PlcR regulon was mostly not affected by the

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des