Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Spectrochimica Acta Part B 65 (2010) 616–626

Contents lists available at ScienceDirect

Spectrochimica Acta Part B


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s a b

Kinetic processes for laser induced plasma diagnostic: A collisional-radiative


model approach☆
L.D. Pietanza a,⁎, G. Colonna a, A. De Giacomo a,b, M. Capitelli a,b
a
CNR-IMIP sect. Bari, via Amendola 122/D, 70126 Bari, Italy
b
Bari University, Chemistry Department, via Orabona 4, 70126 Bari, Italy

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

Article history: A zero-dimensional collisional–radiative (CR) model, coupled self-consistently with the electron Boltzmann
Received 1 December 2009 equation, has been applied to the description of a metallic-laser induced plasma at experimental conditions
Accepted 17 March 2010 typical of LIBS experiment. To take in account expansion effects, the experimental temperature and total
Available online 25 March 2010
number density as function of time have been used as input data. Plasma composition and the simultaneous
time evolution of both heavy particle level distributions and the electron energy distribution function have
Keywords:
Collisional–radiative model
been calculated by taking into account the most relevant collisional and radiative processes. This approach
Kinetic processes estimates the hierarchy of the elementary processes during the expansion and possible deviations from LTE
LIBS conditions. The comparison of the experimental and theoretical results shows a good agreement, but at the
Aluminium same time new questions arise on the analysis of spectroscopic results and on the assumption generally
Local Thermodynamic Equilibrium (LTE) made in LIBS.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction atomic population of level i, determined by the atomic state dis-


tribution, over the same quantity calculated by the Saha equation
Laser induced plasma (LIP) spectral analysis is based on the (NSi ), which sensitivity is of one order of magnitude [5].
assumption of Local Thermodynamic Equilibrium (LTE): the distribu- From a theoretical point of view smaller deviations can be
tions of atomic excited states and electron kinetic energies are estimated and ascribed to specific improper balances of kinetic
considered to be, respectively, a Boltzmann and a Maxwell at the same processes [6]. The same argument can be addressed to the above-
temperature. The validity of LTE approximation in stationary plasmas mentioned McWhirter criterion [1,7] where the rate of collisional de-
is generally deduced by using the well-known McWhirter criterion excitation should be assumed to be at least 10 times the radiative one.
[1], which indicates the minimum value that electron density must Another important aspect is the measurement of the electron
have to ensure the prevalence of collisions on radiative losses. The temperature: Te is usually obtained considering the equilibrium
transient nature of LIP limits the applicability of the McWhirter with the temperature of electronically excited states measured by
criterion and, therefore, other conditions must be considered, to relate means of the Boltzmann plot technique [8]. The main drawback of this
the characteristic time of changing the thermodynamic parameters, methodology is that only the bulk of the atom and ion distributions
due to plasma expansion, with that needed by the kinetic processes can be actually measured by UV–VIS optical emission spectroscopy,
for the establishment of excitation–ionization equilibria [2,3]. when it is reasonable to ascribe most deviations from equilibrium to
Recently, the criteria for the assessment of LTE conditions have the lower lying levels and/or to levels belonging to the distribution tail
been reviewed in Ref. [4], together with experimental methodologies. [9], close to the ionization energy. Thus, despite the optical emission
However, only large deviations from LTE can be detected experimen- spectroscopy can be successfully applied as diagnostic for LIP
tally. In fact, a typical methodology consists in the experimental processes or chemical analysis, it should resort to microscopic kinetic
determination of the parameter bi = Ni / NSi , defined as the ratio of the theory for further insight of possible deviations from LTE.
Collisional–radiative (CR) model [10–17] calculates plasma com-
position and heavy particle (atoms and atomic ions) level distribu-
tions, by solving a system of rate (master) equations. Previous papers
calculate electron impact rate coefficients using a Maxwellian electron
☆ This paper was presented at the 5th Euro-Mediterranean Symposium on Laser energy distribution function (eedf) at a fixed temperature. However,
Induced Breakdown Spectroscopy, held in Tivoli Terme (Rome), Italy, 28 September–1 departure from LTE conditions could induce deviations of eedf from
October 2009, and is published in the Special Issue of Spectrochimica Acta Part B,
dedicated to that Symposium.
Maxwellian and the CR model master equations should be solved self-
⁎ Corresponding author. Tel.: + 39 080 5929502; fax: + 39 080 5929520. consistently with the Boltzmann equation (BE) for free electrons [9].
E-mail address: daniela.pietanza@ba.imip.cnr.it (L.D. Pietanza). This method, also employed in this work, allows the analysis of the

0584-8547/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.sab.2010.03.012
L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626 617

simultaneous temporal evolution of heavy particle level distributions balance principle, which assures that, at equilibrium, the rates of the
and eedf. direct and inverse process are equal.
In this work, the CR model has been built for Aluminium LIP at Radiative processes (Eqs. (3), (4)), lacking of the inverse process,
experimental conditions typical of LIBS experiment. The equations are are responsible of the departure from equilibrium.
solved in a homogeneous system, but, to account for the expansion, Electron impact rate coefficients depend on the electron energy
translation temperature and pressure are assumed to change distribution function n(ε) (eedf) according to
following spatially integrated and time resolved experimental data.

Moreover, investigation has been extended also to different Kij = ∫ε⁎ σij ðεÞnðεÞvðεÞdε ð8Þ
ij
conditions by changing gas temperature and total density input
data, showing the sensitivity of the model to these parameters and in
where ε*ij represents the energy threshold, ν(ε) the electron velocity
which conditions non-equilibrium can be more evident.
and σ the cross section.

2. Collisional–radiative model
2.1. Boltzmann equation for free electrons

The collisional–radiative model here presented refers to a plasma


In the homogeneous and quasi-isotropy conditions, the Boltzmann
containing a generic atom A, its single and double charged ions A+, A++
equation for free electrons becomes [18–23]:
and free electrons. The following kinetic processes are considered
1) electron impact excitation and de-excitation ∂nðεÞ ∂J ðεÞ ∂J ðεÞ ∂Jee ðεÞ
= − E − EL − + San ðεÞ + Ssup ðεÞ ð9Þ
∂t ∂ε ∂ε ∂ε
− Z Z −
e ðεÞ + A ðiÞ↔A ðjÞ + e ðε′Þ ð1Þ
where the J terms, on the right hand side correspond to flux in the
2) electron impact ionization and three-body recombination electron energy space due to the electric field, elastic electron–atom
and electron–electron collisions; the S terms are due to inelastic and
− Z Z + 1 − − superelastic collisions, corresponding to electron impact excitation
e ðεÞ + A ðiÞ↔A ð0Þ + e ðε′Þ + e ð0Þ ð2Þ
and de-excitation of heavy particles as well as ionization and
3) radiative recombination recombination processes.
Electron–electron collisions redistribute energy among the elec-
A
Z+1 −
ð0Þ + e ðεÞ→A ðiÞ + hν
Z
ð3Þ trons forcing the distribution toward a Maxwellian. The number of e–e
collisions per unit time is directly proportional to the electron density
and inversely proportional to the cube of the electron velocity,
4) spontaneous radiative decay
justifying the departure from equilibrium at high energies and for
Z Z low electron density. The departure from equilibrium depends on the
A ðiÞ↔A ðjÞ + hνij ð4Þ
rate coefficients of inelastic and superelastic collisions. Inelastic
collisions, by exciting heavy particles internal levels, tend to bring
where ε's represent the electron energy, Z the charge, hν the emitted higher energy electrons towards lower energies, depopulating eedf
photon and the indexes in parenthesis refers to internal levels (i b j). tails. On the other hand, superelastic collisions transfer energy from
The collisional processes considered are only induced by electron internal levels to electrons.
impact while collisions between heavy particles have been neglected. The CR system of master equations (see Eqs. (5), (6)) and BE (see
The time evolution of ith level population density due to the processes Eq. (9)) are strongly coupled: the electron impact rate coefficients
listed above (Eqs. (1)–(4)) is described by the equation: depends on the electron distribution, while the S terms depend on
plasma composition as well as on atomic and ionic level distributions.
 
dNAZ ðiÞ exc deexc Eqs. (5), (6) and Eq. (9) are solved self-consistently obtaining the
= −ne NAZ ðiÞ ∑ Kij ðnðεÞÞ + ne ∑ Kji ðnðεÞÞNAZ ðjÞ
dt j j simultaneous temporal evolution of level distributions and of the eedf
ion 3−body rec 2 have.
−Ki ðnðεÞÞne NAZ ðiÞ + Ki ðnðεÞÞne NAZ + 1 ð0Þ
The system of differential equations is solved using implicit Euler
ricrad
−NAZ ðiÞ ∑ Aij + ∑ Aji NAZ ðjÞ + Ki ðnðεÞÞne NAZ + 1 ð0Þ method with a time step adaptative procedure [24].
jbi j Ni

ð5Þ 2.2. Energy level scheme

where Kexc, Kde-exc, Kion, K3-body rec and Kric rad represent the rate To build up a CR model, internal energy level scheme for each
coefficients of direct and inverse processes in Eqs. (1)–(3), which species and a complete database for electron impact cross sections
depends on the electron energy distribution function (eedf), and A and radiative transition probabilities are necessary.
the Einstein coefficients of spontaneous decay (Eq. (4)). For Al and Al+ atoms, energy levels have been taken from NIST
The corresponding equation for the electron density is: database [25] while Al++ has been considered only in its ground state.
  h i No cut-off has been considered [26] and the number of levels is kept
dNe ion 3body rec 2 constant.
= ∑ K ðnðεÞÞne NAZ ðiÞ−K ðnðεÞÞne NAZ⁎1 ð0Þ : ð6Þ
dt i

2.2.1. Aluminium
The result is a non-linear system of coupled differential equations. The ground state configuration is:
In this model we are considering also the quasi-neutrality condition,
 
i.e.: Al : ½Ne 3s 3p
2 2
P˚ :
ne = nAlþ + 2nAlþþ : ð7Þ
The excitation of the 3p-electrons results in the configurations 3s2
1
Electron collisions (Eq. (1), (2)) drive the internal level distribu- ( S) nx (with x = s, n ≥ 4; x = p, n ≥ 4; x = d, n ≥ 3; x = f, n ≥ 4; x = g,
tion towards the Boltzmann distribution, since they fulfil the detailed n ≥ 5, etc.).
618 L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626

Another possible configuration is 3s 3p2 and the excitation of the have been calculated from Einstein coefficients of sublevels using the
3p-electrons leads to 3s3p nx with two possible inner core configura- following formula:
tions (3P°) and (1P°). The three possible inner core configurations (1S),
(3P°) and (1P°) have three different ionization energies of 5.985 eV, ∑ gj Aji
10.637 eV, 13.406 eV, respectively. Aul =
i;j
ð10Þ
The observed terms in NIST database are 187 levels. In the present ∑ gj
j
model, only levels with (1S) core configuration have been included,
since excited levels with inner core configurations (3P°) and (1P°) are
unstable due to auto-ionization. The considered levels have been where l and u are the lower and upper lumped levels, while i and j
grouped into 14 lumped levels according to their excitation energy refers to the actual levels respectively in the groups u and l, and gj the
(see Table 1). statistical weight of j-th level.
The excitation energy shown in Table 1 corresponds to the minimum Excitation and ionization electron impact cross sections for Al and
of the levels in the group, while the statistical weight is the sum. Al+ are nearly absent in literature. In Aladdin database [27], only Al
and Al+ ionization cross sections from ground state can be found. To
2.2.2. Single charged aluminium ion have a cross section database as complete as possible, excitation and
The ground state configuration is: ionization cross section data have been calculated according to the
semi-empirical model of Grizinski [28]. By comparison Grizinski
  ionization cross sections with Aladdin database ones, good agreement
þ 2 1 exists being the differences, limited by a factor of two.
Al : ½Ne3s S :
Excitation cross sections of the lumped levels have been calculated
from the individual cross sections as the weighted average over the
The excitation of the 3s-electrons results in the configurations 3s lower individual levels and the sum over the upper individual levels
(2S)3p and 3s(2S)nx. Another possible configuration is 3p2 and the according to:
excitation of the 3p-electrons leads to 3p(2P°)nx. !
The two possible inner core configurations (2S) and (2P°) have two ∑ gi ∑ σij
exc

different ionization energies of 18.828 eV and 25.484 eV, respectively. exc i j


σlu = ð11Þ
The observed terms in NIST database are 217 levels. All these levels ∑ gi
i
have been grouped into 20 lumped levels (see Table 2).
using the same notation as in Eq. (10).
2.3. Radiative transition probabilities and electron impact cross sections
Finally, radiative recombination cross sections have been calcu-
lated according to Kramer formula [29], which holds for hydrogen-
The Einstein coefficients for spontaneous emission have been
like atoms and ions:
taken from NIST database [25]. The database provides 65 transitions
for Al and 135 for Al+. The Einstein coefficients for lumped levels
2 32Z 4 Ry2 1
σrad rec ðεÞ = πao pffiffiffi ð12Þ
3 3ð137Þ3 hνε n3

Table 1 where a0 is the Bohr radius, Z is the charge, Ry is the Rydberg constant,
Al energy levels grouped into lumped levels. Data have been taken from NIST database
h is the Planck constant, ν is the photon frequency, ε is the electron
[25]. Excitation energies correspond to the minimum level energy, which enters in the
group, while the statistical weight is the sum over the levels' statistical weights. energy and n is the principal quantum number of the level in which the
electron recombines. To apply Eq. (12) to Al and Al+ levels, for which the
Lumped Configurations Term Excitation Statistical weight
hydrogen-like approximation is not very suitable, especially for lower
level energy (eV) (2J+ 1)
levels, the dependence directly from the energy level En and not from
1 3s2(1S)3p 2
P° 0.0 6
the principal quantum number n has been introduced:
2 3s2(1S)4s 2
S 3.143 2
3 3s3p2 4
P 3.598 12
4 3s2(1S)3d 2
D 4.021 10 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 3s2(1S)4p 2
P° 4.085 6 Eion
n= ð13Þ
6 3s2(1S)5s 2
S 4.672 2 Eion −En
7 3s2(1S)4d y2D 4.826 10
8 3s2(1S)5p 2
P° 4.993 6
9 3s2(1S)4f 2
F° 5.123 14
where Eion represent the ionization energy of the species considered.
10 3s2(1S)6s 2
S 5.225 6
11 3s2(1S)5d 2
D 5.237 6
12 3s2(1S)6p 2
P° 5.373 6 2.4. Detailed balance principle
13 3s2(1S)5f, g 2
F°, 2G, 2D 5.434 32
14 3s2(1S)6d,f, g 2
D, 2F°, 2G 5.475 454
To calculate de-excitation cross sections from excitation ones, the
3s2(1S)7s, p, d 2
S, 2P°, 2D
3s2(1S)7f, g, h 2
F, 2G, 2H°
detailed balance principle (DBP) has been used:
3s2(1S)8s, f, d 2
S, 2F°, 2D
3s2(1S)9s, f, d 2
S, 2F°, 2D   g ε
deexc exc
3s2(1S)10s, f, d 2
S, 2F°, 2D σji ε−εij⁎ = i σ ðεÞ: ð14Þ
3s2(1S)11s, d 2
S, 2D gj ε−εij⁎ ij
3s2(1S)12s, d 2
S, 2D
3s2(1S)13s, d 2
S, 2D
3s2(1S)14s, d 2
S, 2D
3s2(1S)15s, d 2
S, 2D For ionization and three-body recombination, DBP has been
3s2(1S)16s, d 2
S, 2D applied introducing the equilibrium constant at the gas temperature.
3s2(1S)nd (from n = 17 2
D In this way, at equilibrium conditions, equilibrium molar fraction
to n = 35)
values are reached. By considering a generic A/A+ ionization balance,
L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626 619

Table 2 Table 2 (continued)


Al+ energy levels grouped into lumped levels. Data have been taken from NIST database Lumped Configurations Term Excitation energy Statistical weight
[25]. Excitation energies correspond to the minimum level energy, which enters in the level (eV) (2J + 1)
group, while the statistical weight is the sum over the levels' statistical weights.
1
3s17f F, 3F
Lumped Configurations Term Excitation energy Statistical weight 3s18f 1
F
level (eV) (2J + 1) 3s19f 1
F
1
1 3s2 1
S 0.0 1 3s20f F
3
2 3s3p 3
P° 4.636 9 3p3d D
1
3 3s3p P° 7.420 3
3p2 the three-body recombination cross section for ith level of atom A can
1
4 D 10.598 5
3
5 3s4s S, 1S 11.316 4 be calculated from the following:
6 3p2 3
P 11.665 9
3
7 3s3d D 11.846 15
A
3
P°, 1P° ε gi 1 Q þQ −
8 3s4p 13.071 12
σi
treebodyrec
ðε−Ii Þ = þ
  A e e−Ii = KTgas σiion ðεÞ ð15Þ
9 3s3d 1
D 13.649 5 ε−Ii g0A K T QA
eq gas
10 3p2 1
S 13.841 1
3
11 3s5s S, 1S 14.889 4
3 +
12 3s4d D, 1D 15.062 48 where Ii is the level ionization energy, giA the level degeneracy, g0A the
3
3s4f F°, 1F°
3
ion fundamental state degeneracy, Keq the equilibrium constant
13 3s5p P°, 1P° 15.585 12
14 3s6s 3
S, 1S 16.392 16 calculated from partition functions following the classical thermody-
3s6p 1
P, 3P namic theory [30], QA, QA+, Qe− the partition functions calculated from
3
15 3s5d D, 1D 16.468 84 the real levels included in the model.
3
3s5f F, 1F
3
3s5g G, 1G
3 3. CR model applied to LIP description
16 3s7s S, 1S 17.172 100
3
3s7p P, 1P
3s7d 3
D, 1D Although LIP is an inhomogeneous and transient plasma, in most
3
3s7g G, 1G measurements, system features, experimentally determined at each
3
3s7f F
3
temporal and spatial coordinate, such as temperature, ionization degree,
17 3s6f F, 1F 17.174 84
3s6d 3
D, 1D
spectral intensity etc., are actually the weighted average of the given
3s6g 3
G, 1G physical parameter in the entire spatial distribution. In this connection,
3
18 3p3d F 17.492 21 from a theoretical point of view, the whole spatial distribution can be
3
19 3s8p P, 1P 17.628 309 reduced to one point and typical experimental temporal profiles of
3
3s9s S, 1S
3 spatially integrated temperature and total density [31] can be used as
3s8d D, 1D
3s9s 3
S, 1S input data for the CR model described. By this approach, the effect of the
3s8f 1
F LIP expansion on the kinetic processes is directly obtained by empirical
1
3s9p P considerations, notably simplifying the calculation routines and avoid-
3
3s8g G, 1G
3
ing the use of a multi-dimensional fluid-dynamic description coupled
3s8h H, 1H
3s8f 3
F
with a punctual CR model. With this zero-dimensional CR model, the
3s9p 3
P experimental characteristic times of temperature and density changes
3
3p4s P can be directly compared with kinetic characteristic times, to perform
3
3s9d D, 1D an inspection from a microscopic point of view on the equilibrium
1
3s9f F, 3F
3 assumption generally adopted in LIP diagnostics.
3s9g G, 1G
3s9h 3
H, 1H The input data needed for this model are: gas temperature and total
20 3s10s, 3
S, 1S 18.115 525 density time evolution; initial composition; initial heavy particles and
1
3s10p, P, 3P electron temperatures.
3
3s10d, D, 1D
1
Unfortunately, as a consequence of LIP spectral peculiarities, such
3s10f, F, 3F
3s10g, 1
G, 3G
as the continuum radiation dominated spectrum and the absence of
3s10h 1
H, 3H molecular bands, no accurate experimental data are directly available
3s11s, 3
S, 1S for t b 200 ns after the laser pulse.
1
3s11p, P, 3P Consequently, some assumptions should be made about initial
3
3s11d, D, 1D
1 input parameters considering as t = 0 ns the end of the laser pulse.
3s11f, F, 3F
3s11g 3
G, 1G For gas temperature (Tgas), the initial value has been estimated
3s12p 1
P through the spectral resolved images of aluminium ablation reported
1
3s12s S in Ref. [31]. By those measurements the initial expansion rate has
1
3s12f F, 3F been calculated, as the ratio of the maximum spatial distribution
3
3s12g G, 1G
3s13p 1
P
displacement from the target surface and the detection time
3s13s 3
S, 1S immediately after the laser pulse, resulting in a reasonable velocity
3s13f 1
F, 3F of 8 · 105 cm s− 1. This velocity along the propagation axis can be
1
3s14p P, assumed equal to the total velocity of species, considering that the
1
3s14s S
1 radial component is negligible in the first 50 ns after the laser pulse
3s14f F, 3F
3s15p 1
P, and up to a 400 µm distance from the target surface, since the latter is
3s15s 1
S comparable to the laser spot dimension. Based on this approximation,
3
3s15f F, 1F the corresponding gas temperature has been estimated as 75,000 K by
1
3s16p P, the gas kinetic theory. To deduce gas temperature time evolution, we
1
3s16s S
3s16f 3
F, 1F
start from excitation temperatures obtained from measured emission
3s17p 1
P spectra using the Boltzmann plot technique (see Ref. [31]). The gas
temperature values have been obtained by scaling these measured
excitation temperatures by a factor such to reproduce experimental
620 L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626

Table 3
⁎ and total density Ntot
Gas temperature Tgas ⁎ time profiles used as data input in the
model.

Time (ns) ⁎ (K)


Tgas ⁎ (cm− 3)
Ntot

0 75,000 1.29 · 1020


200 33,728 1.29 · 1020
400 27,160 6.05 · 1019
600 20,368 1.67 · 1019
800 18,528 8.84 · 1018
1000 18,975 9.81 · 1018
1200 18,235 1.13 · 1019
1400 17,022 8.27 · 1018
1600 16,854 9.72 · 1018
1800 17,224 1.12 · 1019
2000 15,869 3.46 · 1018
2200 14,000 2.31 · 1018

ionization degree (see Sections 4.1 and 4.2). These gas temperatures
values together with total density values used in the model are
reported in Table 3. Both data have been interpolated by using a
logarithmic decreasing function (T = a * ln(t) + b; Ntot = c ⁎ ln(t) + d). Fig. 1. Time evolution of gas temperature (Tgas), electron (Te) and excitation heavy particle
Al+
Al and Al+ T0–1 (TAl
0–1 and T0–1) temperatures, in the first case study (Section 4.1).
Results have shown that the choice of the interpolation curve does not
influence the plasma behaviour for t N 200 ns, since experimental
points are close. Concerning the interval [0–200] ns, an adiabatic Electron temperature has been calculated from electron mean
expansion predicts a gas temperature and total density decrease energy using
characterized by a potential law according to blast-wave equation.
However, adiabatic conditions are not fulfilled since the plasma is in 2 εmean
Te = ; εmean = ∫εnðεÞdε ð16Þ
recombination condition with a strong energy loss due to radiation. For 3 k
this reason, gas temperature and total density are expected to decrease
faster than a potential decreasing function and a logarithmic decreasing where k is the Boltzmann constant.
Al Al+
interpolation curve have been used also in this time range. However, Excitation heavy particle temperatures (T0–1 and T0–1 ) have been
this interpolation curve does not pretend to provide a description of gas calculated from the ground state and first excited state level popu-
temperature and total density decreasing mechanism. lations by supposing a Boltzmann law
The time evolution of the particle number density (see Table 3) has
been assumed to coincide with experimental total density time evolution E1 −E0
T0−1 =   ð17Þ
data [31]. As initial value (t=0 ns), the first experimental data at g N
K ln 1 0
t = 200 ns has been used, corresponding to 1.29 · 1020 cm− 3. The g 0 N1
experimental data are given up to 2200 ns. For longer times, the values
of pressure and temperatures have been supposed constant, equal to the where E1, E0, g1, g0, N1, N0 represent, respectively, the level energies,
values at the last available experimental data time, since it is reasonable to statistical weights population of first excited and ground states. This
suppose that the plasma has reached a steady state condition and no more temperature represents essentially the behaviour of the lower energy
expansion occurs. However, there is still radiative energy loss mainly due part of the level distribution.
to spontaneous emission, which has small effect on gas temperature. In As it can be seen from Fig. 1, three different regions could be
this paper, reactive processes with the ambient gas [32,33] have been observed during time evolution. In the first region (t b 10− 10 s), T0–1 Al
,
Al+
neglected, even if their role can be important in long time range. T0–1, Te and Tgas are different. The very high initial ionization degree
Al+
Finally the LIP, in agreement with the general knowledge on metal causes a prevalence of recombination over ionization and TAl 0–1 and T0–1,
laser breakdown, has been supposed to be almost totally ionized become higher than Te, since recombination simultaneously overpop-
(α=0.997). Moreover, initial heavy particle level and electron dis- ulate Al and Al+ higher levels and under-populate Al+ and Al++ ground
tributions have been assumed, respectively, a Boltzmann and a Maxwell states. Equilibration time between the temperatures is comparable with
at the same temperature TAl =TAl+=Te =50,000b Tgas ⁎ (t=0), supposing ionization–recombination characteristic time, given by 1 / (KionNe). In
that most of initial energy is given to the translation degree of freedom. this temporal range (t b 10− 10 s), electron density is, approximately,
These input data appear to be the most reasonable to simulate the 1019 cm− 3, while for temperatures around 50,000 K, ionization rate is of
metal ablation plasma in air and will be used for a first case study (see the order of 10− 8–10− 9 cm3 s− 1, thus resulting in an equilibration time
Section 4.1). To extend the applicability of the CR model approach to of 10− 11–10− 10 s.
different experimental conditions, the effect of changing pressure On the other hand, in the second temporal range (10− 10 b tb 5 10− 6 s),
Al+
(second case study in Section 4.2) and gas temperature (third case equilibration between TAl 0–1, T0–1 and Te has already occurred. These
study in Section 4.2) has been also studied and reported in the following temperatures decrease faster than gas one due to radiative recombina-
discussion. tion. Just before t = 5 · 10− 7 s, the gas and heavy particle temperatures
reaches the equilibrium.
In the last temporal range (t N 5 · 10− 6 s), TAl Al+
0–1, T0–1 and Te reach a
4. Results and discussion steady value lower than Tgas. This behaviour is due to the interplay of
spontaneous emission and radiative recombination. In particular,
4.1. First case study spontaneous emission depletes high energy levels overpopulating the
Al+
ground state, resulting in the decreasing of TAl 0–1 and T0–1 temperatures.
Fig. 1 shows the corresponding time evolution of gas, electron and Moreover, radiative recombination (essentially due to Al+ recombina-
excitation heavy particle temperatures for aluminium plasma by tion), decreasing electron density, weakens the contribution of electron
including all the kinetic processes (Eqs. 1–4). collisions, resulting in the dominance of spontaneous emission.
L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626 621

To show in detail the time regions in which the different kinetic


Al
processes act, Fig. 2 shows the time evolution of T0–1 by adding one by
one the processes in the calculations:

a) only electron collisions


b) electron collisions and spontaneous emission
c) electron collisions, spontaneous emission and Al+ radiative
recombination
d) electron collisions, spontaneous emission and Al+ and Al++
radiative recombination.

Firstly, it can be observed that adding Al+ (case c) and Al++ (case d)
radiative recombinations, the temperature TAl 0–1 decreases in two
different time scales. Al++ radiative recombination is more effective
than that of Al+ at short times (b5 · 10− 6 s). Once Al++ is totally
recombined, Al+ radiative recombination becomes more relevant,
determining the behaviours for longer times (t N 5 · 10− 6 s). This effect
on TAl
0–1 is indirect, because radiative recombination affects the electron
density (see Fig. 3), reducing the contribution of the electron collisions.
Al+
This result is confirmed by the equality of TAl 0–1 and T0–1. It must be also Fig. 3. Electron density time evolution in the first case study (Section 4.1) by including
noted (see Figs. 2 and 3) that the small changes in the electron density in the model: a) only electron collisions, b) electron collisions and spontaneous
emission, d) electron collisions, spontaneous emission and Al+ and Al++ radiative
due to Al++ recombination (about a factor 2) has drastic effects on TAl 0–1.
−6
recombination.
Also spontaneous emission has some effects on TAl 0–1 for t N 5 · 10 s.
Characteristic Einstein coefficients values go from 108–105 s− 1,
but before t = 10− 6 s, electron temperature and density are high The theoretical ionization degree has been calculated through the
enough such that electron collisions overcome spontaneous emission, following formula
and all the temperatures are equal, except when A++ radiative
recombination is included, as described above.
R NAlþ
For tN 5 ·10− 6 s, considering only electron collisions (case a), TAl0–1 = α= with R= ð18Þ
1+R NAl
Tgas, but including spontaneous emission, first and radiative recombina-
tion after, TAl
0–1 settles down to lower values.
Fig. 4 shows the corresponding Al, Al+, Al++, e− molar fraction considering a three-component plasma (Al, Al+ and e−), being Al++
time evolution. Recombination prevails over all the time evolution negligible in the range of interest, where NAl, NAl+ are Al and Al+ total
making Al++ to disappear very rapidly, around t = 5 · 10− 6 s. densities.
The asymptotic molar fractions of ions are smaller than equilib- By including only electron collisions and spontaneous emission
rium values at the given gas temperature due to radiative processes. (cases a and b), the predicted ionization degree is higher than
The corresponding Al and Al+ level distributions and electron experimental values, but with radiative recombination, ionization
energy distribution function (eedf), in this first case, show no degree is lowered and resembles qualitatively well the experimental
deviations from the Boltzmann and Maxwell distributions. Pressure one. Within the experimental errors, the agreement is relatively good
is so high (P N 3 atm) that collisional processes prevail over radiative even if theoretical ionization degree with the inclusion of all the
ones in shaping the distributions. kinetic processes is a little bit lower than experimentally one.
Fig. 5 shows the comparison between experimental and calculated Probably, radiative recombination cross sections in the Kramer
ionization degree in cases a, b, and d. approximation overestimate plasma recombination or gas tempera-
ture values have been underestimated. This aspect will be discussed in
the next section.

Fig. 2. Time evolution of TAl0–1, in the first case study (Section 4.1), by including in the
model: a) only electron collisions, b) electron collisions and spontaneous emission,
c) electron collisions, spontaneous emission and Al+ radiative recombination, d) electron
collisions, spontaneous emission and Al+ and Al++ radiative recombination. Fig. 4. Molar fraction time evolution in the first case study (Section 4.1).
622 L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626

Fig. 5. Experimental (squares) and calculated ionization degree in the first case study Fig. 7. Time evolution of Al level distribution, Ni / gi, where Ni and gi represent the ith
(Section 4.1), by including in the model: a) only electron collisions, b) electron level population density and statistical weight, in the second case study (Section 4.2).
collisions and spontaneous emission, d) electron collisions, spontaneous emission and
Al+ and Al++ radiative recombination.
same over all the temporal evolution with the exception of the last
time range (t N 10− 6 s).
4.2. Second and third case studies: pressure and gas temperature effects For the lower initial pressure, the initial composition is less ionized
than the equilibrium values at the gas temperature, resulting in
In this section, the effect of decreasing total density (and pressure) ionizing behaviour of Al first excited state, with the consequence that
Al Al+
(second case study) and gas temperature (third case study) on LIP T0–1 rapidly decreases as compared to T0–1 and Te.
behaviour has been investigated. The second case study differs from This initial ionization prevalence is confirmed by looking to Al level
the first one in total density time profile, decreased by two orders of distributions (Fig. 7) and to the molar fractions (Fig. 10). In particular,
magnitude. Since the values of the gas temperature used are the same in Fig. 7, a different behaviour between the lower and the upper part
as those in Table 3, the pressure values are reduced as well by two of the distribution is shown: while ionization prevails for ground and
orders of magnitude and range from 3.96 atm at t = 0 ns to 0.03 atm at first excited states, recombination overpopulates the upper levels.
Al Al+
t = 2200 ns. Fig. 6 shows also that T0–1 , T0–1 and Te remain essentially lower
Initial electron density decreases proportionally, while the than Tgas, since electron-heavy particle elastic collisions are not
characteristic time of electron collision increases almost of the same sufficient to equilibrate the two temperatures. By the way, in the
factor. The non-equilibrium, due to radiative processes, is more interval 10− 8–10− 6 s, T0–1Al Al+
, T0–1 and Te are equal.
−6 Al
pronounced, even if it appears at longer time. Finally, for t N 10 s, Te increases slightly with respect to T0–1 and
Al +
Figs. 6–10 show the time evolution of temperature, Al and Al+ T0–1 temperatures (see Fig. 6) due to the contribution of elastic
level distributions, eedf and molar fractions. By looking at the collisions with heavy particles. The non-equilibrium of temperatures
temperatures behaviour (see Fig. 6), several differences respect to has more relevant effects on the distributions (Figs. 7–9), showing
the first case study (Fig. 1) can be observed. The initial temperature non-Boltzmann and non-Maxwell behaviours for t N 10− 6 s. The non-
Al Al+
disequilibrium involves only T0–1 , while T0–1 and Te are essentially the equilibrium is due to the non-linear interplay of radiative recombi-
nation, spontaneous emission and inelastic and superelastic electron

Fig. 6. Time evolution of gas temperature (Tgas), electron (Te) and excitation heavy
Al+
particle Al and Al+ T0–1 temperatures (TAl0–1 and T0–1), in the second case study Fig. 8. Time evolution of Al+ distribution, Ni / gi, where Ni and gi represent the ith level
(Section 4.2). population density and statistical weight, in the second case study (Section 4.2).
L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626 623

Fig. 11. Electron density time evolution for different values of gas temperature
Tgas = βTgas
* and with total density time profile given by Ntot
* (third case study, see
Section 4.2). Tgas
* and Ntot
* are shown in Table 3.
Fig. 9. Electron energy distribution function (eedf) time evolution in the second case
study (Section 4.2).

collisions. For long time, the heavy particle internal distributions are
determined mainly by radiation losses, therefore the presence of
metastable levels produces some spikes.
To investigate the sensitivity of the model to gas temperature
variations (third case study), the gas temperature profile has been
reduced by a constant factor β as Tgas = βTgas
* , with total density time
profile given by N*tot (Table 3). The β values ranges between 0.3 and
1.3.
The main effect of decreasing the gas temperature is relevant
variations on electron density (see Fig. 11) in the time interval 1 ns–
100 μs. The diminishing of electron density causes radiative losses to
be more effective for Al (see Fig. 12) and Al+ level distributions, which
reflect on the eedf (see Fig. 13).
Finally, the effect of gas temperature and pressure change over
ionization degree values has been analysed. Fig. 14 show calculated
ionization degree as a function of time for different gas temperature
(a) and pressure (b) values compared with experimental one. Three
pressure time profiles P = δP* have been shown, with δ equal to 1, 0.1 Fig. 12. Al level distribution at t = 5 10− 5 s for different values of gas temperature
and 0.01 and P* the pressure corresponding to Tgas * and Ntot* . In the Tgas = βTgas
* and with total density time profile given by Ntot * (third case study, see
three different pressure data cases, the gas temperature profile is Section 4.2). Tgas
* and Ntot* are shown in Table 3.
given by Tgas
* (Table 3).

Fig. 13. Electron energy distribution function (eedf) at t = 5 10− 5 s for different values
of gas temperature, Tgas = βTgas
* and with total density time profile given by Ntot * (third
Fig. 10. Molar fraction time evolution in the second case study (Section 4.2). case study, see Section 4.2). Tgas
* and Ntot
* are shown in Table 3.
624 L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626

Fig. 14. Comparison between experimental (squares) and calculated ionization degree
for different values of: (a) gas temperature, Tgas = βTgas
* , with β equal to 0.3, 0.5, 0.7, 1, Fig. 16. Comparison between T0–1 and the Boltzmann fit (TB) temperatures for Al (a)
* ; (b) pressure, P = δP*, with δ equal to 1, 0.1, and
and 1.3 and total density profile Ntot and Al+ (b) in the second case study (Section 4.2).
0.01 and gas temperature profile Tgas* . Tgas
* and Ntot* are shown in Table 3, while P* is the
corresponding pressure profile.
input data (gas temperature and total density), which does not vary
smoothly with time (see Table 3). By decreasing gas temperature and
Ionization degree curve is shifted toward higher values both by pressure (see Fig. 14a and b), ionization and recombination processes
increasing gas temperature and by decreasing pressure, as can be are slowed, reducing the effect of gas temperature and total density
expected also by thermodynamic analysis. Moreover, it can be noted changes in the ionization degree.
that the ionization degree dependence on gas temperature is stronger
than on gas pressure. 4.3. Boltzmann plot
As can be seen for Fig. 5, the computed ionization degree (and the
experimental values) does not vary smoothly with time reflecting the Experimental excitation temperatures are obtained by using the
Boltzmann plot technique, which assumes Boltzmann distributions.

+
Fig. 15. Comparison between calculated Al and Al+ T0–1 (TAl Al
0–1 and T0–1), Al and Al
+

Boltzmann fit (TAl
B and T Al+
B ) and electron temperature (Te ) with experimental excitation Fig. 17. Al level distribution at t = 5 10− 5 s and extrapolation procedure on the
temperature (Texp
exc ) in the first case study (Section 4.1) in the experimental time range. Boltzmann fit to calculate atom density in the second case study (Section 4.2).
L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626 625

been applied to the description of Al metallic-laser induced plasma.


Kinetic processes such as excitation–de-excitation, ionization–recom-
bination by electron impact, radiative recombination and spontaneous
emission have been included in the model. To account for expansion,
time dependent gas temperature and total density have been used as
input data of the model. They have been assumed to follow
characteristic experimental temporal profiles of spatially integrated
excitation temperature and density. A reasonable initial gas tempera-
ture value has been deducted from experimental expansion velocity
data, helping in the description of the preliminary phase of LIP
expansion. The model gives the possibility to analyse the simultaneous
time evolution of heavy particle level distributions, eedf and LIP
composition and to compare characteristic kinetic times with expansion
ones. In this way, LTE departure conditions can be predicted and
analysed giving a valid support to experimental methodologies.
A confirmation of the model validity has been obtained by the
comparison of calculated ionization degree and excitation heavy
particle temperatures with experimental data.
Moreover, the effect of gas temperature and total density change
Fig. 18. Relative difference between exact ionization degree and ionization degree over LIP behaviour has been analysis. The results have shown that, by
calculated through the extrapolation procedure in the second case study (Section 4.2).
decreasing gas temperature and pressure, LTE deviations can occur
also in the ns–ms time range, which is the measurements time range,
However not all the levels are considered, but only radiative ones with showing that, in these conditions, care should be taken in the
emission in the UV–VIS range. Supposing LTE, from experimental experimental data analysed.
results, by extrapolating the Boltzmann plot, the population of the An example of possible misunderstanding in LIBS measurement
ground state is deduced and, as a consequence, the particle density of can be obtained in the experimental determinations of the ionization
a given species. degree from the extrapolation procedure in the Boltzmann plot
In this section the Boltzmann plot technique has been applied to technique: overpopulated level distribution tails, due to radiative
calculated distributions to derive the corresponding temperature TB. recombination, can lead to non-negligible errors in ionization degree
The fitting with the Boltzmann function has been extended to levels values determination.
with index equal or above 3 (see Tables 1 and 2).
In the first case study (Section 4.1), with the only exception of the
Acknowledgements
first time interval (t b 10− 10 s), T0–1 Al Al+
, T0–1 , TAl Al+
B , TB and Te coincide,
showing also good agreement with the available excitation experi-
This research is partially supported by MIUR under project MIUR
mental temperature (Texp exc ) in the range 200–2200 ns (see Fig. 15).
(PRIN 07, 2007K9YPL8_004).
In the second case study (reduction of pressure, see Section 4.2),
comparison between T0–1 and the TB has been reported in Fig. 16 for Al
(a) and Al+ (b). TAl
+
0–1 and TB
Al+
are equal except for t b 100 ns, corre- References
sponding to the time that the Al++ radiative recombination becomes [1] R.W.P. McWhirter, in: R.H. Huddlestone, S.L. Leonard (Eds.), Plasma Diagnostic
negligible. For Al, the differences are appreciable for t b 3 · 10− 8 s and for Technique, Academic, New York, 1965, p. 201.
t N 10− 6 s, being, in the first, more relevant. Looking at Fig. 7, it is evident [2] A. Casavola, G. Colonna, A. De Giacomo, M. Capitelli, Laser ablation of titanium
metallic targets: Comparison between theory and experiment, J. Thermophys.
that Al low energy distribution is determined by electron impact
Heat Transfer 17 (2) (2003) 225–231.
ionization while the distribution tails by recombination. For longer [3] A. Casavola, G. Colonna, A. De Giacomo, O. De Pascale, M. Capitelli, Experimental
times, Al distribution is governed mainly by electron induced transitions and theoretical investigation of laser-induced plasma of titanium target, Appl.
Optics 42/30 (2003) 5963–5969.
between excited states, equilibrating the distribution tails with
[4] G. Cristoforetti, A. De Giacomo, M. Dell’Aglio, S. Legnaioli, E. Tognoni, V. Palleschi,
electrons. Fig. 17 shows Al calculated distribution and the corresponding N. Omenetto, Local thermodynamic equilibrium in laser-induced breakdown
Boltzmann fitting line at t = 5 · 10− 5 s. In the extrapolation procedure, spectroscopy: beyond the McWhirter criterion, Spectrochim. Acta B 65 (2010)
overpopulated Al distribution tail leads to underestimation of ground 86–95.
[5] M.C. Quintero, A. Rodero, M.C. Garcia, A. Sola, Determination of excitation
state and thus total density value and this, in turn, can influence the temperature in a nonthermodynamic-equilibrium high pressure helium micro-
ionization degree determination. wave plasma torch, Appl. Spectrosc. 51/6 (1997) 778–784.
To evaluate the error made by the extrapolation procedure, it is [6] M. Capitelli, F. Capitelli, A. Eletskii, Non-equilibrium and equilibrium problems in
laser-induced plasmas, Spectrochim. Acta Part B 55 (2000) 559–574.
possible to estimate the relative difference between the ionization [7] H.R. Griem, Validity of local thermal equilibrium in plasma spectroscopy, Phys.
degree calculated with actual values and the extrapolated ones (see Rev. 131 (1963) 1170–1176.
Fig. 18) as a function of time: [8] A. De Giacomo, Experimental characterization of metallic titanium-laser induced
plasma by time and space resolved optical emission spectroscopy, Spectrochim.
Acta Part B 58 (2003) 71–83.
jα−αextr j [9] L.D. Pietanza, G. Colonna, M. Capitelli, Coupled solution of a time-dependent
err = ð19Þ
α collisional-radiative model and Boltzmann equation for atomic hydrogen
plasmas: possible implications with LIBS plasmas, Spectrochim. Acta Part B 56
(2001) 587–598.
At low time, due to the high ionization degrees, the errors are [10] D.R. Bates, A.E. Kingston, R.W.P. McWhirter, Recombination between electrons
pretty small, but the error increases for longer times reaching 100% for and atomic ions. I. Optically thin plasmas, Proc. R. Soc. Lond. A 267 (1962)
t = 10− 4 s. 685–689.
[11] H.W. Drawin, Validity conditions for local thermodynamic equilibrium, Z. Phys.
228 (1969) 99–119.
5. Conclusions [12] M. Cacciatore, M. Capitelli, Population densities and ionization coefficient of fast
transient hydrogen plasmas, Z. Naturforsch. 30 (1975) 48–54.
[13] K. Sawada, T. Fujimoto, Temporal relaxation of excited-level populations of atoms
A zero-dimensional collisional–radiative model, coupled self- and ions in a plasma: validity range of the quasi-steady-state solution of coupled
consistently with the Boltzmann equations for free electrons, has rate equation, Phys. Rev. E 49 (1994) 5565–5573.
626 L.D. Pietanza et al. / Spectrochimica Acta Part B 65 (2010) 616–626

[14] A. Bogaerts, R. Gijbels, J. Vlcek, Collisional–radiative model for an argon glow [24] G. Colonna, Step adaptive method for vibrational kinetic and other initial value
discharge, J. Appl. Phys. 84 (1998) 121–136. problems, in: M. Maiellaro, S. Rionero (Eds.), Supplemento ai rendiconti del
[15] A. Bogaerts, R. Gijbels, R.J. Carman, Spectrochim. Acta Part B 53 (1998) Circolo Matematico di Palermo, Klumer Academic Publisher, 1998, pp. 159–163.
1679–1703. [25] National Insitute of Standard and Technology (NIST) at http://physics.nist.gov/
[16] B. van der Sijde, J. van der Mullen, D.C. Schram, Collisional–radiative model in PhysRefData/ASD/.
plasmas, Beitr. Plasmaphys. 24 (1984) 447. [26] G. Colonna, M. Capitelli, A few level approach for the electronic partition function
[17] J. van der Mullen, J. Jonkers, Fundamental comparison between non-equilibrium of atomic system, Spectrochim. Acta Part B 64 (2009) 863–873.
aspects of icp and mip discharges, Spectrochim. Acta Part B 54 (1999) 1017–1044. [27] at http://www-amdis.iaea.org/ALADDIN/collision.html.
[18] S.D. Rockwood, Elastic and inelastic cross sections for electron-Hg scattering from [28] E. Bauer, C. Decker Bartky, Calculation of inelastic electron–molecule collision
Hg transport data, Phys. Rev. Lett. 8 (1973) 2348–2358. cross sections by classical methods, J. Chem. Phys. 43 (1965) 2466–2476.
[19] C.J. Elliott, A.E. Greene, Electron energy distributions in e-beam generated Xe and [29] Zel'dovich, Raizer, Physics of Shock Wave and High-Temperature Hydrodynamic
Ar plasmas, J. Appl. Phys. 47 (1976) 2946–2953. Phenomena, 1966.
[20] G. Colonna, M. Capitelli, Electron and vibrational kinetics in the boundary layer of [30] L.D. Landau, Statistical Physics, Elsevier, Amsterdam, 1979.
hypersonic flow, J. Thermophys. Heat Transfer 10 (1996) 406–412. [31] A. De Giacomo, M. Dell'Aglio, D. Bruno, R. Gaudiuso, O. De Pascale, Experimental
[21] M. Capitelli, G. Colonna, A. Gicquel, C. Gorse, K. Hassouni, S. Longo, Maxwell and and theoretical comparison of single-pulse- and double-pulse-LIBS on metallic
non-Maxwell behavior of electron energy distribution function under expanding samples, Spectrochim. Acta Part B 63 (2008) 805–816.
plasma jet conditions: the role of electron–electron, electron–ion, and super- [32] A.R. Casavola, G. Colonna, M. Capitelli, Kinetic model of titanium laser induced
elastic electronic collisions under stationary and time-dependent conditions, plasma expansion in nitrogen environment, Plasma Source Sci. Technol. 18 (2009)
Phys. Rev. E 54 (1996) 1843–1849. 025–027.
[22] G. Colonna, M. Capitelli, The influence of atomic and molecular metastable states [33] A.R. Casavola, G. Colonna, A. Cristofolini, C.A. Borghi, M. Capitelli, Modeling laser-
in high-enthalpy nozzle expansion nitrogen flows, J. Phys. D Appl. Phys. J. 34 induced plasma expansion under equilibrium conditions, J. Thermophys. Heat
(2001) 1812–1818. Transfer 22 (2008) 407–413.
[23] G. Colonna, M. Capitelli, Boltzmann and master equations for MHD in weakly
ionized gases, J. Thermophys. Heat Transfer 22 (2008) 414–423.

You might also like