Youbo Zhao, Sima Singha, Yaoming Liu, and Robert J. Gordon*
Department of Chemistry, University of Illinois at Chicago, Chicago, Illinois 60607-7061, USA
* Corresponding author: firstname.lastname@example.org
Received October 21, 2008; accepted November 17, 2008;
posted January 9, 2009 (Doc. ID 102966); published February 12, 2009
It is shown that plasma polarization measurements can be used to enhance the sensitivity of laser-induced
breakdown spectroscopy (LIBS). The polarization of the plasma emission is used to suppress the continuum
with only slight attenuation of the discrete atomic and ionic spectra. The method is demonstrated for LIBS
detection of copper and carbon samples ablated by pairs of femtosecond laser pulses. © 2009 Optical Society
OCIS codes: 300.6365, 140.3440, 260.5430.
Laser-induced breakdown spectroscopy (LIBS) is a
versatile tool for elemental analysis with many prac-
tical applications [1,2]. In particular, LIBS may be
used for real-time detection of multiple elements by
tracing their respective atomic (or ionic) and/or mo-
lecular spectral fingerprints emitted by the laser-
generated plasma. With little or no requirements for
pretreatment of the analyte, LIBS may be utilized
universally to measure the composition of samples in
any physical state, including particles and aerosols.
Although LIBS does not have the sensitivity of some
other established methods of chemical analysis, it is
a superior technique for stand-off detection in harsh
environments , such as blast furnaces , nuclear
reactors , biohazardous areas , and the Martian
It is well known that the plasma spectrum pro-
duced by laser ablation of materials consists of a se-
ries of discrete lines and a broadband continuum
[8,9]. Reducing the continuum background is one of
the key challenges for improving the analytical capa-
bility of LIBS. A number of techniques have been in-
troduced to increase the line-to-continuum ratio.
Among them, time-resolved nanosecond LIBS is most
frequently used to extract line emission signals from
the continuum background, taking advantage of their
different temporal-evolution characteristics [10,11].
This technique is very sensitive, however, to the se-
lection of the time gate, because each resonant line
has its own temporal profile that depends strongly on
the laser-excitation conditions and sample properties
[11,12]. Moreover, inevitable variation of experimen-
tal conditions during laser ablation of the samples
greatly deteriorates the reproducibility of the mea-
surement. Spatially resolved detection and buffer-
gas-assisted methods have also been proposed to im-
prove the signal/background capability of LIBS
[13,14]. Nevertheless, because of the strict experi-
mental requirements of these methods, temporally
and spatially integrated spectral measurement is
still the best choice in the practical application of
LIBS, provided that the continuum background can
integrated nanosecond and femtosecond (fs) LIBS
spectra with identical fluences revealed a much
A comparison oftime-
larger ratio of the discrete to continuum emission in
the fs regime .
In this Letter, we describe a new technique for re-
ducing the continuum background. In a previous
study, we found that the plasma continuum emission
produced by ablation of solid materials with pairs of
fs laser pulses is strongly polarized, whereas the dis-
crete fluorescence lines are much less polarized .
Taking advantage of this phenomenon, we show here
that by placing a simple polarizer before the detector,
the ratio of the discrete line to the continuum spec-
trum may be significantly increased in the tempo-
rally and spatially integrated LIBS. We refer to this
technique as polarization-resolved LIBS (PRLIBS).
The apparatus used for PRLIBS is similar to that
used in our previous studies [16–18], with experi-
mental parameters chosen to maximize the polariza-
tion of the continuum . Pairs of fs laser pulses
with durations of ?65 fs and a central wavelength at
800 nm were generated by a Ti:sapphire laser and a
Michelson interferometer. The two pulses had equal
energies of 14 ?J, and the time delay between them
was fixed at 80 ps. The laser was focused onto the
sample surface by a convex lens ?f=100 mm?, and the
total fluence was 8.9 J/cm2. A comparatively long fo-
cal length (and thus larger Rayleigh range) was se-
lected to reduce the sensitivity of the polarization to
the location of the focal point .
Industrial copper (OHFC, 99.9%) and semiconduc-
tor grade graphite (DFP-3-2, Poco) samples were
mounted in air on a computer-driven xyz stage. The
incidence angle of the laser beam was 30° from the
normal direction of the sample surface, and its polar-
ization plane was parallel to the sample surface (s po-
larized). The plasma plume was imaged normal to
the direction of the laser beam by a pair of lenses
with focal lengths of 75 and 50 mm, respectively, onto
the 50-?m-wide entrance slit of a spectrograph (Spec-
300 lines/mm grating blazed at 500 nm was typically
used to disperse the spectrum, which was recorded by
a nongated CCD (PIXIS 400, Princeton Instruments)
camera. A Glan–Thompson polarizer mounted in a
motorized rotation stage (PR50, Newport) was in-
OPTICS LETTERS / Vol. 34, No. 4 / February 15, 2009
0146-9592/09/040494-3/$15.00 © 2009 Optical Society of America
serted in front of the entrance slit of the spectrograph
to measure the polarization of the plasma emission.
The spectrum was measured at each position of the
polarizer by summing the signals generated by 20 la-
ser shots. A fresh sample position was irradiated at
each polarizer angle.
Figure 1 shows the plasma emission spectra of cop-
per and graphite measured with and without the po-
larizer in place. The upper panel is the time- and
space-integrated LIBS spectra, whereas the lower
background ratio (SBR) obtained by PRLIBS. For ex-
ample, the SBR for the green Cu lines at 511, 515,
and 522 nm are increased from 1.5, 1.7, and 2.2 in
LIBS to 4.2, 5.3, and 8.2 in PRLIBS, respectively. Ad-
ditionally, Cu and Cu+lines (especially those at short
wavelengths) that are embedded in the continuum
background of LIBS spectrum are resolved in the
PRLIBS spectrum. For graphite, the C2Swan bands
[14,19] as well as carbon atomic and ionic lines such
as C I at 437 nm, C II at 515 nm, and C III at 388 nm
are much better resolved by PRLIBS.
To understand how the polarizer suppresses the
continuum radiation and thereby enhances the SBR,
we measured the plasma polarization spectra for
both samples. The polarization spectra shown in Fig.
2 were obtained by fitting a Malus function to the sig-
nal intensity in each 0.21 nm interval . A com-
parison of the LIBS emission spectrum and the polar-
ization spectrum, both measured without temporal
gating, shows that the continuum is much more
strongly polarized than the discrete line emission.
The spectra recorded with the polarizer positioned at
different angles are plotted in Fig. 3. When the polar-
izer is positioned with its polarization plane vertical
to that of the plasma continuum radiation, the back-
ground is filtered out significantly, while the emission
lines retain their intensity.
The spectra shown in Fig. 4 demonstrate a practi-
cal application of PRLIBS. Shown here are the LIBS
and PRLIBS spectra of a Cu sample and a U.S. dime
coin. The latter is composed of 95% Cu and 5% Ni. In
the conventional LIBS spectrum, the Ni I 548 nm
peak is visible, but Ni lines at shorter wavelength are
buried in the continuum. In contrast, in the PRLIBS
spectrum the Ni I 386 and 548 nm lines, as well as
the Ni II 403 and 499 nm lines, are readily observed.
The underlying mechanisms of the different polar-
ization states of the discrete and continuous spectra
are not yet fully understood. The continuum is pri-
marily generated by inverse Bremsstrahlung (free–
free) and recombination (bound–free) transitions,
lation of copper and graphite. Upper graphs, LIBS mea-
sured without the polarizer; lower graphs, PRLIBS mea-
sured with the polarizer in place with its polarization plane
perpendicular to that of the laser.
Plasma emission spectra produced by fs laser ab-
curves) and polarization (dotted curves) spectra for (a) cop-
per and (b) graphite samples.
(Color online) Comparison of the intensity (solid
larizer at different angles. 0° corresponds to the polariza-
tion plane of the polarizer parallel to that of the incident
Copper and carbon spectra measured with the po-
February 15, 2009 / Vol. 34, No. 4 / OPTICS LETTERS
whereas the discrete lines stem from resonant
(bound–bound) transitions . Polarization of the
continuum is a consequence of a highly anisotropic
velocity distribution of the free electrons, whereas po-
larization of the discrete spectrum is caused by a
nonstatistical population of the magnetic states of
the emitting species. The origin of these distributions
is under investigation.
In summary, we demonstrated the usefulness of
PRLIBS as compared with conventional LIBS. The
results demonstrate that the SBR is significantly en-
hanced by using a polarizer to suppress the the
strong continuum background produced in femtosec-
ond ablation of materials. Moreover, because tempo-
rally and spatially integrated spectra are employed
by this technique, PRLIBS has good reproducibility,
simplicity and stability and a high signal-to-noise ra-
tio, as compared to time-gated or spatially resolved
We wish to thank Tana Witt for her assistance with
some of the experiments. Support by the National
Science Foundation (NSF) under grant CHE-0640306
and by the U.S. Air Force Research Laboratory
Materials and Manufacturing Directorate is grate-
1. D. A. Cremers and L. J. Radziemski, Handbook of
Laser-Induced Breakdown Spectroscopy (Wiley, 2006).
2. J.P. Singhand S.
Breakdown Spectroscopy (Elsevier, 2007).
3. H. L. Xu, W. Liu, and S. L. Chin, Opt. Lett. 31, 1540
4. R. Noll, H. Bette, A. Brysch, M. Kraushaar, I. Monch,
L. Peter, and V. Sturm, Spectrochim. Acta, Part B 56,
5. A. I. Whitehouse, J. Young, I. M. Botheroyd, S. Lawson,
C. P. Evans, and J. Wright, Spectrochim. Acta Part B
56, 821 (2001).
6. J. Diedrich, S. J. Rehse, and S. Palchaudhuri, Appl.
Phys. Lett. 90, 163901 (2007).
7. B. Salle, J. L. Lacour, E. Vors, P. Fichet, S. Maurice, D.
A. Cremers, and R. C. Wiens, Spectrochim. Acta Part B
59, 1413 (2004).
8. T. Fujimoto and A. Iwamae, eds., Plasma Polarization
Spectroscopy (Springer, 2007).
9. T. Fujimoto, Plasma Spectroscopy (Oxford U. Press,
10. L. Dudragne, P. Adam, and J. Amouroux, Appl.
Spectrosc. 52, 1321 (1998).
11. B. Le Drogoff, J. Margot, M. Chaker, M. Sabsabi, O.
Barthelemy, T. W. Johnston, S. Laville, F. Vidal, and Y.
von Kaenel, Spectrochim. Acta Part B 56, 987 (2001).
12. I. Bassiotis, A. Diamantopoulou, A. Giannoudakos, F.
Spectrochim. Acta Part B 56, 671 (2001).
13. S. J. J. Tsai, S. Y. Chen, Y. S. Chung, and P. C. Tseng,
Anal. Chem. 78, 7432 (2006).
14. L. Nemes, A. M. Keszler, J. O. Hornkohl, and C. G.
Parigger, Appl. Opt. 44, 3661 (2005).
15. J. B. Sirven, B. Bousquet, L. Canioni, and L. Sarger,
Spectrochim. Acta Part B 56, 1033 (2004).
16. Y. Liu, S. Singha, T. E. Witt, and R. J. Gordon, Appl.
Phys. Lett. 93, 161502 (2008).
17. Z. Hu, S. Singha, Y. M. Liu, and R. J. Gordon, Appl.
Phys. Lett. 90, 131910 (2007).
18. S. Singha, Z. Hu, and R. J. Gordon, J. Appl. Phys. 104,
19. C. G. Parigger, J. O. Hornkohl, A. M. Keszler, and L.
Nemes, Appl. Opt. 42, 6192 (2003).
PRLIBS spectra for a copper sample (dotted curves) and a
US dime coin (solid curves).
(Color online) Comparison of (a) LIBS and (b)
OPTICS LETTERS / Vol. 34, No. 4 / February 15, 2009