Spectrochimica Acta Part B 66 (2011) 671–680

Contents lists available at SciVerse ScienceDirect

Spectrochimica Acta Part B

j ourna l homepage: www.e lsev ie r .com/ locate /sab

Study of near infra red femtosecond laser induced particles using transmissionelectron microscopy and low pressure impaction: Implications for laser ablation–inductively coupled plasma-mass spectrometry analysis of natural monazite

François-Xavier d'Abzac a,⁎, Anne-Magali Seydoux-Guillaume a, Jérôme Chmeleff a,Lucien Datas b, Franck Poitrasson a

a GET - UMR 5563 CNRS – Université de Toulouse – IRD – OMP, 14 avenue Edouard Belin, 31400 Toulouse, Franceb TEMSCAN – CIRIMAT – Université de Toulouse, 118 route de Narbonne, 31400 Toulouse, France

⁎ Corresponding author.E-mail address: dabzac@lmtg.obs-mip.fr (F.-X. d'Abz

0584-8547/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.sab.2011.09.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 6 January 2011Accepted 29 September 2011Available online 6 October 2011

Keywords:Laser ablationAerosolParticle countingICP-MSChemical fractionation

The characteristics of infra red femtosecond laser-induced aerosols are studied for monazite (LREE, Th(PO4)) ablation and correlations are established with inductively coupled plasma-mass spectrometry(ICP-MS) signals. Critical parameters are tested within wide ranges of values in order to cover theusual laser ablation -ICP-MS analysis conditions: pulse energy (0.15bE0b1 mJ/pulse), pulse width(60bτb3000 fs), ablation time (t≤10 min) and transport length (l≤6.3 m). Transmission electron mi-croscopy reveals that aerosols are made of agglomerates of ~10 nm particles and 20–300 nm phosphorusdepleted condensed spherical particles. These structures are not affected by any laser ablation parame-ter. Particle counting is performed using electronic low pressure impaction. Small changes on particlesize distribution are noticed. They may be induced either by a peak of ablation rate in the first 15 s athigh fluence (larger particles) or the loss of small particles during transport. We found a positive corre-lation between I (ICP-MS mean signal intensity in cps) and N (particle density in cm−3) when varying E0and t, suggesting that N is controlled by the irradiance (P0 in W·cm−2). Elemental ratio measurementsshow a steady state signal after the initial high ablation rate (mass load effect in the plasma torch) andbefore a late chemical fractionation, induced by poor extraction of bigger, early condensed spherical par-ticles from the deepening crater. Such chemical fractionation effects remain within uncertainties, how-ever. These effects can be limited by monitoring E0 to shorten the initial transient state and delay theattainment of an unfavorable crater aspect ratio. Most adopted settings are for the first time deducedfrom aerosol characteristics, for infra red femtosecond laser ablation. A short transport (lb4.0 m) limitsthe agglomeration of particles by collision process along the tube. Short τ is preferred because of higherP0, yet no benefit is found on ICP-MS signal intensity under 200 fs. Under such pulse widths the increasedparticle production induces more agglomeration during transport, thereby resulting in higher mass loadeffects that reduce the ionization efficiency of the plasma torch. Thus, pulse energy must be set to get anoptimal balance between the need for a high signal/background ratio and limitation of mass load effectsin the plasma torch.

ac).

rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Laser ablation coupled to inductively coupled plasmamass spectrom-etry (LA–ICP-MS) is by now fairly widespread in the field of earth sci-ences. Such systems allow direct in-situ analysis of solid [1–5] or liquid[6] samples. Trace elements can be detected and quantified with highprecision down to a resolution of a few micrometers [7,8], and couplingthe laser to amulti-collection device (MC–ICP-MS) allows direct isotopicanalysis, thus avoiding complex chemical preparations [9–11]. The

technique has been improved continuously [2,12],with the use of shorterwavelengths down to near-UV (Ultra Violet) [13,14], ultra short pulses(b1 ps) [15,16], improved ablation cell [17] and ICP torch [18] technolo-gy. Even though satisfactorymeasurements cannowbeperformed, someof the fundamental issues related to the physics of laser ablation still haveto be addressed. Recently, Horn and von Blanckenburg [19] havewell de-scribed the current limitations of accuracy in laser ablation through theinvestigation of elemental and isotopic fractionation. The phenomenonis usually observed as a temporal deviationwithin raw (MC-) ICP-MS sig-nals [3,20]. Issues concerning non-stoichiometric sampling due tomatrixdependant ablationmechanisms are often highlighted [12]. These distur-bances have been almost suppressedwith the use of femtosecond pulses[20,21], due to different laser/matter interactions [22]. Thermal effects

Table 1Infra red femtosecond laser ablation system and ICP-MS operating conditions.

Laser apparatus Amplitude technologies “Pulsar 10”

Wavelength 800 nmRepetition rate 5 HzPre-ablation laserwarm up

45 min

Pulse duration 60–3000 fsApertured beamdiameter

100 mm

Focusing objective 3.5X, focus length 50 mmSpot size 60–100 μmIncident pulseenergy

0.15–1 mJ/pulse

Ablation cell 25 cm3, cylindrical

ICP-MS model Agilent 7500Ce

Forward power 1550 WGas flows:

Plasma (Ar) 15 l/minAuxiliary (Ar) 0.9 l/minCarrier (He) 0.56 l/minMake up (Ar) 1.16 l/minShield torch Used for all analyses

Data acquisition parameters

Data acquisitionprotocol

Time-resolved analysis

Scanning mode Peak hopping, 1 point per peakDetector mode Pulse counting, dead-time correction appliedMasses scanned 29Si, 31P, 139La, 143Nd, 146Nd, 202Hg, 204Pb,

206Pb, 207Pb, 208Pb, 232Th, 238U, 246ThODwell time 100 msQuadrupole setting time ca. 5 msData acquisition (s) 180 s (20 s gas blank, up to 140 s ablation)

or 600 s (20 s gas blank and 500 s ablation)

672 F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

are considerably reduced when compared to those produced by nano-secondpulses [23] and are limited to the unavoidable temperature gradi-ent due to the laser induced plasma on the bulk sample [24].

Recent attempts to study the laser induced aerosols have led tothe use of transmission electron microscopy (TEM) in order to takeadvantage on the very high resolution of this instrument to obtainqualitative information about the particles chemical composition.Fliegel et al. [25] have collected particles produced by UV–ns-LA aswell as particles remaining after introducing a milled powder of nat-ural zircon crystal into the plasma torch (ICP). Unfortunately, thispowder does not match the real laser induced aerosol, in terms ofmorphology and chemical composition, and the resulting particlesare not what one might expect from the passage of laser inducedaerosols through the ICP torch, even after some incomplete ionizationor disturbing mass load effects. Subsequently, Glaus et al. [26] havecompared aerosols produced by UV–ns-LA and (UV, IR)-fs-LA, makinga direct correlation between particles morphologies and composition,and corresponding signals. Nevertheless, their approach remainedqualitative as it did not take into account the crucial informationabout the quantity of particles produced and their size distribution.

In a previous work [27], we have studied experimentally the fun-damental aspect of femtosecond ablation mechanisms and particlesgeneration processes. Our results have been compared to existingtheoretical studies and it has been possible to refine an analyticalmodel of particles generation processes, based on the main ablationmechanisms, namely vaporization and fragmentation [22]. Hence,the existing model of gas to particle conversion [28], that creates ag-glomerates and spherical particles, has been enhanced by the study ofchemically complex natural and synthetic samples. It has highlightedthe importance of chemical segregation of elements inside theexpanding plasma plume as a function of time and its possible impacton subsequent ICP-MS measurements. The present study builds on

our previous work. It focuses on the link between these aerosolsand the ICP-MS signals, using transmission electron microscopy ana-lyses as well as a counting device similar to the one used by Koch et al.[29], giving access simultaneously to the amount of produced parti-cles and their size distribution. Potential effects on the ICP-MS signalswill be highlighted by varying critical parameters such as pulse ener-gy, pulse width, ablation time and transport length. Results will helpto propose optimal analytical conditions for IR-fs-LA–ICP-MS analysis,based on the numerous existing studies dealing with such issues onvarious types of systems [30,31].

2. Experimental

2.1. Laser ablation

We used a Ti:sapphire femtosecond system operated at its funda-mental wavelength of λ=800 nm (Pulsar 10, Amplitude Technolo-gies, Evry/Seine, France). The system has been described in previousworks [21,32,33]. Specifications are summarized in Table 1. It is ableto provide E0=12 mJ pulses at f=10 Hz with a pulse duration ofτ=60 fs. The laser beam is aimed at the sample through a modifiedoptical microscope (Olympus BX51, Hamburg, Germany), equippedwith a XYZ motorized sample stage. The beam is focused onto the tar-get by a 50 mm focusing lens. The aperture is larger than the beaminitial diameter (~9 mm). Ablation is conducted without defocusingand no masks were used to reduce the crater diameter, of about90 μm.We used a cylindrical 25 cm3 PTFE (polytetra-fluoro-ethylene)ablation cell, mounted with a quartz window, and helium as carriergas. Behavior and reproducibility of ablation events are similar tothese quantified in Courtieu et al. [34] on a synthetic quartz sample.

2.2. Sample

Monazite is a monoclinic phosphate enriched in light rare earth el-ements [35] (LREE, Th(PO4)) commonly occurring as accessory phasein a wide variety of rocks. It is composed notably of P (27.81 wt.%), Si(1.42 wt.%), La (14.51 wt.%), Ce (30.59 wt.%), Pr(3.14 wt.%), Nd(10.2 wt.%), Sm (2.05 wt.%), Th (6.92 wt.%), U (0.13 wt.%) and Pb(0.16 wt.%). That is why it is considered as a good alternative to zirconfor U–Th–Pb geochronology [36]. It has been extensively studied forits annealing properties [37] and low solubility [38] for nuclearwaste storage purposes [39]. The Moacyr specimen, from the pegma-tite of Itambe (Brazil), has been used throughout this study. Charac-terization of its chemistry [40], microstructures and crystallinityunder various temperature conditions [37,39] have been previouslyconducted. Laser induced damage (crater morphology and ablationmechanisms) [41] and aerosols (particles generation processes) [27]have been studied using the same sample and the specimen hasbeen used as reference material for LA–ICP-MS analysis [8].

2.3. Collection

Collections of particles were achieved using an in-house-builttrapping system [27] consisting of a modified 5 ml syringe equippedwith an aluminum stub, usually used as sample holder for scanningelectron microscopy. A TEM sample holder was placed on the stub,to be impacted by aerosols. Before use, all syringes were cleaned24 h in an HCl solution (1 mol·l−1) and rinsed 24 more hours inmilliQ water. Final mounting was realized in a class 10000 room toavoid contaminations on the collecting surfaces.

2.4. Observation

Observations were conducted at the TEMSCAN service of the Uni-versity of Toulouse, using a JEOL 2100F (Tokyo, Japan) transmissionelectron microscope operated at 200 keV. Themicroscope is equipped

673F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

with a field emission gun (FEG) as electron source, a high annulardark field (HAADF) STEM (scanning transmission electron micro-scope) detector, an energy dispersive X-ray analyzer (EDX) systemfrom Princeton Gamma Tech (Princeton, NJ, USA), and a CCD cameraGatan (Pleasanton, CA, USA) (1Kx1K). X-ray analyses are reliablefrom sufficiently high concentration of element, empirically deter-mined to be ~1 wt.%.

2.5. Counting

Particle counting and sizing have been conducted using an elec-tronic low pressure impactor (ELPI, DEKATI, Tampere, Finland). Thesystem has been described in Kuhn et al. [42] and used for the samepurpose by Koch et al. [29]. Particles in the gas flux are electricallycharged in a corona cell, then impacted on 12 successive stages, as afunction of their charges, themselves dependant on the effective con-tact surface. Test acquisitions were made without corona cell and nosignal was recorded. Thus, no detectable interfering electrostaticcharge is created at aerosols surface before particles reach the count-ing device and they can be considered as neutral. Aerosols are mea-sured from da=0.007 μm to 6 μm (aerodynamic diameter) under a10.0 l·min−1 gas flow rate. Our experimental 0.5 l·min−1 carrier gasflow rate has then been completed by a filtered air entrance. In conse-quence, a dilution factor of 20 has been taken into account for dataconversion. One must keep in mind that the measured da is not repre-sentative of the real particle diameter. According to DeCarlo et al. [43],da does not correspond to the volume equivalent diameter (dve) in ourexperimental conditions because it does not take into account the par-ticle density (ρp) and the shape factor (χ). Considering a mean densityof monazite (ρp=5.5), da overestimates dve. Unfortunately, we cannotapply a single ρp and χ to our aerosols, because they are composed ofboth high χ (clusters) and low χ (round beads) particles which arechemically different (different ρp), but nevertheless belong to a singlemono-modal size distribution [27]. Results should then be consideredas globally overestimated, but valuable for comparisons. Acquisitionsare performed over t=10 min long ablations (3000 shots atf=5 Hz), after completing a 40 s blank measurement/backgroundsuppression. Uncertainty on each size class is defined as a standard de-viation (σ) and depends on the particles concentration (C'i). The latteris the normalized value of particles density on every impactor stage (i),calculated from themeasured (Ci) and the total particles density (ΣiCi)following Eq. (1):

C0i ¼ Ci=∑iCi ð1Þ

Relative standard deviations (RSD) for C'i have been empirically calcu-lated from daily acquisitions on NIST610 synthetic glass standard ref-erence material [44] (8 values over 8 days): σc=30.0%RSD forC'ib10.0%, σc=18.0%RSD for 10.0%bC'ib20.0% and σc=7.0%RSD forC'iN20.0%. We consider these as the most robust uncertainties for C'i,as (i) they are determined for a homogeneous synthetic sample and(ii) they are calculated using a large amount of acquisitions over thewhole campaign. Values of σ are rounded up to the nearest integer.Values of C'i under 8·10−4 (most depleted fractions) are not takeninto account, as well as the aberrant uncertainty of 207%RSDmeasuredon the smallest fraction. The uncertainty for total particle density isalso calculated from NIST610 acquisitions, as σ of the total amount ofparticles. For 8 values over 8 days, σN=16.6%RSD.

2.6. Sampling protocols

The basic laser settings are: E0=1mJ/pulse, τ=60 fs, f=5 Hzrepetition rate, l=0 or 1.5 m (ELPI connection) and helium as carriergas. Some of these and other parameters were varied to see their ef-fect: transport distance (l), ablation time (t) which is associatedwith crater development, pulse energy (E0) and pulse duration

(τ).All particle collections were performed by placing the trap at thecell output (l=0), through 3 runs of 120 s ablations, in order to col-lect as many observable particles as possible. The influence of t isalso studied through ELPI acquisitions, because density and size dis-tributions are measured as a function of time, so consequently andqualitatively as a function of depth.

2.7. ICP-MS

Ablation products were analyzed by an Agilent 7500ce quadrupoleICP-MS. The ICP parameters (Table 1) were optimized to obtain themaximum sensibility over scanned masses. The detector was set in“analog” mode for all masses to avoid all “pulse/analog” correctionfactors potentially degrading the precision of measurements and sig-nal stability. Moreover, the “pulse” mode cannot be used because ofthe high concentration of most of the measured masses. Thoriumoxide (ThO, mass=248) was constantly checked and ThO/Th ratio re-mains around 0.5%. Twelve other masses were scanned (Table 1)under a time resolved analysis protocol. Ablation was triggered after20 s background signal, and stopped after 140 s, except for one mea-surement that has been carried on over 10 min (20 s background+580 s ablation). All signals are reported in counts per second (cps).The temporal uncertainty, representing the transient signal stabilityon a single element acquisition, has been calculated from a slidingstandard deviation, representing an arithmetic progression of stan-dard deviations σx calculated from 5 replicates (1 s of ablation) anddefined as follow (Eq. (2)):

uσn¼ uσ0

þ nr ð2Þ

where uσ0 is the first term calculated for the first 5 replicates, n=R-4with R is the total number of replicates, and r=1 the common differ-ence representing a shift of 1 replicate for each uσn iteration. The effec-tive temporal uncertainty is the mean value of uσx, calculated overn=2500 iterations (500 s) on a random signal, and equal to 8%RSD.The temporal relative standard deviation (TRSD%) as defined by Gonza-lez et al. [45] is used for signals implying elemental ratios.

3. Results

3.1. Morphologies and composition of particles

Fig. 1a and b shows two lowmagnifications bright field TEM imagescorresponding to τ=60 fs and τ=3000 fs respectively, both withE0=1mJ/pulse. High magnifications observations show two types ofparticles: agglomerates of 10–15 nm particles (Fig. 1d) with phospho-rus and silicon dominating thorium and LREE, and spherical particles(Fig. 1c), which are almost free of phosphorus and enriched in thoriumand LREE [27]. Transmission electron microscopy reveals that the sizesof the latter spread from diameters of dp~20 nm to several hundredsof nanometers. These morphologies are unchanged whatever the set-tings tested. On all samples, apparent density and particle size distribu-tion observed by TEM are not representative of the total ablationproduct and it must be quantitatively characterized by ELPI.

3.2. Transport distance (l)

Fig. 2 reports ELPI results obtained with the 3 different tubelengths after the ablation cell. It shows that particles density decreasesby nearly 50% between 1.5 m and 3.9 m, and then holds for longerdistance (6.3 m). Furthermore, size distributions (Fig. 2, lower graph-ic) show that particles collected at 1.5 m are enriched in small sizeparticles (~30% of 7bdab55 nm) relatively to the recorded stable dis-tribution at 3.9 m and 6.3 m (~10% of 7bdab55 nm). The relative dif-ference is well visible for these two later acquisitions, for which thefraction of particles such as daN94 nm has increased by 20%.

0.5µm0.5µm

τ=60fs E0=1mJ/pulse τ=3000fs E0=1mJ/pulsea b

50nm 50nm

c d

PSi

O

Th

Cu

Cu

La

CeNd

Sm

Gd

0

0.2

0.4

0.6

0.8

1

1.2

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0KeV

I[X]/I

[Cu]

Th

Th

0

0.5

1

1.5

2

2.5

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0KeV

I[X]/I

[Cu]

P

SiO

Th

CuLa

CeNdSm

Gd

Th

Fig. 1. Transmission electron microscopy bright field images of particles produced during monazite ablation by high energy (E0=1mJ/pulse) pulses combined with ultra-short(τ=60 fs) (a) and short (τ=3000 fs) (b) pulse widths. For all laser settings, two types of particles are observed: (c) round shaped specimens enriched in Th and LREE and (d)agglomerates with P dominant. EDX spectra are normalized to the intensity of Cu, which is the main sample grid component.

0

0.5E+06

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

3.5E+06

4.0E+06

4.5E+06

0 100 200 300 400 500 600

t (s)

N(c

m-3

)

1.5m3.9m6.3m

1.5m3.9m6.3m

C'i

(%)

0

0.1

0.2

0.3

0.4

0.5

0.01 0.1 1

da (μm)

Fig. 2. Electronic low pressure impaction results showing the influence of transport lengthon particle density (upper graphic) and particle size distribution (lower graphic). Lasersettings are E0=1mJ/pulsewith τ=60 fs and f=5Hz. Error bars are 1σ standard deviationcalculated from 8 acquisitions over 8 days on NIST610. Note that the signal falls for almost50% within the first 3.9 m and this is partially related to a lower relative abundance ofsmall particles falling from ~30% to ~10%.

674 F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

3.3. Aerosol characteristics and ICP-MS signals

Fig. 3 combines ELPI and ICP-MS signals, measured over 10 minwith identical laser settings and the same tube length (6.3 m). Parti-cle production rapidly increases to a maximum density in the firsttwenty seconds of ablation, and decreases until the laser shuts off(t=500 s) (Fig. 3, upper graphic). The associated ICP-MS signal,shown for 143Nd, undergoes a peak of intensity before sharply losingabout 80% of its highest value within 100 s. Afterward, the signal re-mains above the background intensity by one order of magnitude,for the whole ablation time. To investigate the correspondence be-tween the particle production and ICP-MS signal, we monitored the31P/143Nd ratio over the whole ablation time (lower part), because31P is mostly enriched in agglomerates and 143Nd in beads.

Analyses show a transient signal during the first 15 s and a steadyvalue (0.093±0.007; σ=4.7%TRSD) during the next 45 s (1), followedby an unstable signal progressively increasing (0.106±0.019;σ=11.4%TRSD) (2) before reaching a steady value until ablation ends(0.111±0.024;σ=17.7%TRSD) (3). Finally, themean 31P/143Nd ratio in-creases by 14% from areas (1) to (3), showing a relative enrichment in31P. However uncertainties show that this increase is not significant.

The first 120 s was also examined in detail, as it is usually theexploited time range for LA–ICP-MS measurements. Fig. 4 shows206Pb/238U ratio on this time range. Time slices are represented in associ-ation with their respective particles size distribution. Area (a) shows amonomodal pattern centered on 94bdab157 nm. This fraction repre-sents 50% of the total amount of particles. Area (b) shows a shift tolower sizes. The monomodal pattern is now centered on 55bdab94 nmand larger size fractions are reduced to ~33% of the total amount of aero-sol. Area (c) shows no change in size distribution. The latter has beenrecorded within the remaining 8 min of ablation. The 206Pb/238U ratiovariations are similar to 31P/143Nd. We measure 206Pb/238U=0.084±0.006 (σ=3.7%TRSD) in area (b), and a higher value of 0.095±0.018(σ=12.9%TRSD) in area (c), showing a chemical fractionation of ~14%from (b) to (c), although remaining within uncertainties.

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

0 200 300 400 500 time (s)0

N (

cm-3

)

Inte

nsity

(cp

s)

0.05

0.1

0.15

0.2

0.25

0.3

0

2E+08

4E+08

6E+08

8E+08

1E+09

1.2E+09

143Nd

N

31P

/143 N

d

σ

σ

1 2 3

100

0.05

0.1

0.15

0.2

0.25

0.3

31P

/143 N

d

Fig. 3. Combination of ELPI and ICP-MS 146Nd acquisitions over 10 min of ablation atτ=60 fs, E0=1mJ/pulse, f=5 Hz and l=6.3 m. Note that ICP-MS signal loses intensi-ty much faster than particles production. The lower part represents a measured and aschematic 31P/143Nd signal illustrating the 3 different stages observed along the signal:(1) a steady state with a mean value of 0.093±0.007 (4.7%TRSD) after the transientinitial peak of ablation, correlated to the high peak of signal, (2) a transient ratio of0.106±0.019 (11.4%TRSD) marking a first loss of stability and chemical fractionationcorrelated with the end of the transient ICP-MS signal, and (3) a constant unstablestate at 0.111±0.024 (17.7%TRSD) with a slightly chemically fractionated signal.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.01 0.1 1 10 0.01 0.1

0.4

0.45

da

C'i

(%)

0400 20

206 P

b/23

8 U

a

a b

b

0.05

0.1

0.15

0.2

0.25

0.3

Fig. 4. Comparison of 206Pb/238U signal and particle size distributions measured by ELPI recorrelated with a large particles (94 nmbdab157 nm) enriched aerosol stabilized after 20lower size aerosol (55 nmbdab94 nm), before a progressive loss of stability and accuracy (

675F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

3.4. Influence of E0 and τ

Fig. 5 shows a data compilation obtained by particle counting andICP-MS measurements while varying pulse width (τ). Fig. 5a showsthe measured density of particles (N in cm−3) for all pulse durationtested, at a distance of 1.5 m from the cell output which providesthe highest quantities of particles. After 40 s of background recording,ablation starts and N strongly increases until a maximum densityshortly after t=100 s, in all cases. Then, N shows a decreasing signaleventually falling under the limit of detection. The raw 238U ICP-MSsignals (Fig. 5b) have a similar evolution. Mean particles production(in cm−3·s−1) and mean ICP-MS signal of 238U, defined as the inte-grated intensity of the whole acquisition divided by the acquisitiontime (in counts per second), are correlated in Fig. 5c. Uncertaintieson both signals are described above and plotted as 1σ.

The general trend of signal intensity increasing when τ shortens isconfirmed, even if a closer look indicates that ICP-MS signal reaches amaximum for τ=200 fs and below. On the widest range of pulse du-rations tested (τ=60 fs to τ=3000 fs) the aerosol production is di-vided by 5 while the ICP-MS signal is divided by 2.5. The mean sizedistribution, measured over the whole ablation, is unchanged andcorresponds to the middle transient distribution shown in Fig. 5d.

Fig. 6 is analog to Fig. 5, but dedicated to the influence of pulse ener-gy (E0). Fig. 6a records a global increase of particle density with increas-ing E0, followed by a similar increase on ICP-MS signal (Fig. 6b). Thecorrelation between ELPI and ICP-MS signals suggests a linear trend,at least from 0.3 mJ upwards (Fig. 6c). Error bars are 1σ and representthe same uncertainty as for the pulse duration study. Themean size dis-tribution (Fig. 6d) remains unchanged, except for E0=0.15 mJ/pulse,for which the 94bdab157 nm fraction is increased by ~5% and thefine fraction (28bdab55 nm) is reduced by ~10%. Nevertheless, it re-mains monomodal and centered on 55bdab94 nm.

4. Discussion

Particles morphologies and compositions are not affected by anyparameters probably because in the sub-picosecond regime, ablationmechanisms [22,27,46] andparticles generation processes [27,28,47]are unchanged. This may also explain why particle size distributionsshow no significant differences over most experiments. However,

1 10 0.01 0.1 1 10

(μm)

10060 80 120

time (s)

c

c

latively to three areas determined by the signal behavior (Fig. 4). (a) A transient ratios and followed by (b) a steady state of smooth signal during 30 s, associated with ac) with unchanged particle size distribution.

0

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

3.5E+06

4.0E+06

4.5E+06

5.0E+06

0 100 200 300 400 500 600

60fs100fs200fs500fs1000fs3000fs

da (μm)

0

0.1

0.2

0.3

0.4

0.5

0.01 0.1 1

60fs100fs200fs500fs1000fs3000fs

20 30 40 50 60 70 80 90

t (s) t (s)

Inte

nsity

(cp

s)

1000

0.5E+06

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

3.5E+06

0

5000

10000

15000

20000

25000

0 5.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06

60fs100fs200fs

500fs

1000fs

3000fs

60fs100fs200fs500fs1000fs3000fs

N (

cm-3

)

iΣiNi/t (cm-3.s-1)

60fs100fs200fs500fs1000fs3000fs

C'i

(%)

60fs100fs200fs500fs1000fs3000fs

d

60fs100fs200fs

500fs

1000fs

3000fs

Inte

nsity

(cp

s)

c

60fs100fs200fs500fs1000fs3000fs

ba

Fig. 5. Electronic low pressure impaction and ICP-MS results showing the influence of laser pulse width (τ) during monazite ablation at E0=1 mJ/pulse on (a) particles density N(cm−3) and (b) ICP-MS signal of 238U as a function of time (t); (c) ICP-MS intensity (238U in cps) as a function of particles mean flow rate (ΣiNi/t in cm−3·s−1) and (d) particle sizedistribution where C'i (%) is the normalized amount of particles. Error bars in are 1σ standard deviation, calculated from 8 acquisitions over 8 days for particle size distribution data(c) and 1σ standard deviation calculated from a sliding uncertainty (d) defined in Section 2.7. Note the unchanged particle size distribution whatever the pulse width.

676 F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

aerosol density is significantly affected by system setting variationsand the following explanations may be proposed:

(1) Loss of particles during transport has been extensively dis-cussed and explained by deposition under electrostatic orgravitational forces occurring along the tube [48]. Neverthe-less, such processes cannot explain entirely the very strongdecrease of particle density observed in the present study(Fig. 2). The relative loss of small aerosol (dab55 nm) relativeto larger particles (daN94 nm) between 1.5 m and 3.9 m(Fig. 2), lead us to suggest agglomeration processes occurduring transport. This would be favored by the high collisionprobability due to inhomogeneous dispersion of aerosols inthe vortex-like geometry of the transport flow [49].

(2) The influence of pulse duration on accuracy and repeatabilityhas been clearly shown. These parameters remain unchangedwhen pulse width is set under 300 fs at the laser output [33],and can be significantly degraded above this value [21]. Thepresent study confirms this. This may result from a loweraerosol production (Fig. 5c), probably induced by (i) a fall ofirradiance (P0 in W·cm−2) at equivalent fluence (F0 inJ·cm−2) because of the lengthening of pulse duration(P0=F0/τ) and (ii) thermal diffusion inside the sample [50].Nevertheless, melting effects should remain limited even atthe longest pulse width of our study (3000 fs). Accordingly,we have not observed any melt droplet during this studyand we have shown that particles grow in the laser inducedplasma [27,28]. This is confirmed by the unchanged particlesize distribution whatever the pulse width. Both the decreaseof irradiance and the increase of thermal diffusion durationdeteriorate the initial multiphoton absorption of the targetand at the same time the following avalanche ionization pro-cesses [30,51], leading to a lower ablation efficiency.

Moreover, this interpretation may also explain the exponen-tial decrease of particle production,which follows a trend sim-ilar to the loss of irradiance when lengthening pulse duration(Fig. 7). It can then be related to the increase of signal withlowering τ from 3000 to 200 fs (Fig. 5c), and we also showthat a maximum response from the ICP-MS is reached forτ=100 fs. Further lowering of τ greatly increases the particleproduction, probably by themean ofmore efficient avalancheionization processes, but there is apparently no further benefitfor ICP-MS signal intensitywith our experimental setup. Inter-estingly we also found in previous work that accuracy andrepeatability do not improve notably when lowering pulsewidths under 200 fs [21,33]. As discussed above (i), producingmore particles should lead to increased agglomeration pro-cesses during transport. More massive particles are eitherlost via deposition or incompletely ionized in the ICP.We could therefore attribute the maximum response of theICP-MS detector to an increase of themass load in the plasmatorch, declining its ionization efficiency. The increase of parti-cle mean size is not visible on Fig. 5 because ELPI is placedtoo close to the ablation cell (l=1.5 m) for aerosols toagglomerate significantly.

(3) Pulse energy has also an influence on the ablation rate thataffects the evolution of particle density and ICP-MS signal.Indeed, previous authors [52,53] have described the ablateddepth per shot (z) as a function of the applied fluence (F)and energy penetration (leff

−1):

z ¼ l−1eff ln F=Fthð Þ ð3Þ

where Fth is the threshold fluence. This is in agreement withour results as particle density follows a logarithmic trend as

1mJ0.8mJ0.6mJ0.3mJ0.15mJ

1mJ/pulse0.8mJ/pulse0.5mJ/pulse0.3mJ/pulse0.15mJ/pulse

0.15mJ

0.3mJ

0.6mJ

0.8mJ

1mJ

1mJ0.8mJ0.6mJ0.3mJ0.15mJ

0

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

3.5E+06

4.0E+06

4.5E+06

5.0E+06

0 100 200 300 400 500 600

da (μm)

0

0.1

0.2

0.3

0.4

0.5

0.01 0.1 1

20 40 60 10080 120

t (s) t (s)

Inte

nsity

(cp

s)

00

3.0E+07

2.0E+07

1.0E+07

4.0E+07

5.0E+07

6.0E+07

7.0E+07

8.0E+07

9.0E+07

00 1.0E+06 2.0E+06 3.0E+06

N (

cm-3

)

iΣiNi/t (cm-3.s-1)

C'i

(%)

Mea

n In

tens

ity (

cps)

c

ba

d

Fig. 6. Electronic low pressure impaction and ICP-MS results showing the influence of laser pulse energy (E0) during monazite ablation at τ=60 fs on (a) particles density N (cm−3)and (b) ICP-MS signal of 238U as a function of time (t); (c) ICP-MS intensity (238U in cps) as a function of particles mean flow rate (ΣiNi/t in cm−3·s−1) and (d) particle size distri-bution, where C'i (%) is the normalized amount of particles. Error bars in are 1σ standard deviation, calculated from 8 acquisitions over 8 days for particle size distribution data (c) and1σ standard deviation calculated from a sliding uncertainty (d) defined in Section 2.7. There seem to be a linear dependency between signal intensity and particle production follow-ing: 238U=0.88 (ΣiNi/t)−165,221 with R2=0.85. Note the relative loss of ~10% of 7 nmbdab55 nm particles when E0=0.15 mJ/pulse.

677F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

a function of pulse energy (Fig. 8). This correlation allows usto consider that E∝F (F=E/S with S the irradiated surface),and N∝z, i.e. the particle density is proportional to the abla-tion depth. The linear dependency observed on Fig. 6c showsthat the ICP-MS signal can be directly correlated to the abla-tion efficiency and confirms that this parameter is controlledby irradiance P0 (in W·cm−2).

(4) Finally, the ICP-MS signal evolution as the ablation proceeds isalso directly linked to the density of particles produced.

0 500 1000 1500 2000 2500 3000 3500

τ (fs)

0

2.0E+09

4.0E+09

6.0E+09

8.0E+09

1.0E+10

1.2E+10

1.4E+10

1.6E+10

1.8E+10

P(W

)

0

5.0E+05

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

iΣiN

i/t (

cm-3

.s-1

)

Fig. 7. Particlesmean flow rate (SiNi/t in cm−3.s−1) as a function of pulse duration t (in fs).The grey curve represents the evolution of irradiance P (inW) as a function of t, following:P=1012t−1. The similar behaviour of particle production suggests that irradiance plays akey role in the ablation rate. Error bars are 1s standard deviation, calculated from 8 acqui-sitions over 8 days.

During thefirst shots, the effectivefluence ismaximal becausethe exposed surface at the focal point of the laser beam is flat.This induces a peak of ablation rate, and thus a rapid increaseof aerosol density to amaximumvalue. The transient elemen-tal ratios measured during the first 15 s could then be inter-preted by significant mass load effect in the plasmatorch because of high particle density of large mean size(94bdab157 nm) (Fig. 4a) undergoing less efficient ioniza-tion. Thereafter, the evolution of the crater aspect ratio

5.0E+05

0.0E+00

1.0E+06

1.5E+06

2.0E+06

2.5E+06

3.0E+06

3.5E+06

0 0.2 0.4 0.6 0.8 1 1.2

E0 (mJ/pulse)

iΣiN

i/t (

cm-3

.s-1

)

Fig. 8. Particles mean flow rate (SiNi/t in cm−3.s−1) as a function of pulse energy E0(mJ/pulse). The logarithmic trend (SiNi/t=106ln(E0)+3.106; R2=0.90) is probablycorrelated to the established relationship between the ablation depth and the effectivefluence on sample. Error bars are 1s standard deviation, calculated from 8 acquisitionsover 8 days.

678 F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

[54,55] (depth to diameter ratio) controls the particle produc-tion. In fact, progressive defocusing of the beamand extensionof the irradiated surface (cone shape development) lead to adecrease of the applied fluence and, as a consequence, of theablation rate. Thus, less matter is ablated from the substrateas the crater deepens, explaining the decrease of both particledensity and ICP-MS signal intensity and stability (Fig. 3). Thiseffect is particularly marked in fs-LA relative to ns-LA becausethe ablation threshold is attained also in the borders of theGaussian beam. At first, this evolution yields stable chemicalratio, but at some point, a too high crater depth becomes anissue. In fact, the laser induced plasma has an increasingly re-duced space to expand as the crater deepens. Hence, earlycondensed matter, supposed to form spherical particles [27],should have a high probability to impact and adhere to thepit walls before being flushed off the crater. At the end, thelate condensing fraction of the plasma would preferentiallybe extracted from the pit and formagglomerates. This eventu-ally leads to an enrichment of the bulk aerosol in late condens-ing elements, and then to the temporal chemical fractionationobserved in Fig. 3 between 31P (contained in agglomerates)and 143Nd (rather contained in early condensed beads). An in-teresting observation is that a similar fractionation is observedon Fig. 4 with 206Pb/238U, suggesting that 206Pb would ratherbe part of agglomerates. Themore refractory nature of U rela-tive to Pb, and of Nd relative to P, suggests that this segrega-tion is likely to occur. This result is of primary importancefor U–Th–Pb dating. It is supported by a recent study wherewe have shown that chemical segregation between twodifferent types of particles can be the source of chemical frac-tionation in femtosecond LA–ICP-MS [27]. Indeed, EDX ana-lyses (Fig. 1) cannot resolve such low concentrations(b1 wt.%) and thus cannot determine inwhich type of particleand inwhat extent uraniumand lead arepartitioned.Here,wecan argue on this segregation by observing similar behavioron highly concentrated elements in monazite (P and Nd) forwhich both EDX analysis and ICP-MS measurements arepossible. However, although systematic, such chemical frac-tionation effect is much smaller than for nanosecond laserablation ICP-MS [20], and remains within uncertainties.

From a more general point of view, these results and related inter-pretations provide optimal conditions for infra red femtosecond laserablation–ICP-MS analysis. Transport should be as short as possible toavoid loss of ablation products, especially if this loss is inhomogeneousand corresponds to particles with a specific composition [27], and pre-vent the agglomeration of particles by collision processes. This can becritical if it leads to a limitation of the ICP-MS signal as observed whenτb200 fs. Pulse duration should be reduced below 300 fs in order toenhance ablation efficiency and perform homogeneous sampling. Wealso show that a greater ablation rate does not involve a much higherandmore accurate ICP-MS signal. Ablation time should also be carefullymonitored to avoid high crater aspect ratio that induce temporal chem-ical fractionation as previously observed in nanosecond LA [54,55],although this effect is much reduced in femtosecond LA–ICP-MS analy-sis. We can also define a more precise time window of analysis, andgive hints to enlarge it. In fact, our results show that for high fluenceand low repetition rate, the signal is transient (~15 s) before becomingstable and remains accurate until around 60 s. Yet, until 90 s, stabilitydecreases but themean value is not much affected. These 75 s of steadysignal window can be extended by adjusting E0. Indeed, previous work[33] has shown that temporal fractionation can be reduced by loweringpulse energy, thus delaying the attainment of the critical crater aspectratio. A larger analysis time windows is then reachable. Hence, we findthat contrary to previous conclusions based on nanosecond LA [56], agreater freedom is allowed in the choice of pulse energy without risk

of chemical fractionation in femtosecond LA. Settings only depend onthe ICP-MS signal intensity required for counting statistics and thesample constraints. Lower energies can also be expected to produceless mass load effect at the beginning of ablation, and thus help stabiliz-ing the signal in less than15 s. The only limitationwould be a low signal/background ratio after some point, due the fall of effective fluence bydefocusing of the laser beam. If longer signals are required, the problemmay be solved by active focusing (refocusing at each shot) or rasteringablation modes. Fractionation issues potentially related to beam defo-cusing could be expected due to the deterioration of pulse width afterthe focal point where non-linear optical effects may occur in the femto-second regime. Nevertheless, it has to bementioned that, in our case, nodeterioration apparently occurs over the whole range of pulse widthstested. In fact, high fractionation effects suchas those reported by Jeffrieset al. [57]with IR-nanosecond laser ablation systems and typical of thesesystems were never detected here. To summarize, searching for theoptimal settings in femtosecond LA–ICP-MS comes to search for theoptimal particle production. This can be achieved by favoring the laserinduced plasma expansion, so that condensation/coalescence processesoccur faster, without any loss of material. Sufficient volumes are neededand low pressure cells designs are a promising technology to beexplored.

5. Conclusion

Key parameters in IR-fs-LA–ICP-MS experiments have been testedfrom the laser induced aerosols point of view in a wide range of values.Transmission electron microscopy is a powerful technique to study par-ticles morphologies and compositions, even if chemical analyses viaEDX are still limited to concentrations above ~1%. Pulse energy, pulse du-ration, crater development and transport length have a visible effect onparticle density, much less on particle size distribution and no effectson particle morphology. Results obtained by varying τ and E0 show thatICP-MS response is strongly correlated to the particle production. Similarresults on both parameters suggest that particle density is controlled byirradiance. At constant E0, themaximum ICP-MS signal intensity reachedwhen τ=200 fs. No benefit is noticed when using shorter pulse widths,probably because higher particles production resulted into increased ag-glomeration during transport and higher mass load effects, critical forplasma torch ionization efficiency. The coupling of particle countingand ICP-MS measurements has provided a link between the 31P/143Ndbehavior, detectable by EDX and well identified in the two types of par-ticles observed with TEM, to the much less concentrated 206Pb and 238U.This result provides a hint that lead may rather be contained in agglom-erates and uranium in beads, which is critical for U–Th–Pb dating usingLA–ICP-MS. Finally, thiswork brings physical arguments to analytical ob-servations and is able to provide optimal conditions for infra red femto-second laser ablation–ICP-MS analysis. Most of them fit well to thepreviously empirically preferred settings, often defined for nanosecondLA although they are much less stringent for analysis by femtosecondLA–ICP-MS. Transport should be as short as possible (lb4 m) to reducethe probability of agglomeration along the tube. Pulse duration increasessignal intensity when shortened but no benefit is noticed when settingunder 200 fs. This lower limit should be confronted to analysis on othermaterials like metals, in regard to the singular particles characteristicslighted up by previous TEM observations [27] and to their particularlyshorter thermal relaxation time. Ablation time is linked to crater devel-opment and must be carefully watched over, via a good monitoring ofpulse energy and an optimization of the steady signal time window.Working at low energies would help reducing the transient signal atthe beginning of ablation (mass load effect in the ICP) and delaying thetemporal chemical fractionation (reaching of critical crater aspect ratio)by a lower ablation rate. However, it should be stressed here that con-trarily to nanosecond LA–ICP-MS, chemical fractionation observed byfemtosecond LA–ICP-MS is very small and within analytical uncer-tainties. Futureworks should focus on the importance of favoring plasma

679F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

expansion speed, in order to generate the most possible homogeneousaerosols, which is essential for an efficient ionization within the plasmatorch, and therefore an enhanced accuracy and reproducibility of themeasurements. In this case, specific ablation cells should be used tofavor low pressure at the ablation site.

Acknowledgments

The authors would like to thank Rémi Freydier, Frédéric Candaudapand Aurélie Lanzanova for their availability, David le Dur fromECOMESURE (Janvry, France) and his great help during the Low Pres-sure Impaction experiments. Luc Vigroux, Pierre-Marie Paul and Pier-rick Leroy are thanked for their support and advising in femtosecondlaser handling. This work is co-founded by CNRS, Amplitude Technolo-gies and a PPF from the Ministry of Research.

References

[1] A.L. Gray, Solid sample introduction by laser ablation for inductively coupledplasma source mass spectrometry, Analyst 110 (1985) 551–556.

[2] R.E. Russo, X. Mao, H. Liu, J. Gonzalez, S.S. Mao, Laser ablation in analytical chem-istry—a review, Talanta 57 (2002) 425–451.

[3] J. Kosler, Laser ablation ICP-MS—a new dating tool in Earth science, Proc. Geol.Assoc. 118 (2007) 19–24.

[4] P. Arrowsmith, Laser ablation of solids for elemental analysis by inductivelycoupled plasma mass-spectrometry, Anal. Chem. 59 (1987) 1437–1444.

[5] R.E. Russo, X. Mao, O.V. Borisov, Laser ablation sampling, Trends Anal. Chem. 17(1998) 461–469.

[6] C.A. Heinrich, T. Pettke, W.E. Halter, M. Aigner-Torres, A. Audétat, D. Günther, B.Hattendorf, D. Bleiner, M. Guillong, I. Horn, Quantitative multi-element analysisof minerals, fluid and melt inclusions by laser-ablation inductively-coupled-plasma mass-spectrometry, Geochim. Cosmochim. Acta 67 (2003) 3473–3497.

[7] S.P. Chenery, J.M. Cook, Determination of rare earth elements in single mineralgrains by laser ablation microprobe−inductively coupled plasma mass spectrom-etry: preliminary study, J. Anal. At. Spectrom. 8 (1993) 299–303.

[8] J.L. Paquette, M. Tiepolo, High resolution (5 [mu]m) U–Th–Pb isotope dating ofmonazite with excimer laser ablation (ELA)-ICPMS, Chem. Geol. 240 (2007)222–237.

[9] T. Hirata, T. Ohno, In-situ isotopic ratio analysis of iron using laser ablation-multiplecollector-inductively coupled plasma mass spectrometry (LA–MC–ICP-MS), J. Anal.At. Spectrom. 16 (2001) 487–491.

[10] I. Horn, F. von Blanckenburg, R. Schoenberg, G. Steinhoefel, G. Markl, In situ ironisotope ratio determination using UV-femtosecond laser ablation with applica-tion to hydrothermal ore formation processes, Geochim. Cosmochim. Acta 70(2006) 3677–3688.

[11] J. Chmeleff, I. Horn, G. Steinhoefel, F. von Blanckenburg, In situ determination ofprecise stable Si isotope ratios by UV-femtosecond laser ablation high-resolutionmulti-collector ICP-MS, Chem. Geol. 249 (2008) 155–166.

[12] D. Günther, B. Hattendorf, Solid sample analysis using laser ablation inductivelycoupled plasma mass spectrometry, Trends Anal. Chem. 24 (2005) 255–265.

[13] R.E. Russo, X.L. Mao, O.V. Borisov, H. Liu, Influence of wavelength on fractionationin laser ablation ICP-MS, J. Anal. At. Spectrom. 15 (2000) 1115–1120.

[14] D. Günther, R. Frischknecht, C.A. Heinrich, H.J. Kahlert, Capabilities of an argonfluoride 193 nm excimer laser for laser ablation inductively coupled plasmamass spectometry microanalysis of geological materials, J. Anal. At. Spectrom.12 (1997) 939–944.

[15] R.E. Russo, X. Mao, J.J. Gonzalez, S.S. Mao, Femtosecond laser ablation ICP-MS, J.Anal. At. Spectrom. 17 (2002) 1072–1075.

[16] T. Hirata, Y. Kon, Evaluation of the analytical capability of NIR femtosecond laserablation–inductively coupled plasma mass spectrometry, Anal. Sci. 24 (2008)345–353.

[17] T. Hirata, Development of an on-line low gas pressure cell for laser ablation–ICP-mass spectrometry, Anal. Sci. 23 (2007) 1195–1201.

[18] Q. Bian, C.C. Garcia, J. Koch, K. Niemax, Non-matrix matched calibration of majorand minor concentrations of Zn and Cu in brass, aluminium and silicate glassusing NIR femtosecond laser ablation inductively coupled plasma mass spectrom-etry, J. Anal. At. Spectrom. 21 (2006) 187–191.

[19] I. Horn, F. von Blanckenburg, Investigation on elemental and isotopic fraction-ation during 196 nm femtosecond laser ablation multiple collector inductivelycoupled plasma mass spectrometry, Spectrochim. Acta Part B 62 (2007) 410–422.

[20] F. Poitrasson, X. Mao, S.S. Mao, R. Freydier, R.E. Russo, Comparison of ultravioletfemtosecond and nanosecond laser ablation inductively coupled plasma massspectrometry analysis in glass, monazite, and zircon, Anal. Chem. 75 (2003)6184–6190.

[21] R. Freydier, F. Candaudap, F. Poitrasson, A. Arbouet, B. Chatel, B. Dupré, Evaluationof infrared femtosecond laser ablation for the analysis of geomaterials by ICP-MS,J. Anal. At. Spectrom. 23 (2008) 702–710.

[22] D. Perez, L.J. Lewis, Molecular-dynamics study of ablation of solids under femto-second laser pulses, Phys. Rev. B Condens. Matter Mater. Phys. 67 (2003) 184102.

[23] J. Bonse, J. Krüger, Probing the heat affected zone by chemical modifications infemtosecond pulse laser ablation of titanium nitride films in air, J. Appl. Phys.107 (2010).

[24] A. Ben-Yakar, A. Harkin, J. Ashmore, R.L. Byer, H.A. Stone, Thermal and fluid pro-cesses of a thin melt zone during femtosecond laser ablation of glass: the forma-tion of rims by single laser pulses, J. Phys. D Appl. Phys. 40 (2007) 1447–1459.

[25] D. Fliegel, M. Klementova, J. Kosler, Phase and composition changes of titaniteduring laser ablation inductively coupled plasma mass spectrometry analysis,Anal. Chem. 82 (2010) 4272–4277.

[26] R. Glaus, R. Kaegi, F. Krumreich, D. Günther, Phenomenological studies on struc-ture and elemental composition of nanosecond and femtosecond laser-generatedaerosols with implications on laser ablation inductively coupled plasma massspectrometry, Spectrochim. Acta Part B 65 (2010) 812–822.

[27] F.X. D'Abzac, Effets d'une impulsion laser Infra-Rouge femtoseconde sur lesmicro-nano-structures des minéraux. Implications pour les analyses in-situ parLA–ICP-MS, Minéralogie, Université Paul Sabatier, Toulouse III, 2010, p. 241.

[28] R. Hergenröder, A model for the generation of small particles in laser ablation ICP-MS, J. Anal. At. Spectrom. 21 (2006) 1016–1026.

[29] J. Koch, A.v. Bohlen, R. Hergenröder, K. Niemax, Particle size distributions andcompositions of aerosols produced by near-IR femto- and nanosecond laser abla-tion of brass, J. Anal. At. Spectrom. 19 (2004) 267–272.

[30] K. Niemax, Laser ablation—reflections on a very complex technique for solid sam-pling, Fresenius J. Anal. Chem. 370 (2001) 332–340.

[31] D. Bleiner, D. Günther, Theoretical description and experimental observation ofaerosol transport processes in laser ablation inductively coupled plasma massspectrometry, J. Anal. At. Spectrom. 16 (2001) 449–456.

[32] A.M. Seydoux-Guillaume, R. Freydier, F. Poitrasson, R. Wirth, Effects of femtosec-ond laser ablation on monazite micro- and nano-structure, Geochim. Cosmochim.Acta 70 (2006) A574.

[33] F.X. D'Abzac, F. Poitrasson, R. Freydier, A.M. Seydoux-Guillaume, Near infra redfemtosecond laser ablation: the influence of energy and pulse width on ICP-MSsignal of natural monazite, J. Anal. At. Spectrom. 25 (2010) 681–689.

[34] C. Courtieu, F.X. D'Abzac, J. Chmeleff, D. Guillaume, A.M. Seydoux-Guillaume, Per-formances of 800 nm femtosecond laser ablation on natural and synthetic quartz,Eur. J. Mineral. 23 (2011) 391–400.

[35] Y. Ni, J.M. Hughes, A.M. Mariano, Crystal chemistry of the monazite and xenotimestructures, Am. Mineral. 80 (1995) 21–26.

[36] R.R. Parrish, U–Pb dating of monazite and its application to geological problems,Can. J. Earth Sci. 27 (1990) 1431–1450.

[37] A.M. Seydoux-Guillaume, R. Wirth, L. Nasdala, M. Gottschalk, J.M. Montel, W.Heinrich, An XRD, TEM and Raman study of experimentally annealed naturalmonazite, Phys. Chem. Miner. 29 (2002) 240–253.

[38] F. Poitrasson, E. Oelkers, J. Schott, J.-M. Montel, Experimental determination ofsynthetic NdPO4 monazite end-member solubility in water from 21 °C to300 °C: implications for rare earth element mobility in crustal fluids, Geochim.Cosmochim. Acta 68 (2004) 2207–2221.

[39] A.-M. Seydoux-Guillaume, R. Wirth, A. Deutsch, U. Scharer, Microstructure of 24-1928 Ma concordant monazites; implications for geochronology and nuclearwaste deposits, Geochim. Cosmochim. Acta 68 (2004) 2517–2527.

[40] A.M. Seydoux-Guillaume, J.L. Paquette, M. Wiedenbeck, J.M. Montel, W. Heinrich,Experimental resetting of the U–Th–Pb systems in monazite, Chem. Geol. 191(2002) 165–181.

[41] A.M. Seydoux-Guillaume, R. Freydier, F. Poitrasson, F.X. D'Abzac, R. Wirth, L.Datas, Dominance of mechanical over thermally induced damage during femto-second laser ablation of monazite, Eur. J. Mineral. 22 (2010) 235–244.

[42] H.-R. Kuhn, J. Koch, R. Hergenroder, K. Niemax, M. Kalberer, D. Gunther, Evalua-tion of different techniques for particle size distribution measurements onlaser-generated aerosols, J. Anal. At. Spectrom. 20 (2005) 894–900.

[43] P.F. DeCarlo, J.G. Slowik, D.R. Worsnop, P. Davidovits, J.L. Jimenez, Particle morphologyand density characterization by combined mobility and aerodynamic diameter mea-surements. Part 1: theory, Aerosol Sci. Technol. 38 (2004) 1185–1205.

[44] N.J.G. Pearce, W.T. Perkins, J.A. Westgate, M.P. Gorton, S.E. Jackson, C.R. Neal, S.P.Chenery, A compilation of New and published major and trace element data forNIST SRM 610 and NIST SRM 612 glass reference materials, Geostand. Newsl. 21(1997) 115–144.

[45] J.J. Gonzalez, A. Fernandez, X. Mao, R.E. Russo, Scanning vs. single spot laser abla-tion (λ=213 nm) inductively coupled plasma mass spectrometry, Spectrochim.Acta Part B 59 (2004) 369–374.

[46] D. von der Linde, K. Sokolowski-Tinten, The physical mechanisms of short-pulselaser ablation, Appl. Surf. Sci. 154–155 (2000) 1–10.

[47] J.J. Gonzalez, C. Liu, S.-B. Wen, X. Mao, R.E. Russo, Metal particles produced bylaser ablation for ICP-MS measurements, Talanta 73 (2007) 567–576.

[48] C.C. Garcia, M. Wälle, H. Lindner, J. Koch, K. Niemax, D. Günther, Femtosecondlaser ablation inductively coupled plasma mass spectrometry: transport efficien-cies of aerosols released under argon atmosphere and the importance of the focusposition, Spectrochim. Acta Part B 63 (2008) 271–276.

[49] J. Koch, M. We, R. Dietiker, D. Günther, Analysis of laser-produced aerosols by in-ductively coupled plasma mass spectrometry: transport phenomena and elemen-tal fractionation, Anal. Chem. 80 (2008) 915–921.

[50] P. Lorazo, L.J. Lewis, M. Meunier, Short-pulse laser ablation of solids: from phaseexplosion to fragmentation, Phys. Rev. Lett. 91 (2003) 225502.

[51] B.C. Stuart, M.D. Feit, S. Herman, A.M. Rubenchik, B.W. Shore, M.D. Perry, Nano-second-to-femtosecond laser-induced breakdown in dielectrics, Phys. Rev. B Con-dens. Matter Mater. Phys. 53 (1996) 1749–1761.

[52] A. Ben-Yakar, R.L. Byer, Femtosecond laser ablation properties of borosilicateglass, J. Appl. Phys. 96 (2004) 5316–5323.

680 F.-X. d'Abzac et al. / Spectrochimica Acta Part B 66 (2011) 671–680

[53] S. Nolte, C. Momma, H. Jacobs, A. Tünnermann, B.N. Chichkov, B. Wellegehausen,H. Welling, Ablation of metals by ultrashort laser pulses, J. Opt. Soc. Am. B 14(1997) 2716–2722.

[54] A.J.G.Mank, P.R.D.Mason, A critical assessment of laser ablation ICP-MS as an analyticaltool for depth analysis in silica-based glass samples, J. Anal. At. Spectrom. 14 (1999)1143–1153.

[55] O.V. Borisov, X. Mao, R.E. Russo, Effects of crater development on fractionationand signal intensity during laser ablation inductively coupled plasma mass spec-trometry, Spectrochim. Acta Part B 55 (2000) 1693–1704.

[56] J. Koch, I. Feldmann,N. Jakubowski, K. Niemax, Elemental composition of laser ablationaerosol particles deposited in the transport tube to an ICP, Spectrochim. Acta Part B 57(2002) 975–985.

[57] T.E. Jeffries, N.J.G. Pearce, W.T. Perkins, A. Raith, Chemical fractionation during infra-red and ultraviolet laser ablation inductively coupled plasma mass spectrometry—implications for mineral microanalysis, Anal. Commun. 33 (1996) 35–39.