Space- and time-resolved dynamics of plasmas generated by laser

double pulses interacting with metallic samples{

Reinhard Noll, Ralph Sattmann, Volker Sturm and Stefan Winkelmann

Fraunhofer-Institut für Lasertechnik (ILT), Steinbachstrasse 15, 52074 Aachen,

Germany. E-mail: [email protected]; Fax: 49 241 8906 121; Tel: 49 241 8906 0

Received 8th December 2003, Accepted 11th March 2004

First published as an Advance Article on the web 19th March 2004

Significant improvements to the analytical performance of laser-induced breakdown spectrometry (LIBS) were
achieved by the use of laser double pulses to ablate and excite the sample material to be analysed. To clarify
the underlying physical phenomena the dynamics of the laser-induced plasma in air is studied using a high-
speed electro-optic camera to observe the spatial and temporal development of the plasma geometry. A Mach–
Zehnder interferometer was set-up to detect the spatio-temporal changes of the refractive index of the plasma.
The velocities of the luminous plasma front and shock waves after the impact of the first laser pulse are
consistent with the prediction of Sedov’s model for spherical expansion. The dynamics caused by the second
laser pulse of the double pulse differ significantly, indicating a modified state of the local atmosphere in the
interaction region. Electron densities determined interferometrically amount to about 3 6 1018 cm23 at the
center of the hemispheric plasma geometry at a delay time of 2 ms. Whereas the electron density for single or
double pulses is approximately the same, the volume of the plasma is greater by more than a factor of 3 after
the interaction with the double pulse of the same total energy.

1. Introduction directed orthogonally to the sample surface.12 For orthogonal

re-heating the sequence of the pulses is exchanged, i.e., the laser
The dominant excitation source for laser-induced breakdown pulse irradiates the sample at first and then the re-heating pulse
spectrometry (LIBS) is single Q-switched pulses generated by propagating parallel to the sample surface is applied.13 In the
pulsed solid state lasers.1–5 These pulses have, typically, a width dual pulse crossed beam approach the first laser pulse is split into
(FWHM) in the range of 5–20 ns. The repetition rate amounts two beams, irradiating the sample under an angle of incidence of
to 2–50 Hz for flashlamp-pumped lasers. Double pulses and 45u; the second pulse follows in the same directions.14
multiple-pulses were studied to improve the analytical per- Investigations of collinear double pulses6,15 and triple pulses8
formance of LIBS, for instance for the determination of traces generated within a single flashlamp discharge of a Nd:YAG
in metallic matrices or to enable special applications such as, laser for plasma excitation showed a significant improvement
for example, the analysis of samples under water,6–10 or to of the detection limits. With this approach detection limits for
increase the ablation rate in laser material processing.11 The the elements C, P, S, Mn, Ni and Cr in low-alloy solid steel
term ‘‘double pulse’’ is defined here as follows: two laser pulses samples of less than 10 mg g21 were achieved for the first time
each having a width in the range of about 5–30 ns, which are using LIBS. In a recent paper a LOD of 5 mg g21 has been
separated in time by several 100 ns up to several microseconds. reported for the direct analysis of trace elements in liquid steel
This interpulse separation is denoted by Dt in the following. using a collinear triple-pulse configuration.16
Several methods are available to generate these double pulses: Collinear double and multiple pulses lead to a significant
(a) a single pulsed laser with a modified electronic control of the increase of the ablated mass.6 Comparing the ablated mass of a
Q-switch to generate two pulses within a single flashlamp steel target using double pulses with 2 6 40 mJ and Dt ~ 6 ms
discharge, (b) superposition of the beams of two pulsed lasers. with a single pulse of 1 6 80 mJ yields an increase of about a
For case (a) only one laser is necessary and the two pulses factor of 2.5. Whereas for single pulses the ablated mass
are emitted collinearly. However, the interpulse separation is strongly saturates with increasing pulse energy, no such
limited to about 0.5 ms v Dt v 180 ms. Case (b) requires two
lasers and and a precise adjustment of the spatial superposition
of the two pulses. This approach offers a high flexibility in
terms of the wavelengths of the pulses, the pulse widths and the
sequence of pulses. Fig. 1 illustrates schematically the various
configurations studied so far for LIBS with double pulses. The
arrows depict the direction of propagation of the laser pulses,
the numbers show the temporal sequence. In the collinear case
both pulses have the same axis of propagation and are directed
orthogonally to the sample surface.6–9 Various pulse combina-
tions in terms of different wavelengths and different pulse
widths have been studied.10,11 In the case of orthogonal pre-
DOI: 10.1039/b315718k

ablation a first laser pulse is irradiated parallel to the sample

surface, generating a laser-induced plasma above the sample
surface in the ambient atmosphere, and the second pulse is

{ Presented at the Second Euro-Mediterranean Symposium on Fig. 1 Double pulse configurations studied for LIBS. The arrows
Laser Induced Breakdown Spectroscopy, Hersonissos, Crete, Greece, depict the laser pulses and their direction of propagation and the
September 30th–October 3rd, 2003. numbers 1, 2 their temporal sequence.

This journal is ß The Royal Society of Chemistry 2004 J. Anal. At. Spectrom., 2004, 19, 419–428 419
 saturation behaviour is observed for double and multiple
pulses up to burst energies (denoting the sum of pulse energies
within a double or multiple pulse) of 320 mJ. Electron
temperatures in the laser-induced plasma determined spectro-
scopically are higher by 1000 K to 1500 K using double pulses
with respect to those measured in the plasmas generated by
single pulses of the same energy (i.e., the energy of the single
pulse is equal to the total energy of the double pulse). Hence the
increased line emission observed using double pulse excitation
is linked with the greater mass ablated and the higher
temperatures achieved in the laser-induced plasma.6
For collinear double pulses studied for the analysis of solid
metallic samples under water an increase of the ablated mass of
aluminium samples by a factor of 3.8 was reported.7 The Fig. 2 Experimental set-up for high-speed photography of the
observed increase of line emissions is attributed to the luminous laser-induced plasma. M, Mirror; GT, Glan–Thompson
formation of a bubble in the water layer above the sample polarizer; CL, volume absorbing disc calorimeter; L, lens; P, laser-
surface generated by the first pulse into which the emitting induced plasma; S, sample; OA, optical axis of streak camera; LM, long
atoms can expand. Owing to the absence of atomic emissions distance microscope; SC, streak camera.
from the plasma produced by the single pulse it was not
possible to determine the temperature and electron density for interaction with single and double Q-switched Nd:YAG laser
underwater excitation. The authors conclude that in this case pulses is investigated. The pulse width of a single pulse amounts
the temperature is less than 3000 K. For double pulse excitation to y15 ns, whereas each pulse of a double pulse has a width of
an electron temperature of 8900 K was determined.7 y25 ns. In the case of double pulses the energy ratio adjusted is
Studies of collinear double pulses interacting with aluminium 1 : 1. The interpulse separation of the double pulses amounts to
samples in air showed higher line intensities lasting longer than Dt ~ 6 ms. The pulses are focused by a planoconvex lens of
corresponding emissions generated by single laser pulses.9 The 200 mm focal length on pure iron targets (Fe concentration w
time evolution of the line emission follows an exponential 99.99%). The beam waist of the laser focus lies inside the
decay, where the decay time constant for the Al II line sample, 5 mm beneath its surface (D s ~ 5 mm). The diameter
624.34 nm and double pulse excitation is more than twice that of the laser beam on the surface of the sample is 280 mm. The
determined for single pulse excitation. Looking at the ionic pulse energies are adjusted by use of a Glan–Thompson
nitrogen emission N II at 500.1 nm revealed greater line polarizer and measured with a volume absorbing disc calori-
intensities in the case of single pulse excitation than for double meter (Scientech 38-0101). The optical axis of the streak
pulses, indicating that the pre-plasma formation induced by the camera is oriented perpendicularly to the propagation direction
first pulse reduces the air present nearby the surface and hence of the laser beam irradiating the sample. The laser-induced
the concentration of nitrogen available for emission.9 plasma is imaged by a long distance microscope on the streak
In the orthogonal pre-ablation configuration an enhance- camera (Hadland, Model Imacon 790) with an S 20 photo-
ment of Fe lines of about a factor 30 was observed comparing cathode having the dimensions 9 6 8 mm2. The distance
pre-ablation double pulses with single pulses.12 A temperature between the laser-induced plasma and the front end of the
increase by up to 5000 K was observed for pre-ablation double microscope is about 160 mm. The streak camera can be
pulses. The signal enhancement is correlated with an increased operated in framing mode and in streak mode. In streak mode
crater volume, i.e., the ablated mass is greater for pre-ablation the slit of the streak camera is oriented collinearly to the
double pulses than for single pulses.12 propagation axis of the irradiated laser beam. The magnifica-
From a practical point of view, the collinear configuration is tion object-to-image was determined using a line graticule
the most simple approach for double pulses, offering an easy placed at the position of the laser propagation axis just above
access to the sample to be analysed. Hence, the investigations the sample and observed with the streak camera in the focus
presented in this paper are focused on the collinear double mode. The experiments are carried out under normal atmos-
pulse configuration. The aim of the investigations is to clarify pheric conditions.
the underlying physical processes of the observed improve- The experimental set-up used for the interferometric investiga-
ments achieved with collinear double pulses, studying the tions is shown schematically in Fig. 3. The components to
space- and time-resolved dynamics of the plasmas generated generate the laser-induced plasma are identical to Fig. 2, but
by laser double pulses interacting with metallic samples. The the sample is now located in a gas tight measuring chamber
experimental methodology is based on high-speed framing allowing experiments with air pressures of p0 ~ 100 mbar and
and streak images to observe the spatial and temporal develop- 1013 mbar. The Mach–Zehnder interferometer is oriented in
ment of the plasma using the optical emission of the plasma. such a way that one arm passes through the interaction volume
Furthermore the spatio-temporal changes of the refractive above the sample surface perpendicular to the irradiated
index of the plasma are investigated using a Mach–Zehnder Nd:YAG laser beam. The radiation source for the interfero-
interferometer, where the shifts of the interference fringes are metry is a cw argon ion laser (Coherent, Model Innova 100–10)
detected with a high-speed electro-optic camera. operated at an emission wavelength of 514.5 nm. The inter-
ference fringes are observed with the streak camera. To avoid
damage to the streak camera due to continuous illumination by
2. Experiment the argon laser beam, a mechanical shutter and an acousto-
optic modulator (AOM) are used to extract pulses of y10 ms
The experimental set-up for the time-resolved photography of duration of the argon laser beam synchronized to the triggering
the luminous laser-induced plasma is shown in Fig. 2. A of the Nd:YAG laser. The first order diffracted beam of the
Nd:YAG laser (Lumonics, Model HY 1200, modified) with AOM is used for the interferometer, whereas the zeroth and
multi-pulse option, operated at 10 Hz and 1064 nm emission higher orders are blocked by an aperture (see A in Fig. 3). The
wavelength, is used to generate the plasmas on the surface of telescope (L2 and L3 in Fig. 3) serves to adjust the diameter of
a plane metallic sample in ambient air. Details of the laser the illuminated area at the location of the plasma. To cover
are described in ref. 8. For the experiments presented in this the total plasma geometry during the expansion phase, an
paper, the plasma dynamics resulting from the laser-sample illumination diameter of 20 mm was choosen. The second

420 J. Anal. At. Spectrom., 2004, 19, 419–428

Fig. 3 Experimental set-up for the interferometric measurement of the laser-induced plasma. SH, Shutter; AOM, acousto-optic modulator; A,
aperture; BS1 to BS4, beam splitters; MC, measuring chamber; EF, edge filter; IF, interference filter. For further abbreviations see Fig. 1.

telescope (L4 and L5), and the internal optic of the streak synchronized using a personal computer as master trigger and
camera, form an image of the plasma on the photocathode. The three digital delay generators. Before the start of each mea-
beam deviation of the argon laser beam due to refractive index surement a minimum of 50 flashlamp discharges of the laser
gradients in the interaction region of the Nd:YAG laser beam are activated to stabilize the pulse energy to a standard
may cause uncertainties in the Abel inversion of the inter- deviation of 2%.
ferograms: however, accurate results are achieved by proper
imaging of the plasma plane on the photocathode of the streak
camera.17 The interference filter has a maximum transmission 3. Results and discussion
at the wavelength of the argon ion laser and a spectral width of
3.1. Framing and streak photography
10 nm (FWHM). The edge filter (Schott, type OG 515)
attenuates additionally the intense emission of the laser- The framing and streak images shown in this section are gained
induced plasma. with the set-up shown in Fig. 2. Figs. 4 and 5 show framing
The Nd:YAG laser, shutter, AOM and streak camera are photographs of the evolving plasma induced by a 2 6 40 mJ

Fig. 4 Framing images of the laser-induced plasma after the first pulse of a double pulse. Exposure time 10 ns per frame, time between subsequent
frames 50 ns. The time sequence of the frames is shown on the top left with the numbers 1–10. The view detected by the frames is illustrated
schematically on the top right. The white horizontal line in each frame indicates approximately the position of the sample surface. Laser parameters:
2 6 40 mJ, interpulse separation 6 ms.

J. Anal. At. Spectrom., 2004, 19, 419–428 421

 Fig. 5 Framing images of the laser-induced plasma after the second pulse of a double pulse. Experimental settings and arrangement of frames are
as for Fig. 4.

laser double pulse in air at 1 bar. The streak camera is operated

with a framing rate of 2 6 107 frames s21, i.e., the temporal
separation between subsequent frames amounts to 50 ns. The
exposure time per frame is 10 ns. The magnification and the
setting of the aperture of the microscope are kept constant to
assure comparability of the recordings. Fig. 4 shows the frames
after the first pulse of the double pulse. The first pulse reaches
the sample between frames 1 and 2. (Numbers refer to the
sequence of frames as indicated on the top left of Fig. 4. The
arrangement of frames shown in Fig. 4 was choosen according
to the usual orientation of the time axis from left to right. The
original arrangement of the frames yielded by the streak
camera is different.) Frames 2–4 are overexposed due to the
high luminosity of the plasma at this phase. Fig. 5 shows the
frames after the second pulse of the double pulse. The second
pulse of the double pulse reaches the sample at the exposure of
frame 1 (frame numbering as for Fig. 4).
The plasma geometry and dynamics observed in Figs. 4 and
5 differ significantly from each other. After the first pulse a Fig. 6 Streak photographs after first and second laser pulse of a
nearly hemispherical plasma geometry appears staying in double pulse. Experimental parameters are the same as for Fig. 4.
contact with the sample surface for times up to 400 ns after
arrival of the first laser pulse. The extension of the plasma in
the direction of the laser beam axis amounts to y1 mm (frame upper image shows the laser-induced plasma after the first
5 corresponding to an elapsed time of 150 ns after the first pulse, the lower image shows the plasma after the second pulse.
visible plasma emission in frame 2). After the second pulse the After the first pulse, the plasma stays in contact with the sample
plasma emission starts at the sample surface (see frame 1 in surface, while the boundary of the luminous plasma expands to
Fig. 5). This is a clear indication that the second laser pulse about 1 mm above the sample, as observed in the framing
primarily interacts with the sample surface and not with a images of Fig. 4. The luminous layer evolving after the second
residual plasma originating from the first pulse. The plasma pulse starts at the sample surface and expands up to a distance
then splits into two regions, where one region lasts at the of about 2 mm from the sample surface. After 300 ns the front
sample surface and dissappears after about 100 ns. The other of the luminous layer changes direction and propagates back
region attains a layer-like shape, propagating towards the towards the sample. Approximately 600 ns after the irradiation
incoming laser beam. At y150 ns after the irradiation of the of the second laser pulse, the luminous front again reaches the
second pulse of the double pulse an inhomogeneous distribu- sample surface (marked by a white triangle in the lower image
tion of the luminosity in this layer is observed, looking like a of Fig. 6). At the surface a partial reflection of the plasma front
turbulent structure, which tends to decrease at later times. is visible. The observed dynamics of the plasma expansion after
After 200 ns the expansion towards the incoming laser beam the second pulse is a clear indication that the first laser pulse
stops. After 350 ns the luminous plasma starts to expand in the has led to a transient modification of the ambient particle
backwards direction towards the sample surface. At 450 ns the density above the sample, thus enabling a greater expansion of
plasma shows a more homogeneous luminosity distribution the sample material ablated by the second pulse. The observed
and its shape differs significantly from the one observed in the reflection-like behaviour of the luminous front at about 300 ns
time interval between 150 ns and 250 ns. The plasma has nearly after the second laser pulse may be caused by a radially
reached the sample surface again. increasing particle density in the surrounding atmosphere as a
The further temporal development can be observed by streak result of the interaction of the first laser pulse. Streak photo-
photographs as shown in Fig. 6. The observation geometry is graphs taken with a greater aperture of the microscope show
the same as for Figs. 4 and 5. The streak velocity amounts to that for times beyond 1 ms after the second laser pulse the
50 ns mm21 on the screen of the streak camera. The time scale luminosity of the plasma becomes more and more homo-
runs from left to right. (In the original streak images delivered geneous and the structure and geometry of the plasma stays
from the streak camera the time scale ran from right to left. nearly constant for more than 5 ms.
These images are presented here side-inverted according to the In the time range shown in Figs. 4–6 the emission spectrum
usual direction of the time axes.) The white line illustrates the of the plasma changes from a continuum emission at early
position of the surface of the sample. The luminous regions times to pronounced line emissions after about 800 ns.6 The
beneath the white line are due to an overexposure of the streak exponential decay of the continuum emission can be described
images for times close to the arrival of the laser pulse. The by a decay time constant of about 90 ns corresponding to the

422 J. Anal. At. Spectrom., 2004, 19, 419–428

time range covered by the frames 2 and 3 of Fig. 4 and frames
1, 2 and 3 of Fig. 5.
Framing and streak images of Figs. 5 and 6 show clearly that
the dominant effect of the second pulse is again an interaction
with the sample surface and not a re-heating of a residual
plasma above the sample originating from the first pulse. In a
simplified approach we do not consider the non-linear
character of the interaction of the laser beam with the residual
plasma, which may lead to heating and ionisation processes.
The absorption of laser radiation in a plasma can then be
described by the absorption coefficient using the following
formula,18 assuming nc % v and vp % v:
nc v2p
a~ qffiffiffiffiffiffiffiffiffiffiffiffi2ffi (1)
v
c0 v2 1{ vp2

where nc is the collision frequency, vp the plasma frequency and

v the laser frequency. For this estimate we neglect absorption
by microscopic particulates originating from condensed iron
vapour or solidified melt droplets thrown out of the interaction
region by the first pulse. The collision frequency nc is given by
nc ~ nei 1 nen, where nei is the electron-ion collision frequency
and nen the electron-neutral collision frequency. The electron-ion
collision frequency nei can be estimated using the formula given
in ref. 19 for the electron-ion momentum transfer in a plasma
using the screened Rutherford cross section and assuming a Fig. 7 Height z of the luminous plasma front above the sample
surface as a function of time determined from side-on streak images.
Maxwellian electron distribution. The electron-neutral colli-
Top: after the first laser pulse; bottom: after the second laser pulse.
sion frequency is estimated by taking the geometric cross Experimental parameters are the same as for Fig. 4.
section of neutral iron atoms. With the following plasma
parameters determined spectroscopically at a time delay of bars in the z-direction indicate the reading error in determining
1 ms after a single pulse of 80 mJ: electron density ne ~ 7 6 the front position on the enlarged streak images. Uncertainties
1016 cm23, neutral iron density nFel ~ 6 6 1016 cm23, electron on the time axis are caused by deviations from the nominal
temperature T e ~ 8300 K, we obtain nei ~ 1.3 6 1012 s21 and streak velocity, which are specified to be less than 5%, and by
nen ~ 1.4 6 109 s21. Using relation (1) the estimated reading errors. The corresponding error bars are not shown
absorption coefficient due to ablated material originating in Fig. 7 since they are significantly smaller than those in the
from the first pulse amounts to a ~ 0.3 m21. Taking as an
z-direction. After the first pulse two regions of different
upper limit a plasma dimension of 5 mm, this corresponds to a
dynamics can be distinguished. For times up to 25 ns after the
transmission of more than 99% of the irradiated laser intensity
start of the interaction of the first laser pulse the data points
of the second pulse. In this case a dominant part of the laser
can be fitted by a linear regression described by the function
energy may propagate to the sample surface again where it
z 3 t(0.76 ¡ 0.02). A linear fit to the data points after 25 ns yields
interacts with the pre-ablated sample surface. Hence, the
z 3 t(0.44 ¡ 0.04). Under the assumption that the luminous
plasma emission induced by the second pulse should start at the
plasma front coincides with the position of the shock front at
sample surface, which is consistent with the experimental
the early expansion phase, we can compare the determined dyna-
observations of Figs. 5 and 6.
mics with the prediction of Sedov’s model. For times ¡ 25 ns
The expansion of the laser-induced plasma can be described
Sedov’s model cannot be applied, since the laser pulse still
by the Sedov-model for shock waves after a strong explosion in
persists and couples energy into the interaction region. Beyond
a homogeneous atmosphere.20,21 A dimensional analysis leads
25 ns the determined exponent is close to the theoretical value
to the following relationship for the expansion z of the shock
for a spherical expansion of a shock front. For the second laser
wave as a function of time t:
pulse the determined data points are shown in the lower dia-
1
z~l0 ðE0 =r0 Þ2zf t1zf
2
(2) gram of Fig. 7. Applying a single fit using the function z 3 tb
yields an exponent of b ~ 0.67 ¡ 0.02, which is consistent with
where E0 is the energy deposited in a small volume and r0 is the a planar shock front expansion. (A close look at the data points
density of the homogeneous undisturbed ambient gas, l0 in the lower diagram of Fig. 7 shows that there might be a small
depends on the specific heat ratio of the surrounding gas and j change in the dynamics after a time of about 60 ns. Applying
describes the geometry of the energy release and the shock wave the fit only to the data points with t w 60 ns yields b ~ 0.53 ¡
expansion with j ~ 3 for a point-like energy release and a 0.02. However, the change in the dynamics after the second
spherical shock wave expansion and j ~ 1 for a planar shock pulse at a time of about 60 ns is significantly smaller than the
wave expansion. Hence, for a spherical shock wave the expan- one oberved at about 25 ns after the first laser pulse.) These
sion should follow the relation z 3 t0.4 and for a planar values indicate different expansion geometries for the plasmas
expansion z 3 t0.66. The exponent of t is always expected to be induced by the first and the second laser pulse and are in
twice that of (E0/r0). qualitative agreement with the results of the framing images
To determine the expansion velocity of the luminous plasma shown in Figs. 4 and 5.
with a higher accuracy we have taken streak images of The derivatives of the regression curves of Fig. 7 yield the
the expanding plasma with streak velocities of 1 ns mm21 and expansion velocities of the luminous plasma front in the
5 ns mm21, while keeping the other experimental parameters z-direction. In the time interval from t ~ 30 ns to t ~ 60 ns after
constant. Fig. 7 shows a double-logarithmic plot of the deter- the first laser pulse this velocity decreases from 0.8 6 106 cm s21
mined expansion velocities after the first and the second laser to 0.6 6 106 cm s21 (corresponding to Mach numbers of 23
pulse of a double pulse. The time zero was chosen at the start of and 17), whereas in the same time interval after the second
the plasma emission detected in the streak images. The error laser pulse the expansion velocities are 1.5 6 106 cm s21 and

J. Anal. At. Spectrom., 2004, 19, 419–428 423

 1.2 6 106 cm s21 respectively. Hence, after the second pulse
the expansion is, by about a factor of 2, faster than after the
first pulse.

3.2. Interferometry of shock wave propagation

The experimental setup shown in Fig. 3 enables the observation
of changes of the refractive index n caused by the expanding
plasma in the measuring arm of the Mach–Zehnder inter-
ferometer. In a first step, without the laser-induced plasma, the
mirrors of the interferometer are tilted to adjust the number
and orientation of the interference fringes. We have chosen a
fringe orientation parallel to the sample surface. The slit of
the streak camera crosses these fringes perpendicularly. Fig. 8
shows streak images of the interferograms for a laser double
pulse with 2 6 80 mJ, Dt ~ 6 ms and Ds ~ 5 mm. The sample
surface is located at z ~ 0 mm. The expanding plasma and
shock front leads to a displacement of the interference fringes.
After the first laser pulse a disturbance of the fringe pattern
propagates in the positive z-direction. At the leading edge of
this disturbance a strong displacement of the interference
fringes towards the sample surface is observed. Reference
measurements with glass plates in the measuring arm of the
interferometer show that this direction of fringe displacement
corresponds to an increased index of refraction. Hence the
Fig. 9 Height z of the shock front above the sample surface as a
displacement of the fringes observed at the leading edge of the function of time determined from interferograms observed in side-on
expanding fringe disturbance is attributed to an increased direction with the streak camera, cf.
particle density caused by the shock wave. After the second
laser pulse an analogue displacement of the interference fringes exponent describing the expansion of the shock front in the
is observed at the leading edge of the evolving disturbance in time range from 200 ns to 12 ms after the first pulse amounts to
the fringe pattern, indicating the formation of a second shock 0.45 ¡ 0.01. This result agrees with the exponent determined
front propagating behind the shock front induced by the from the streak images of the luminous plasma at the early
first laser pulse. The reflection-like behaviour of the luminous times between 30 ns and 65 ns after the first laser pulse, cf.
plasma observed in Fig. 6 at about 300 ns after the impact of Fig. 7. The time exponent of the shock front expansion after
the second pulse does not show up in a corresponding structure the second laser pulse is 0.35 ¡ 0.04 and hence it is significantly
of the fringe patterns observed in Fig. 8. A potential reason smaller than the one determined from the luminous front at
might be a too small change in the refractive index caused by
early times, cf. Fig. 7.
the back-travelling plasma, which cannot be resolved by the For the determination of the time exponent for the shock
set-up used.
front propagation after the first laser pulse, functions of the
For a comparison of the experimental data with the Sedov
following form are fitted to the measuring data:
model, the expansion z of the shock front as a function of time
t is shown in a double-logarithmic plot in Fig. 9.
z ~ Ata (3)
The distortion of the interference fringes can be detected
clearly for more than 12 ms after the laser pulse interaction
Since the shock front propagates into an undisturbed
(see Fig. 8 bottom, shock front induced by the first laser pulse
ambient atmosphere the pre-conditions of Sedov’s model are
at about z ~ 7 mm) in contrast to the luminosity of the plasma,
fulfilled and a comparison of eqn. (3) with eqn. (2) yields the
which can be observed for a few microseconds only. The
following relations: a ~ 2/(2 1 j) and A ~ l0(E0/p0)a/2. For the
latter parameter the relation A 3 (E0/p0)a/2 also holds, where
p0 is the ambient pressure, which can be changed in the experi-
ment, cf. Fig. 3. In the following we assume that the energy E0
deposited locally is proportional to the irradiated laser pulse
energy E1 (E1 denotes the pulse energy of a single pulse or the
pulse energy of the first pulse of a double pulse). The fit
parameters A and a are determined for measurements with the
following sets of parameters: (a) single laser pulses with E1 ~
20 mJ, 40 mJ, 80 mJ, 120 mJ and 160 mJ at p0 ~ 1 bar ambient
pressure; (b) single laser pulses with E1 ~ 40 mJ at ambient
pressures of p0 ~ 100 mbar, 500 mbar and 1013 mbar. For
these parameters the determined exponent a varies between
0.36 and 0.50, with an average value of ā ~ 0.43 ¡ 0.05. This
value agrees with the expected exponent of 0.44 for a spherical
shock front propagation according to Sedov’s model. Relation
(3) describes well the dependence of the shock wave position as
a function of time for distances of up to 7 mm from the sample
surface, which is still below the critical radius21 rc ~ (E/p0)1/3,
which amounts to rc ~ 9.3 mm for E ~ 80 mJ and p0 ~
Fig. 8 Interferograms registered with the streak camera after the first 1013 mbar. At rc the counterpressure of the ambient gas can
and second laser pulse of a double pulse. Laser parameters: 2 6 80 mJ, no longer be neglected and the self-consistent solution of the
interpulse separation 6 ms, p0 ~ 1013 mbar. problem of a strong explosion no longer holds.

424 J. Anal. At. Spectrom., 2004, 19, 419–428

 The propagation of shock waves in an inhomogeneous
atmosphere was described by Sedov assuming the following
relation for the density variation of the ambient atmosphere by
a power law:20

r ~ r(r) ~ Dr2Q (4)

where D is a positive constant, r is the radius and Q is a

dimensionless constant. The propagation of a spherical shock
wave as a function of time is then given by the following
power law:

r 3 t2/(52Q) (5)

Taking the experimentally determined exponent for the

shock wave after the second pulse and relation (5), we obtain
Fig. 10 Fit parameter A as a function of the ratio of the laser pulse
Q ~ 21.25. Eqn. (4) then yields r(r) 3 r1.25, which describes a
energy E1 and the ambient pressure p0 for the first laser pulse.
density increasing radially from the center. Whereas the first
shock wave propagates into a spatially homogeneous ambient
Fig. 10 shows the values of the fit parameter A determined atmosphere, the second shock wave encounters, as a result of
for the parameter sets described above in a double- the interaction of the first laser pulse, a locally rarefied ambient
logarithmic plot. The linear regression shown in Fig. 10 yields: atmosphere, allowing for higher propagation velocities of the
A 3 (E1/p0)0.23 ¡ 0.02. As expected by the Sedov model, the second shock wave.
exponent is about half of ā. Hence the dynamics of the shock
front propagation after the first laser pulse can be described in a 3.3. Index of refraction of laser-induced plasma
wide range of laser pulse energies and ambient pressures by
Sedov’s model for a spherical expansion. The index of refraction caused by the laser-induced plasma is
For the description of the expansion of the shock wave after determined from the interference patterns detected in the
the second laser pulse, we use in analogy to relation (3) the framing mode of the streak camera. Figs. 12 and 13 show
following function: z ~ Btb. Since the ambient atmosphere is framing images of the interference patterns after the first and
not homogeneous for the second shock wave the parameters B the second laser pulse of a 2 6 80 mJ double pulse and an
and b cannot be interpreted with respect to eqn. (2). The ambient pressure of p0 ~ 1013 mbar. The framing rate is
following measuring parameters were used to determine B and 106 s21, corresponding to a time separation between
b: (a) double pulses with E1 ~ E2 (E2 denotes the pulse energy subsequent frames of 1 ms. The exposure time amounts to
of the second pulse of a double pulse) and E2 ~ 20 mJ, 40 mJ, 200 ns per frame. For a better use of the image format of
60 mJ, 80 mJ and 100 mJ at p0 ~ 1013 mbar ambient pressure; the streak camera the camera is rotated by 90 degrees around
(b) double pulses with E1 ~ E2 ~ 40 mJ at ambient pressures the axis of observation. The mutual distance between the
of p0 ~ 100 mbar, 500 mbar and 1013 mbar. The determined fit interference fringes is increased in relation to the one chosen
parameter b varies between 0.27 and 0.41, with an average for the streak images in the previous section to simplify the
value of b̄ ~ 0.32 ¡ 0.05. Fig. 11 shows the determined evaluation of the interferograms. The geometry and sequence
parameter B as a function of the laser pulse energy of the of the frames is shown on the right side of Fig. 12. The time of
second pulse E2 at a pressure of p0 ~ 1013 mbar (pressure in irradiation of the laser pulses cannot be determined exactly
from the framing images. In Fig. 12 the first pulse arrives at the
the ambient atmosphere before the impact of the first pulse). A
sample between the frames nos. 1 and 2, and the second pulse
regression curve yields: B 3 (E2)(0.30 ¡ 0.02). The parameter B is
between frames 7 and 8. In Fig. 13, the second pulse hits the
not plotted as a function of (E2/p0) since there is no homo-
sample between frame nos. 1 and 2.
geneous pressure distribution in the interaction region at the
The boundaries of the displaced interference fringes show a
time the second laser pulse is irradiated on to the sample.
well defined hemispherical shape. At the front of the shock
The determined exponent is approximately equal to b̄. This
wave the fringes are displaced towards the sample surface,
behaviour differs significantly from the shock wave propaga-
which is due to the increase of the refractive index caused by the
tion after the first laser pulse.
compressed air and sample particles. Inside the sphere, defined
by the leading edge of the shock wave, the fringes are displaced
in the opposite direction, i.e., towards the positive z-direction,
compared with their original position. This displacement
corresponds to a refractive index v1, indicating the presence
of free electrons and/or a depletion of the ambient atmosphere.
The fringe displacements are symmetrical with respect to the
z-axis, hence the method of Abel inversion can be applied to
determine the refractive index distribution.22
The fringe displacement is given by:
ð
1  
dðy,z,tÞ~ np {na dx (6)
l
p

where x, y, z are the coordinates shown in Fig. 3, l is the

wavelength of the argon ion laser, np is the refractive index of
the plasma and na is the refractive index of the ambient atmos-
phere in the reference arm of the interferometer. The integra-
Fig. 11 Fit parameter B as a function of the ratio of the energy E2 of tion runs along the line of sight across the plasma volume. The
the second laser pulse at an initial ambient pressure of p0 ~ 1013 mbar. refractive index of the plasma can be described by the following

J. Anal. At. Spectrom., 2004, 19, 419–428 425

 Fig. 12 Framing images of inteferograms detected in a time interval between the first and the second laser pulses of a double pulse. In the upper
right a schematic view of the orientation of the fringes relative to the plasma and the sample is shown. The diagram in the lower right shows the
temporal sequence of the frames with the delay times tdelay, starting at 0 ms for the first frame. Experimental parameters are the same as for Fig. 8.

relation:22 the literature. However, calculations of polarizabilities for the

e2 l2 X transition element copper showed that the polarizability of a
np {1~{ 2 2
Ne z 2pNj aj (7) copper ion is less than one-tenth of the value calculated for
8p me c e0 j
copper atoms.24 Assuming that similar relations hold for iron
where N e is the electron density, Nj are the densities of neutral we expect only a small contribution of the iron ions to the
atoms and ionic species in the plasma and aj the polarizabilities. refractive index, which is in the order of the one estimated for
The polarizability of atomic iron23 is aFel ~ 13 6 10230 m23. the iron atoms.
The contribution of the iron atoms to the refractive index can A reduction of the ambient gas particle density behind
then be estimated with eqn. (7): nFel 2 1 ~ 18.2 6 10223 N Fel, the shock wave will lead to a reduced refractive index. The
where N Fel is the particle density of atomic iron in cm23. The order of magnitude of this effect is estimated assuming that
contribution of the free electrons amounts to ne 2 1 ~ 212 6 the shock wave reduces the ambient gas particle density
10223 N e, with N e in cm23. With an average degree of ioniza- completely, i.e., the shock wave leaves a vacuum behind. Under
tion in the plasma, determined spectroscopically, of 90% the this simplified assumption the ratio of the fringe displace-
ratio |nFel 2 1|/|ne 2 1| amounts to 0.08, hence the contribution ments due to vacuum dv and free electrons de amounts to:
dv
of the atomic iron particles can be neglected. No data for the de ~j1{na j=jne {na j~0:7for na ~ 1.000 277 9, l ~ 514.5 nm
polarizability of singly ionized iron atoms have been found in and N e ~ 1018 cm23. If a complete displacement of the ambient

Fig. 13 Framing images of inteferograms detected after the second laser pulse of a double pulse. All other experimental parameters are the same as
for Fig. 12.

426 J. Anal. At. Spectrom., 2004, 19, 419–428

Fig. 14 Refractive index distribution determined from the framing interferograms. Left: 4 ms after a single 80 mJ laser pulse. Right: 5 ms after a
single 80 mJ laser pulse.

atmospheric particle density occurs, then the expected fringe ambient pressure of 500 mbar show a reduced depth of the
shift is of the same order of magnitude as the one expected by refractive index trough.
the free electrons of the plasma. Since both effects go into the In the following we assume that only the free electrons lead
same direction, the experimentally determined fringe shifts can to the reduced refractive index region with n v 1; then the
only be attributed to the free electrons of the plasma by the relation (7) can be used to calculate the electron density dis-
de
ratio de zd v
~0:6. tribution. Fig. 16 shows the result. At the centre of the plasma
Fig. 14 shows the determined distribution of the refractive sphere the electron density attains a value of y3 6 1018 cm23.
index 4 ms (left) and 5 ms (right) after the irradiation of one The estimated relative error is 20%. Taking into account the
80 mJ laser pulse. The formation of the shock front in a above discussed reduction of the ambient gas particle density
spherical geometry can be clearly seen by the values of the behind the shock wave, the electron density is only y60% of
refractive index n w 1. Behind the shock front a volume of this value, i.e., 2 6 1018 cm23. The electron density determined
reduced refractive index becomes visible. Fig. 15 shows refrac- interferometrically after a single pulse of 80 mJ and 2 ms is
tive index distributions 2 ms after a single 80 mJ laser pulse (left) about the same. Comparing these values with those gained
and after the second pulse of a 2 6 40 mJ double pulse (right). from Stark broadened atomic iron lines measured by spatially
The formation of two concentric shock waves can be seen in the integrating the whole plasma emission6 shows that the inter-
right diagram, where the second shock wave is weaker than the ferometric values are more than an order of magnitude greater.
first. Interpreting the central region with n v 1 as the plasma A potential reason for this discrepancy could be that the Stark
volume, the volume ratio amounts to 3.3, i.e., the plasma broadening of iron atoms (Fe I line at 538.337 nm) was
volume after the double pulse is by a factor of 3.3 greater than measured, hence the outer regions of the plasma contributed
in the case of the single pulse (for the same total laser pulse predominantly to the spectroscopic signals, where the electron
energy irradiated). Framing interferograms taken at a reduced density is lower than in the centre.

Fig. 15 Refractive index distribution determined from the framing interferograms. Left: 2 ms after a single 80 mJ laser pulse. Right: 2 ms after the
second pulse of a 2 6 40 mJ double pulse.

J. Anal. At. Spectrom., 2004, 19, 419–428 427

 References
1 L. Berman and P. Wolf, Appl. Spectrosc., 1998, 52, 438–442.
2 H. Zhang, F. Yueh and J. Singh, Appl. Optics, 1999, 38, 1459–
1466.
3 F. Wallis, B. Chadwick and R. Morrison, Appl. Spectrosc., 2000,
54, 1231–1235.
4 D. Anglos, Appl. Spectrosc., 2001, 55, 186A–205A.
5 M. Kurihara, K. Ikeda, Y. Izawa, Y. Deguchi and H. Tarui,
Appl. Optics, 2003, 42, 6159–6165.
6 R. Sattmann, V. Sturm and R. Noll, J. Phys. D: Appl. Phys., 1995,
28, 2181–2187.
7 A. Pichahchy, D. Cremers and M. Ferris, Spectrochim. Acta, Part
B, 1997, 52, 25–39.
8 V. Sturm, L. Peter and R. Noll, Appl. Spectrosc., 2000, 54, 1275–
1278.
9 F. Colao, V. Lazic, R. Fantoni and S. Pershin, Spectrochim. Acta,
Part B, 2002, 57, 1167–1179.
10 L. St.-Onge, V. Detalle and M. Sabsabi, Spectrochim. Acta, Part B,
2002, 57, 121–135.
11 S. Klimentov, S. Garnov, T. Kononenko, V. Konov, Pl. Pivovarov
and F. Dausinger, Appl. Phys. A, 1999, 69, S633–S636.
12 D. Stratis, K. Eland and S. Angel, Appl. Spectrosc., 2001, 55,
Fig. 16 Electron density distribution calculated from Fig. 15, right, 1297–1303.
assuming that the refractive index inside the second shock wave is 13 D. Stratis, K. Eland and S. Angel, Appl. Spectrosc., 2000, 54,
caused by free electrons only. 1270–1274.
14 A. Kuwako, Y. Uchida and K. Maeda, Appl. Optics, 2003, 42,
6052–6056.
Conclusions 15 R. Noll, R. Sattmann, V. Sturm, S. Lüngen and H.-J. von
Wachtendonk, Stahl und Eisen, 1997, 117, 57–62 (in German).
The investigations have shown that the effect of collinear 16 L. Peter, V. Sturm and R. Noll, Appl. Optics, 2003, 42, 6199–6204.
double pulses is a local transient reduction of the particle 17 C. M. Vest, Appl. Optics, 1975, 14, 1601–1606.
density in the vicinity of the sample surface caused by the first 18 P. Mulser, Z. Naturforsch., 1970, 25a, 282–295.
pulse. The second pulse interacts predominantly with the 19 A. Anders, A Formulary for Plasma Physics, Akademie-Verlag,
sample surface. No re-heating of a residual plasma from the Berlin, 1990, p. 51.
first pulse was observed. The material ablated by the second 20 L. I. Sedov, Similarity and Dimensional Methods in Mechanics,
Gostekhizdat, Moscow, 4th edn., 1957. English transl. ed.
pulse expands faster into the cavity of reduced particle density M. Holdt, Academic Press, New York, 1959.
originating from the first pulse attaining a significantly 21 Y. B. Zel’dovich and Y. P. Raizer, Physics of Shock Waves and
enlarged plasma volume. The electron densities are approxi- High-Temperature Hydrodynamic Phenomena, eds. W. Hayes and
mately the same for single and double pulses of the same total R. Probstein, Academic Press, New York, 1966, vol. I.
energy. The larger plasma volume and the fact that the mass 22 R. Boxmann, J. Appl. Phys., 1974, 45, 4835–4846.
ablated using double pulses is also increased significantly6 23 S. Fraga, J. Karwowski and K. Saxena, Atomic and Nuclear Data
favours a more intense and longer lasting emission of element Tables, 1973, 12.
24 S. Fraga, K. Saxena and B. Lo, At. Data, 1971, 3, 323–332.
specific radiation to be used for LIBS analysis.

428 J. Anal. At. Spectrom., 2004, 19, 419–428