Вы находитесь на странице: 1из 6

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/234024970

A one-dimensional plume plasma expansion:


Self-similar approach

Article in Physics Letters A · February 2012


Impact Factor: 1.68 · DOI: 10.1016/j.physleta.2011.11.018

CITATIONS READS

2 82

3 authors, including:

Djamila Bennaceur-Doumaz Mourad Djebli


Centre Algérien de Développement des Tec… University of Science and Technology Houa…
42 PUBLICATIONS 54 CITATIONS 57 PUBLICATIONS 160 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Djamila Bennaceur-Doumaz
letting you access and read them immediately. Retrieved on: 18 May 2016
JID:PLA AID:21021 /SCO Doctopic: Plasma and fluid physics [m5Gv1.5; v 1.58; Prn:25/11/2011; 15:54] P.1 (1-5)
Physics Letters A ••• (••••) •••–•••

Contents lists available at SciVerse ScienceDirect

Physics Letters A
www.elsevier.com/locate/pla

A one-dimensional plume plasma expansion: Self-similar approach


Rachid Fermous a , Djamila Bennaceur-Doumaz b , Mourad Djebli a,∗
a
Theoretical Physics Laboratory, Faculty of Physics, USTHB, Algiers, Algeria
b
Centre de Développement des Technologies Avancées, Cité du 20 Août 1956, BP. 17 Baba Hassen, Algiers, Algeria

a r t i c l e i n f o a b s t r a c t

Article history: The expansion of a plasma induced by laser ablation is investigated using a single-fluid model combined
Received 14 June 2011 with Saha’s equation. The space coordinates x and time t are combined to a one self-similar variable ξ =
Received in revised form 31 October 2011 x/(ct ). To obtain ordinary differential equations, two different transformations for the density are used.
Accepted 1 November 2011
The density profiles during the expansion are found to be completely different, one corresponds to the
Available online xxxx
Communicated by F. Porcelli
common results i.e., the density decreases monotonically with ξ , while with the second transformation,
the profile shows a density increasing for certain interval of the self-similar variable. This effect is
enhanced with higher initial ionization fraction. The role of the initial velocity which corresponds to
the start of the expansion from an unperturbed plasma or from an expansion already going on is pointed
out.
© 2011 Elsevier B.V. All rights reserved.

The self-similar analysis of nonlinear partial differential equa- plasma into the vacuum for which ion velocity with respect to the
tions sometimes leads to rigorous solutions for problems in fluid acceleration can be obtained numerically [11]. In the present Let-
mechanics, diffusion and wave propagation [1]. Herein, two major ter, we focus our investigation on the plume plasma expansion for
classes of self-similar transformations have been used. One dealing which a one-dimensional hydrodynamical model is used. Although
with wave propagation, in terms of it the wave is depending only the self-similar variable ξ = x/ct is the same as in previous works,
on the combination of the space coordinates and time, i.e. x − ct. our attention is paid to the ansatz leading to normalize the physi-
So that the solution of the partial differential equation governing cal parameters, in particular the density. In this case, two different
the wave propagation keeps its exact shape and therefore, called transformations are used. Not only the density transformation that
solitary waves [2]. Instead of following the one wave motion in its plays an important role on the expanding profiles but also bound-
frame of reference, the governing equations of the expansion from ary conditions. As the self-similar approach gives an asymptotic
one medium to vacuum or to another medium, are transformed solution the question, whether the true evolution, obtained by a
into a form that allows to follow the expanding front. Lagrangian code for the ion fluid dynamics, being the exact one,
Self-similar approach is important in plasma expansion. In this which one wants to approach by a simplified model such as quasi-
case, the flow variables do not depend on the coordinates and neutrality or even self-similarity, can only be decided by a direct
time separately, but on the ratio x/t, for a one-dimensional situ- comparison. This means that without knowledge of the exact solu-
ation. Therefore, the fluid does not have any characteristic length tion, it is difficult to make any proposal about which simplification
or time [3]. The self-similar solution for an expanding plasma into holds after which time and in which region. Therefore, it is appro-
a vacuum was given by Gurevich et al., based on the ion Vlassov priate to investigate different schemes and compare the obtained
equation, the analysis had been developed in planar geometry [4]. solutions to make a statement about the more accurate, using a
Self-similar solutions have been used in different situations as long physical argument when the exact solution is not available.
as the expansion occurs with a quasi-neutral plasma expansion [5], The removal of material from an irradiated target surface by a
such as the power-driven expansion of a foil of negligible thick- high-intensity laser pulse, i.e., laser ablation involves mainly two
ness [6] or in dusty plasma expansion [7,8]. Moreover, in laser stages:
ablated plasma the self-similar solution has been found for isother-
mal or adiabatic situations [9] and taking into account the pres- 1. absorption of laser light that leads to heat and ionize the
ence of nonthermal electrons [10]. In a previous paper, Huang et al. formed vapor;
obtained a general solution for an isothermal expansion of neutral 2. plume expansion.

As the characteristic time of the expansion of the laser-induced


* Corresponding author. plasma after the end of the laser pulse is much longer than the
E-mail address: mdjebli@usthb.dz (M. Djebli). time of plasma formation and initial plasma expansion duration,

0375-9601/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.physleta.2011.11.018
JID:PLA AID:21021 /SCO Doctopic: Plasma and fluid physics [m5Gv1.5; v 1.58; Prn:25/11/2011; 15:54] P.2 (1-5)
2 R. Fermous et al. / Physics Letters A ••• (••••) •••–•••

the two stages can be considered separately [12]. Moreover, the x/C s t. However, we also need to transform the plasma parameters,
propagation of plasma can be considered one-dimensional only up i.e., the density, the velocity, the pressure and the temperature.
to a distance from the target of the order of the laser spot size [13]. We focus on density transformation. There are mainly two differ-
For plasmas consist of electrons, ions and atoms, and seeking sim- ent ways to normalize the density, as shown in Appendix A:
plicity, it can be assumed that there is only one component. This
is the single-fluid description. So, the density n and the velocity v • For collisionless expansion, we normalize the density to its ini-
are for a fluid element positioned at x at time t. The fluid equa- tial value no , so we write n = no N I , N I corresponds to the
tions for the single-fluid model are obtained by adding the fluid transformed normalized density.
equations of all species [14] • When fluid equations involve sink/source terms, the previ-
ous transformation can no longer lead to a self-similar set of
∂n ∂
+ (nv ) = 0 (1) equations. In this case,we choose the transformation such as
∂t ∂x n = no N II /ωt, ω = 2π no e 2 / M is the ion plasma frequency
  
∂ ∂ T which determines the expansion time scale. The plasma fre-
(nv ) + nv 2 + n (1 + η) =0 (2)
∂t ∂x M quency depends on the density at equilibrium no .
 2  
∂ v 3 T Ui
n +n (1 + η) + η It is important to note that the last transformation gives also a set
∂t 2 2M M of self-similar equations even in the absence of sink/source terms.
 2  
∂ v 5 T Ui Therefore, it is reasonable to question the choice of the self-similar
+ n v + nv (1 + η) + η = 0 (3) transformation. Now, we focus just on that difference and re-write
∂x 2 2M M
the set of Eqs. (1)–(4). All parameters corresponding to the case
where M and T are the mass of the plume atoms (ions) and the n = no N I are labeled by ( I ). For the second case, we label the
temperature, respectively. The plume is assumed to behave as an parameters by (II). To obtain the set of differential equations in
ideal gas of pressure p = n(1 + η)k B T , η is the ionization fraction. dimensionless form we normalize the thermal and the ionization
At thermal equilibrium the ionization-recombination is gov- energy by MC s2
erned by Saha’s equation
  32   ∂ NI ∂VI
ηe ηi 2 U + 2π me k B T Ui (V I − ξ ) + NI =0 (6)
= exp − (4) ∂ξ ∂ξ
ηo n Uo h2 kB T ∂VI TI ∂ NI ∂TI ∂ ηI
(V I − ξ ) + (1 + η I ) + (1 + η I ) + TI = 0 (7)
where the ionization fractions are ηo = nn /n, ηe = ne /n, ηi = ni /n. ∂ξ Ni ∂ξ ∂ξ ∂ξ
nn , ne , ni refer to neutral atoms density electron density and ion   
1 5 ∂VI
density, respectively. h is Planck’s constant, U i is the ionization (V I − ξ ) + T I (1 + η I ) + U i η I
energy and U + and U o are the electronic partition functions. k B VI 2 ∂ξ
 
is the Boltzmann constant. The initial gas density is no = nn + ni , 5 3ξ ∂TI
which is equal to the density n(t = t o ) when the expansion starts.
+ − (1 + η I )
2 2V I ∂ξ
Using the quasi-neutrality condition for ions with one ioniza-     
tion level ni = ne and the mass conservation ηi + ηo = 1, Saha’s
5 3ξ ξ ∂ ηI
+ − TI + 1 − Ui
equation (Eq. (4)) can be written in the form 2 2V I V ∂ξ
    
  32   5 3ξ ξ 1 ∂ NI
η2 2U + 1 2π m e T Ui + − T I (1 + η I ) + 1 − U i ηI = 0 (8)
= exp − (5) 2 2V I VI N I ∂ξ
1−η U0 n h2 T  
ηI ∂ N I ηI 3 U i ∂ T I 2 − ηI ∂ ηI
where ηi = η and T is given in energy unit. − + + =0 (9)
N I ∂ξ TI 2 T I ∂ξ 1 − η I ∂ξ
The finite expansion is studied in the limit of half space ex-
pansion. At the initial stage, the gas is supposed to fill the half For the second transformation one obtains
space x < 0. In the time scale corresponding to the ion motion,
( V II − ξ ) ∂ N II ∂ V II
the expansion front is initially√ moving with the same velocity as + =1 (10)
the ion acoustic wave C s = 5T /3M. For the isothermal case, the N II ∂ξ ∂ξ
electric field is space independent and is different from zero when ∂ V II 1 ∂ N II
x → 0. When the expansion starts with an unperturbed plasma ( V II − ξ ) + T II (1 + ηII )
∂ξ N II ∂ξ
(n(ξo ) = no ), ions located at the discontinuity x = 0 are accelerated
∂ T II ∂ ηII
to infinity by the infinite electric field as soon as t > 0, the ion ve- + (1 + ηII ) + T II =0 (11)
locity increases linearly in x for a given t. However, a quasi-neutral ∂ξ ∂ξ
adiabatic plasma expansion has a front velocity which can reach ∂ V II 3 ∂ T II
a maximum speed of 2C s /(γ − 1), in our case this corresponds
T II (1 + η) + ( V II − ξ )(1 + ηII )
∂ξ 2 ∂ξ
to 3C s (Refs. [5,14]). For an adiabatic expansion with ionization-  
3 U i ∂ ηII
recombination, we expect this limit to be higher when η = o. The + ( V II − ξ ) T II + =0 (12)
kinetic energy of the plume is redistributed and the increase of 2 T II ∂ξ
 
entropy results in a higher expansion maximum velocity. ηII ∂ N II ηII 3 U i ∂ T II (2 − ηII ) ∂ ηII ηII
The fluid equations (1)–(3) combined with Saha’s govern the − + + =− (13)
N II ∂ξ T II 2 T II ∂ξ (1 − ηII ) ∂ξ ξ
expansion of a plasma with a moving boundary. Solving the set of
partial differential equations that depends on x and t is not an easy The sets of Eqs. (6)–(9) and (10)–(13) are now depending only on
task in this case. The existence of a transformation(s) of variable one variable (ξ ), so the partial differential derivative can be seen as
which achieve a reduction of independent variable in a system of an ordinary one. To solve both sets of equations, we need bound-
equations is a direct consequence of the self-similarity [1]. In the ary conditions that correspond to ξ = ξo . The numerical solution is
present study, we adopt the standard transformation used in ex- implemented by using starting values for the density, the velocity,
panding front to combine time and space coordinates, i.e., ξ(x, t ) = the temperature and the ionization fraction which corresponds to
JID:PLA AID:21021 /SCO Doctopic: Plasma and fluid physics [m5Gv1.5; v 1.58; Prn:25/11/2011; 15:54] P.3 (1-5)
R. Fermous et al. / Physics Letters A ••• (••••) •••–••• 3

Fig. 1. Normalized density versus the self-similar variable with different ionization
fractions at equilibrium: ηo = 0.1 (—), 0.25 (− · −), 0.5 (− − −), 0.9 (· · ·). Plots cor-
respond to the case (I), the density transformation is n = no N I . Boundary conditions Fig. 2. Normalized density versus the self-similar variable with different ionization
correspond to ξo that ensures V (ξo ) = 0 and N (ξo ) = 1. fractions at equilibrium: ηo = 0.1 (—), 0.25 (− · −), 0.5 (− − −), 0.9 (· · ·). Plots cor-
respond to the case (II), the density transformation is n = no N II /ωt. The boundary
conditions are taken at ξo = 0.
ξ = ξo . The choice of ξo is also subject to deep analysis. By com-
bining the space coordinate and time, we loose the information
that corresponds to the experimental setup. So, the boundary con- of the self-similar solution. Any change on the initial self-similar
ditions correspond to an equilibrium state for which the plasma value turns out to be the same translation of the limited value of
starts to expand. There are two different approaches to implement the self-similar solution.
the numerical simulation: For the second ansatz (II), as the set of differential equations
(10)–(13), is nonhomogeneous, it is impossible to use the con-
• Give any starting value for ξo (say 0 or 1) that ensures no ditions: V (ξo ) = 0 and N (ξo ) = 1. Therefore, we start from an
vanishing denominator in any terms of Eqs. (6)–(9) or (10)– arbitrary value of ξo which is 0. Besides the limit of the self-
(13). For this ξo , the density, speed and temperature are taken similar solution which is different in comparison with the case (I),
from experimental data. The value given to ξo is not important, the curves (Fig. 2) show concavity instead of convexity as it is in
the transformation preserves the shape under translation. the first case. The profile shown by Fig. 2 can be split into two
• Instead of imposing any value to ξo , we use the fact that the parts, a density increasing that can be attributed to radiation en-
plasma is initially unperturbed which means V (ξo ) = 0 and ergy transfer which is used to ionize more neutral atoms and thus
N (ξo ) = 1 [15]. Solving the latter equation gives ξo which it increases the density of the ions instead of being used as kinetic
turns out to be depending on the equilibrium values. energy. Far away from the source region, the radiation energy is
less than the ionization energy and therefore, the ions are accel-
In the present work boundary conditions correspond to the ab- erated to overcome the tendency to break the quasi-neutrality due
lation of a titanium target, just after the end of the laser pulse, to electron motion ahead of the expanding front. The end of the
which gives only one ionized species in the presence of the pulsed expansion corresponds to the beak-down of quasi-neutrality. The
laser [16]. main difference between cases (I) and (II) may be attributed to
Ordinary differential equations (6)–(9) and (10)–(13) are solved the transformation itself. In case (I) the differential equations lead
numerically to plot the normalized density versus the self-similar to homogeneous set of equations, for nontrivial solutions the de-
parameter. In Fig. 1 we plotted the density profile for the expan- terminant of the system must vanish. The √ velocity as a function
sion of a plasma with different ionization fractions at equilibrium. of ξ has a linear profile, i.e., V − ξ = ± 5T /3, when T is con-
The starting self-similar value is ξ ∼ 1.3 for η = 0.1 (solid line), but stant (isothermal case). The case (II) leads to a nonhomogeneous
it is ξ ∼ 1.5 for η = 0.25 (dashed dotted line). All curves have the set of differential equations, there is no analytical expression for
same intersection point ξo ∼ 0.86 obtained by solving the set of V , the profile is obtained from the numerical solution. Although
equations with V (ξo ) = 0 and N (ξo ) = 1. The expansion is associ- both density transformations lead to a set of self-similar equations
ated with a rarefaction wave propagating in the opposite direction in our opinion, the second one is more appropriate for the case un-
to the expanding front, the result given by Fig. 1 corresponds to der investigation. Let us remember that equation of motion of the
common profiles of a plasma expansion into the vacuum. The ion- single-fluid model is obtained by adding the equations of motion
ization effect is seen only on the expansion velocity and the limit of all species where collision terms are included. As the collisional
of the self-similar parameter ξl corresponding to N I → 0 which is term describes the transfer of momentum from ions to electrons
ξl ∼ 6.5 for ηo = 0.25 and ξl ∼ 7.5 for ηo = 0.5. Now if one uses and from electrons to ions, conservation of momentum requires
an arbitrary choice of the starting ξo but with the same condi- that collision terms are not included in the single-fluid equation
tions for N o and V o , the plot is similar to Fig. 1 but starting from of motion. However, the effect of collisions is considered through
ξo = 0. The change concerns the limit of the self-similar parame- Saha’s equation and the energy conservation equation.
ter. For example, it is ξl ∼ 8.5 when η0 = 0.25. The combination of It is important to note that boundary conditions play an im-
having both initial conditions on ions density and velocity did not portant role in the plume expansion. We have assumed that the
yield new physical insight. The entire self-similar domains are the formed plume is initially held in the reservoir. At the starting of
same, the only changes concern the limit of the validity domain the expansion, the gate is opened at t = 0. The plume has an ini-
JID:PLA AID:21021 /SCO Doctopic: Plasma and fluid physics [m5Gv1.5; v 1.58; Prn:25/11/2011; 15:54] P.4 (1-5)
4 R. Fermous et al. / Physics Letters A ••• (••••) •••–•••

the plume kinetic energy. Therefore, in the light of what the first
two transformations learn, the second self-similar transformation
may be more accurate to investigate plume plasma expansion, in-
cluding ionization effects.
To conclude, the one-dimensional plasma expansion has been
investigated using the self-similar approach. The plasma of ions,
electrons and atoms is governed by the single-fluid equation com-
bined with Saha’s equation. The set of differential equations is
transformed by using the similarity variable ξ = x/C s t. Two trans-
formations are applied to the density, the first is n = no N I and
the second is n = no N II /ωt. The first transformation gives a con-
vex profile while the second corresponds to a concave profile.
Although both transformations lead to a self-similar solution, to
make any conclusion about the appropriate asymptotic solution,
one has to compare the given solution with the exact one other-
wise physical consideration has to be the only possible criterion
for any statement. Transformation (I) may give the exact profile
for vapor expansion or plume expansion with very low ionization
rate where both the one-fluid and the individual profiles are self-
similar. Higher ionization rates means that a momentum transfer
is important between the plasma different species. In this case the
second transformation (II) is more appropriate. Moreover, the ve-
locity shows deviation from the known linear profiles versus the
self-similar variable. The deviation is enhanced by a higher initial
ionization rate as a consequent of kinetic energy redistribution.

Acknowledgements

One of the authors M.D. thanks Pr. Hans Schamel for fruitful
discussion on self-similar formulation. The constructive sugges-
tions of the anonymous referee are also gratefully acknowledged.

Appendix A. Density transformation

For the density self-similar transformation, we write the con-


tinuity fluid equation in its general form (the same can be done
with momentum equation):

∂ n ∂(nv )
+ = ±a j n j n (A.14)
∂t ∂x
where ν j = a j n j stands for the collision frequency between the
species of density n j and the species of density n. When we nor-
malize the density by its equilibrium value no and the velocity by
C s , one obtains:

∂ NI ∂( N I V I )
Fig. 3. (Color online.) Normalized velocity versus the self-similar variable (a) cor- + Cs = ±a j no N j N I (A.15)
responds to the self-similar transformation (I) and (b) corresponds to the case (II). ∂t ∂x
Red curves with initial ionization rate ηo = 0.1 and black curves with ηo = 0.9. The
But with the transformation n = no N II /(β t ), where β is a constant,
initial velocity is V o = 0 (—) and 1 (− − −).
we obtain:
   
tial ionization rate ηo , with this scenario the front initial velocity is ∂ N II ∂ N II V II a j no
+ Cs = ± 2 N II N j (A.16)
V o = 0. However, it is well known that the expansion takes place ∂t t ∂x t βt
just after the vapor formation in this case the ion front velocity
Let us recall the self-similar variable ξ = x/(C s t ), for Eq. (A.15) one
is V o = 0. In Fig. 3 is plotted the normalized velocity versus the
obtains:
self-similar variable, (a) corresponds to the case (I) and (b) to the
case (II). We have combined two different initial ionization rates ∂ NI ∂( N I V I )
ηo = 0.1 (red curves) and ηo = 0.9 (black curves) with two differ- −ξ + = ±a j no N I N j t (A.17)
∂ξ ∂ξ
ent values of V o , solid lines correspond to V o = 0 and dashed ones
to V o = 1. The latter case means that the expansion front is equal with the same self-similar variable ξ = x/(C s t ) Eq. (A.16) gives,
to the ion-acoustic speed. This is corrected if ξo = 0 and the expan- ∂ N II ∂( N II V II ) a j no
sion is isothermal [5]. We use this value to make a comparison. For −ξ − N II + =± N II N j (A.18)
∂ξ ∂ξ β
the self-similar transformations (II) the velocity profile is linear for
V o = 0, whatever is the initial ionization rate and it is linear for It is clear from Eq. (A.17) that the transformation labeled by ( I ) is
low ionization rate with the transformation (I). When V o = 1 and self-similar only if a j = 0, no sink/source terms in continuity equa-
with higher initial ionization rate the profile is nonlinear, the max- tion, otherwise time appears explicitly. The second transformation
imum velocity is higher as a consequence of the redistribution of (II) gives a self-similar equation whatever is a j = 0 or = 0, in this
JID:PLA AID:21021 /SCO Doctopic: Plasma and fluid physics [m5Gv1.5; v 1.58; Prn:25/11/2011; 15:54] P.5 (1-5)
R. Fermous et al. / Physics Letters A ••• (••••) •••–••• 5

sense the second transformation of the density is more general. [5] Ch. Sack, H. Schamel, Phys. Rep. 156 (1987) 311.
The constant β is an arbitrary parameter used to normalize the [6] A.V. Farnsworth, M.M. Widner, M.J. Clauser, P.J. McDaniel, K.E. Lonngren, Phys.
Fluids 22 (1979) 859.
collision frequency, so it must have the same unit as frequency.
[7] K.E. Lonngren, Planet. Space Sci. 38 (1990) 1457.
We can choose to normalize ν j by the plasma frequency at equi-
 [8] M. Djebli, Phys. Scr. 81 (2010) 025902.
librium β = ω = 2π no e 2 / M. The latter depends on the density [9] M. Murakami, Y.-G. Kang, K. Nishihara, S. Fujioka, H. Nishimura, Phys. Plas-
at equilibrium. mas 12 (2005) 062706.
[10] D. Bennaceur-Doumaz, M. Djebli, Phys. Plasmas 17 (2010) 074501.
[11] Y. Huang, Y. Bi, X. Duan, N. Wang, X. Tang, Y. He, Appl. Phys. Lett. 92 (2008)
References 031501.
[12] B. Wu, Y.C. Shin, Phys. Lett. A 371 (2007) 128.
[13] M. Capitelli, A. Casavola, G. Colonna, A. De Giacomo, Spectrochim. Acta, Part
[1] H.C.S. Hsuan, K.E. Lonngren, W.F. Ames, J. Eng. Math. 8 (1974) 303. B 59 (2004) 271.
[2] A. Wazwaz, Partial Differential Equations and Solitary Waves Theory, Springer, [14] K.R. Chen, J.N. Leboeuf, R.F. Wood, D.B. Goehegan, J.M. Denato, C.L. Liu, A.A.
Dordrecht, 2009. Puretzky, Phys. Rev. Lett. 75 (1995) 4706.
[3] Ya.B. Zel’dovich, Yu.P. Raizer, Physics of Sock Waves and High-Temperature Hy- [15] D. Anderson, M. Bonnedal, M. Lisak, Phys. Scr. 22 (1980) 507.
drodynamic Phenomena, Academic Press, New York, 1966. [16] S. Abdelli-Messaci, T. Kerdja, A. Bendib, S. Messaoud-Aberkane, S. Lafane,
[4] A.V. Gurevich, L.V. Pariĭskaya, L.P. Pitaevskiĭ, JETP 22 (1966) 449. S. Malek, J. Appl. Surf. Sci. 252 (2005) 2012.

Вам также может понравиться