Degenerate Four Wave Mixing on FeI atomic vapours during thermal decomposition of Fe(CO)5 : Saturation and absorption effects

Article abstract of DOI:10.1016/S0009-2614(01)01140-X

Thermal decomposition of Fe(CO)₅ vapours diluted in a buffer gas (Ar) was performed in a static optical cell heated up to 673 K. The nascent FeI formation in the gas phase was monitored by recording the Degenerate Four Wave Mixing (DFWM) spectrum of the a⁵D_j-y⁵D⁰_j ^′ atomic transitions around 302 nm. Line-splitting phenomena, occurring upon strong saturation conditions at high laser power, were investigated on the DFWM lineshape. The role played by the absorption of exciting and signal beams in altering the relative lines intensities, predicted assuming an optically thin medium, is discussed and modelled in order to account for the experimental results.

Full text of DOI:10.1016/S0009-2614(01)01140-X

9
November 2001
Chemical Physics Letters 348 /2001) 209±216
www.elsevier.comꢀlocateꢀcplett
Degenerate Four Wave Mixing on FeI atomic vapours
during thermal decomposition of FeꢀCO† : saturation
5
and absorption e€ects
a
L.De Dominicis , M.Di Fino , R.Fantoni
a
a,
a
*
, S.Martelli , O.Bomat ꢀõ
b
Miguel , S.Veintemillas Verdag u er
b
a
ENEA Applied Physics Division, Via E. Fermi 45, I-00044 Frascati ꢀRome), Italy
Instituto de Ciencia de Materiales de Madrid, Cantoblanco, 28049 Madrid, Spain
b
Received 2 August 2001; in ®nal form 25 September 2001
Abstract
Thermal decomposition of FeꢀCO† vapours diluted in a bu€er gas /Ar) was performed in a static optical cell heated
5
up to 673 K.The nascent FeI formation in the gas phase was monitored by recording the Degenerate Four Wave
0
5
5
j
j
0
Mixing /DFWM) spectrum of the a D ±y D atomic transitions around 302 nm.Line-splitting phenomena, occurring
upon strong saturation conditions at high laser power, were investigated on the DFWM lineshape.The role played by
the absorption of exciting and signal beams in altering the relative lines intensities, predicted assuming an optically thin
medium, is discussed and modelled in order to account for the experimental results. Ó 2001 Elsevier Science B.V. All
rights reserved.
1
. Introduction
collected.Most of the spectroscopic measurements
were performed on atomic Fe vapours by means of
Laser Induced Fluorescence /LIF) [2], a technique
currently used for the quantitative space resolved
detection of FeI [3].
Degenerate Four Wave Mixing /DFWM)
spectroscopy, a non-linear technique based on
light scattering from laser induced gratings, is an
e€ective alternative to LIF in order to detect traces
of chemical species, since to a comparable sensi-
tivity, DFWM adds a better space resolution.In
DFWM experiments two pump laser beams with
The atomic structure of neutral iron /FeI) has
been exhaustively studied by several spectroscopic
methods.Laser based techniques played a primary
role in developing the accurate description of en-
ergy levels, lifetime and transition probabilities in
FeI.The results of a large number of studies on
iron are summarised in the National Institute of
Standard and Technology /NIST) database [1],
where as much as 2092 spectral line-positions are
both the same wavelength k
state, travel through the investigated medium
forming with respect to each other an angle h.If k
P
and polarisation
*
Corresponding author.Fax: +39-6-9400-5312.
E-mail address: fantoni@frascati.enea.it /R. Fantoni).
P
0
009-2614ꢀ01ꢀ$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 6 1 4 / 0 1 ) 0 1 1 4 0 - X

210
L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
5
5
0
j
matches with one of the allowed transitions of one
or more chemical species in the medium, the gen-
erated pattern of optical fringes induces a sinu-
soidal modulation of the molecular internal energy
and a modulation of the refraction complex index.
This modulation originates, in turn, a grating able
to di€ract a fraction of a third degenerate probe
study the a D
FeI, located around 302.06 nm, were investigated
j
±y D ꢀj ˆ 2; 3; 4† transitions of
in an optical cell ®lled with a mixture of FeꢀCO†
5
vapours diluted in Ar, where the thermal decom-
position of the precursor was induced.In order to
5
5
0
2
investigate the saturation e€ects the a D ±y D
2
DFWM lineshape has been recorded at di€erent
values of pump laser energy.A theoretical simu-
lation, based on the radiative re-normalisation
method reported in [17], is proposed to explain the
splitting mechanism.
The DFWM spectra collected far from satura-
tion intensity showed, both at the high tempera-
ture /up to 673 K) and after cooling down the
thermally decomposed mixture, line intensities
characterised by strong deviations from the values
expected for an optical thin medium at thermal
equilibrium.A simpli®ed analytical model is fur-
ther presented to account for the wavelength de-
pendent signal reduction, due to the absorption of
incoming pump beams and of the generated signal
beam.
beam /at wavelength k ) which is crossing the in-
P
teraction region at the Bragg angle.In the last
years DFWM on small molecules and radicals has
been extensively studied, due to its potential for
quantitative measurements of physical parameters
/
as for instance gas concentration and tempera-
ture) in chemically reacting media /e.g., ¯ames and
plasma) [4±6].It has been found that DFWM
signal is considerably less sensitive to collision
quenching as compared to LIF, hence well suited
for the detection of species at nearly atmospheric
or higher pressure [7].
At the beginning in atomic vapours DFWM
was mainly investigated because of the extremely
high re¯ectivity /up to about 100%) observed from
laser-induced gratings [8,9], a promising condition
for the development of phase-conjugate mirrors
based on atomic gases.Furthermore DFWM on
atomic vapours was the object of special care due
to laser beam saturation e€ects in the lineshape,
which, ®rst theoretically predicted, were experi-
mentally observed in sodium vapours [10±12].
Highly saturated atomic DFWM lineshapes show,
together with the usual broadening, a character-
istic splitting when the laser is tuned at exact res-
onance with the probed transitions.Despite the
partial knowledge on the physical mechanisms at
the basis of the saturated lineshape splitting, this
e€ect should be carefully accounted for any time
DFWM is used for quantitative measurements in
atomic vapours.
2. Theoretical background
The DWFM signal intensity is proportional
to the square of the third-order non-linear sus-
ceptibility and therefore depends on the square
of the population di€erence Nif between the
initial and ®nal atomic level coupled by the
resonant pump laser.Furthermore, in the limit
of optical thin medium condition, it was found
[5] that the integrated DFWM line intensity Iif
is proportional to the square of the atomic line
strength Bif
Â
Ã
2
2
I / B N I I ;
ꢀ1†
if
if if
P
pr
In the present contribution we report the results
of DFWM detection of atomic iron produced by
thermal dissociation of iron pentacarbonyl,
where I and I are, respectively, the pump and
P
pr
probe beam intensity.In thermal equilibrium at a
temperature T, not suciently high for populating
the upper level of the transition, Nif can be ap-
proximated by the numbers N of atoms in the
lower state given by the Boltmann distribution
FeꢀCO† , a volatile oily complex which decom-
5
poses into Fe ‡ 5ꢀCO† upon heating above 523 K.
i
The FeI formation due to FeꢀCO† photo-frag-
5
mentation /photolysis) has been the subject of a
large number of spectroscopic investigations [13±
ÀꢀEi=kT†
N
i
/ Ng
i
e
;
ꢀ2†
15], but only scant information is available for
thermally induced dissociation [16].In the present
i i
where g and E are, respectively, the degeneracy
and the energy of the lower atomic level, and N is

L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
211
the atomic number density.From Eq./2) results
that the ratio between two DFWM spectral lines
follows the thermal distribution of atoms in the
excited levels probed.
results at di€erent temperatures /i.e., to de®ne a
thermometric curve or to evaluate relative con-
centrations).
It should be mentioned that in the theoretical
treatment neither the particle motion nor satura-
tion e€ects on the DFWM signal intensity are
considered.The hypothesis that particle motion
can be neglected implies supposing that the atoms
are frozen in their positions during the time du-
ration of the exciting laser pulse and ascribing the
observed line broadening entirely to collisional
events.These assumptions are valid provided that
Anyway, if the optically thin medium approxi-
mation aꢀx†L ꢁ 0, with aꢀx† absorption coe-
cient and L length of the active medium, is not
ful®lled, strong deviations in the DFWM line in-
tensities ratio can be observed due to absorption
e€ects on pumps, probe and signal beams.In
particular, since the absorption coecient is a non-
linear function of the number of atoms in the
lower level, the signal reduction e€ects are more
intense the stronger the DFWM lines /originated
by the most populated levels at the temperature T)
are.Furthermore, the nearly cubic dependence of
the DFWM signal intensity on the exciting laser
beam energy /when pumps and probe beams are
assumed equally intense) contributes to enhance
the absorption e€ects by altering line intensity ra-
tio's.Although a complete theoretical treatment of
pumps, probe depletion and signal absorption
e€ects has been already proposed in the limit of
non-saturating pump intensities [18], it requires
awkward numerical solutions drastically limiting
the range of application.
the experimental condition K ꢂ k
T
is ful®lled,
where K ˆ k =2 sinꢀh=2† is the fringe spacing and
P
k is the mean thermal displacement of an atom
T
during the time duration of the laser pulse /split in
contemporary pumps and probe beams).
Saturation e€ects in Forward DFWM
/FDFWM) lineshape were formerly studied by
using the radiative re-normalisation technique [17].
The relevant parameter S introduced in the model
is de®ned as
2
X
S ˆ
;
ꢀ4†
c c
0
ab
where c and c are the population and collisional
0
ab
In the present DFWM experiment two pump
beams with an intensity sucient to saturate the
one-photon probed transitions were used /see
Section 3) together with a weaker probe beam,
whose intensity is far below the saturation
threshold.In order to model the experimental data
we used a simpli®ed model in which the absorption
e€ect is comprehended, as a whole, in a single at-
tenuation factor
relaxation rate, respectively, while X is the Rabi
frequency of the transition
l E
P
ij
X ˆ
ꢀ5†
ꢁh
with l the induced dipole moment of the probed
ij
transition and E the electric ®eld strength of the
P
pump beam.From Eqs./4) and /5) the parameter
S turns out to be non-dimensional.If the partial
gas pressures and the temperature are kept ®xed, S
increases as pump laser power is raised.Any time
the pump energy exceeds the saturation threshold
level /corresponding to S ˆ 1), the line broadening
due to Rabi oscillations dominates the collisional
width.
Â
à Á
2
b
I
if / Bif
N
i
1 À aif
N
i
:
ꢀ3†
A further simpli®cation was introduced by re-
leasing the wavelength dependence of aif and
substituting this coecient with an average value
a, which assumes that only the absorption e€ect
due to population di€erences between atomic
levels is taken into account.The a-dimensional
a; b parameters /the latter introduced as the ex-
ponent in the right factor of Eq./3)) can be cal-
culated from the measured lines intensities ratio
at a given temperature and values obtained can
be successively used to predict experimental
The theoretical FDWFM spectrum of the
0
5
5
a D ±y D transition in FeI at di€erent degree of
2
2
saturation /i.e., di€erent values of S) is reported in
Fig.1.
As predicted by the model, the ®ngerprint for
the saturation of FDFWM lineshape is the ap-
pearance of a dip located at exact resonance of the

212
L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
À1
0
:2 cm by manufacturer.The FDFWM phase
matching geometry [17], chosen for an ecient
signal generation, required the use of suitable op-
tics to generate three co-propagating beams trav-
elling at the corners of a square before focalisation
by a lens /f ˆ 1000 mm).The two pump beams
had the same energy, while the probe beam was
four times weaker.The interaction region was a
3
cylinder with a volume of 7:85 mm with 1 cm of
length along the beam propagation direction.
The FDFWM signal emerged from the inter-
action region, located at the centre of the 60 cm
long optical cell, as if originating from the fourth
corner of the square.The signal, after being spa-
tially ®ltered, was detected by a photomultiplier
tube connected to a Boxcar Averager /SRS245).
The dye laser scan and data acquisition were si-
multaneously controlled by a PC /Epson 386DX).
In the recorded spectrum each data point corre-
sponded to the average of signals from 10 subse-
quent laser shots.
5
5
0
2
2
Fig.1 .F DWFM spectrum of the a D ±y D transition in FeI
calculated for di€erent values of the saturation parameter S.
Details about the radiative re-normalisation model used are
given in [17].
transition frequency and increasing with S.The
model is capable of reproducing the line-splitting
of saturated DFWM lines observed in the crossing
pump beams geometry for the excitation of atoms,
including iron, despite a comprehensive descrip-
tion of the physics underlying this phenomenon
being still lacking.
In the absence of an adequate theory, the
modelling of saturation e€ects must be based on
an exhaustive experimental characterisation of
each atomic species used as active medium, in or-
der to avoid introducing errors in quantitative
FDFWM measurements.
FeI atomic vapours was generated by raising up
to 673 K the temperature of the cell ®lled with a
mixture of 10 mbar of FeꢀCO† and 190 mbar of
5
Ar.The cell was equipped with a PtꢀRh thermo-
couple to measure the gas temperature.
The e€ect of disregarding the atomic motion on
DFWM line intensities has been considered for the
present experimental parameters.The derivation
of Eq./3) does not include the washing-out e€ect
of the induced grating due to the motion of the
resonant species, which is more and more impor-
tant as temperature increases.Here the angle
h
under which the two pump beams were crossed
was 0.74°, thus resulting in a fringe spacing equal
to 23 lm.Since at 673 K the mean velocity of the
FeI atom is 350 mꢀs, which corresponds to a mean
free path of 5:25 lm during a 15 ns laser pulse, the
assumption made in neglecting the atomic motion
is justi®ed.
3
. Experimental apparatus
The laser radiation for the excitation of the FeI
0
5
5
j
j
0
a D ±y D transitions was generated by frequency
doubling in a BBOI crystal the output of a narrow-
band dye laser operating with Rhodamin B.The
dye laser was pumped by a doubled Nd:YAG laser
/
JK2000) providing up to 250 mJ per pulse at 532
4. Experimental results
nm with a repetition rate of 1 Hz.In this experi-
mental scheme laser pulses in the spectral range
between 300 and 310 nm were produced, with 15
ns of time duration and 600 lJ of energy.The
bandwidth of the dye laser is speci®ed to be
Iron pentacarbonyl vapours exhibit a strong
broad absorption band between 200 and 400 nm
[19].The thermal decomposition of Fe ꢀCO† was
5
monitored in the temperature range 297±573 K by

L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
213
measuring the transmission through the cell of a
weak laser pulse /10 lJ energy) tuned at 302.2 nm
on this broad band.The experimental data, shown
in Fig.2, indicate that at room temperature the
laser radiation is completely absorbed by
the FeꢀCO† vapours and that the transmission
5
of the laser pulse increases with temperature due to
the thermal decomposition of the FeꢀCO† mole-
5
cule, which is completed around 523 K in good
agreement with previous results [16].
5
5
0
The lineshape associated to the a D
transition has been recorded at a temperature of
73 K with I
2
±y D2
6
P
ˆ 100 lJ and I ˆ 200 lJ.Synthetic
P
lineshapes through the experimental curves have
been evaluated according to [17], the results are
5
5
0
2
2
Fig.3.Experimental lineshape /dots) of the a D ±y D tran-
sition in FeI at 673 K.Pump beam energy was 100 lJ.The
continuous curve is a theoretical simulation with S ˆ 0:93.
shown in Fig.3 for I ˆ 100 lJ.
P
The calculated Rabi frequency in this experi-
À1
mental con®guration is 0:034 cm , with S laser
beam diameter in the interaction region of 1.0 mm
excited state population lifetime c .The closest
aa
À1
agreement was achieved for cab ˆ 0:3 cm , which
and l ˆ 0:028 D as determined by the relation
ij
corresponds to S ˆ 0:93.In Fig.4 the spectrum
s
2
1
3 ꢁh e fif
recorded at I
P
ˆ 200 lJ is reported together with
lif ˆ g
;
ꢀ6†
the theoretical simulation performed with
S ˆ 1:86.The presence of a dip at the centre of the
experimental FDFWM lineshape is clearly ob-
servable.The adopted model reproduces with
good accuracy the lineshape, and the apparent
discrepancies are ascribable to the multi-mode
axial structure of the laser emission.In fact, due to
i
g
f
e if
2m x
where fif ˆ 0:024 and xif ˆ 4:09 eVꢀh are the os-
cillator strength and the frequency of the transi-
tion, respectively, and g the involved level
5
0
2
À3
degeneracy.The lifetime of the y D level is 6.5 ns
À1
[
2], which gives a value of 4:1 Â 10 cm for the
5
5
0
Fig.2 .T ransmission of a laser pulse at 302 2. nm through a cell
2
2
Fig.4.Experimental lineshape /dots) of the a D ±y D tran-
®
Transmission increases with temperature due to the thermal
lled with 10 mbar of FeꢀCO† and heated up to 600 K.
sition in FeI at 673 K.Pump beam energy was 200 lJ.The
continuous curve is the result of theoretical simulation with
S ˆ 1:86.
5
decomposition of FeꢀCO† absorbing the laser beam.
5

214
L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
the non-homogeneous energy spatial distribution
in the focal volume, regions which contribute to
FDFWM signal with di€erent degrees of satura-
tion, may coexist.It is evident that saturation in-
duces strong lineshape distortions in Fe DFWM
spectrum, which completely hinder the application
of this technique for quantitative measurements of
physical parameters at high laser intensity.
The lines intensity ratios, in both spectra of Fig.
5, show divergences from the expected values in
the approximation of optically thin medium and
thermally distributed atomic population.As a
5
matter of fact, at 298 K the population of the a D
4
5
3
level is ten times the one of the a D level, thus the
5
5
0
4
a D ±y D transition, appearing slightly weaker,
4
should be nearly 100 times stronger than the
0
5
5
Consequently, FeI FDFWM spectrum of the
0
a D ±y D line.
3
3
5
5
a D
j
±y D band was recorded at an excitation
Deviations from the expected distribution are
even more pronounced at 673 K, where the
j
energy I
P
ˆ 100 lJ, below the saturation thresh-
5
5
0
5
old.The normalised spectra recorded at 673 K and
after cooling the dissociated gas mixture at 298 K
are reported in Fig.5.From lines assignment the
a D ±y D line becomes stronger than the a
2
0
2
5
D ±y D fundamental transition.
4
4
The possible occurrence of spurious FeI
DFWM signals due to FeꢀCO† photolysis [20±22]
5
5
0
j
a D ±y D ꢀj ˆ 2; 3; 4† transitions of FeI could be
j
5
identi®ed.An unassigned broad structure, not
originating from any FeI transition, was also ob-
served.The experimental evidence that its intensity
drops to zero at room temperature suggests that it
was analysed by repeating the experiment with 1
mbar of pure FeꢀCO† at room temperature.Since
5
no DFWM signal generation was observed, it can
be concluded that the amount of FeI produced by
belongs to a hot band of some FeꢀCO†
mo-
FeꢀCO† laser photolysis and in non-thermal dis-
xˆ1;...;4
5
lecular fragments.
tribution remained below the detection limit /ꢁ10
(
a)
(
b)
5
5
0
Fig.5.Spectrum of the a D
j
±y Dj system of FeI recorded at 673 K /a) and after having cooled the vapours at 298 K /b).Lines
0
5
5
0
assignment allowed to identify the a D
relevant transitions marked by arrows is shown on the right side.
j
±y D ꢀj ˆ 2; 3; 4† transitions of FeI.A scheme of the pertinent energy levels of FeI with the
j

L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
215
ppm) of the experimental apparatus.This experi-
mental ®nding suggests that the thermally pro-
duced FeI atoms populate the excited levels
according to the Boltzmann law and that the
anomalous line intensity ratio's can be ascribed
only to the strong e€ect of pump beams depletion
and signal re-absorption.
Accordingly, the FeI DFWM spectrum mea-
sured at room temperature was used to ®t the
parameters in Eq./3).The obtained values a ˆ 1
and b ˆ 2:7 were subsequently used to predict in-
tensity ratio at 673 K.The results are compared in
Table 1 with the results of calculations performed
without ꢀb ˆ 0† and with ꢀa ˆ 1; b ˆ 2:7† absorp-
tion e€ects.The applied model provides a rea-
sonable estimate of the experimental data /within
To quantify the degree of approximation, which
the simpli®ed treatment implies, for a given pair of
a and b values, a function Gꢀx† can be de®ned as
ꢀ
ꢀ
Gꢀx† ˆ ꢀꢀ1 À 0:1x† À eÀ0:2xꢀ1 À 0:1x† ꢀ;
ꢀ
2:7
2
ꢀ
ꢀ8†
where x ꢀ0 < x < 0:5† is the fraction of atoms in an
excited level normalised to the ground state pop-
ulation.This function represents how much the
approximated correction factor disagrees from the
exact value.In the present case we have
Gꢀx† ꢁ 0:1x.Since in our experimental conditions
we observed at most x ꢃ 0:33; Gꢀx† can only span
between 0 and 0.033, thus con®rming the validity
of the proposed simpli®ed approach.
1
0% accuracy).Possible local temperature gradi-
5
. Discussion and conclusions
ents inside the cell, likely to occur in the proximity
of the optical windows, can be responsible for the
observed di€erences.
5
5
0
j
The spectrum of the a D
j
±y D
0
system of FeI
recorded with a non-linear spectroscopic technique
DFWM) based on scattering from laser induced
gratings in a gaseous medium is reported for the
rst time.FeI atomic vapours were produced by
heating up to 673 K iron pentacarbonyl vapours.
The ratio between the length of the region in which
signal is generated and the length of the optical cell
The physical meaning of the b parameter de-
serves further attention.In our experimental set-
up the pump beams, with an intensity very close to
the saturation value, show a linear attenuation
di€erent from both the weak probe and the signal
beams, which obey an exponential absorption law.
It follows that the exact factor, which should be
used in Eq./3) is:
/
®
®
lled with FeI vapours being nearly equal to
À2
1
:6  10 , strong absorption e€ects on DFWM
2
À2aNi
signal were observed in spectra at 673 K and room
temperature.This observation is consistent with
the analysis of Ewart [23], who demonstrated that
Na vapours, in the same pressure range as used in
our experiment, becomes optically thick over a
length of about 5 cm.With the purpose to dem-
onstrate that absorption e€ects can strongly alter
the expected line intensities, a simple model was
ꢀ
1 À aN † e
:
ꢀ7†
i
It is impossible to estimate a with simple mathe-
matical approaches, therefore the composite ab-
sorption e€ects have been replaced with the
approximating factor of Eq./3).This procedure
greatly simpli®ed the calculations despite the need
of an additional parameter ꢀb†.
Table 1
5
5
0
j
Experimental and calculated lines intensity ratio of a D
j
±y D DFWM spectrum of FeI /at 298 and 673 K)
b
b
Ratio designa-
a
tion
Measured
/T ˆ 298 K)
Calculated
/T ˆ 298 K)
Calculated with Measured
c
Calculated
/T ˆ 673 K)
Calculated with
c
correction
correction
/
/T ˆ 673 K)
T ˆ 298 K)
/T ˆ 673 K)
1.68
1.33
1.27
À2
À5
À2
À3
À2
I
I
I
22=I44
33=I44
22=I33
4 Â 10
7 Â 10
3:3 Â 10
1.54
1.14
1.39
4 Â 10
À3
1.6
2:5 Â 10À2
4 Â 10
1.54
2:3 Â 10À2
4 Â 10
À2
1:7 Â 10
0.1
a
b
c
5
5
0
j
j
In this notation the intensity of the a D ±y D line is designed as Ijj.
Value calculated assuming thermal distribution and no absorption e€ects /optically thin medium).
Value calculated with Eq./3), a ˆ 0:1 and b ˆ 2:7 /optically thick medium).

216
L. De Dominicis et al. / Chemical Physics Letters 348 ꢀ2001) 209±216
proposed to reproduce the experimental ®ndings.
This model gives reasonable agreement with data
References
[
[
1] J.Sugar, C.Corliss, J.Phys.Chem.Ref.Data 14 /Suppl.2)
1985).
/
awkward numerical calculations.
10% accuracy), without the need of introducing
/
2] T.R. O'Brian, M.E. Wickli€e, J.E. Lawler, W. Whaling,
J.W. Brault, J.Opt. Soc. Am. B 8 /1991) 1185.
5
5
0
The a D ±y D transition was chosen to moni-
2
2
tor the saturation e€ect on FeI DFWM lineshape
at 673 K since among the recorded lines it is less
a€ected by pump beams depletion.The evidence of
line-splitting due to saturation indicates that
atomic vapours are targets which can signi®cantly
contribute to shed light on the physical nature of
the saturation line-splitting in DFWM.It is
worthwhile to mention that in our previous ex-
[3] F.Orsitto, M.Borra, F.Coppotelli, G.Gatti, Rev.Sci.
Instrum.66 /1995) 597.
[
[
[
4] P.Ewart, S V. .O'Leary, Opt.Lett.11 /1986) 279.
5] T.Dreier, D J. .Rakestraw, Appl.Phys.B 50 /1990) 479.
6] S. Williams, R.N. Zare, L.A. Rahn, J. Chem. Phys. 101
/
[7] P.Ljungberg, O.Axner, Appl.Phys.B 63 /1996) 69.
1994) 1093.
[8] D.H. Bloom, P.F. Liao, N.P. Economou, Opt. Lett. 2
/
9] N.Tan-no, T.Hashimiya, H.Inaba, IEEE J.Quant.
1978) 58.
[
periment dealing with saturated DFWM on NO
molecule no evidence for such a dip was observed
2
Electron.QE 16 /1980) 147.
10] G.P. Agrawal, A. Van Lerberghe, P. Aubourg, J.L.
Boulnois, Opt.Lett.7 /1982) 540.
[
[
24].The induced dipole moment in FeI
À2
ꢀ
ꢁ 3  10 D† is considerably lower than in NO
2
[11] G.Grynberg, M.Pinard, P.Verkerk, Opt.Commun.50
1984) 261.
12] G.Grynberg, M.Pinard, P.Verkerk, J.Physique 47 /1986)
17.
/
ꢀ
ꢁ 0:4 D†, this di€erence is rather general in
[
[
comparing atoms with molecules and in our
opinion may help towards a deeper comprehension
of their di€erent saturation behaviour.Finally it
has to be pointed out that we choose Ar as bu€er
gas because it has a lower quenching cross-section
6
13] Z.Karny, R.Naaman, R N. .Zhare, Chem.Phys.Lett.59
/1978) 33.
[14] D.W. Trainor, S.A. Mani, J. Chem. Phys. 68 /1978) 5481.
[
15] J.T. Yardley, B. Gitlin, G. Nathanson, A.M. Rosan, J.
Chem.Phys.74 /1981) 370.
for FeI with respect to N [25], thus allowing to
2
[
16] S.Pignataro, F .P .Lossing, J.Organomet.Chem.11 /1968)
71.
minimise the contribution to signal due to thermal
grating formation.Nevertheless, by suitably de-
laying the probe beam, signi®cant information on
the collisional energy redistribution processes in
FeI, as a function of di€erent bu€er gases, can be
obtained from DFWM signals.
5
[17] B.Attal Tretout, H.Bervas, J P. .Taran, S.Le Boiteux, P.
Kelley, T.K. Gustafson, J. Phys. B 30 /1997) 497.
[
[
18] W.P. Brown, J. Opt. Soc. Am. 73 /1983) 635.
19] M.Kotzian, N.R o sch, H.Schr o der, M.C. Zerner, J. Am.
Chem.Soc.111 /1989) 7687.
[
20] S.A. Mitchell, P.A. Hackett, J. Chem. Phys. 93 /1990)
7
21] K. Lee, J.S. Goo, J.K. Ku, Chem. Phys. Lett. 244 /1995) 213.
813.
[
Acknowledgements
[22] L.B ~a nares, T.Baumert, M.Bergt, B.Kiefer, G.Gerber, J.
Chem.Phys.108 /1998) 5799.
[
[
23] P.Ewart, S V. .O'Leary, J.Phys.B 17 /1984) 4595.
24] R.Fantoni, L.De Dominicis, M.Giorgi, R B. .Williams,
Chem.Phys.Lett.259 /1996) 342.
We would like to thank S.Pignataro, B.Attal
Tretout and G.Gatti for helpful discussions.The
contribution of M.Giorgi at the early stage of the
work is gratefully acknowledged.
[
25] S.A. Mitchell, P.A. Hackett, J. Chem. Phys. 93 /1990)
7813.