laser-induced material ejection from model molecular solids and liquids: mechanisms, implications,...

Laser-Induced Material Ejection from Model Molecular Solids and Liquids:Mechanisms, Implications, and Applications

Savas Georgiou* and Antonis Koubenakis#

Institute of Electronic Structure and Laser, Foundation for Research and Technology-Hellas, P.O. Box 1527, 71110 Heraklion, Crete, Greece

Received July 24, 2002

ContentsI. Introduction 349II. Studies on Cryogenic (van der Waals) Molecular

Solids352

A. Delineation of Photodesorption Processes 352B. The Phenomenon of Explosive Desorption 353

1. Cluster Ejection 3532. Ejection Mechanisms 3553. Implications for Relevant Applications

(MALDI, MAPLE)359

4. Energy Dissipation Processes 3605. Laser Pulse Dependence 3616. Substrate-Mediated Ejection/Confined

Ablation362

C. Desorbate Translational Distributions 363D. Chemical Processes and Effects 370

1. Photochemical Processes and Effects 3712. Thermal Decomposition Effects 375

III. Studies on Liquids 377A. Transparent Liquids on Absorbing Surfaces:

Liquid Superheating377

B. Absorbing Solutions and Liquids 3811. Photoinert Systems: Photomechanical

Mechanism of Material Ejection381

2. Photolabile Liquids: PhotochemicalMechanism of Material Ejection

383

C. Optical Processes in the Irradiation at HighIrradiances

386

IV. Applications 387A. Cryogenic Films-Based Techniques 387B. Liquid-Based Applications 387

1. “Steam Laser Cleaning” Technique 3872. Liquid-Assisted Material Processing and

Nanostructure Formation389

V. Acknowledgments 390VI. References 390

I. IntroductionUpon irradiation of condensed phases with laser

pulses at high laser irradiances, massive materialejection is observed. This phenomenon has beennamed “ablation”. A plot of the quantity of theremoved material as a function of laser fluence

usually has a sigmoidal dependence, as in Figure 1,though the exact shape may differ considerablydepending on system properties and irradiationparameters. Despite its apparent violent nature, thisefficient material removal method has provided thebasis for a wide spectrum of highly successful ap-plications, ranging from analytical chemistry1-4 tomicroelectronics,5-9 medicine,1,7 restoration of paintedartwork,10 etc. In some of these applications, UVablation is used to effect ejection of material in thegas phase for its analysis or subsequent deposition.2-4,9

In the other applications, UV ablation is employedto effect the appropriate shaping or processing of thesubstrate via the removal1,5-8,10,11 of unwanted mate-rial. In either case, ablation offers the crucial advan-tage of micrometer precision in the removed depth,with little thermal or other degradation to the ejectedor remaining material. Further advantages includea high degree of reproducibility, the capability ofinterfacing with a variety of laser-based techniquesfor on-line monitoring of the process, etc.

Despite the widespread and highly successful ap-plications, several aspects of laser ablation of molec-ular substrates induced with nanosecond or shorterpulses remain poorly understood. The first andforemost problem to be elucidated remains the issueof fundamental mechanisms responsible for the ma-terial ejection, i.e., how the absorbed light energyresults eventually in material ejection. Generally, thephenomenon is described phenomenologically fromthe dependence of the etching depth (or of the amountof the removed material) on laser fluence (laser pulseenergy per unit irradiated area, i.e., Flaser). Usually,the fluence at which the sharp increase in the etchingdepth is observed, or the Flaser intercept of theextrapolation of the rising section of the curve, isconsidered to be the threshold for ablation. However,despite their predominant use, such curves do notyield much physical insight, especially since theydiffer quantitatively and even qualitatively accordingto irradiation and system parameters. Even theexistence of a threshold is very difficult to ascertainfrom examination of these curves. Clearly, the ques-tion raised is if ablation can be identified with specificphysical characteristics that distinguish it fromprocesses at lower fluences or if it is adequatelyspecified only in terms of the amount of materialremoved. Unfortunately, for several applications, onlythe amount of removed material is of concern, andas a result, this issue has been emphasized over thatof the involved physical processes.

* To whom correspondence should be addressed. Phone: +30 810391121. Fax: +30 81 0391318. E-mail: [email protected].# Present address: Department of Chemistry, Swiss FederalInstitute of Technology (ETH), Honggerberg HCI, CH-8093 Zurich,Switzerland.

349Chem. Rev. 2003, 103, 349−393

10.1021/cr010429o CCC: $44.00 © 2003 American Chemical SocietyPublished on Web 02/12/2003

The difficulties encountered in trying to establishthe fundamental physical processes underlying thephenomenon can be easily appreciated by consideringa schematic of the various steps leading to materialejection. In the first step, energy must be coupled intothe system via absorption. Typically, fluences of ∼100mJ/cm2 (3 MW/cm2) or higher are required experi-mentally to effect material ejection by nanosecondlaser pulses. Given a typical cross section of ∼10-18-

cm2/molecule, multiphoton processes and other non-linearities in the absorption process, such as anni-hilation of electronically excited states, saturationeffects, etc., become highly likely. Establishing theabsorbed energy experimentally is no easy task, sincethe extensive material ejection during the laser pulseresults in significant scattering and absorption of theincident light by the ejecta. In fact, the extensive ma-terial ejection severely limits the use of any spectro-scopic technique in probing processes in the substrate(at least up to microsecond time scales after thenanosecond laser pulse). Following light absorption,material ejection can be envisaged to be induced inat least four different ways. At least for nanosecondpulses, a good percentage of the absorbed energy canbe expected to decay into thermal energy, since typi-cal radiationless decay constants for organic mol-ecules are ∼10-12-10-9 s. In view of the low thermalconductivity of molecular systems, high surface tem-perature changes may be attained, and desorption/evaporation rates can be substantial (thermal mech-anism). However, given the very short time scale ofirradiation, phase transformations under nonequi-librium conditions may occur (explosive boiling/phaseexplosion), resulting in quantitative and even quali-tative differences from “simple” thermal processes.Unfortunately, the dynamic optical, thermophysical,etc. properties under these conditions can be distinctfrom those of a uniformly heated system, whichfurther complicates the quantitative analysis andeven the interpretation of the results. The fast changein temperature implies also a volume change andthus a high-amplitude pressure generation, which

Savas Georgiou received his B.Sc. in chemistry and mathematics fromKnox College, Illinois (1983), and his Ph.D. in physical chemistry (on thestudy of photodissociation dynamics of organometallic compounds) fromthe University of Utah (1988). He subsequently performed postdoctoralwork on Raman spectroscopy of biomolecules at Princeton University.After a two-year military service in the Greek Army, in 1993 he joined theInstitute of Electronic Structure and Laser of Foundation for Researchand TechnologysHellas, where he is now a Senior Researcher. He hasalso held positions as a Visiting Assistant Professor at the ChemistryDepartment of the University of Crete and as an Assistant Professor atthe Physics Department of the University of Ioannina, Greece. He hasreceived various awards and has participated in several European Unionresearch projects. His current interests focus on studies of laser ablation,laser photochemistry and biophysics, and laser material processingschemes.

Antonis Koubenakis was born in 1973 in Heraklion, Crete, Greece. Hereceived his B.Sc. degree in physics in 1996 and his M.Sc. degree inatomic and molecular physics in 1997 from the University of Crete. Hesubsequently performed work on the mechanisms of laser desorption/ablation of van der Waals solids at the Foundation for Research andTechnologysHellas (F.O.R.T.H.) and received his Ph.D. from theUniversity of Crete in March 2002. For this work, he received a prizefrom the Gordon Conference of Laser Interactions with Materials (June2000). He is currently a postdoctoral fellow in the research group of Prof.Renato Zenobi at the Swiss Federal Institute of Technology in Zurich(ETH). A. Koubenakis has coauthored 15 articles in international journals.His research interests include the study of laser desorption processes,mass spectroscopy, and near-field optical analytical techniques.

Figure 1. Etching curves typical of the ones determinedin the UV ablation (nanosecond pulses) of molecularsystems (the particular ones concerning irradiation ofpolyimide at the indicated wavelengths). Absorption coef-ficients: R(193 nm) ) 4.25 × 105 cm-1; R(248 nm) ) 3.1 ×105 cm-1; R(308 nm) ) 105 cm-1; R(351 nm) ) 0.32 × 105

cm-1. The measurements rely on mass-loss measurementsusing a quartz crystal microbalance. Changes in thefeatures of the curves depending on laser wavelength areclearly noted. The curves exhibit the so-called “Arrheniustails” that have been a matter of controversy. At highfluences, the curves tend to saturate (observed for thecurves at the shorter, more strongly absorbed wave-legenths). This is often ascribed to the increasing absorp-tion/scattering of the incident light by the material ejectedduring the laser pulse. Reprinted with permission fromKuper, S.; Brannon, J.; Brannon, K. Appl. Phys. A 1993,56, 33. Copyright 1993 Springer-Verlag.

350 Chemical Reviews, 2003, Vol. 103, No. 2 Georgiou and Koubenakis

http://dontstartme.literatumonline.com/action/showImage?doi=10.1021/cr010429o&iName=master.img-000.jpg&w=125&h=155


http://dontstartme.literatumonline.com/action/showImage?doi=10.1021/cr010429o&iName=master.img-002.png&w=239&h=159

can result in material ejection via essentially “me-chanical” rupture (spallation) of the upper layers ofthe substrate. To complicate things even further,upon UV excitation, most organic molecules photo-dissociate with rather significant quantum efficiency.The formation of a high number of gaseous photo-products that exert a high pressure upon expansionand/or the high amount of energy that is liberatedby exothermic reactions can contribute to materialejection (photochemical mechanism).

These processes necessarily occur in parallel; thus,the question is raised of which one is responsible formaterial ejection. Clearly, the different processes willresult in different features/effects, and thus thespecification of their contribution is intimately re-lated to the question of the nature of ablation. Therelative importance of the processes can be ap-proximately assessed via some simple criteria. As-suming a “thermal energy pulse” and photoinertsystems, the relative importance of thermal versusmechanical processes depends on the ratio of pulseduration to thermal and stress relaxation times.These times are given, respectively, by tthermal ) labs

2/Dand tacoustic ) labs/cS, where labs is the optical penetra-tion depth (assumed to be much smaller than thelaser beam radius, so that a one-dimensional analysisapplies), D is the thermal diffusivity, and cS is thespeed of sound. The two terms represent, respec-tively, the time for heat diffusion (“thermal equilibra-tion”) and the time for acoustic wave (and thusmechanical relaxation) within the optical penetrationdepth. Typically, for molecular solids, labs in UV is inthe range from 100 nm to 5 µm, and D ≈ 10-7 m2/s,so tthermal ≈ 1 µs and tacoustic < 1 ns. Of course, in thecase of UV excitation, the previous delineation as-sumes that the time scale of electronic energy deac-tivation, τelec-deact, is fast enough compared to the pre-vious times. The τelec-deact e tthermal condition is ex-pected to be fulfilled for most molecular systems, butthe τelec-deact e tacoustic condition may not be (sectionsII.B.4 and II.B.5). Neglecting for the moment thisissue, the following delineation can be drawn:

(a) For laser pulse durations shorter than tthermal(thermal confinement regime), the temperature pro-file is determined by the laser light distribution, i.e.,

where ∆T(z) is the temperature “jump” induced atdepth z from the surface, R is the absorption coef-ficient, F the mass density, Flaser the laser fluence,and CP the heat capacity, light reflection and scat-tering being neglected. Since the maximum possibletemperature is attained in the target volume, ther-mal desorption/evaporation rates can be significantand can account for the high material removal ratesobserved at high laser fluences. Furthermore, underthis condition, the extent of thermal dissipation anddamage adjacent to the irradiated area is limited.This has been one of the key factors in the success ofnanosecond UV laser processing of molecular sub-strates.

(b) The previous argument, however, does notexplain some of the unique features of the phenom-

enon that have been crucial for its widespreadimplementation, e.g., how fragile molecules are ejectedwith minimal fragmentation. It now appears that, atleast for simple systems (photoinert and low cohesiveenergy) and nanosecond pulses, these unique featuresmay be associated with “nonequilibrium” phase trans-formation, namely explosive boiling. In this case, anextra criterion may be introduced to establish whethermaterial ejection is due to simple thermal desorptionor to ablation, namely the rate of homogeneousbubble formation being competitive with evaporativecooling rates (section II.B.2) or with the rate of energyconsumed for bubble formation in the presence ofnuclei promoting heterogeneous nucleation (sectionIII.A).

(c) For even shorter laser pulses, such that τpulse <tacoustic (stress-confinement regime), heating is effectedunder nearly isochoric conditions, leading to the mostefficient possible generation of a thermoelastic wave.In the presence of an interface with a lower acousticimpedance medium (e.g., air), this wave contains, dueto reflection at the interface, both compressive andtensile stresses. Such a high-amplitude bipolar stresswave can lead to the ejection of material via es-sentially mechanical rupture of the upper layers(spallation) (sections II.B.5 and III.B.1).

Despite their appealing simplicity, the above cri-teria clearly serve only as approximate guidelines.In most cases, there is a competition or synergybetween the various processes, and which one domi-nates will depend sensitively on the characteristicsof the substrate (in particular, cohesive energy).

Evidently, it is very difficult to specify or controlexperimentally all the different aspects involved inthe phenomenon. For the theoretical description,analytical treatments similarly have to resort toconsidering the individual processes, but unfortu-nately, this simplification in mathematical formalismmay miss essential aspects of the phenomenon.Several interesting features of material ejection inthe irradiation of molecular solids at high laserfluences have been indicated by molecular dynamics(MD) simulations based on the breathing spheremodel. However, even here, because of the enormouscomputing power required to simulate the laser/matter interaction, several simplifications are madein the representation of the molecules and irradiationconditions. In view of these difficulties, it is easilyunderstandable that a wide range of uncertaintiesand controversies exist concerning the mechanismsof ablation.

These problems aside, the previous discussiondirectly indicates that the study of UV ablation ishighly interdisciplinary, encompassing questions fromthe fields of thermodynamics, photophysics/chemis-try, hydrodynamics, etc. As a result of this interdis-ciplinarity, the study of UV ablation can be expectedto result in new information about molecular photo-physics/chemistry. This is clearly underscored by thefact that the wide range and high success of theapplications of the phenomenon would not have beenexpected by conventional photochemistry. To thecontrary, by any conventional photophysical/chemicalcriterion, the employed high irradiances would be

∆T(z) )RFlaser

FCPe-Rz (1)

Laser-Induced Material Ejection from Solids and Liquids Chemical Reviews, 2003, Vol. 103, No. 2 351

expected to result in ill-defined effects. The fact thatthis is not the case suggests the need to introducenew concepts that may be of far-reaching scientificimpact. Indeed, important issues concerning non-equilibrium phase transformations, electronic excita-tion and deactivation processes, material dynamics,etc. have already been raised and studied in the field.

Though the “power” of UV ablation is illustratedin the processing of the complex polymers, biopoly-mers, tissues, painted artwork, etc.,1-11 the study ofthe processes underlying the phenomenon in realisticsystems raises significant difficulties. Studies onsimple molecular systems may provide a usefulstarting point by enabling detailed probing andelucidation of the processes. Simple compounds atambient conditions are usually liquids or gases.Laser-induced material ejection from liquids has beenextensively studied in its own right. It has alsoprovided a basis for understanding processes incomplex systems, in particular tissues. On the otherhand, to simulate the solid state of the substratescommonly encountered in practice, while retainingthe advantage of molecular simplicity, a number ofstudies have been performed on films/solids of simplecompounds condensed at low temperatures, namelycryogenic or van der Waals films.

In the following, the work on cryogenic films ispresented first, and subsequently that on liquids.Certainly, laser-induced material ejection processesin these systems are expected to be closely related.The first reason for adopting this separate presenta-tion is that no fully satisfactory general model(s) forthe dependence of the phenomenon on the phase/state of the substrate has thus far been developed.In retrospect, it is clear that in both cases, similarproblems/questions/mechanisms have been addressedall along, but this has been recognized only in recentstudies. Furthermore, even if the same basic mech-anisms are largely involved in both phases, there aresignificant differences in terms of their energy con-tent, tensile strength, efficiency to propagate stresswaves, species diffusivity and reactivity, etc. thatmay affect various aspects of the phenomenon. Someof these differences will be noted. The second andmain reason for the separate presentation is thatdifferent experimental techniques have been em-ployed in the case of solids versus liquids. As a result,different aspects of the phenomenon have beenlargely emphasized in the corresponding studies.

II. Studies on Cryogenic (van der Waals)Molecular Solids

A. Delineation of Photodesorption ProcessesIn the past decade, there has been considerable

work on photodesorption processes from cryogenic orvan der Waals films. In these studies, the van derWaals films serve as model systems for examiningmolecular dynamics in condensed phases and forcomparing to corresponding processes in the gasphase. The studies exploit the fact that these filmscan be prepared in a vacuum, thereby enablingpowerful techniques developed previously in therealm of molecular dynamics and surface science to

be applied to probe the photoinduced processes. Thesetechniques will not be described here, since they havebeen reviewed in detail in the literature.12 By far, themost usual one has been12 quadrupole mass spec-troscopy (QMS) in a time-of-flight (TOF) arrangementto establish the nature and the intensity as well asthe translational distributions of the ejected species.

In the framework of these studies, it was earlydemonstrated that the processes induced in theirradiation of cryogenic films at high enough fluencesexhibit distinct differences from those observed at lowlaser fluences.14,18,24 At very low fluences (e1-5 mJ/cm2, depending on adsorbate absorptivity), onlyphotofragments or a few parent molecules are ob-served to eject in the gas phase12,13 (Figure 2). The

desorption signal originates exclusively from absorb-ing molecules, thus establishing the existence ofmolecular selectivity in the ejection process. Onlyexcited molecules or fragments from the topmost fewlayers of the film contribute to the signal. Ejectionis generally13,14,18,23 ascribed to electronically medi-ated processes, e.g., to the repulsion between frag-ments during photolysis or to the repulsion developedbetween the electronically excited molecules andtheir neighboring molecules. Generally, the photo-desorption yield is observed to increase linearly withFlaser (Figures 2 and 3a). Furthermore, at thesefluences, the translational distribution (e.g., shape

Figure 2. Intensities and most probable translationalenergies (Em values) of the gas-phase ejected CH2I andCH2I2 species as a function of laser fluence in the irradia-tion of CH2I2 multilayer film (where θ is the number ofmonolayers) at λ ) 308 nm. Up to ∼2 mJ/cm2, only CH2Iis ejected in the gas phase. Its intensity scales linearly withthe laser fluence, and its most probable translationalenergy is constant. This corresponds to electronicallymediated photodesorption. At higher fluences, ejection ofthe parent molecule CH2I2 becomes increasingly signicant.In this fluence range, the translational energies of bothparent and fragment scale sharply with laser fluence.Reprinted with permission from ref 15. Copyright 1989American Institute of Physics.



and ⟨Etrans⟩) of the photodesorbed molecules/frag-ments is independent of the laser fluence (Flaser) andcan be related to the energy dissipation during thedesorption process (Figure 2).

In contrast, at higher fluences and for thicker solids[Flaser g 10 mJ/cm2 for Θ g 30 monolayers (ML);Figures 2 and 3b], ejection of as much as a fewmonolayers per pulse is generally observed. Accord-ingly, the term “explosive photodesorption” wascoined for this phenomenon.15 The fluence necessaryfor its observation decreases with increasing surfacecoverage, apparently because of the increased rateand density of energy deposition in the film. Noselectivity is indicated in this fluence range, and evenconstituents that do not absorb at the irradiationwavelength are observed to desorb. For example, inthe irradiation of CH2I2/NH3 mixture at λ ) 308 nm,significant desorption of NH3 is observed, despite thefact that NH3 does not absorb at this wavelength.14

In contrast, at lower fluences, only CH2I or I frag-ments are detected in the gas phase, while NH3desorption is negligible. The translational energiesare generally found to scale with incident laserfluence (Figure 2), which sharply contrasts with theinvariance of the translational distributions observedfor surface-mediated processes. Domen and Chuang13

first noted that these features of “explosive photode-

sorption” are very similar to those observed in thephotoablation of polymers. Due to the lack of selec-tivity and the supralinear (nearly exponential) (Fig-ures 2 and 3b) dependence of desorption intensity onlaser fluence, the process has been considered to be“thermal” in nature.

This “explosive” regime has been observed for awide number of systems (adsorbates) on surfaces,such as CH3I,17 NO,18,21,22 Cl2,19,20 CH3Br,24 CH2I2,13-16

C6H6,25,26,37 C6H5Cl,37,38 C6H5CH3,38-41 etc. [In theearliest studies, there is some ambiguity about theexact regime in which desorption is effected, e.g., thecontradictory explanations advanced23,24 for the pho-toejection from H2O and NH3 films at 248 and 193nm]. However, surface scientists largely neglected thephenomenon, evidently because of its “apparent”unselective nature. Nevertheless, even in the firststudies, evidence was provided18 that explosive de-sorption cannot be considered a “simple” thermalprocess. In the UV irradiation (λ ) 220-270 nm) ofcondensed NO films at low laser fluences, electronic-ally mediated photodesorption is observed18 (Figure3a). At higher fluences, the desorbing NO signalexhibits an additional translational component withvelocities corresponding closely to the estimated filmsurface temperature (Figure 4). In agreement withthe presumed thermal nature of the process, theintensity of this component shows a nearly exponen-tial dependence on laser fluence (Figure 3b). Mostinterestingly, at even higher fluences, the character-istics of this translational component change (Figure4b). This change was ascribed exclusively to theadiabatic-like expansion of the ejected material, asa result of its high quantity (section II.C). No othersignificance was ascribed to this difference, thoughin retrospect, the change may have direct mechanisticimplications (section II.C).

B. The Phenomenon of Explosive Desorption

1. Cluster Ejection

At present, in the whole field of ablation of molec-ular substrates, well-defined physical criteria fordescribing the phenomenon remain to be established.Several different features have been associated withit, but thus far, none has proven to be generalenough. As indicated in the preceding article in thisissue, molecular dynamics simulations based on thebreathing-sphere model28-30 suggest that there is awell-defined fluence threshold, above which massiveejection of material occurs largely in the form ofclusters. In contrast, at lower laser fluences, ejectionof material occurs exclusively in the form of mono-mers and (distinctly different from the electronically-mediated process discussed in section II.A) is ascribedto thermal surface vaporization. The change frommonomer to cluster ejection is suggested to reflect achange in the mechanism of material ejection, i.e.different processes may be delineated within the“explosive desorption regime”. In fact, cluster ejectionhas often been implicated11,28 to be a particular, ifnot an inherent, feature of the processes induced athigh laser fluences. However, the current evidenceas well as the MD simulations indicates that cluster

Figure 3. NO desorption signal vs laser fluence in theUV irradiation of condensed NO films (250 ML thick). (a)Fast peak due to electronically mediated desorption (λ )247 nm). (b) Slow peak ascribable to laser-induced thermaldesorption (λ ) 273.7 nm). Reprinted with permission fromref 18. Copyright 1988 American Institute of Physics.



ejection is due to different mechanisms according tothe laser pulse width. For this reason, the discussionherein is limited only to the irradiation with typicalnanosecond pulses, whereas the influence of the laserpulse width is discussed in section II.B.5. The plau-sible limitations resulting from the neglect of theelectronic degrees of freedom in the MD simulationsare addressed in sections II.B.4, II.B.5, and III.C.

Experimentally, the suggested intense cluster ejec-tion is not documented in most cases. For the major-ity of systems mentioned in section II.A to have beenstudied in the explosive regime, signal for evendimers of the parent molecules accounts for ≈5% orless of the total signal.18,20-22 Evidence for clusterejection deriving from changes in the translationaldistributions with successive laser pulses28 is incon-clusive, as these may be ascribed to film structuralchanges. However, the failure to detect clusters maybe partly due to the limitations of the employedionization techniques. In most cases, ionization of the

neutral desorbates is effected with high electron-impact energies (Eelectron > 50 eV) or high laserintensities. These conditions may result in extensivefragmentation of the clusters.33

Nevertheless, (HBr)n clusters (n e 4) have beendetected27 in the irradiation of HBr multilayers at193 nm with 0.3-5 MW/cm2 irradiances (τpulse ≈ 30ns). Though the study did not recognize it, the irra-diation condition clearly corresponds to “explosive de-sorption” (ejection yield of 100 langmuirs per pulse).The relative intensities of the clusters (1.00:0.01:0.001:0.001, for cluster size n from 1 to 4) were notedto be inconsistent with their formation in adiabaticexpansion of a gas. Thus, the clusters were suggestedto eject directly from the surface. The possibility ofcluster ejection in the irradiation at high fluences wasrecognized by Knutzer et al.17 in the irradiation ofcondensed CH3I films at 266 nm. The resonantmultiphoton ionization spectra of the desorbates werenoted to become broad and featureless in the “explo-sive photodesorption” regime. These spectral featuresare strongly indicative of cluster ionization. Evidencefor cluster ejection in the irradiation of 2,5-dihy-droxybenzoic acid and poly(ethylene glycol) films(Flaser ) 4-12 mJ/cm2 at λ ) 337 nm) has beenobtained by trapping the ejected material on a plateand examining the morphology of the depositedmaterial by atomic force microscopy (AFM).34 Thestudy has carefully avoided the caveats of the trap-ping method. At low fluences, the collected films aresmooth, indicating molecular deposition. In contrast,at higher fluences (>4 mJ/cm2), the films are roughand composed of particles (∼200 nm diameter),indicating the deposition of material in the form ofclusters or droplets. Cluster contribution is indicatedto decrease at higher laser fluences, in good agree-ment with the results of simulations.30 Cluster for-mation has been reported occasionally for othersystems but has not been studied in detail.42

Haskin and John32 have examined the desorbatetime-of-flight spectra as a function of the delay timeand position of the focused postdesorption ionizationlaser beam in the 266-nm irradiation (I ) 15 MW/cm2) of films of chrysene-d12 (Figure 5). When theionization beam is focused ∼100 µm away from thesurface, a well-defined parent ion peak of highresolution is observed. In sharp contrast, whenionization is effected close to the surface (∼50 µm),the recorded parent ion signals are found to be broadand of very low resolution (Figure 5). These changesin the mass spectra resolution can be attributed tothe fact that, close to the surface, clusters are ionized,whereas at larger distances, the ejected molecularclusters have disintegrated into their monomers.Disintegration may be due to the high internalenergy excitation of the ejected clusters, thus result-ing in significant evaporation (section II.B.2), and/or due to the numerous collisions that the clusterssuffer in the plume (section II.C). At any rate,according to the previous result, a second reason forthe failure of most studies to detect clusters may berelated to the fact that desorbate detection is effectedat relatively large distances (>10-1 m) from thesubstrate. The time scale of cluster ejection was also

Figure 4. “Evolution” of the desorbate translationaldistributions with increasing laser fluence in the UVirradiation of condensed NO films (>400 ML thick) at λ )273.7 nm in the “explosive photodesorption” regime. (a)Onset of collisional regime (Flaser ≈ 4.5 mJ/cm2). Fit of theslow peak to a Maxwellian distribution at 300 K. The totalyield is about 1.7 × 1013 molecules/shot. (b) Supersonicexpansion. The fit consists of a Maxwellian at 135 K anda shifted Maxwellian at 150 K with a 620 m/s streamvelocity. Note that the fast peak (maximum at ∼80 µs) thatis ascribed to electronically mediated desorption is distinctfrom the supersonic peak. Reprinted with permission fromref 18. Copyright 1988 American Institute of Physics.



examined. At relatively low laser fluences, clusterejection is indicated to last for 0.5-0.8 µs, after whichdesorption is largely in the form of monomers,whereas at high fluences, cluster ejection continuesup to ∼1.5 µs. The temporal evolution of the ejectionprocesses will be considered further in the nextsection.

Cluster observation has been common in the ir-radiation of frozen aqueous solutions of salts. In the266-nm irradiation of frozen CeCl3/H2O solutions(absorption is ascribed to electronic excitation of thehydrated metal35) at 3 × 108 W/cm2 (τpulse ≈ 8 ns),CeO(D2O)x

+ (x e 5) and D3O+(D2O)n (n ) 1-3)complexes are detected.35 Fragmentation of largerlaser-ejected clusters and/or condensation processesin the plume are implicated. (H3O)+(H2O)n andNa+(H2O)n (n ) 1-30) ionic clusters series have alsorecently been reported43 in the IR ablation of frozenaqueous solutions of a protein, but the mechanismof cluster formation in this case remains to be fullyelucidated. In the irradiation of frozen aqueoussolutions of XMnO4 (X ) Na, K) salts at 532 nm,36

neutral cluster ejection, established via photoioniza-tion at 322 nm, is observed upon irradiation at I ≈ 6× 107 W/cm2, while direct ejection of ions is observedat 2.5 times higher irradiances. Peaks correspondingto M(H2O)n

+ (M ) Na, K) clusters, with n values ashigh as ∼16, have been detected. Interestingly,though the cluster size distribution resembles thatobtained in supersonic jets, the estimated Machnumber for the plume expansion in the laser processis too small (section II.C.). Thus, as in the irradiationof HBr films,27 the observed cluster series is indicatedto represent or derive largely from clusters directlyejected from the film. Probably, in the case of thesesystems, the strong interaction of the metallic ionwith the solvent molecules prevents the clusters frombeing completely fragmented/evaporated to theirmonomers. Other factors, such as degree of internalexcitation, differences in the collisional rates in theplume, etc., may also affect the stability of the ejected

clusters in the various systems, but at present suchconsiderations are highly speculative. The impor-tance of the strength of the intermolecular interac-tions for the common observation of adduct peaks inMALDI has often been indicated.44,45

In all, cluster ejection in the irradiation at highirradiances is by now demonstrated for a number ofmolecular solids. In contrast, at lower fluences, it isa rather rare observation and is largely accountedfor by system-specific (substrate-mediated) mecha-nisms.14 Certainly, the generality of the statementis far from being proven, but elucidation of the factorsthat contribute to the evolution and survival rate ofthe clusters in the plume will help in addressing thisissue further. Yet, even if established, cluster ejectiondoes not provide a mechanistic understanding of thephenomenon.

2. Ejection MechanismsIn a different approach, the operation of two

different, competing mechanisms in the UV irradia-tion of van der Waals solids at high irradiances hasbeen shown through the comparative examination ofthe ejection efficiencies of nonabsorbing dopants(alkanes such as c-C3H6, c-C6H12, C10H22, and ethers/alcohols (CH3)2(CH2)nO, D2O) of varying bindingenergies to the matrix.39-41 C6H5CH3 is employed asa matrix because of its minimal fragmentation,46

thereby avoiding complications due to any photore-activity. Since the dopants do not absorb at theirradiation wavelength (λ ) 248 nm),46 their relativeejection efficiencies provide information on the natureof the energy dissipation in the film and of themechanisms by which it results in material ejection.Essentially, in the comparison of these systems, theparticularities of the excitation step(s) remain thesame, so the nature of the ejection process can beexamined independently of the photoabsorption.

The desorption efficiencies are probed as a functionof laser fluence via time-of-flight quadrupole massspectrometry (i.e., neutral desorbates are detected viaelectron impact). Interestingly, a fluence range (1 ×105-3 × 106 W/cm2, τpulse ≈ 30 ns) can be deline-ated in which the desorption signal as a function oflaser fluence (Figure 6) clearly follows an exponen-tial (Arrhenius-type) dependence, or more strictly,

The activation energies (∆Edes) for the laser-inducedejection of the various dopants are found to be inrather good correspondence with their binding ener-gies to the matrix, as these are determined byconventional thermal desorption spectroscopy (TDS)(Table 1).47 As a result, at these fluences, onlydopants that are weakly bound to the matrix areobserved in the gas phase.40 Thus, it is stronglysuggested that desorption at these fluences is con-sistent with surface thermal vaporization. Othermechanisms of ejection are shown to be unimportant.It is noted that for weakly bound dopants, thedesorption signal can be very high, a point whoseimportance will be underlined in section II.B.3.

Figure 5. Mass spectra of laser-ejected chrysene-d12recorded by postdesorption laser ionization with the focusedionizing beam 100 µm (upper graph) and 50 µm (lowergraph) from the surface. The desorption wavelength is 266nm, and the delay between desorption and ionization laserpulses is fixed at 0.4 µs. The parent ion envelope consistsof the molecular ion (m/z ) 240), the 13C isotopomer (m/z) 241), and satellite [M - nD] peaks associated with theloss of 1-4 deuterium atoms from the parent chrysene-d12. Reprinted with permission from ref 32. Copyright 1999American Chemical Society.

S ∝ x 12πmkBT

exp(- ∆Edes

kBFlaser) (2)



In contrast, in the irradiation at higher laserirradiances (I > 3.3 × 106 W/cm2), the ejectionintensities of the various dopants show no correlationwith their binding energies to the matrix.39,40 Evendopants that are strongly bound to the matrix (suchas C10H22, with a binding energy to the matrix of ∼0.8eV) are found to be ejected efficiently. Furthermore,

the dopant-to-matrix signal ratios reach values closeto the initial film stoichiometry (but, note thatdeviations are usually observed in the case of thestrongly bound dopants) (Figure 7a). It is demon-

strated that there is no change in the light absorptionprocess. Thus, these changes in ejection efficiencies,as compared to the observations at lower fluences,signify a change in the mechanism.

For the strongly bound dopants, the ejection signalrelative to that of C6H5CH3 remains nearly constantwith successive laser pulses. Furthermore, the Flasercorrespondence of their ejection efficiency closelycorresponds to that of the matrix.40,41,47 Both featuresindicate the unselective ejection of a film thicknessaccording to

where R represents the (effective) absorption coef-ficient. The dependences of both toluene and dopantsignals are well described by the same thresholdfluence, Fthr. These features establish that ablation

Figure 6. (a) Desorption intensity of neutral toluene(C6H5CH3) recorded from freshly deposited films as afunction of the incident laser fluence at 248 nm. (b)Semilogarithmic plot of the desorption yield as a functionof 1/Flaser. Reprinted with permission from ref 47. Copyright2002 University of Crete.

Table 1. Comparison of Desorption EnergiesDetermined by Thermal Desorption Spectroscopyand Activation Energies for the Laser-InducedDesorption below the Ablation Threshold (Reprintedwith permission from ref 47. Copyright 2002University of Crete.)

systema compoundETDS

(kJ/mol)bEdes

(kJ/mol)c

neat C6H5CH3 C6H5CH3 41 ( 2 30 ( 3(CH3)2O/C6H5CH3 (CH3)2O 16 ( 3 17 ( 3

C6H5CH3 35 ( 3 25 ( 3c-C3H6/C6H5CH3 c-C3H6 15 ( 3 14 ( 3

C6H5CH3 35 ( 3C6H12/C6H5CH3 C6H12 33 ( 4 22 ( 6

C6H5CH3 39 ( 3 25 ( 3C10H22 /C6H5CH3 C10H22 d 77

C6H5CH3 45 ( 5 31 ( 2a Mixtures of the indicated compounds with a 1:5 molar ratio

of the dopant vs C6H5CH3. b Binding energy of the compoundsas determined by thermal desorption spectroscopy. The bind-ing energy of C6H5CH3 differs in the different systems, due tothe influence of the relatively high concentration of thedopants. c Activation energies determined from conventionalsemilogrithmic plots of the laser-induced desorption signalsvs 1/Flaser. d C10H22 desorbs thermally at temperatures wellabove that at which all C6H5CH3 has desorbed.

Figure 7. Concentration [i.e., Idopant/(Idopant + Itoluene)] of(CH3)2O and C10H22 dopants in the plume as function ofthe laser fluence (Flaser) in the irradiation of frozen mixturesof these dopants with C6H5CH3. (a) Mass spectrometricmeasurements (λ ) 248 nm) and (b) molecular dynamicssimulations. The horizontal lines indicate the initial con-centration of dopants in the sample. The solid and dashedvertical lines mark the ablation thresholds for the corre-sponding systems (different due to the different heatcapacities and cohesive energies of the two systems).Reprinted with permission from ref 40. Copyright 2001American Institute of Physics.

lejected ) 1R

ln(Flaser

Fthr) (3)




entails the nearly nonselective expulsion of a volumeof material. Thus, UV ablation is qualitatively adifferent phenomenon from thermal desorption (usu-ally presumed for explosive photodesorption), and itcan be experimentally identified through the ejectionof strongly bound species. In other words, in the“explosive desorption” regime, two different fluenceranges can be distinguished where qualitatively dif-ferent mechanisms operate.

The previous conclusion corresponds well to thedelineation drawn by the MD simulations (for nano-second pulses) for the operation of different mecha-nisms above versus below a well-defined, specificlaser fluence (the threshold). Indeed, simulations40

on the (CH3)2O/C6H5CH3 and C10H22/C6H5CH3 sys-tems yield results that are in very good correspon-dence with the experimental observations (Figure7b). Interestingly, the simulations show that thestrongly bound dopants are ejected exclusively withinclusters of the matrix, whereas the weakly boundones are ejected largely in the form of monomers.Survival of a percentage of the clusters at thedetection volume will affect the ionization process31

and may account, at least partly, for the fact thatthe mass-spectroscopically determined ratio of dopantand matrix intensities appears to be lower than thefilm stoichiometry (Figure 7a). Thus, though still notfully proven experimentally, the previously definedcriterion for the ejection of dopants that are stronglybound to the matrix appears to be intimately cor-related with the criterion advanced by the simula-tions for the onset of cluster ejection.

The change in the mechanism of material ejectionat a specific fluence suggests the operation of aprocess that is competitive with the thermal surfaceevaporation and which becomes dominant above thethreshold. The nature of this process is indicated bythe abrupt change in the slope of Figure 6b at afluence (∼40 mJ/cm2) well below the determinedablation threshold. At this fluence, the surface filmtemperature is estimated to correspond to the melt-ing point of the matrix.41 In support of this cor-respondence, above this fluence, the mobility ofweakly bound dopants within the C6H5CH3 matrixis indicated to become quite high (diffusion constantindicated to be g10-12 m2/s). In parallel, as estab-lished by optical examination, structural changes areinduced to the film (neat C6H5CH3 films) (Figure8a).47 Changes in the substrate have been monitoredby the transient transmission and reflection of aHeNe beam incident on the film. A sharp decreasein the transmitted signal (Figure 8b) is observed100-200 ns after the UV laser pulse. By means ofgated imaging techniques, the transmission decreaseis established to be due to light scattering and notabsorption. The time scale of the light scattering isconsistent with that of the formation of bubbles inliquids overheated above their boiling point (sectionIII.A.). For fluences below the threshold, the bubblesfinally collapse, and the morphology of the film ispermanently modified. With increasing laser flu-ences, the bubbles grow in size and number, asindicated by the exponentially increasing scatteredlight intensity (Figure 8b). At a specific laser fluence,

their growth becomes high enough to result inmaterial ejection.

For organic molecules, the optical penetrationdepth (labs), at least at λ < 300-350 nm, is in the100 nm-10 µm range (e.g., for the C6H5CH3 films,labs ≈ 4 µm). The thermal relaxation time, tthermal )FCPlabs

2/4D ≈ 10 µs (where F is the density, C the heatcapacity, and D the thermal conductivity), is muchlonger than the typical nanosecond excimer laserpulses. Thus, energy remains confined (thermalconfinement regime) and can result in melting and,at higher fluences, in the strong overheating of theliquid. Liquids heated above the temperature corre-sponding to the external pressure at their saturationtemperature are thermodynamically metastable (Fig-ure 9a), since their chemical potential µL is higherthan that of the vapor, µV

48-60 (in the present case,the external pressure is specified by the recoilmomentum of the evaporating gas). However, thetransformation (boiling) requires bubble formation,which is limited by the work necessary for theformation of a new interface within the liquid (i.e.,the surface tension, σ).52,53 According to thermo-dynamics, the free energy for bubble formation isgiven by

Figure 8. (a) Transmission images of HeNe beam onC6H5CH3 films following irradiation with (left) ∼2000pulses at ∼30 mJ/cm2 and (right) 50 pulses at ∼55 mJ/cm2 (i.e., below the ablation threshold, but at a fluence atwhich melting should occur). (b) Time-resolved transmis-sion at λ ) 633 nm upon irradiation of condensedC6H5CH3 films with one UV pulse (λ ) 248 nm) at theindicated fluences. The signal has been normalized to thetransmitted intensity before the UV pulses. Reprinted withpermission from ref 47. Copyright 2002 University of Crete.

∆G ) N[µV(PV) - µL(PL)] - (PV - PL)VV + Aσ

)4πRbubble

3

3(psat - PL)(1 -

νL

νV) + 4πRbubble

2 σ (4)



PV and PL are respectively the pressure inside thebubble and the ambient pressure of the liquid, µ isthe chemical potential of the corresponding states,A is the surface area of the bubble, N is the numberof molecules within the bubble, and νV and νL are,respectively, the specific volume of the vapor and ofthe liquid. In the first equation, the second termrepresents the work directed against the pressureforces and the third the work for the bubble interfaceformation. The second equation derives under theassumption of constant P and T and relatively lowdegree of superheating.53 For bubbles exceeding acritical value, Rcr ) 2σ/(PV - PL), given by therequirement of mechanical stability of the bubble, ∆G< 0 and subsequent growth is thermodynamicallyunavoidable. However, for smaller Rbubble, the surfacetension term dominates and ∆G > 0. Consequently,the number of homogeneously nucleated bubblesformed via thermal fluctuations is given by

where No is the number density of the molecules inthe liquid. The corresponding rate J (number pervolume and time) for the spontaneous formation ofbubbles with the critical size Rcr is

Here, kf denotes the molecular evaporation rate(which obtains somewhat different values dependingon the derivation), η relates the pressure within thecritical bubble with the saturation pressure psat(T)at the corresponding temperature, and vL is thespecific volume of the liquid (assumed to be incom-pressible). The mean time of formation of a criticalnucleus in a volume V is τ ) (JV)-1 and determinesthe lifetime of the metastable state. Both σ [ap-proximated usually by σ ) σo(1 - T/TC)x, where TC isthe critical temperature of the compound]52,53 and PV

- PL factors depend sensitively on temperature.Thus, bubble growth and the lifetime τ dependcritically on the maximum attained film temperatureand its subsequent temporal evolution.

The surface film temperature drops rapidly afterthe end of the laser pulse as a result of the evapora-tive cooling. For small overheating, the work providedby the evaporating gas (i.e., the first term in eq 4) isinsufficient to compensate for the surface tensionlimitation, and thus bubble growth eventually halts.However, at higher fluences, the surface temperaturereaches high enough values for bubble growth tobecome significant. In other words, the higher theattained temperature, the deeper the excursion of theliquid into the “metastability region” indicated inFigure 9b. Due to the sharp decrease of σ and theincrease of (PV - PL)2 factors with temperature, Jincreases sharply (by orders of magnitude) over asmall temperature range, which accounts for the“threshold-like” behavior of laser ablation. Further-more, since σ and PV are related to the cohesiveenergy of the system,48-53 the threshold for materialejection can be expressed as

where Lp represents the optical penetration depth,Ecr the critical energy density (J/cm3), and CTo (C,heat capacity; To, initial temperature) the initialenergy content. The increase of the threshold withincreasing system cohesive energy has been repro-duced by both MD simulations40 and experimentalstudy39 (Figure 7). However, these qualitative aspectsaside, the quantitative analysis of the dynamics ofexplosive boiling on the nanosecond time scale doesnot appear to be well established. For instance, thedegree of overheating necessary for material ejectionis found both experimentally40,41,47 and by the MDsimulations28-30 to be lower than that suggested byconventional thermodynamic considerations.49-51,58,59

The latter usually consider that volume ejectionoccurs at the spinodal decomposition limit. At thislimit, the stability criteria are violated (e.g., (∂P/∂V)T) 0, (∂T/∂S)P ) 0, which is physically unacceptable)

Figure 9. (a) Schematic of the temperature dependence(at constant pressure) of the chemical potential of theliquid (1) and of the gas phase (2) of a one-componentsystem. The crossing corresponds to the equilibrium point(i.e., projection of the binodal for the given pressure). (b)The binodal and spinodal curves and the region of meta-stability for toluene. C is the critical point, and Tc ) 593 Kand Pc ) 41 bar are respectively the temperature and thepressure of the compound at the critical point. Reprintedwith permission from ref 47. Copyright 2002 University ofCrete.

Nnhom ) No exp[- ∆G(Rbubble)

kBT ] (5a)

J ) kfNo exp[- ∆G(Rcr)kBT ] )

kfNo exp[- 16πσ3

3kBTL(ηpsat - PL)2]η ≡ exp[ νL

RTL(PL - psat(TL))] (5b)

Fthr ) Lp(Ecr - CTo) (6)



and the liquid phase becomes unstable, decomposingspontaneously (i.e., in an activationless process) intoa gaseous and liquid droplet mixture. The fact thatthe massive ejection associated with ablation isobserved at much lower temperatures indicates theprocess to be under kinetic rather than thermody-namic constraints. [It is noted that recent densityfunctional analysis61 and biased Monte Carlo simula-tions62 on superheated liquids have indicated severalshortcomings of the classical nucleation theory. Infact, even the assumption of a spherical form for thecritical bubbles is indicated to be a simplification (atleast for large overheatings).]

According to the above, there is essentially acompetition between surface and volume vaporiza-tion, with the latter becoming dominant above aspecific fluence. A similar delineation has also beenconsidered by Wu et al.64 in the irradiation of frozenglycerol at 193 nm, largely on the basis of theobserved desorbate translational distributions (sec-tion II.C.). High enough temperatures (∼550 K),suggestive of “phase explosion”, are also estimatedin the irradiation of frozen aqueous solutions ofsalts.36 As will be discussed in the next section,explosive boiling directly accounts for the indicatedunique features associated with material ejection athigh laser fluences (for nanosecond pulses). Thus, inaddition to the usual criterion of thermal confine-ment, the criterion of the rate of homogeneous bubbleformation being faster than the rate of evaporativecooling must be employed to establish whether mate-rial ejection represents really a volume ejection (abla-tion) process or simply a surface thermal evaporationprocess.

3. Implications for Relevant Applications (MALDI, MAPLE)

There are a number of immediate implications ofthe suggested delineation and, in particular, of theidentification of nanosecond-induced laser ablationwith explosive boiling. If the upper layers of thesubstrate are superheated, the underlying materialshould also melt. This is illustrated for dopants thatare weakly bound to the matrix by the contributionthrough diffusion and thermal desorption from theunderlying (nonejected) layers.38,41 The contributionof the prolonged (“postablation”) thermal evaporationto the total ejection signal in UV ablation has oftenbeen noted.1,11,14,32,36 This may result in discrepanciesfrom the previously suggested ejection of a well-defined volume of material (e.g., for the previouslynoted deviations for the strongly bound dopants).Furthermore, the consequent segregation effectswithin the melt must be carefully considered whenmultipulse irradiation protocols are employed, sincethey can be severe and can seriously complicateinterpretation of the results.41 These effects on thestructural integrity of films and their implication forpractical applications63 have been noted.

The most important result of the studies on simplesystems appears to be the finding that dopants/analytes that are strongly bound to the substrate/matrix can be ejected only in the ablative regime.Therefore, in MALDI and MAPLE techniques, theobservation of the biopolymers in the gas phase only

at high fluences appears to be due to the existenceof an actual, physically significant threshold. Incontrast, the matrix can be detected at lower fluences,due to the fact that it can desorb thermally. Accord-ingly, the desorption/evaporation terms in theMALDI and MAPLE acronyms are misnomers. InMALDI studies, there is still controversy about thisissue65,66 on the basis that the detection limit isdependent on the employed spectroscopic technique.We note that this does not affect the validity of theargument for the presence of a threshold, but itrather is related to the question of the specificationof its exact value.

Along the same lines, the previous delineationsuggests that terms such as “thermal ablation” maybe misleading.1 The term “thermal ablation” is em-ployed to describe extensive desorption, largely con-sistent with thermodynamic properties (e.g., ∆Hsubl,∆Hvap, etc.), whereas the “nonthermal” term is usedto describe deviations from this behavior. However,it is clear that the amount of material removed, evenif extensive, does not define a new phenomenon, andinstead the term “ablation” should be reserved for theunique characteristics/features that are observedabove a specific fluence.

With all caution, it is interesting to note that adistinction relevant to the thermal surface desorptionversus volume material ejection delineation is oftendrawn, even in the case of polymers, between a lowlaser fluence range in which mass loss occurs due tothe depletion of light volatile species (photoproducts)and the fluence range in which actual etching (abla-tion) is effected.1,11 This similarity suggests thatcommon features become evident for a variety ofmolecular systems, thus providing the basis for thedevelopment of more “universal” laser -material-removal models. On the other hand, as the studiesin section II.A show, the yield due to direct electronic-ally mediated processes is quite low, indicating thatsuch processes (invoked within a so-called “photo-physical” model of ablation) cannot be relevant forthe extensive material ejection observed at highfluences.

Another implication concerns the physical signifi-cance of the activation energies determined fromthe semilogarithmic plots of ejection signals versus1/Flaser (Figure 6b). It is clear that the value deter-mined in the ablative regime must be related tothe activation energy of the process (related to thebubble nucleation energy within the “explosive boil-ing model”) and cannot be related with the bindingenergy of specific molecules. In fact, even the valuesdetermined below the threshold are related to ∆Edesfrom superheated liquids and may differ from thetabulated ∆Hevap under ambient pressure. This dis-cussion illustrates plausible pitfalls in the inter-pretation of Arrhenius-type fittings to the ejectionefficiencies of analytes in UV ablation (a usualapproach in the study of more complex systems). Thisaspect has been well emphasized by Zhigilei andGarrison.28-30

The explosive boiling model can also explain thesuggestion of the MD simulations for the ejection ofmaterial in the ablative regime mainly in the form


of clusters. Since the sensible heat (CP∆T) that hasbeen “stored” in the metastable liquid suffices for thevaporization of only a fraction of the liquid,57,58 a largepercentage of the material is ejected in the form ofdroplets, exactly as observed in the molecular dy-namics simulations for the longer laser pulses (150ps). (Thermodynamically, this results from the factthat explosive boiling occurs under nonconstant PLand T.52) As a result of the ejection of material largelyin droplets/clusters, the requisite energy may be quitelower than ∆Hevap. The explosive boiling model canalso account for the ejection of the weakly bounddopants mainly as monomers and the ejection of thestrongly bound analytes within clusters of the ma-trix.40 During homogeneous bubble formation, thedesorption rate of the strongly bound dopants intothe bubbles is very slow, and as a result, these speciesare ejected only within the liquid droplets formedduring the process. Explosive boiling of mixtures hasbeen considered previously.52,53 A plausible implica-tion is that the fast “exploitation” of the excess energyfor the vaporization of the volatile component (in thecase MALDI, of the matrix itself) into bubbles mayresult in the rapid cooling of the material, therebylimiting thermal degradation of any labile dopantsand also heat conduction (and consequently thermaldegradation) to the substrate. This possibility isunder examination.

Several implications of the suggested cluster ejec-tion in relationship with protein stabilization andionization processes are discussed in the other ar-ticles of this issue. The discussion therein furtherhighlights the importance of considering clusterejection in ablation. Herein, we note only that if alarge number of clusters survive at the detectionvolume, their presence may seriously affect thedetection efficiency of the cluster-embedded spe-cies.34,52 Calibrating the spectroscopic techniques fortheir relative sensitivity toward clusters versus mono-mers is practically impossible, especially since thedistribution of cluster size is not known. Because ofthis, quantitative gas-phase measurements may tendto be “artifactual”, and discrepancies between studiesmay appear according to the employed detectiontechnique.

4. Energy Dissipation Processes

Despite their neglect in the previous discussion, thenature of the initial electronic excitation and subse-quent deactivation step(s) is expected to be of directrelevance to the material ejection process. Thoughthese processes are usually implicated in the ablationof photolabile systems (section III.B.2.), their consid-eration is nevertheless important, even for photoinertsystems. In fact, it has led to the prediction of theablation threshold on a basis quite different from thatassumed in the MD simulations. Fain and Lin67 haveargued that the initial population of highly excitedphonons increases further the rate of the electronicrelaxation, thereby introducing a mechanism of posi-tive feedback. The rate of energy transfer from theelectronically excited molecules to the phonons of themolecular crystal increases with decreasing energyseparation (∆Edif). Consequently, electronic energy

decay will accelerate as higher and higher phononsare populated (Figure 10). When energy flow into the

phonons exceeds their decay rate, their numberbegins to increase exponentially, i.e., an “avalanche”phenomenon, thus resulting in explosive-like vapor-ization. It was later argued68 that the avalanchephenomenon may alternatively be due to the interac-tion of the electronic excited states with localizedvibrational excitons (vibrons), which should decaymuch slower than phonons, thereby making “hotspot” formation more likely.

The model successfully accounts for the observed13-15

dependence of the threshold on film thickness. Fur-ther experimental support for the model derives fromthe study68 of the dependence of the luminescencedecay of films of Cl2 on laser fluence (λ ) 308 nm).As the ablation threshold is approached, the decayis found to exhibit distinctly nonexponential temporalbehavior. The deviation from the exponential decayis demonstrated to be due to the accelerated internalconversion of the assessed S1 state of Cl2, rather thanto energy transfer or other plausible decay mecha-nisms. (However, the possibility that the formationof a metastable liquid may also affect the electronicenergy decay process was not recognized.) Similarquenching of the luminescence close to the thresholdis also reported for 2,5-dihydroxybenzoic acid andferrulic acid.69,70 The plausible importance of excitonpooling mechanisms for ionization in MALDI hasoften been underlined.70-73 As discussed in sectionIII.C, similar processes are indicated to be importantin the ablation of liquids and also of polymers.

Quite early, Levine and Vertes74-77 argued that therates of vibrational energy transfer between analytesand matrix may be one of the key factors in MALDI.In the initial studies, the limitation in the rate ofenergy transfer was invoked mainly to account forthe stability of the ejected proteins within an es-sentially simple thermal desorption mechanism (sec-tion II.D.2). The importance of the energy-transfer

Figure 10. Representation of the de-excitation processesaccording to Fain and Lin. Optical excitation occurs fromthe vibronic levels of the ground electronic state, φo, to theexcited electronic state, φs. Isoenergetic with φs is a set ofvibronic levels, φl, belonging either to another electronicstate or to the ground state, φo. A set of excited states, φl,interacts with phonons, giving rise to the stimulatedemission of phonons, pωq. Reprinted with permission fromref 67. Copyright 1989 American Institute of Physics.



processes has been ramified recently by moleculardynamics simulation78 of the vibrational-to-transla-tional energy-transfer processes in the irradiation ofa solid O2 system. For irradiation at low laserfluences, this rate is rather slow because the crystalthermally expands and, as a result, the couplingefficiency of the vibrational-to-phonon modes de-creases. However, at much higher excitation rates,the sample melts, and diffusion of the vibrationallyexcited species occurs. The interaction of excitedspecies results in very fast energy transfer to trans-lational modes. This results in a pressure pulse (thesimulation is performed for 30-ps pulses, closelycorresponding to the stress-confinement regime),which causes rapid material expansion and eventu-ally ejection in the form of clusters. The importanceof the interaction of vibrationally excited moleculesfor the energy-transfer rates has been demonstratedexperimentally in picosecond studies in liquids.79

In all, despite the limited experimental evidence,novel aspects of energy decay/dissipation processesare indicated that are of direct relevance to thematerial ejection process. These aspects have notbeen considered in the “explosive boiling” modeldiscussed previously. Yet, the models may not becontradictory to each other, and it is plausible thatthey have just captured different aspects of thephenomenon. The studies in the present section maywell describe the initial stage of energy dissipationinto vibrational/phonon modes, which is, in any case,the starting point of the MD simulations based onthe “breathing-sphere” model.28-30 The good agree-ment of the MD simulations with experiments isprobably due to the fact that, for nanosecond irradia-tion, electronic decay is faster than or at mostcomparable with the laser pulse width. Thus, mate-rial ejection is mainly determined by the energydistribution following electronic state deactivation(note also that the processes considered in refs 67 and78 suggest an acceleration of the electronic energydecay and dissipation). In fact, optical properties ofsuperheated liquids are indicated to differ greatlyfrom those at equilibrium conditions (section III.C),and the same can be expected for the electronic de-excitation rates. Thus, even if the explosive boilingmodel is in principle correct, the quantitative analysisof the experiments must eventually take into accountthese effects. To this end, it is important in futurestudies to examine energy decay processes in parallelwith the formation of the metastable liquid phase.Substrate-mediated laser superheating of liquids(section III.A) can provide a most convenient methodto study photophysical properties under metastableconditions. Furthermore, more elaborate theoreticalwork and simulations including the electronic de-grees of freedom are evidently required.

The importance of electronic processes may be atleast partly elucidated via the study of the wave-length dependence of the ejection processes. For the“phase explosion” mechanism, material ejection shouldbe wavelength independent, aside from any (effective)absorption coefficient corrections. Some early studieson IR ablation have been reported,80 but no systemhas been studied systematically enough in both UV

and IR to enable a meaningful comparison. Similarly,very little has been reported on the comparison atdifferent UV wavelengths. A study on the compara-tive examination of benzyl derivatives at 193 and 248nm was of a preliminary nature.37 Examination ofwavelength dependence is also important in assess-ing the relative contribution of photochemical versusthermal processes to material ejection.1,11

5. Laser Pulse Dependence

The previous results hold for irradiation with UVnanosecond pulses (typically excimer laser). In theirradiation with shorter pulses, material ejection isconsidered to be in the stress-confined regime andto be due to the “mechanical” rupture (spallation) ofa surface layer under the action of the tensilecomponent of the generated stress, as described indetail in section III.B and in the contribution byPaltauf and Dyer in this issue. MD simulationsprovide further support for this model.30 In compari-son with thermal evaporation or explosive boiling,photomechanical ablation requires less energy perunit volume for material removal, since only thebonds at the interface between the ejected layer andthe underlying one need to be broken without exten-sive volatilization. Therefore, ablation can be effectedat lower substrate temperatures and thus lowerincident laser fluences. For this reason, extensiveattention has been given to the exploitation of this“cold ablation” mechanism. A number of additionaladvantages are expected for this mechanism, e.g.,minimization of the segregation/diffusion effects de-scribed for nanosecond pulses (section II.B.3), re-duced photolysis (due to the lower fluences required),and reduced recombination product formation (sec-tion II.D) in the case of photolabile systems, etc.

In the few reports thus far on the pico- andfemtosecond ablation of molecular solids, the thresh-old is sometimes,47,82 though not in all cases,81,83,84

found to decrease as compared with nanosecondirradiation. For laser pulses that are not shortenough (e.g., ∼50-100 ps), the invariability of thethreshold84 is understood, since the change in thepulse duration is insufficient to effect a switch fromthe thermal to the stres-confinement regime. How-ever, the invariability for femtosecond pulses ob-served in few studies remains difficult to account for.On the other hand, neither the decrease observed inthe other studies can be unequivocally ascribed to thesuggested mechanistic change. In femtosecond ir-radiation, electronic multiphoton processes can beexpected to become highly likely and may be partlyresponsible for the decrease. Indeed, nitrogen filmscan be ablated efficiently with picosecond pulses (λ) 263 nm, τpulse ) 8 ps, 5 J/cm2)89,90 and frozenacetone and methanol with femtosecond pulses (λ )790 nm, τpulse ) 130 fs, at 4 × 1015 W/cm2),91 athoughthese compounds are transparent at the irradiationwavelengths (however, these studies were performedat very high irradiances, well above the threshold).The increasing importance of multiphoton processesis indicated in the ablation of polymers1,92,93 as wellas in surface science (desorption induced by multipleelectronic transitions, DIMET).88


Even in the absence of multiphoton processes,electronic energy processes, which are neglected inthe thermal versus stress-confinement delineation,will be of determining importance in the irradiationwith femtosecond pulses. For instance, the rate ofelectronic energy decay (usually g10 ps) will deter-mine the degree of effected stress-confinement. Fur-thermore, energy de-excitation processes may differgreatly from the nanosecond case, with significantimplications for the material ejection process. Theimportance of electronic processes appears to be alsoindicated by the observation of enhanced desorbatefragmentation in the first few femtosecond stud-ies.47,83 This result is incompatible with the photo-mechanical model, since this model suggests materialto eject largely in the form of cold, “large” chunks.On the basis of the discussion in section II.D.1, thisshould result in reduced fragmentation, at variancewith the experimental results. It is also noted that,in the case of polymers, a most dinstinct differencein the photoproducts that remain in the substrate hasbeen observed85,86 between nanosecond and femto-second ablation. It appears that the examination ofthe chemical products may be most informative aboutthe mechanisms in the irradiation with short laserpulses.

A most interesting indication on the interplay ofthe various factors that become important withdecreasing laser pulses has been reported in the caseof copper-phthalocyanine films.94 The ablation thresh-old is found to decrease with decreasing laser pulsewidth (35 mJ/cm2 for 170-fs pulses; 80 mJ/cm2 for250-ps pulses; 140mJ/cm2 for 100-ns pulses; in allcases, λ ) 780 nm). Multiphoton processes areindicated to occur. Interestingly, the etching depthin the femtosecond and picosecond cases remainsnearly constant with increasing Flaser above thethreshold, which differs distinctly from the usualgradual increase observed for the nanosecond case(e.g., the one in Figure 1 or 6a). The differencesuggests a mechanistic change. A pressure-drivenmaterial ejection is suggested for the femtosecond orpicosecond case.

6. Substrate-Mediated Ejection/Confined AblationThe previous discussion has focused on processes

induced in the irradiation of the “free” surface ofabsorbing films. In several experiments, materialejection is effected via irradiation through the trans-parent substrate that supports the condensed filmor, in other cases, via light absorption by the sub-strate itself. This mode of irradiation presents anumber of experimental advantages, in particular ifablation of cryogenic films is to be employed for theproduction of molecular beams for use in kineticstudies or for surface experiments.19-22 However, itis now understood that the mechanisms of materialejection may differ from those described above for theirradiation of the free surface of the substrates. Anumber of different terms, such as confined ablation,backward irradiation ejection (or even the apparentlycontradictory term “transmission irradiation mode”),and laser-induced forward transfer, have been em-ployed to describe the laser-induced material ejectioneffected in these cases.

Substrate-mediated laser-induced desorption hasbeen extensively studied for (sub)monolayers, mainlywith a view toward analytical applications.95-97 Theinterest derives from the fact that, for small orweakly bound molecules, laser-induced substrate-mediated desorption can be effected with minimaldecomposition. Thus, combined with gas-phase diag-nostic methods, it provides a powerful tool for thetrace analysis of adsorbates on a variety of surfaces.This process will not be discussed further, since ageneral consensus96,97 on its purely thermal equilib-rium mechanism appears to have been reached.Instead, the discussion will focus on the UV-inducedejection observed in the irradiation of multilayers.

For transparent multilayer films on an absorbingsubstrate, heat conduction from the substrate (incompetition with conduction into the substrate, theratio of energy conducted to the film versus thesubstrate being xDf/Ds) can result in the heating ofthe overlying layer. Assuming that the interfacetemperature rise is instantaneous, the temperatureprofile in the film can be approximated by98

where z represents the distance from the interface,R the surface reflection, F the density, and Cp theheat capacity, and the subscripts f and s refer to filmand substrate, respectively. Analytical solutions ofthe temporal and spatial (depth) temperature evolu-tion in films of N2 and H2O condensed on an absorb-ing metallic substrate have been presented.100,101 Ifthe film heat conductivity is high enough, the surfacetemperature of the condensed film will rise suf-ficiently to result in material loss via thermal vapor-ization. At even higher fluences, the interface tem-perature increase may result in local melting andsubsequent (heterogeneous) vaporization of the ma-terial adjacent to the substrate in a process similarto that considered in section III.A. For moderateenergy input, the vapor formation may be insufficientto overcome the film adhesion and to provide thenecessary kinetic energy for the ejection of theoverlying layer. In this case, only deformation of thefilm may occur. However, at higher energy absorptionrates, extensive bubble growth will occur, resultingin the detachment of the film and the developmentof a high pressure, leading to spallation of the wholematrix.34,37 The importance of the induced pressurehas also been indicated by MD simulations.102,103 Adetailed analytical description has been reported109

but for different type of films.The previous features have been been recently

visualized by novel ultra-high-speed optical micros-copy in the irradiation of viscous liquid R-terpineoldoped with highly absorbing BaTiO3 nanoparticles(because of the liquid state, the dynamics differssomewhat from that of solids).104,105 The film thick-ness is 10-15 µm, and absorption is limited to 1 µmat the support-film interface. At low fluences (∼20mJ/cm2), deformation of the film can be detected, butno material ejection occurs. At somewhat higherfluences, however, the vigorous bubble formation

∆T )(1 - R)Flaser

Ffcp,fxDf + Fscp,sxDs

1xπt

e-z2

4Dft(7)


close to the interface results subsequently in materialejection. Material removal is initiated at ∼500 ns,well after the laser pulse (τpulse ≈ 150 ns).105 The delayis acribed to the slow formation of the bubbles dueto the high viscocity of the medium. Most interest-ingly, two distinct regimes104 could be identified inrelation with the dynamics of the plume ejection,likely due to the different degree of bubble growthand impulse strength provided for material ejection.For intermediate laser fluences, the plume collapsesinto a jet (“jetting regime”) after 1 µs. In contrast, athigher fluences, a dense plume consisting of vaporand micrometer-sized particles/droplets is ejected.

The ejection of large, micrometer-sized particles inthe irradiation at high fluences has also been dem-onstrated34 in the irradiation (λ ) 10.6 µm, 2 J/cm2)of 2,5-dihydroxybenzoic acid/poly(ethylene glycol)films on silica substrate by AFM characterization ofthe trapped ejected material. The indicated volumeejection should result in the ejection in the gas phaseof embedded dopants/analytes independently of theirbinding energy. Indeed, efficient ejection of bio-polymers106-108 via substrate-mediated absorption/heating has been demonstrated. Advantage is takenof this irradiation mode for the deposition of polymersand biomolecules in the so-called laser-induced for-ward transfer deposition described in the article byChrisey et al. in this issue. However, difficulties inthe reproducibility of the signals, at least in theanalytical applications, have been noted. Likely, thisis due to the high sensitivity of the ejection processto film thickness,19,22,103 a dependence that is not socritical for irradiation of the free surface of the films.

Several other studies have been performed underconfined ablation conditions and will be mentionedin the appropriate sections. Interest presents theattempt110 to use surface plasmon resonance spec-troscopy (i.e., probing the surface plasmon frequen-cies of the metal substrate on which the films arecondensed) to monitor with nanosecond time resolu-tion the laser-induced (λ ) 248 nm, τpulse ) 30 ns,8-26 mJ/cm2) material ejection process from con-densed films (2-propanol, acetone, and tetrafluo-romethane). Differences in the time delay of theejection onset from a few monolayers (thickness of∼50 Å) were observed according to the nature of theadsorbates. Unfortunately, the observed differencesare difficult to analyze, because of the combined effectof substrate heating and absorption by the film.Despite its indicated potential, the technique has notbeen pursued further in the case of frozen films.

Finally, a first intriguing report87 on femtosecond-induced substrate-mediated material ejection hasbeen published. Near-infrared (λ ) 800 nm, τpulse ≈150 fs, I ≈ 5 × 1011 W/cm2) laser desorption ofmultilayer benzene films on Pt results in a hyper-thermal translational component. Ejection was as-cribed87 to impulse transferring from the metallicsurface to the topmost desorbate layers via elasticcollisions of the molecules, though a pressure-inducedmaterial ejection process also seems likely. In otherstudies,99 ejection upon femtosecond laser irradiationis indicated to be highly forward peaked, much moreso than in the corresponding nanosecond case. This

high forward peaking has been taken advantage offor the deposition of submicrometer structures ofbiopolymers.99

In all, for nanosecond pulses and at least forphotoinert systems, through the parallel contributionof experiments, analytical considerations, and simu-lations, a unifying understanding of the processesleading to the laser-induced material ejection seemsto be attained. Certainly, several aspects remain tobe addressed for a fully satisfactory model to beestablished. Furthermore, it will be important toexamine how these processes evolve with increasingmolecular size and cohesive energy of the systemfrom these model solids to more complex ones.Nevertheless, it is clear that new directions offundamental importance, e.g., in thermodynamics,molecular photophysics, and molecular structuralrearrangements, have opened up. As for the materialejection induced by ultrafast irradiation, thoughdetailed molecular dynamics and analytical work hasalready been performed, very little can be assumedat present, as clearly experimental work has beenvery limited.

C. Desorbate Translational DistributionsIn principle, detailed information on the mecha-

nisms of material ejection would be expected fromthe examination of the desorbate translational, an-gular, and internal energy distributions. Typically,ionization of the desorbates with an electron beamor with a second laser pulse is employed to establishtheir velocity distribution from the time of their flightfrom the substrate to the detection volume. Thecorrections necessary for the proper analysis of thesespectra are well described in the literature111 (thoughin the case of ablation, arguments for its modificationhave appeared114). Internal energy distributions canbe conveniently probed via state-selective ionizationof the desorbates, usually via resonantly enhancedmultiphoton ionization (REMPI). Determination ofthe angular distributions is quite demanding on thevacuum system design if detection is based on massspectrometric techniques, but more straightforwardapproaches employing optical imaging have beendeveloped.152

Characterization of the desorbate distributions hasproven to be a most powerful approach in the studyof surface-mediated photodesorption, yielding de-tailed information about energy distribution/dissipa-tion in the desorption process.111 However, in abla-tion, due to the large amount of ejected material,numerous collisions occur between the desorbates,resulting in significant modification of their initialdistributions. Thus, the translational distributionsrecorded in laser ablation are expected to representa convolution of the initial “impulse” given duringejection and the subsequent postdesorption dynamicsin the plume. However, at present, the relativecontribution of these processes is not clear, so modelsand explanations usually deal with the one or theother extreme.

Elucidation of the factors affecting translationaldistributions is of major importance, because they areone of the main experimental results available in


ablation studies. Furthermore, the energy transferbetween the various degrees of freedom that iseffected by the postdesorption collisions may becrucial for the various features of ablation (e.g., forthe stability of ejected analytes in MALDI, for theresolution of the mass spectra in laser ablation massspectroscopy and MALDI, for the structural qualityof films deposited by MAPLE, etc.). In the case ofrealistic substrates, interpretation of the transla-tional distributions is hampered by the formation ofplasma. In contrast, in the case of van der Waalsfilms, due to their low cohesive energies (and thelarge ionization potential of the molecules), materialejection can be effected at fluences at which plasmaformation is minimal. Thus, comparison of experi-mental data and theoretical models is more straight-forward. [In this section, no distiction is madebetween IR- and UV-induced processes, since therelevant information is very scarce. Yet, this is a mostimportant question that needs to be addressed in thefuture.]

Illustrative examples of the change in the de-sorbate translational distributions with increasinglaser fluence are provided in the UV irradiationof condensed NO films18 and of frozen glycerolfilms.64 At intermediate laser fluences (i.e., corre-sponding to thermal desorption), the spectra areusually well described by Maxwell-Boltzman distri-bution with a temperature close to the estimatedfilm temperature. In contrast, at high laser fluences,the desorbate translational distributions (Figure 4)can generally be described by “shifted” Boltzmandistributions (though several exceptions are alsonoted26,139,158,166):

The drift velocity υdrift can be related to the modelsdescribed below. In several cases, the description ofthe distributions requires two shifted Maxwellians,20-22

but the physical significance of the two componentsis not clear. The change of the translational distribu-tions to shifted-Boltzmann distributions is oftenascribed to an adiabatic-like expansion of the mate-rial due to the high ejected quantity.18-22,64,112-114

With increasing laser fluence, the average (⟨Etrans⟩)and the most probable translational energies(E trans

mp ) shift to higher values (Figures 2 and 11),consistent with the idea that the higher amount ofejected material results in a “tighter” adiabaticexpansion. A forward-peaked angular distribution ofthe desorbates is often observed, described usuallyby a cosp ϑ (with p as high as 25) dependence,111

where ϑ is the angle from the perpendicular to thesurface.

State-specific ionization of the desorbates has beenused to probe the internal energy distributions, thuscomplementing translational distribution studies. Inthe 193-nm ablation of condensed NO films22 (under“confined ablation” conditions), the average rotationalenergy is found to be more than ∼10-fold smaller

than ⟨Etrans⟩ for both υ ) 0 and υ ) 1 (Table 2).Additionally, the vibrational energy is low, e.g., formolecules with Etrans ) 0.6 eV, the υ ) 1/υ ) 0vibrational distribution was 3 ( 1%. The observationof translationally fast but rotovibrationally colddistributions is strongly indicative of the developmentof a supersonic expansion. The cooling effect as aresult of the expansion of the material has also beendemonstrated in the “laser evaporation” of anilinefrom a cryogenic CO2 matrix on an absorbing sub-strate115 (acting as a heat source) as well as in theCO2 laser-induced ejection of benzimidazole dissolvedin a glycerol/water matrix.116 A significant decreasein the vibrational temperature is demonstrated as afunction of the time delay or position between theablating and ionizing laser beams. The degree ofcooling is indicated to depend on the matrix, withmatrix molecules of fewer internal degrees of freedomproviding more effective cooling.116

Analytical descriptions and computer simulationsof the postdesorption dynamics provide further in-formation on the extent of the collisional perturba-tion. Usually, a 1D geometry (planar front of the

dN(υb,T,υdrift) ) N( m2πkBT)3/2

exp- m2kBT

[υx2 +

υy2 + (υz - υdrift)

2] d3υ (8)

Figure 11. (a) Fluence dependence of the fitted center-of-mass velocities of the TOF spectra of the HOCH2CHOH(61 amu) and CH2OH (31 amu) peaks detected in the UVirradiation (λ ) 193 nm) of frozen glycerol films (the twopeaks are considered to derive, respectively, from theparent glycerol and photofragment HOCH2CHOH viaelectron impact dissociation in the mass spectrometerionizer). The data show an onset of the center-of-massvelocity at Flaser ) 200 mJ/cm2 for both peaks. The center-of-mass velocity becomes supersonic above 300 mJ/cm2. (b)Temperature fitted from TOF spectra as a function offluence for the two peaks. Trend lines are added as a visualaid and have no scientific significance. Reprinted withpermission from ref 64. Copyright 2001 American Instituteof Physics.



ejected material) is assumed, since the laser spotwidth on the substrate is large compared with allother relevant lengths of the flow. The approximationloses validity with increasing distance from thesurface and increasing dimensions of the plume.

Analytical models rely on the concept of the Knud-sen layer.120-128 For moderate densities of desorbedparticles (e.g., 0.5 monolayer in a 10-ns desorptionperiod), a Knudsen layer is assumed to form close tothe target surface, in which the velocity distributionevolves from “half-range” Maxwellian (all desorbingparticles streaming away from the surface, i.e., υx >0) to a “full-range” distribution (i.e., - ∞ < υx < ∞)in a center-of-mass coordinate system. The width ofthe Knudsen layer is indicated to be ∼15-20 freemean paths.122 Based on mass, momentum, andenergy conservation across the layer, the followingconditions are derived between parameters at theboundary of the Knudsen layer (denoted by thesubscript KL) and those at the film surface (denotedby the subscript s):

where R ) (γ/2)1/2M [M is the Mach number, M )uKL/(γRTKL)1/2], erfc is the complementary error func-tion [erfc(R) ) (2/xπ)∫R

∞ e-s ds], T denotes tempera-ture, P pressure, and F number density, and γ )CP/CV ) (j + 5)/(j + 3), where j is the number ofaccessible internal degrees of freedom. M is some-times considered to be a variable (e1 for one-dimensional flow) but to a good approximation canbe set equal to 1, so that the flow velocity equals thesound speed R: uKL ) R ) xγkBTKL/m (m is particlemass). In this case, the relation kBTs ) E trans

mp /2 isreplaced by kBTs ) E trans

mp /ηK, with ηK ) (γ/8)[1 + (1 +16/γ)1/2]2 (ranging from 2.52 for a monatomic particle

to 3.28 for a polyatomic molecule). Thus, even for fewgas-phase collisions (as low as three per particle), asignificant percentage of the initial thermal energymay be transformed into forward-directed kineticenergy, and the temperature of the vapor will beindicated to be quite lower than the initial temper-ature in the film. In parallel, the collisions arepredicted to result in partial sharpening of theangular distribution to cos4 ϑ,121 and in the partialbackscattering of desorbed particles (∼18% for mon-atomic gases up to ∼25% for polyatomic gases).

At even higher quantities of ejected material, theKnudsen layer is assumed to be followed by anunsteady adiabatic expansion. Plume dynamics isdescribed by the conservation equations for mass(continuity equation), momentum (Euler equation),and energy:

where F represents density, V the velocity vector, Ppressure, and E the energy). Φ is the external heatinput into the gas and is assumed to be zero in theadiabatic approximation. The expansion results inthe further decrease of the vapor temperature andthe parallel increase of the Mach number (M).Kelly121 assumes an analogy to a firing gun (i.e., thematerial is held initially within a reservoir of aspecific depth and then at t ) 0, the gate at x ) 0 isremoved and the material expands adiabatically) inorder to derive

for the resulting flow velocity and the temperature.In parallel, the angular distribution narrows evenfurther, to cosp ϑ, where p ) (1 + M)2 (for j ) 0).Detailed mathematical treatment of gas expansionwithin the Knudsen layer model has been pre-sented.120-128

Table 2. Average Values of the Desorbate Translational and Internal Energy Distributions in the 193-nm LaserVaporization of NO Multilayer Films (Flaser ≈ 20-25 mJ/cm2) (Reprinted with permission from ref 22. Copyright1989 The American Physical Society.)

probe delaya,b

(µs) ((3µs) ⟨ET⟩ (eV) ⟨ER⟩c (eV) Trotd (K) ⟨F2/F1⟩e ⟨υ ) 1/υ ) 0⟩

85 0.71 0.024 ( 0.006 220 ( 20 1.1 ( 0.40.5

100 0.56 0.017 ( 0.004 180 ( 20 0.70 ( 0.30.4

100 (υ ) 1) 0.56 130 ( 30 0.03 ( 0.01135 0.31 0.009 ( 0.002 108 ( 15 0.50 ( 0.2

0.3

160 0.22 0.011 ( 0.003 120 ( 15 0.20 ( 0.10.2

210 0.14 0.014 ( 0.003 160 ( 20 0.35 ( 0.10.2

a The energy distributions are determined via state-selective multiphoton ionization of the NO desorbates. The probe delayrepresents the delay between the desorption/ablating and the ionizing laser pulses. b All results are for υ ) 0 except where noted.Reported errors represent 2σ. c Average rotational energy calculated from analyzed rotational populations. d Rotational best-fittemperature for low J. e Average spin-orbit ratios (Π3/2/Π1/2) for Eint < 600 cm-1.

TKL

Ts) [x1 + π(γ - 1

γ + 1R2)2 - xπ γ - 1

γ + 1R2]2

(9a)

FKL

Fs) x Ts

TKL [(R2 + 12) eR2

erfc(R) - Rxπ] +

12

Ts

TKL[1 - xπ R eR2

erfc(R)] (9b)

pKL

ps)

FKLTKL

FsTs(9c)

∂F∂t

+ ∇(F‚V) ) 0 (10a)

∂V∂t

+ (V‚∇)V + 1F∇P ) 0 (10b)

∂E∂t

+ (V‚∇)E + PF

(∇‚V) ) ∇Φ (10c)

uM ) MRM ) M(γkBTM/m)1/2;

TM ) TKL[ γ + 12 + (γ - 1)M]2

(11)


Computer simulations have been employed inparallel to examine the validity of the analyticalsolutions. Sibold and Urbassek have employed theMonte Carlo method to simulate the one-dimensional(1D)122 and the three-dimensional (3D)123 single-component pulsed-gas expansion of atoms in thevacuum under the assumption of constant flux. The3D problem is shown to be uniquely specified by thenumber of ejected monolayers per square meter persecond (Θ) and by the aspect ratio of the desorbedplume (b ) ro/υτpulse, where ro is the laser spot width).As either Θ or b increases, the average number ofcollisions experienced per particle (Ncoll) increases,with a significant increase of the translational energyalong the axis of desorption. In a different approachby NoorBatcha et al.,130-132 particles are consideredto desorb according to a time-dependent flux, φ(t) ∝exp(-t/τf). It is shown that the number of collisionsincreases sharply when τf < 10-8 s (Table 3). Accord-ing to both models, the distributions are best de-scribed by two temperatures, Tz (temperature alongthe axis perpendicular to the surface) and Txy (tem-perature along the radial axes), resulting in the so-called “elliptical distributions”, though the physicalsignificance of these parameters has been questioned:120,121,127,138

Both υdrift and Tz are predicted to increase, whileTxy decreases with increasing Ncoll. Due to the kine-matics of the collisions, fast molecules focus in thecenter of the jet, whereas slow ones focus at large

angles. For given Θ, ⟨Etrans⟩ and Ncoll increase and theangular distribution broadens with the number ofinternal modes because of the higher internal energythat can be “dumped” into the translational motions(Table 3). These simulations are in qualitative agree-ment with the predictions of the Knudsen layermodel, i.e., that the translational distributions of thedesorbates for different systems but with the samenumber of internal degrees of freedom can be welldescribed by a single parameter (ηK in the Knudsenmodel). However, with increasing desorbing amount,deviations are observed between the Knudsen modeland the simulations, indicating that the descriptionof the transaltional distributions must incorporatethe influence of the surface coverage.

The previous theoretical studies appear to pro-vide a quantitative understanding of the experi-mental results mentioned at the beginning of thissection. The changes in the translational distribu-tions with increasing laser fluence are often consid-ered18,64,112,113,134 to illustrate the evolution fromvaporization to Knudsen layer formation and finallyto the unsteady adiabatic expansion. For instance,in the 193 nm irradiation of frozen glycerol films(Figure 11), the nonzero υdrift value observed atfluences above ∼200 mJ/cm2 is ascribed to theenhanced contribution of collisions, and the super-sonic center-of-mass velocities observed at even higherfluences (300 mJ/cm2) are ascribed to adiabatic ex-pansion (however, note the very high parallel in-crease of the time-of-flight Tz between 200 and 300mJ/cm2 in Figure 11b). A number of experimentalstudies rely on the Knudsen layer concept to estimatesurface temperatures, consistent with estimatedabsorbed laser density. Furthermore, the indicatedvery efficient energy “removal” from the internaldegrees of freedom may result in the stabilization of

Table 3. Results of the Monte Carlo Simulation on the Distributions of Rapidly Desorbed Particles from aSurfacea (Reprinted with permission from ref 132. Copyright 1991 The American Physical Society.)

system Θ (ML) Z Ncoll ⟨cos θ⟩ Txy (K) Tz (K) υdrift (104 cm/s) Tint (K) ηK

Xe 0.1 0 2.18 0.7355 283 ( 1 946 ( 66 1.47 ( 0.10 2.531.0 0 12.17 0.8927 53 ( 1 1273 ( 44 1.57 ( 0.06 2.53

NO 0.1 2 1.86 0.7158 337 ( 3 997 ( 46 3.00 ( 0.07 426 ( 3 2.801.0 2 1.84 0.8560 113 ( 1 2063 ( 149 2.27 ( 0.29 153 ( 1 2.80

valine 0.1 3 2.50 0.7319 322 ( 3 1171 ( 108 1.45 ( 0.16 414 ( 3 2.881.0 3 16.53 0.8634 120 ( 3 2291 ( 168 1.34 ( 0.20 168 ( 3 2.88

valine 0.1 13 2.54 0.7002 253 ( 3 1237 ( 105 1.47 ( 0.16 468 ( 2 3.121.0 13 18.47 0.8393 151 ( 2 3186 ( 265 1.54 ( 0.28 326 ( 1 3.12

(NO, He) 0.1 8.41 0.8669 119 ( 3 1022 ( 88 5.81 ( 0.22 87 ( 20.9 10.63 0.8428

(NO, Ar) 0.1 9.13 0.8762 69 ( 2 1512 ( 107 2.51 ( 0.26 95 ( 20.9 9.34 0.8452

(NO, Xe) 0.1 10.00 0.8551 87 ( 7 1449 ( 210 2.23 ( 0.61 121 ( 40.9 10.28 0.8453

(valine, He) 0.1 14.95 0.9026 138 ( 6 954 ( 79 4.01 ( 0.10 46 ( 30.9 16.87 0.8645

(valine,b He) 0.1 15.57 0.8733 181 ( 7 1200 ( 101 3.89 ( 0.14 114 ( 30.9 18.80 0.8390

initial distribution 0.6667R Θ, number of monolayers assumed to desorb. In the case of the neat films (upper section of the table), the numbers represent

the total amount desorbed. In the case of mixtures (lower section of the table), the total desorbing amount is, in all cases, onemonolayer and the values 0.1 and 0.9 represent the surface coverage by the molecules only. Z, number of internal degrees offreedom assumed to participate in collisions. nj, average number of collisions experienced by a molecule. ⟨cos θ⟩, average angulardistribution of the desorbates. Txy, Tz, Tint, and υdrift, parameters obtained by fitting the simulated velocity distributions of thedesorbed particles to elliptical translational distributions. ηK, the values that are obtained by fitting the simulation results to theequation E trans

mp ) nKkBTs based on Knudsen layer formation. b Simulations performed with Z ) 13.

I(υ,θ) ∝ υ3 exp[- m2kB

[(υ cos θ - υdrift)2

Tz] +

υ2 sin2 θTxy

] (12)


the ejected thermally labile molecules. In fact, thisis the basic idea of the “cooling plume model” forrationalizing the reduced thermal degradation ofbiopolymers in MALDI. A hydrodynamic analysiswas employed75,136,137 to examine this point for typicalMALDI conditions (in combination with the bottle-neck model advanced by the same team). The “coolingplume model” has been invoked in a number of cases,as will be discussed in detail below.

Yet, despite the evidence from the previous studies,it is not clear how far the analogy to the adiabaticexpansion can be extended. In the previous theoreti-cal studies, the desorbates are assumed to originatefrom the surface with a thermal Maxwellian velocitydistribution. However, as shown in section II.A, thenature of the ejection process in ablation is indicatedto be much more complex (prolonged ejection ofmaterial in the form of clusters and monomers froma superheated liquid). Chen et al.142-145 have notedthat the prolonged vaporization of material providesa source of mass and energy inflow to the plume,which makes the unsteady expansion nonadiabatic.This contribution was modeled by adding an ap-propriate mass and energy source in the hydrody-namic equations (eqs 10a and 10c). It is shown thatthe consequent increase of entropy results in trans-lational energies 2-3 times higher than those ex-pected from the previous models. Furthermore, theacceleration by this “dynamic” source in a directionperpendicular to the surface may result in the moreforward-peaked expansion of the plume. In someanalogy to the previous suggestion, the simulationsby the Garrison30,138 and Vertes45 groups indicate thatthe velocity of the ejecta correlates with their initialdepth within the film. Surface molecules are ejectedwith higher velocities due to the “push” they get fromthe molecules below. The axial velocity distributionis shown30,138 to be well described by a modifiedMaxwell-Boltzmann (MB) distribution with a rangeof stream velocities, whereas a simple MB distribu-tion with the same temperature can well describe theradial distribution. An important point of this sug-gestion is that, in contrast to other models, a singletemperature can consistently describe the differencesbetween the axial and radial translational distribu-tions without resorting to the adoption of “elliptical”parameters.

The difficulties in accounting consistently for allfeatures of the experimental translational distribu-tions have been detailed in a number of studies. Itshould be cautioned, however, that a number of“trivial” factors may affect the experimental trans-lational distributions and may account for some ofthe observed discrepancies from theoretical models.A nonhomogeneous laser profile or even the use of aGaussian beam, ill-defined amorphous structure ofthe deposited films, etc. will result in desorbates witha convolution of different translational and angulardistributions. Furthermore, a sensitive dependenceof the translational distributions on number of pulses,angle of laser incidence, etc. has been noted.25 Inparticular, there has been an inexcusable lack ofattention to the mode of irradiation (i.e., confinedablation or ejection effected by a combination of direct

adsorbate absorption and surface-mediated heatingemployed in several experiments), though it is clearthat this parameter may critically affect the desor-bate observables. However, the major limitation inthe evaluation of the theoretical models derives fromthe fact that a complete set of translational, angular,and internal energy distributions as a function ofnumber of ejected particles is needed. Aside from theevident immense undertaking that such measure-ments represent, the usual direct way to vary ex-perimentally the number of ejected particles is througha change of the laser fluence. However, a change inthe fluence affects both the quantity of ejectedmaterial and the initial temperature, thereby com-plicating the comparison of experimental results withtheoretical models.

In the UV irradiation of indole films139 (λ ) 266nm, τpulse ) 10 ns) under conditions resulting in theejection of 2.0 monolayers per shot, the neutralmolecule is characterized by a high translationaltemperature (3400 K), while the vibrational temper-ature is only 210 K, and the angular distribution isforward-peaked, with the major component beingproportional to cos7 θ. These features would appearto suggest an adiabatic-like expansion. However, thevibrational temperature is similar for different ve-locities. Furthermore, both “cold” and “hot” moleculesare found to be described by the same angulardistribution. These features are inconsistent with a“jet-like” expansion, since molecules in the densestregion of a jet are expected to experience morecollisions than the average and thus should be cooledto lower vibrational temperatures. The number ofcollisions is estimated to be only 2-7, and collisionalperturbation is indicated to account only for 10% ofthe vibrational cooling. These discrepancies weresuggested to be due to the non-Boltzmann nature ofthe initial desorbate translational distribution. Simi-larly, in various studies,36,141 the estimated Machnumber is much lower (∼2) than that expected foradiabatic expansion (g5). It has been suggested36

that the discrepancy can be ascribed to the fact thatejection lasts for microseconds, and thus it should beconsidered to be of the “steady” expansion type, i.e.,TM ) TKL(γ + 1)/[2 + (γ - 1)M2] instead121 of eq 11.Similar difficulties are recognized in accounting forthe angular distributions,139 which are often bimodal,at variance with the predicted simple cosp ϑ depen-dence. The bimodality has been ascribed to thecontribution of different desorbate populations (e.g.,“ablative” versus “postablative” ejection)140 and/or tothe reduced number of collisions (thus less forward-peaked expansion) of the particles ejected from theperiphery of the irradiated spot.127

In the IR ablation of neat C6H6 films (on nonab-sorbing substrates), Braun and Hess25 have foundthat the desorbate’s most probable translationalenergy, E trans

mp , does not increase monotonically withlaser fluence, but most interestingly it shows a “phasetransition”-like dependence (Figure 12). The charac-teristic “plateau” of the diagram appears at fluencessuggestive of film temperatures close to the meltingpoint of the compound. In view of the similarity ofthe Etrans vs Flaser diagram to a P, T thermodynamic


diagram (but not necessarily a valid correspondence),ablation, at least in IR, was claimed to be photother-mal in nature. Evidently, the implicit assumption isthat the most probable desorbate velocities aremainly determined by the film temperature changes.However, the fitting of the time-of-flight spectrarequires nonzero and rather large υdrift values, whichwould appear to indicate substantial gas-phase col-lisions. It should also be noted that the exact depen-dence of Figure 12 has not thus far been reportedfor any other system.

In an approach similar to that described above, inthe 248-nm irradiation of neat C6H5CH3,41,47 CH3I,190

and C6H5Cl films,38,141 at low fluences, the mostprobable velocity (υmp) of the desorbates shows a veryweak dependence on Flaser, but above the threshold,the slope changes abruptly (Figure 13a). It is clearthat υmp does not scale smoothly as a function of thedesorbing signal (Figure 13b), and thus the abruptchange cannot be ascribed exclusively to enhancedcollisional effects in the plume. Similarly, the parallelhigh increase of both center-of-mass υdrift and Tz inFigure 11 does not appear to be fully compatible withgas-phase collisional models. In view of these results,the change from MB distributions to shifted MBdistributions that is observed18,64 as the ablativeregime is approached indicates an extra “component”of the desorbate velocities, besides the collisionalperturbation. The extra component may be relatedto the “explosive boiling” process or the dynamic“impulse” effect.143

In all, even for simple one-component systems, theexperimental desorbate distributions do not appearto be fully accounted for by existing models. Theinterplay of a number of plausible factors has alreadybeen underlined. Furthermore, if we accept the tenetthat ejection occurs largely in the form of clusters,which subsequently disintegrate, then the accuratetheoretical description of the desorbate distributionswill depend critically on the elucidation of the factorsthat affect cluster evolution in the plume. It is

interesting to note that, in the laser-induced ejectionfrom HBr multilayers,27 distinctly different angulardistributions have been observed for the differentcluster sizes. Though the exact regime of irradiationin this study is unclear, their result, nevertheless,underlines the complexity of the cluster ejectionprocess and of their subsequent dynamics in theplume. Thus far, very little theoretical attention hasbeen paid to the cluster ejection/translational distri-butions correlation.

For bicomponent systems, additional features re-lating to the characteristics of the two components(masses, collisional cross sections, relative concentra-tions, etc.) become crucial. Once more, the variousmodels suggest different degrees of desorbate distri-bution pertubations. For Knudsen layer formation,the sputtered material tends to move together (i.e.,in a single gas cloud), independently of the mass (i.e.,different species develop a similar flow velocity),129

in close analogy to the case of supersonic beams. Thelight component is characterized by a low TOFtemperature and the heavy one by a high one.129 InMonte Carlo simulations, the influences of the rela-tive concentration, the number of internal degrees(Table 3),132 and the different masses133 or desorptionenergies135 of the two ejected compounds have beenstudied for thermal desorption from binary mixtures.Because of the difference in the initial velocity of thetwo components, segregation effects in the ejectedplume are demonstrated, with its front (outer) partbeing composed preferentially of the “lighter” spe-

Figure 12. Most probable translational energy as afunction of the incident laser fluence in the IR laser(λ ) 1033.48 cm-1) ablation of 2-µm-thick benzenefilms condensed on a NaCl window. The dashed straightlines indicate the separation of the measured data intothree regions with different slopes. Reprinted with permis-sion from ref 25. Copyright 1993 American Institute ofPhysics.

Figure 13. (a) Flaser dependence of the most probabletranslational energies, Etrans, of the neutral parent mol-ecules detected in the irradiation of thick C6H5CH3 filmsat 248 nm (τpulse ≈ 30 ns). (b) The same data as a functionof the desorbing material. Reprinted with permission fromref 47. Copyright 2002 University of Crete.




cies.133 As a result of the collisional energy transfer,the average translational energy of the light speciesbecomes lower than that of the heavy species (Figure14). Furthermore, depending on the degree of inter-

mixing of the two components in the plume, thetranslational distributions may be a convolution ofpopulations that have suffered different degrees ofcollisional perturbations and, as a result, becomebimodal. On the other hand, in the molecular dy-namics simulations, the translational distributionsof the analytes have been correlated with their initialposition in the film.30,45,138 “Heavy” analytes areshown30,45,138 to be ejected with a translational dis-tribution nearly identical with that of the matrix(major component of the mixture). Time-of-flightspectra of the analytes are suggested to be sharpestat the threshold, deteriorating at high fluences. Thedeterioration with increasing fluence is ascribed tothe release of the analytes from a higher film thick-ness, resulting in a higher variation of the velocitydistribution, and to the higher attained surfacetemperatures. A more detailed discussion is pre-sented by Zhigilei et al. in the preceding article inthis issue.

Experimentally, differences between the two com-ponents ejected in the irradiation of mixtures maybe due to the nonuniform mixing/segregation effectsin the film.112 However, differences are also docu-mented in cases where there is no evidence forsegregation. In the 248-nm irradiation of C6H5CH3/CnH2n mixtures below the threshold,47 the E trans

mp ofthe dopants weakly bound to the matrix (e.g., (CH3)2Oand C3H6) are lower than that of C6H5CH3, despitethe fact that at these fluences desorption is thermal(section II.B), and both the dopant and the matrixshould desorb with equal E trans

mp . Clearly, there iscollisional energy transfer in the plume from the“light” dopant to the matrix molecules, which arethus accelerated to velocities higher than those

attained in the irradiation of the neat C6H5CH3 film.This result appears to be in qualitative correspon-dence to the Monte Carlo simulations,133 but quan-titative comparison requires further examination.

Different translational features are observed in theablative regime.141 As clearly indicated by Figure 15,

well-defined differences are observed between thestrongly and the weakly bound dopants in terms oftheir translational distributions.39,47 The indicatedfeatures are observed for all fluences above theablation threshold. Similar differences are also ob-served between amino acids and phenol desorbatesin the CO2 laser-induced ejection from their frozensolution within an ethanol/glycerol matrix.185 Therather broad distributions of the weakly bounddopants ((CH3)2O, phenol) may be ascribed to colli-sional effects in the plume. Alternatively, the distri-bution broadness may reflect the extensive contri-bution of “postablation” thermal desorption thatoccurs for these compounds (section II.B.3). On theother hand, for the strongly bound dopants (decane,amino acids), which are supposed to be ejected withinclusters of the matrix molecules, the time-of-flightspectra are nearly superimposable with those of thematrix.39 This closely corresponds to the findings inMALDI that proteins are ejected with the samevelocity as the matrix independently of theirmasses,147-152 and to the prediction of the moleculardynamics simulations.30,45,138

In the very few reported examples, the angulardistributions of the “heavy” analytes are indicated tobe even more forward-peaked than those of thematrix molecules. Puretzky et al.153 managed tomonitor the time evolution of the angular distribu-tions of DNA and of the matrix in UV MALDI (λ )248 nm, Flaser ) 50 mJ/cm2) via laser-induced fluo-rescence imaging of the plume. To this end, DNA wastagged with a fluorophore. A sharpening of the angu-lar distribution of DNA as compared with that of thematrix has been observed. It has been well accountedfor by a hydrodynamic description of the flow.

The same mechanisms as those described abovehave also been invoked to account for the transla-

Figure 14. Mean energy Ehϑ of particles arriving at anangle ϑ with respect to the surface normal as calculatedby Monte Carlo simulation in the thermal desorption of asystem composed of two particles with a 1:5 mass ratio.Solid line, heavy particles; dashed line, light particles;dotted line, mean energy for the collision-free flow, Ehϑ )Eo as a reference. Reprinted with permission from ref 133.Copyright 1993 The American Physical Society.

Figure 15. Time-of-flight curves recorded in the ablationof C10H22/C6H5CH3 and (CH3)2O/C6H5CH3 (in both cases,1:5 molar ratio) solids (λ ) 248 nm, τpulse ≈ 30 ns, in eachcase, at a fluence ∼1.5 above the corresponding threshold).Reprinted with permission from ref 47. Copyright 2002University of Crete.




tional distributions of the products ejected in theablation of photoreactive compounds. For instance,for products that are strongly bound to the matrix,the translational distributions in the ablative regimeare38,39,141 nearly identical with those of the parentmolecule, exactly as observed for strongly bounddopants discussed above. On the other hand, thevelocities of the “light” fragments or photoprod-ucts38,64,141 have been usually correlated with thepicture indicated by 3D Monte Carlo simulations.133

It is understandable that, due to the massive mate-rial ejection, photoproduct translational distributionsin ablation are mainly determined by the plumedynamics, and there is minimal influence of theinitial fragmentation or reactivity process (in contrastto electronically mediated photodesorption processes).Yet, in a few cases, photofragments may be observedwith very high velocities and translational distribu-tions deviating considerably from the previous mod-els. It is likely that these are formed by photofrag-mentation in the gas phase (plume),38,64,176 resultingin additional energy deposition on the fragment. Theeffect of chemical reactions (atom recombination andmolecule dissociation) that may occur in the plumeon the desorbate distributions has also been exam-ined by Monte Carlo simulations. It is shown thatchemical reactions in the plume result in significantspread of the angular and translational distribu-tions.154

Summarizing previous studies, it is clear that somegeneral trends have by now been established for anumber of simple molecular solids/films. The trendsappear to be extrapolable to more complex systems,at least MALDI. However, the interpretation of theobserved trends and features in terms of gas phase(i.e., plume) and primary processes (i.e., film) of theejection remains to be fully elucidated. It is noted thateven the description of distributions for desorptionfrom a metastable liquid (i.e., below the threshold)calls for additional theoretical work. This is an aspectthat has been largely overlooked in the literature.Studies on solids of very simple (monatomic anddiatomic) molecules may be most useful and providethe most stringent test of the various models, since,understandably, their extension to polyatomic mol-ecules presents a number of additional complexities.

It has been usual in studies on complex systemssuch as polymers to base rather detailed and strongmechanistic statements (e.g., temperatures attainedin the film, thermal versus nonthermal nature ofejection, etc.) on translational distribution fittings.However, in view of the interpretational uncertain-ties discussed above and of the sensitive dependenceof the translational distributions on a number ofexperimental parameters, caution should be exertedin trying to extract mechanistic conclusions from suchfittings. Additionally, the interpretation of the trans-lational distributions in the irradiation of thesesystems is further complicated by the formation of aplasma, extensive secondary fragmentation and re-activity in the plume, etc.

D. Chemical Processes and EffectsThe issue of chemical processes and effects in the

ablation of molecular substrates is closely related to

the question of the mechanism(s) of the phenomenon(sections II.A and II.B), since photochemical pro-cesses have been suggested1,11,155 to contribute to, oreven dominate, material ejection in the irradiationof photolabile systems. In fact, the issue of photo-chemical versus thermal mechanisms has been themost hotly debated topic in the field of ablation ofmolecular substrates. However, even in the case of athermal mechanism (e.g., explosive boiling), theenergy released by the induced exothermic reactionscan be substantial, thus affecting quantitatively thefeatures of the material ejection process157 (e.g.,threshold, ejection efficiencies, etc.).

Studies dedicated to addressing the issue of pho-tochemical-induced material ejection have been de-tailed in the case of liquids (section III.B.2). In thissection, we take a more general view of the chemicalprocesses in the UV ablation of molecular substrates.Specifically, independently of the specific mecha-nisms responsible for material ejection, how canchemical processes in UV ablation be described?Conventional photochemistry considers bond break-ing and formation in a well-defined environmentunder well-defined conditions. In contrast, in UVablation, a large number of electronically excitedmolecules are formed, high temperatures are at-tained, high-amplitude stress waves are induced,and, as described above, the “physical” condition ofthe substrate is rather unusualsall these within atime-varying framework. Therefore, the real chal-lenge raised in UV ablation of molecular substratesis the development of appropriate models with pre-dictive power to describe chemical processes underthe above conditions.

The issue of the nature and the extent of chemicalmodifications is also crucial for the optimization ofthe various applications of UV ablation. The pro-cessed molecular substrates (polymers, tissues, etc.)generally include a wide variety of chromophores,which upon photoexcitation may dissociate into highlyreactive fragments. Additional species may be formedby the thermal or stress-induced breakage of weakbonds. In material-processing schemes, it is impor-tant to minimize accumulation of these species in thesubstrate, since they may form in the short- or long-term byproducts (e.g., oxidation products) with det-rimental effects on the integrity of the substrate. Inanalytical diagnostics of laser-ejected material, it isimportant that fragmentation does not compromiseparent molecule signal intensity. On the other hand,control of the fragmentation may be useful for thegroup analysis/characterization of the compounds.

In the following sections, we consider first thequalitative (product patterns) and quantitative as-pects (extent of photolysis and product yields) in theirradiation of photolabile/photodissociable systems,whereas section II.D.2 considers the issue of thermaldegradation in the ablation of molecular solids. Asclearly indicated by the previous discussion, thisdivision is somewhat formalistic (since a high degreeof coupling between the thermal and (photo)chemicalprocesses is expected), and the reason for adaptingit is for simplicity of presentation.


1. Photochemical Processes and EffectsThe studies on van der Waals films have indicated

a number of unexpected features about chemicalprocesses and effects in the irradiation of molecularsolids with high-intensity laser fields. A most inter-esting result in the study by Domen and Chuang13-16

was the observation of intense parent peak in the“explosive desorption” from CH2I2 films (λ ) 308 nm)(Figure 2). In the ablation of Cl2 films at 355 and 193nm, the Cl signal comprises less than 7-10% of thetotal desorption signal20 (Table 4). These findings are

most surprising given that, upon excitation at thecorresponding wavelengths, the studied moleculesdissociate in the gas phase with near-unity quantumyield. No evidence for fragmentation has been ob-tained in the irradiation of NO2 films158 (λ ) 248 nmunder a “confined ablation” scheme) and of H2O films(λ ) 193 nm).176 These results bear directly on thefeature of ablation of molecular substrates that hasbeen crucial for its use for analytical purposes, e.g.,in MALDI. Stability of dopants (e.g., no decomposi-tion or reaction with the surrounding polymer) is alsoobserved in the UV ablation of doped polymers, atleast for photoinert dopants (e.g., naphthalene,phenanthrene, etc.).159

It can be suggested that the observed parent peakrepresents molecules that are ejected without havingabsorbed a photon themselves. Since photodissocia-tion occurs only for molecules that have absorbed aphoton, whereas even molecules that do not them-selves absorb are ejected, an “apparent” fragmenta-tion yield is expected, given by

where R is the absorption coefficient, Flaser is theincident fluence, F, NR, and MW are the density, theAvogadro number, and the molecular weight, respec-tively, and lejected is the ejected layer thickness. Fortypical values as for the systems studied above, thisratio ranges from 10-2 to 10-1, thus accounting atleast semiquantitatively for the reduced apparent

fragmentation yield. [In some of the previousstudies,20,158-161 the employed mode of irradiationcorresponds to “confined ablation”. As discussed insection II.B.2, in confined ablation, material maylargely be ejected intact. So, in these experiments,the apparent degree of desorbate photofragmentationmay be greatly reduced from that in the irradiationof the free surface of the substrates.] Another impor-tant implication becomes apparent by comparisonwith eq 3. It is clear that lejected is energy-densitydependent, whereas the photodissociation/productnumber is photon flux dependent. Thus, differencesin the apparent fragmentation yield with irradiationwavelength alone do not constitute sufficient evidencefor a change in mechanism, as sometimes argued inpolymer studies. Furthermore, assuming all factorsbeing the same, the apparent percentage of photolysisof a photolabile chromophore can be reduced bymixing it with another strongly absorbing (photoin-ert) chromophore. This is, of course, one of the evidentfactors underlying the success of MALDI. On theother hand, concerning the extent of photofragmen-tation in the remaining substrate, eq 3 suggests itto be constant with laser fluence (since the nonejectedlayers are subject to the same Fthr). However, asdiscussed below, both eq 13 and the previous impli-cation neglect the possibility of “postablation desorp-tion” of fragments and/or photoproducts that areformed below the etched depth (section II.B.3).

Though eq 13 provides a basis for understandingthe extent of desorbate photolysis in UV ablation, itdoes not appear sufficient to account for all observa-tions concerning chemical effects. Important experi-mental information in this direction was provided bythe examination of the product patterns in theablation of neat films of ICl and XeF2.160,161 Uponelectronic excitation to a dissociative state, efficientformation of Cl2 and I2 (as well as Cl and I fragments)is observed in the ablation of ICl films (λ ) 532 or266 nm), and formation of F2 in the case of XeF2(λ ) 266 nm) (Table 4). Clearly, the fragmentsproduced by the photolysis of the parent moleculesreact extensively with each other (though in the caseof ICl, the observed products may partly derive fromdirect, concerted reactions of the excited parentmolecule). It is thus likely that the reduced fragmen-tation in ablation is due, at least partly, to theoperation of efficient recombination processes; i.e., apercentage of the fragments recombines with re-formation of the parent molecule. This possibility wasstressed already in the first studies by Domen andChuang.14,15

The question is raised, however, about the factorsresponsible for the efficient operation of the recom-bination processes/cage effects. Initially, materialejection was generally considered to occur largelyduring the laser pulse,17-24 in which case the ef-ficiency of the recombination processes is difficult toaccount for. Furthermore, in this case, secondary ab-sorption in the gas phase would be expected to besignificant, thus resulting in further fragmentation.The secondary gas-phase photolysis has been stronglyemphasized in polymer studies.1,6,11 As illustrated bythe discussion in the other articles in this issue, the

Table 4. Relative Yields of Species Detected in theAblation of Photolabile Systemsa (Reprinted withpermission from ref 161. Copyright 1996 AmericanChemical Society.)

parent molecule species neutral ratiob,c

Cl2 (λ ) 355 nm) Cl2 1Cl 0.10 ( 0.03

ICl (λ ) 532 and 266 nm) ICl 1I <0.18Cl <0.84I2 0.19Cl2 0.60

XeF2 (λ ) 266 nm) XeF2 1XeF 0Xe 9.11F 4.31F2 1.154

a The study corresponds to “confined ablation”. b The re-ported ratios have been corrected for the relative detectionefficiencies of the various species/fragments in the massspectrometer. c Parent peak normalized to 1.

Nfragment

Nparent)

qFlaser(1 - exp( -Rlejected))hνFlejectedNa/MW

(13)


relative contribution of photolysis and of reactivityin the gas phase (i.e., plume) versus in the film is anissue of high importance in both MALDI and polymerablation.

The extent of the relative contribution of photolysisin the gas phase versus in the film evidently dependscritically on the time scale of plume ejection (thisdelineation may be somewhat of an oversimplifica-tion, since there is not really a defining line in theevolution from the condensed phase to the gas phase).Even for simple thermal evaporation, desorptioncontinues for 100 ns after the laser pulse, until theeffected cooling reduces the temperature to lowenough values. In the case of explosive boiling (sec-tion II.B.2.), at least for fluences close to the ablationthreshold, material ejection can be considerablydelayed after the end of the nanosecond laser pulsedue to the time necessary for the bubbles to grow toa large enough size.53 A considerable delay in thematerial ejection onset has similarly been observedin the irradiation of liquids (section III). Since mo-lecular photolysis occurs very fast (usually on pico-second or nanosecond level), the delay in plumeejection and the prolonged material ejection after thelaser pulse suggest that the largest percentage ofdesorbates is photolyzed in the film before it has beenejected in the gas phase. As a result, their photolysisshould be subject to the efficient electronic deactiva-tion and radical recombination processes that arecharacteristic of condensed phases,181 thereby ex-plaining, at least qualitatively, the operation of “cageeffects” advanced by Domen and Chuang. However,in the case of ablation, the quantitative aspects arefar from clear, since the condensed phase is presum-ably a superheated liquid. Ventzek et al.162 suggestthat the degree of fragmentation can be predicted onthe basis of the corresponding equilibrium dissocia-tion/recombination constants within the critical fluid(but such constants are largely unavailable).

The issue of photolysis in the film versus in theplume was specifically addressed in the 248-nmablation of condensed C6H5Cl films.164 Interestingly,the apparent photolysis yield, as determined from thegas-phase [HCl]/[C6H5Cl] ratio, exhibits two dinstinctdependences on Flaser (Figure 16). The small yieldobserved at fluences close to the threshold appearsto be consistent with the assumption of reducedphotolysis efficiency within a condensed phase (but,the validity of the argument of the photolysis ef-ficiency depends critically on the extent of HCl vsC6H5Cl “postablation” desorption, see section II.B.3).38

At any rate, at higher fluences, at which extensivematerial ejection during the laser pulse is indicatedto occur (Figure 8a), the photolysis yield exhibits asharp and nearly linear increase with Flaser. Thecorrelation strongly suggests the increasing impor-tance of secondary gas-phase photolysis of the com-pound in the plume. Furthermore, at these fluences,free chlorine becomes significant, whereas this spe-cies is hardly detected at lower fluences. Most prob-ably, this represents Cl produced by photolysis of theejected compound in the initial stages of the plumeejection, when the density of C6H5Cl is still low.Multiphoton processes and weak plasma formation

may also become important in determining the extentof desorbate photolysis at higher fluences.

A dependence similar to that shown in Figure 16is observed for the O2 photofragment versus parentsignal in the irradiation of condensed O3 films166 at248 nm (Table 5). A reduced parent peak at theexpense of higher fragment intensity is also foundfor higher laser fluences in the UV (λ ) 266 and 213nm)189 irradiation of polycyclic nitrate films. Thedegradation of the mass spectra at high fluences, wellabove the threshold, is a common observation inMALDI, though in this case, enhanced thermaldecomposition (section II.D.2) may also be involved.The only case for the opposite trend is reported inthe irradiation of frozen glycerol64 at 193 nm. TheHOCH2CHOH photofragment (Table 5) is found todecrease relative to the ejected parent molecule withincreasing laser fluence. The reason for the differentbehavior observed for this system is not understood.

In the ablation of indole films139 at 266 nm (atwhich both solid and vapor indole absorb) at 75 mJ/cm2 (probably close to the ablation threshold, whichhowever was not determined), the fraction of mol-ecules that interact with the laser light is a percent-age of the ejected material, and thus the ionizationefficiency is lower than expected. Following ejection,

Figure 16. (a) Fluence dependence of the intensities of(neutral) C6H5Cl and HCl detected in the UV (λ ) 248 nm,τpulse ≈ 30 ns) irradiation of neat C6H5Cl condensed films.The right-hand ordinate refers to C6H5Cl and the left-handone to the intensity of the HCl product. (b) The gas-phaseHCl concentration as a function of the incident laser flu-ence. The change in the dependence of the HCl yield atlaser fluences >150 mJ/cm2 has been ascribed to the in-creased contribution of secondary photolysis in the ejectedplume at these high fluences. Reprinted with permissionfrom ref 164. Copyright 1997 Elsevier Science B.V.



16% of the desorbates absorb one ablating laserphoton and subsequently fluoresce back to highervibrational levels of the ground electronic state (withthe vapor heated from Ti ) 173 to 233 K), whereas10-5 molecules absorb two photons and ionize. Incontrast, in the irradiation at λ ) 290 nm, which isstrongly absorbed by the solid but not by the vapor,only 0.1% of the molecules are heated by the ablatinglaser, and the ionized fraction is reduced from 10-5

to 10-7.In all, the studies thus far have provided strong

evidence for the importance of recombination pro-cesses in the ablation of moleculer photosensitivefilms. However, the extent of their influence on thedesorbate photofragmentation extent is expected todepend greatly on the specific reactivity of theproduced radicals, i.e., to be system-specific. On theother hand, the detrimental effects of secondaryabsorption in the plume for the apparent photosta-bility of the molecules have been clearly illustratedby the previous studies. This contribution may, infact, be partly responsible for differences, such as forthe enhanced fragmentation and a higher percentageof “internally hot” molecules observed in the “front”part of the plume versus in the main part.140,152,153

These differences have alternatively been related tothe degree of thermal degradation of the correspond-ing populations,156,157 but the gas-phase contributionappears to be an equally likely factor in the case ofUV ablation. Finally, it is stressed that the previousdiscussion of film versus plume contribution pertainsonly to photolysis. The degree of reactivity in the filmversus in the plume is a much more involved aspect,which has been a particular consideration in the caseof MALDI, as described in the article by Zenobi andKnochenmuss in this issue.

Besides the question of the extent of photofrag-mentation, another most important question con-cerns whether product patterns in UV ablation differfrom those in conventional photochemistry.

In the few systematic studies of photoproduct for-mation as a function of laser fluence, different specieshave been detected in the gas phase above versusbelow the threshold. In the irradiation of condensedfluorophenyl azide169,177,180 (F5PhN3) at 248 nm,F5PhN and N2 are detected above the ablation thresh-old (the smaller detected fragments may be largelydue to the cracking at the ionization source). Incontrast, below the threshold, only N2 is detected inthe gas phase. Similarly, in the 248-nm irradiation

Table 5. Examples of Photoproducts Observed in the Ablation of Condensed Films of Photolabile Compoundsand Reaction Pathways Accounting for Their Formation

systemwavelength

(nm)observed

products/fragments proposed reaction scheme refs

CH2I2 308 CH2I, I, CH2 CH2I2 + hν f CH2I + I 13-15CH2I2 + hν f CH2 + 2I

ICl 532, 266 I, Cl, I2, Cl2 ICl + hν f I + Cl 160, 161Cl + Cl f Cl2I + I f I2

Cl2 355 Cl Cl2 + hν f Cl + Cl 19-21XeF2 266 XeF, F, F2 XeF2 f XeF + F 161

F + F f F2C6H5Cl 248 Cl, C6H5, HCl, C12H10, C6H5Cl + hν f C6H5 + Cl 38, 47, 156,

(C6H6),a C6H4Cl2, C6H5Cl + Cl f C6H4Cl + HCl 157, 164,C12H9Cl, C12H8Cl2 C6H5Cl + C6H5 f C6H4Cl + C6H6 167, 172

C6H4Cl2 + C6H5 f C6H4Cl + C6H5ClC6H5 + C6H5 f C12H10C6H4Cl + C6H4Cl f C12H8Cl2

b

C6H4Cl + C6H5 f C12H9Clb

C6H4Cl + Cl f C6H4Cl2CH2(OH)CH(OH)CH2OH 193 HOCH2CH2OH, C3H8O3 + hν f CH2OH + HOCH2CHOH 64

(glycerol) CH2OH, CH3 (CH3 is suggested to be formed viaprotonation in postdesorption collisions)

CHCl3 193 CHCl2, HCl, CCl2, CHCl3 + hν f CHCl2 + Clc 176, 184Cl, CCl3 CHCl3 + hν f CCl2 + HCl

CHCl3 + Cl f CHCl2 + HClCl + CHCl3 f CCl3 + HCl(a number of additional products may be

formed with successive laser pulses)C6F5N3 (FPA) 248 C6F5N (FPN), N2, C6F5N3 + hν f C6F5N + N2 169, 177, 180

(C6F5N)2 C6F5N + C6F5N f (C6F5N)2O3 248 O2, O O3 + hν f O2 + O 166

(O2 may also derive fromO + O f O2 in the film)

Si3N3C6H21(hexamethylcyclotrisilazane)

193 H6C2SiNH, Si2C4N2H14 Si3N3C6H21 + hν f H6C2SiNH +(Si2C4N2H14)

265

H6C2SiNH + hν f H3C-SiH2-NCH2

a C6H6 could not be ascertained experimentally. b These products may also be formed, respectively, by the direct reaction of Cland C6H5 with the parent molecule via the so-called “ipso subsitution. c The Cl time-of-flight spectrum differs significantly fromthose of CHCl3 and CHCl2 both in width and peak arrival time. (A similar observation has been also reported in the irradiationof C6H5Cl films.) The Etrans is comparable to that determined in the gas-phase photolysis of the compound, thereby suggestingthat it has undergone very few collisions. It appears likely that it is produced via photolysis of the ejected CHCl3 in the plume,similar to the suggestion advanced in the studies of C6H5Cl38 and indole film.139 Additional support for this suggestion is relatedto the high reactivity of the chlorine atom, which suggests that any Cl formed in the film will react before it is ejected.


of neat C6H5Cl films38,172 above the ablation thresh-old, a number of different products are detected inthe gas phase (Table 5), whereas at lower fluences,only HCl is observed from freshly deposited films. Itwould be tempting to suggest that the observationof the various products above the threshold reflectsthe reduced reaction selectivity in ablation. However,in both cases, the examination of the irradiated filmsclearly shows additional photoproducts formed in thesubstrate below the threshold. In the F5PhN3 sys-tem, the formation of the dimer, fluoroazobenzene(F5PhNdNPhF5), in the irradiated film169 is estab-lished by FTIR and absorption measurements. Theproduct is formed quantitatively in the film, and allprecursor is chemically modified after a sufficientnumber of pulses. In the C6H5Cl system, formationof the phenyl derivatives in the film at fluences belowthreshold has been established by transmissionmeasurements,38,172 pump-probe172 studies, and ther-mal desorption spectroscopy.182 Thus, the detectionof a higher variety/number of different products inthe gas phase above versus below the threshold mustbe related mainly to the different material ejectionmechanisms in the corresponding fluence ranges(section II.B.2). Below the threshold, only the volatileproducts (i.e., N2, HCl, etc.) desorb, whereas the onesthat are strongly bound to the matrix are ejected onlyin the ablative regime. This delineation must betaken into careful consideration in studies on complexsubstrates (e.g., polymers) that rely largely on theexamination of gas-phase ejected species to extractmechanistic information.

The difference in ejection mechanisms may haveimportant implications for the extent of photoproductformation in the corresponding fluence ranges. Clearly,in the irradiation below the threshold with successivelaser pulses, there is accumulation in the substrateof “heavy” radicals that may result in side reactionsand extensive “polymeric” formation. This possibilityis indicated in the examination of the C6H5Cl38,182 andhalocarbon films.184 The deleterious effects of ac-cumulating chemical modifications to the integrityof the substrate and/or efficiency of material ejectionhave been noted in MADLI173 and in the irradiationof polymers11,86,174,178 (usually observed for irradiationclose to the threshold). In contrast, these side effectsmay be reduced in extent in the ablative regime, dueto the efficient removal of even the larger, stronglybound species. Indeed, subsequent irradiation athigher fluences removes the modified layers, withrestoration of the ejection signal. As indicated above,eq 3 suggests the extent of photofragmentation andphotoproduct formation in the remaining substrateto be constant with laser fluence. However, thisneglects the possibility that fragments and/or photo-fragments that are formed below the etched depthand are weakly bound to the matrix may diffuse tothe surface and thermally desorb (“postablation”desorption; section II.B.3). This effect may be acrucial factor for the success of the various lasermaterial-processing schemes, since it indicates thatsmall (usually the most reactive) species formed evenbelow the etched depth may be efficiently removed.Thus, we have stressed178,179 that the importance of

high substrate absorptivity for successful laser mate-rial processing may be related, besides the well-known criterion for efficient material removal andgood surface morphology, to the optimal effectedetching rate versus depth of product accumulation(“confinement” of photoproducts remaining in thesubstrate very close to the surface and efficientremoval).

Another effect of the photoproduct accumulation atfluences below the threshold may be the enhance-ment of light absorption with successive laser pulses.This results in an induction effect; that is to say, thedesorption signal from freshly deposited films isrelatively low, but it increases with successive laserpulses. This effect is most pronounced in the irradia-tion of films that absorb weakly at the irradiationwavelength (e.g., CH2Cl2 and C6H12 at 248 nm),168 butit is also observed38,167 for compounds such asC6H5Cl, in that their photolysis results in productsthat are stronger chromophores than the precursor.The observation of incubation even for the simple vander Waals films clearly shows that, in the corre-sponding phenomenon in polymers,1,11,170,171 the ne-cessity of chain breakage is of minor importance forits observation. Analytical modeling of the opticalchanges effected to the substrate with successivelaser pulses has been described in ref 171.

Turning now to the formation mechanisms of thephotoproducts observed in the ablative regime, in allsystems mentioned thus far, these can be fullyaccounted for by well-known reactions of the radical/fragments produced by the photolysis of the precur-sor. There is no indication that new reactivity path-ways open up in the ablative regime. A few selectedexamples are summarized in Table 5. For the indi-cated compounds, thermal decomposition is indi-cated160,161 to be minimal due to the strong bonds(Ebond g 3 eV) of the compounds. In the irradiationof C6H5Cl films at 248 nm (τpulse ) 30 ns), thesuggested radical mechanism (Table 5) is furthersupported by the fact that changes in the productpattern upon dilution of other compounds (C6H12,Freon, etc.) in the C6H5Cl matrix are found tocorrelate with the degree of reactivity of C6H5 andCl photofragments toward the added compound.47,172

The observed product pattern in the irradiation ofneat C6H5Cl films has been quantitatively modeledby MD simulations.156,157 However, the modelingrelies on reaction rate constants adapted from con-ventional solution studies. At present, the validityof this (implicit) assumption is unclear.172,183 Furtherlimitations in MD simulations may result from thenecessary scaling of the constants due to computingpower restrictions. We have suggested190 that thestudy of the influence of deuteration on photoproductpatterns, a well-known methodology in organic ki-netic studies, may be particularly useful in address-ing such questions.

Importantly, reactions induced in the UV ablationof photolabile systems may be highly exothermic. Ifthese reactions are fast enough to compete withmaterial ejection due to explosive boiling, then theenergy release into the system may contribute sig-nificantly to material ejection156,157 (discussed in more


detail in section III.B.2). In the case of van der Waalsfilms, this possibility is indicated in the com-parison of the UV (λ ) 248 nm) ablation of C6H5CH3and C6H5Cl films. Much more direct evidence hasbeen provided in the 248 nm ablation of films of thehighly photoreactive pentafluorophenyl azide (FPA).In the ablation of films of the compound dispersedwithin Ar, the pentafluorophenylnitrene (FPN) radi-cal is clearly detected177,180 in the plume, whereas thisintermediate is not observed in the ablation of neatFPA films.169 The difference is ascribed to the en-hanced thermal decomposition of the radical in thelatter case due to the high temperatures resultingfrom the high exothermicity of the reactions that takeplace in the neat films (Table 5). Furthermore, theblackbody radiation spectrum of the plume in thecase of the neat film is consistent with a temperatureas high as 2900 K. Further studies in this directionare expected to provide detailed information on the“coupling” between thermal and (photo)chemicalprocesses in the UV ablation of photolabile systems.

Despite their importance, ionization processes havenot been specifically examined in the UV ablation ofsimple molecular systems. A number of early workshave studied ionization processes for a variety ofrather complex organic molecules. Karas and Hill-enkamp have delineated different categories of com-pounds in relation with the observed ions and plau-sible ionization processes.195 The various factors thatmay be involved in ionic desorbate formation inMALDI are discussed in the subsequent articles. Incontrast, in the case of van der Waals films, thestudied compounds generally lack the functionalunits that would promote ion formation. As a result,observed ionic desorbates either represent preformedions of the salts mixed in the substrate35,36,43,191 orare formed via secondary absorption of photon(s) bythe ejected desorbates in the plume.18,64,139,189 In theformer case, it is important that the clusters that arecommonly observed in the irradiation of these sys-tems are generally not related to the embedded salts,but instead to the solvated ions (i.e., M(solvent)n areobserved rather than MX(solvent)n, where X ) an-ion). This result supports the suggestion that theclusters are directly ejected from the substrate (sincethe salts dissociate upon dilution into hydrated ions).Alternatively, the final form of the desorbed speciesmay be dictated by thermodynamic stability factors,as discussed by Knochemuss and Zenobi in this issue.As for the ionization effected by secondary photonabsorption in the plume, this has been exploited forthe desorption and state-resolved detection of thedesorbates18 in a one-laser experiment (i.e., by theablating laser pulse) and for analytical purposes.189

In the irradiation of frozen glycerol films,64 ion signalis observed at a well-defined threshold, above whichit increases linearly with laser fluence. A combinationof thermal ionization process (described by the Sahaequation197) and ionization via a single-photon ab-sorption in the plume was proposed to account forthese features.

Studies on ion formation in simple systems exhib-iting acid-base reactivity, electron-transfer pro-cesses, etc. evidently would help greatly in the elu-

cidation of the ionization processes in MALDI. Mash-ni and Hess80 studied ion formation and ejection inthe IR irradiation of films of simple organic com-pounds, and this work could provide the basis forcorresponding studies in UV. Such studies may bemost useful in addressing the importance of chemicalcharacteristics of the matrix and analyte for desor-bate ionization in the UV irradiation, the role that“exciton pooling” 68-71 processes (section II.B.4) areindicated to play, the relative contribution of theprocesses in the film versus in the plume, the timescale of ion formation versus that of material ejectionand plume expansion, etc.

2. Thermal Decomposition EffectsIn all previous cases, the compounds are thermally

stable, and the attained film temperatures are nothigh enough to result in any significant decomposi-tion. For thermally labile compounds, however, fur-ther chemical modifications via thermal decomposi-tion can be expected. The issue of thermal degradationto the substrate is crucial in the laser-processing/structuring of materials, and thermal decompositionof the desorbates is crucial in analytical applications(laser ablation mass spectrometry and MALDI).

Concerning the extent of thermal degradation tothe substrate, this is usually discussed in terms ofthe thermal diffusion time (tthermal, section II.B.2) ascompared with the nanosecond laser pulse width.However, as noted in the previous sections, the timescale of material ejection can be considerably longerthan the laser pulse width, and so the simplecomparison may underestimate somewhat the extentof thermal diffusion.185 A more detailed descriptionstarts from the consideration of the heat conductionproblem in the irradiation of the systems. Usually,the enthalpy formulation of the heat equation isemployed because of its convenience in dealing withphase changes including melting.1,187 Written in theframe of reference of the receeding surface (assumedto be along the z-direction), this becomes

with the main boundary condition concerning theenergy loss at the surface due to evaporation,

Here, υ represents the rate of material removal, andthe term in the parentheses represents the enthalpydifference between the vapor and the condensedphase (evaluated at the surface temperature, Ts). Cis the heat capacity, F the density, L the evaporationlatent heat, Hs the enthalpy of the substrate at theinteface, and H s

V the vapor enthalpy, H sV ) F∫T0

Ts

CP(V)(T ′) dT ′ (all expressed per unit volume). The

solution of the equation for the “nonstationary case”relevant to the irradiation with nanosecond pulseshas been discussed amply in the literature.1,187 The

∂H∂t

- υ∂H∂z

) ∂

∂z(k∂T∂z ) -

∂(I e-Rz)∂z

,

H(T) ) F ∫T0

TC(T ′) dT ′ (14)

k∂T∂z

|z)0 ) D∂H∂z

|z)0 ) υ(L - Hs + HV(Ts))


extent of thermal decomposition or any “pseudo-unimolecular” reaction up to time t is then assumedto be given by a time-integrated Arrhenius equation,∫0

t A exp(-Eact/RT(t ′)) dt ′ (where A and Eact are thecorresponding pre-exponential factor and activationenergy).185,186 However, no satisfactory analysis hasyet been reported for the case of explosive boiling(though similar to the so-called “volume” modelsconsidered in the case of polymer ablation anddiscussed in the contribution by Bityurin et al. in thisissue). Furthermore, the estimation of the extent ofthermal decomposition neglects that reaction con-stants may be significantly affected by the parallelchange in the physical state and structure of thesubstrate. Initial studies178 on doped polymers indi-cate that these effects are manifest in unexpectedway in the kinetics of photoproduct formation.

It has been even more challenging to account forthe thermal stability of the desorbates in the ablationof molecular solids. Evidently, the thermal confine-ment factor by itself is insufficient to account for thereduced thermal degradation of the ejected desor-bates. A number of different suggestions have beenadvanced to account for this observaion. Early ion,Vertes and Levine74-76 suggested that the reducedfragmentation of the biopolymers in MALDI is dueto a bottleneck in the energy transfer from the matrixto the protein. The suggestion was advanced toreconcile the observed stability of the biopolymerswith the very high temperatures necessary for theirejection by a simple thermal desorption/evaporationprocess. It was suggested that, because of the mis-match between the vibrational frequencies of thematrix and those of the biopolymer, an energy-transfer bottleneck is formed, and the phonons pumpreadily only the hydrogen bonds between the matrixand the protein, but not the “internal” biopolymervibrations. A competitive kinetic model74,76 and com-puter simulations77 indicate that the analyte can be,indeed, ejected with an internal temperature muchlower than that of the matrix. However, since inter-and intramolecular energy-transfer processes occurefficiently on a picosecond time scale, it is question-able whether irradiation with typical nanosecondpulses is short enough to preclude them. “Bottleneck-ing” effects appear more likely in the irradiation withpicosecond pulses; in fact, strong evidence for suchan effect has been indicated in polymers.196 However,even in the irradiation with nanosecond pulses, theconcept of energy bottlenecks may be a significantone, even if their physical origin may not be the oneunderlined by Vertes and Levine. Indeed, in the caseof explosive boiling, the formation of a high numberof bubbles within an initially homogeneous mediumcan introduce barriers of various sorts. The possibilityof the fast “exploitation” of the excess energy by the“volatile” matrix in bubble formation/growth contrib-uting to the cooling of the material has already beennoted (section II.B.3).

An interesting comparison was early reported byBuck and Hess26 that in the ablation of neat C6H6films at 248 nm, no naphthalene is formed; signifi-cant product is formed instead, presumably via priordecomposition of C-H bond(s), for substrate-medi-

ated heating of the adsorbate layer. Unfortunately,the reported information about the irradiation condi-tions is insufficient to establish the factors respon-sible for the observed difference. The use of a metallicsubstrate may have been responsible for the naph-thalene formation in the heating experiment. A moreconcrete example of thermal decomposition wasobtained in the CO2 laser ablation of frozen aqueoussolutions of the tryptophan and tyrosine aminoacids.188 At laser fluences slightly above the ablationthreshold, the molecular peak is stronger than thatof the fragments formed by the loss of the side chains.With increasing fluence, the molecular and the frag-ment peaks increase in parallel. However, above alaser fluence about twice the ablation threshold, thefragment signal increases at the expense of theparent molecule peak in a way very similar to thatdepicted in Figure 16. It was shown that heating ofthe plume by secondary absorption is insufficient toaccount for the observed trend. The fragmentationwas ascribed to the enhanced degradation of theanalyte molecules within the highly heated cavitationbubbles formed in the explosive boiling process (inwhich case fragmentation precedes any cooling effectsdue to postdesorption collisions). The study wasperformed in the IR. Nevertheless, given the similar-ity of the mechanisms to those described in sectionsII.B.2 and II.B.3, the findings of the study may bealso relevant to UV ablation, though in this case thephotoionization/dissociation in the plume, as de-scribed above, will probably set in at lower fluences.

A systematic attempt at assessing thermal decom-position in the UV ablation of molecular solids wasreported by Vertes and Gijbels.191 To this end, theionic and neutral desorbates in the irradiation ofphosphonium salts, neat or dissolved in matrices (nic-otinic acid, NH4Cl, 3-nitrobenzyl alcohol, glycerol),at λ ) 266 nm with τpulse ≈ 15 ns were studied. Inthe irradiation of the neat solids, fragments compat-ible with thermal decomposition as well as productsdue to recombination reactions are observed. In con-trast, upon dilution of the salts in UV-absorbing ma-trices, the yield of intact ions is highly increased, andrecombination products are greatly reduced (evi-dently due to the high degree of dilution within thematrix). A detailed characterization of the depen-dence of the extent of decomposition on irradiationconditions, matrix properties, and analyte concentra-tion was performed. Reduced decomposition is ob-served for matrices that absorb strongly at the irradi-ation wavelength (an effect probably sufficiently ac-counted for by eq 13) and for matrices of a low boiling/sublimation point. On the basis of their results, theauthors suggest that triphenylphosphonium saltswith polycyclic aromatic substituents can be used as“molecular thermometers”, but evidently simpler sys-tems would be preferable. Furthermore, the photo-chemistry of these systems is not well characterized,and it is not clear to what extent the observed prod-ucts may also derive from direct photodissociation.

As indicated in section II.C, particular emphasishas been placed on the possibility that the reduceddesorbate degradation is largely due to the “plumecooling effect”. In fact, this has been the most often


invoked21,22,30,75,115,116,138,191-193 factor accounting forthe “softness” (i.e., low internal energy excitation) ofanalyte ejection effected in their laser-induced ejec-tion from a matrix. More recently, in view of the MDsimulation results (section II.B.2), the rapid evapora-tion from the clusters (in the plume) has also beensuggested as a plausible factor leading to the coolingof the incorporated macromolecules. Yet, in othercases, the importance of the plume cooling has beenquestioned. For instance, in the UV ablation oftryptophan-glycine and tryptamine (λ ) 532 nmwith τpulse ) 10 ns at I ) 108 W/cm2),140 thermaldecomposition of the internal “hot” parent moleculeis demonstrated by a high contribution of metastablefragment in the post-ionization spectra. The frag-ment/parent molecule ratio in ablation is 4 timeshigher than that for jet-cooled compound.140 If, in-stead, postdesorption collisions are promoted byintroducing a gas flow, then decomposition is foundto be highly reduced. However, the experimentinvolved absorption by both the adsorbates and thesubstrate (in this case, the gas-phase collision num-ber may be much lower). Furthermore, the coolingefficiency will depend greatly on the presence, if any,of a matrix, relative concentration, etc.120-123,133-135,154

Indeed, in the irradiation of analytes within simpleinert matrices, the cooling degree is indicated115,116

to “exceed” that effected via laser ejection of neatcompounds into a supersonic jet, as employed inmany cases.117-119 Thus, the importance of plumeexpansion to reduce internal excitation appears tohave much evidence in support, but its determiningrole remains to be firmly established. Study ofbicomponent simple systems such as those used insection II.B.2, in which the analytes have thermallylabile/weak bonds, may enable a quantitative assess-ment to be made of the importance of the variousfactors for the thermal stability of the desorbates.

In all, the studies on van der Waals films illustratea number of interesting features about chemicalprocesses and effects in the UV ablation of molecularsubstrates. These studies provide a basis for under-standing the fragmentation/reactivity processes inthe UV ablation of more complex molecular sub-strates. However, evidently many issues remain tobe addressed, even in the case of simple compounds.First, it is not clear how reactivity in metastableconditions may be described and how it may differfrom that under equilibrium conditions. Thoughmetastable liquids have been studied extensivelyfrom a thermodynamic standpoint, very little con-sideration has been given to their photophysical/chemical properties (e.g., absorption and photolysiscross sections, reactivity, etc.). Furthermore, if vi-bronic “hot spot” formation occurs,67,68 the usualconcept underlying reaction kinetics in solutions,according to which photofragment “thermalization”occurs very rapidly before any reaction, may not bevalid in UV ablation. It would be interesting toexamine the applicability of Kasha’s rule,194 whichhas been shown to underline most photophysics/chemistry in solution studies (de-excitation to the S1state is usually ultrafast, and fragmentation yieldsand reaction patterns are nearly wavelength inde-

pendent). This is an important point to study, be-cause in polymers, different mechanisms (e.g., pho-tochemical versus thermal) of material ejection aresuggested to operate at different wavelengths, de-pending on the nature of the excited chromophores.Furthermore, thus far, no study of photoproductformation kinetics has been reported. In the case ofpolymers, there is evidence that the kinetics ofproduct formation (in the substrate) differs distinctlyand in unexpected ways from that at low laserfluences.178 Such kinetic studies will be definitelymuch more informative than just the characterizationof the ejected products. Unfortunately, addressingthese questions via gas-phase diagnostics, even insimple systems, may not be straightforward, sincethe ejection of some photoproducts within clustersmay hinder their quantitative analysis and compari-son. Complementary study of the species remainingin the substrate may be much more enlightening.

III. Studies on Liquids

Laser-induced material ejection from liquids hasbeen intensely pursued in its own right. Further-more, it has provided the basis for understandinglaser-induced ablation of more complex systems. Inparticular, early on it was indicated that the physicalmechanisms underlying the “soft” tissue ablationmay bear a strong resemblance to that of absorbingliquids. Further strong interest was generated byinitial studies showing that laser irradiation of liquidfilms can have a high technological impact, in par-ticular for the removal of submicrometer particles,laser micromachining, etc.

Laser-induced material ejection from liquids caninvolve a number of processes, such as evaporation,spallation, photochemical processes, dielectric break-down, and electrostriction. Herein, we confine de-scription to only some of these processes. In particu-lar, dielectric breakdown and related processes arenot discussed, since they have been amply re-viewed197-199 in the literature and they are more ofa physical rather than a chemical nature. However,it is noted that laser-induced plasma and breakdownprocesses may be crucial in various implementationsof the phenomenon. Excluding these mechanisms, theliquid-to-vapor transition can be effected by two basicways, either by boiling (normal boiling or explosiveboiling) or by cavitation effected upon application oftensile pressure to the liquid. Finally, the possibilitythat photochemical processes may induce materialejection has been examined in detail in the case ofliquids and this work is discussed in section III. C.

A. Transparent Liquids on Absorbing Surfaces:Liquid Superheating

For a fast enough heating rate such as that at-tained by an electric current pulse, superheating ofliquids well above the saturation point correspondingto their ambient pressure is known to occur.52,53 Incomparison with the conventional techniques, laserirradiation provides the possibility of attaining ul-trafast heating rates and studying nucleation phe-nomena with a high temporal resolution. In this re-


spect, the most well defined case occurs for transpar-ent liquids adjacent to/spread on UV absorbingsurfaces (such as polyimide, silicon or metallic film).The system offers the advantage that energy deposi-tion is clearly decoupled from the subsequent phasetransformation.

Upon irradiation with a UV pulse, the absorbedenergy diffuses from the substrate into the adjacentliquid. Since a short pulse duration implies a smallthermal diffusion length, strong overheating of theadjacent liquid layer can be effected even at moderateenergy densities far below any damage threshold ofthe substrate. The transient temperature field es-tablished in the liquid adjacent to the absorbingsurface is given by eq 7 on the basis of 1-D heatflow206,207,210 (energy “consumption” for phase changeand heat-transfer resistance at the liquid/substrateinterface are neglected). The temporal evolution ofthe interface temperature has been confirmed ex-perimentally by monitoring the reflectance of asemiconductor substrate (e.g., p-Si), which exhibitshighly temperature-dependent optical proper-ties.203,206,207 In particular, the semiconductor layercan be attached on the back surface of the laser-heated metallic substrate.206,207 In this arrangement,the temperature and the phase change phenomenacan be monitored independently (and in parallel) withnanosecond resolution via the optical reflectance,respectively, of the back and of the front side of thesystem. The development of this technique resolvesa main problem in the older conventional studies onsuperheating of liquids, namely that of accuratelyknowing the temperature of the vigorously boilingliquid. On the basis of such measurements, it isestablished that, at sufficiently high laser fluences,peak temperatures above the boiling point of theliquid are attained, in which case bubble formationis expected to occur.

A variety of techniques have been employed toprobe bubble nucleation and growth in the super-heated liquids adjacent to the absorbing surface. Themost direct method has turned out202-206 to be thetemporally resolved monitoring of the reflectance andscattering of an incident probing laser beam (Figure17). However, the quantification as well as thedetection sensitivity of these optical techniques issomewhat limited. To overcome the first limitation,Kim et al.212 have resorted to interferometric mea-surements relying on a Michelson-type arrangementto quantify the effective (optical) thickness of thebubble layer. To overcome the second restriction,Yavas et al.210 have relied on surface plasmon reso-nance spectroscopy (SPR). The technique is based onthe fact that the bubbles represent effective scatter-ers for the surface plasmons of the metallic substrate,thus resulting in the broadening of their resonance.In the absence of bubble formation or followingbubble collapse on the microsecond time scale, onlyan angular shifting of the resonance curve is observedas a result of the temperature rise in the liquid.

On the basis of these techniques, distinct regimesof the bubble growth have been delineated. In thefirst stage, separate bubble nuclei (embryos) beginto grow on the surface upon irradiation above a

liquid-specific threshold fluence. The bubble nucleiformation is evidenced by a characteristic initialincrease in the specular reflectance of a probing laserbeam incident on the liquid/substrate interface205

(Figure 17 (bottom)) and by the appearance of a humpin the SPR signal. These changes can be ascribed tothe transient change in the dielectric constant, ε, ofthe medium that occurs upon the formation of afoamy layer (corresponding to a thin layer of smallbubbles on the metallic surface). Maxwell-Garnetteffective medium theory can be employed205 to cal-culate εeff of the foam as a function of the fractionalvolume of the bubbles. A simple ray optics analysisof the three-layer system consisting of liquid, foam,and the metallic substrate satisfactorily accounts forthe observed transient increase in the reflectivity.

The exact threshold for embryo formation dependson the sensitivity of the employed probing techniqueto the bubble size. A particularly sensitive way ofmonitoring is afforded by SPR, with an indicateddetection limit of ∼20-nm bubble size. Embryos arefound to nucleate for surface temperatures onlyslightly above the boiling condition (∆Tsurf ≈ 10

Figure 17. (Top) Experimental setup for the opticalreflectance and scattering monitoring of the bubble forma-tion/growth that occurs in the transient laser heating.(Bottom) Transient reflectance in the irradiation of 2-pro-panol (a) and water films (b) adjacent to a Cr-plated (0.2µm thick) quartz upon irradiation with an excimer (248nm) laser pulse. The probe beam has λ ) 752 nm, iss-polarized, and is incident at a 7° angle. Reprinted withpermission from ref 205. Copyright 1993 The AmericanPhysical Society.



K).206,210 The very small overheating necessary forbubble nucleation is due to the heterogeneous natureof the process, with the solid surface providing thenucleation sites. The heterogeneous nature is con-firmed by the dependence of the required superheat-ing on the degree of smoothness of the metallicsurface. Due to the heterogeneous nature of nucle-ation, the nucleation rate of bubble formation ismodified52 from that in the homogeneous case (sectionII. B.2) into

where F(θ) ) 1/4 (2 + 3 cos θ - cos3 θ). N o2/3 is the

number of molecules per unit area at the interface,and θ is the contact angle. As a result of theheterogeneous nature, the degree of penetrationeffected into the metastable region may be greatlylimited.

The embryos are formed during the nanosecondlaser pulse, and their initial growth is largely inertia-controlled. The growth of (spherical) bubbles can beestimated from the Rayleigh equation that describesthe mechanical energy balance for a growing bubblein an incompressible (i.e., constant density) liquid:52,53,200,201

In the early stage, R(d2R/dt 2) is negligible and bubblegrowth is inertia controlled, so that the bubble radiusas a function of time can be approximated as (contactangle 0°)

where the pressure difference term is approximatedusing the Clausius-Clapeyron equation. PV and TVrepresent the pressure and temperature inside thebubble, PL the ambient pressure, Tsat(PL) the satura-tion temperature corresponding to PL, F the density,and hLV the vaporization heat). The formula yieldsan upper limit, since it assumes constant superheat-ing of the liquid, a single spherical bubble on a solidsurface, and reversible bubble expansion. Thus, thegrowth rate is estimated212 to be 10-14 m/s (for H2Oat peak interface temperatures in the range of425-500 K), whereas experimentally, it is found tobe ∼3 times lower (0.5-3.6 m/s).205,206,208,212 At 10 nsfollowing the onset of the excimer laser pulse, theupper limit for the fractional volume of bubbles inthe superheated water layer amounts to f ≈ 0.05-0.1, corresponding to an estimated bubble numberdensity of ∼1013 m-2 (∼1% mass fraction of vaporiza-tion).

Further growth (stage II) of the bubbles can occuronly if sufficient heat is supplied from the surface.This growth is reflected by the decrease in thespecular reflectance (Figure 17) with the concomitantincrease of the scattered light. These optical changesare due to the fact that bubble size (Rbubble) grows toa size ≈ λprobe/2πn (where λprobe is the probe beamwavelength and n is the refractive index), and as aresult, Mie scattering dominates (in which case theextinction coefficient of the foam layer Cext ) ∫Nbubble-(Rbubble)2πReffective

2 dRbubble, with Reffective2 ≈ Rbubble

2 forlarge enough bubbles). Thus, the transition to Miescattering has been taken to represent the thresholdfor the bubble growth, distinct from the threshold forbubble nucleation described above. For atmosphericpressure, such growth is observed to occur for peaksurface temperatures of ∼100 K above the boilingtemperature (for H2O and CH3OH).205,208,212 As dis-cussed in section II.B.2, spontaneous growth canoccur once the embryonic bubble radius reaches acritical radius, Rcr ) 2σ/(PV - PL).52,53 It is demon-strated that growth is observed when the super-heated liquid thickness exceeds the critical radius.Thus, the growth rate in this regime is limited bythe speed of heat diffusion from the solid surface intothe liquid (∝ xDt). If this condition is met, thenboth the size and the number density of the bubblesincrease for times up to ∼150 ns. The bubbles remainconfined within the superheated liquid layer, and thebubble size does not exceed the superheated liquidlayer thickness.

In the 150-400-ns range (stage III), the effectivebubble layer thickness, as monitored via interferom-etry, increases, but the number density of the bubblesdecreases, manifested by recovery of the light-scat-tering loss.212 Thus, in this regime, bubbles coalesce,and the bubble layer can be well approximated as athin vapor film. For high enough fluences, when thefilm thickness exceeds a certain limit, instabilitiesset in, resulting in the violent ejection of the overlyingliquid layer, which subsequently breaks down intojets. Finally, after 500 ns (stage IV), the size and thenumber density of the bubbles are reduced, resultingin the decrease of the light scattering in the restora-tion of the initial width of the surface plasmonresonance. The decay rate of the bubbles is muchslower than their growth rate, with their collapsebeing complete after ∼1 µs. This result sharplycontrasts the simple ideal bubble dynamics solu-tions,201 which predict the collapse time of a bubbleto be smaller than the expansion time. Plausibly, theslow decay rate is due to the high density of cavita-tion bubbles211 formed at these irradiation conditions.In this case, nonlinear interaction between thebubbles can be expected through the liquid pressurefluctuations that they induce. It is shown that thisinteraction may result in an increase of the bubblelifetime as compared to that expected from idealsolutions.

The dependence of the phenomenon on external(applied) pressure has been studied.206,207 For pres-sures up to ∼2.2 MPa, the nucleation of the embryosis found to be nearly insensitive to the pressure,consistent with the fact that this process is domi-

J )No

2/3(1 + cos θ)

2F(θ) (3F(θ)σπm )1/2

exp[- 16πF(θ)σ3

3kBTL(ηpsat - pL)2] (15)

R d2Rdt 2

+ 32 (dR

dt )2 ) 1FL

(PV - PL - 2σR ) (16a)

R ) x2(PV - PL)3FL

t with

PV - PL )FVhLV(TV - Tsat(PL))

Tsat(PL)(16b)


nated by inertial factors. Similarly, the rates for thesubsequent bubble growth and collapse (in the Miescattering regime) are independent of pressure, butthe total volume of the bubbles (as measured by thetotal optical scattering signal) decreases with in-creasing pressure. This can be ascribed to the factthat, with increasing pressure, the boiling pointincreases, and thus the corresponding degree ofsuperheating, Tsurf(Flaser) - Tboil, decreases. At highenough external pressures, bubble formation is in-hibited.

Bubble nucleation and formation results in acoustictransients that propagate through the liquid209,212

with sonic velocity. Below the onset of vaporizationthreshold, a low-amplitude pressure is generated onlyvia thermal expansion. However, above the threshold,the collective and synchronous expansion of multiplebubbles results in a strong compression pulse. Es-sentially, the superposition of the pressure waveletsthat are produced by the numerous growing bubbles[the pressure generated by a single sphere being P∝ FRbubble

2 (d2Rbubble/dt 2)] results in a pulse that canbe approximated as a plane wave. Indeed, the tran-sients have been detected by piezoelectric transduc-ers,209 by an optical beam deflection technique basedon the “mirage effect”,209 and by SPR.210 Because ofdifferences in their time resolution and integratingnature, these techniques yield somewhat differentvalues, with the more sensitive SPR reporting a ∼40-ns width of the acoustic pulse and ∼2 MPa peakamplitude. Because the generated pressure is pro-portional to d2Rbubble/dt 2, the acoustic enhancementin the laser-induced vaporization process is causedby the fast bubble expansion in the growth stage andnot by the much slower bubble collapse. This en-hanced acoustic excitation is taken advantage of in“steam laser cleaning” and in the liquid-assistedmaterial ablation (section IV).

Most interestingly, even after the bubbles collapse,repeated bubble formation may occur213 on a longer,microsecond time scale. This secondary bubble for-mation is due to cavitation at the liquid-solidinterface, induced by the reflected acoustic pulsesgenerated by the primary bubble growth. By using adouble-laser-pulse experiment and probing differentareas of the sample surface, it is shown213 that anessential requirement for the observation of this“memory effect” is the preceding bubble formation atthe liquid-solid interface. With increasing delay timebetween the two laser pulses, the “memory effect”decays, and eventually no cavitation enhancement isobserved. Thus, even upon collapse, the primarybubbles induced by the UV irradiation do not com-pletely disappear but survive for a few hundredmicroseconds and can serve as embryonic bubbles.The survival rate depends on the nature of thepresent solutes, with salts (e.g., NaCl) providing aparticularly effective stabilization of the embryos.This result is directly related to the issue of thenature and stabilization of microscopic bubbles (nu-clei) within liquids, a question that has not yet beensatisfactorily resolved.

Liquid superheating can be similarly observed inthe case of absorbing particles dispersed within a

nonabsorbing solvent for irradiation with short enoughpulses (τp < 10-6 s).229 For the case of particle sizesmaller than their thermal diffusion length, so thatthey are thermally homogeneous, the fluence for fullydeveloped bubble formation around the particle is

where T0 is the initial temperature, R, Fg, γ, and Cpg

represent the size, density, integral absorption coef-ficient (such that particle absorption equals 4πR2γ),and heat capacity of the particle, Λ is the specific heatof evaporation, and D is the thermal diffusivity of thesolvent. The term within the parentheses gives thefluence necessary for the heated layer to reach theboiling point, and the third term is related to theextra heat required for its vaporization. A furtherdetailed description may be found in ref 230.

The bubble formation around individual micropar-ticles (melanosomes) heated by a 30-ps laser pulsehas been visualized by time-resolved optical micros-copy. Bubbles are observed as early as 0.5 ns after-ward, expanding to a few micrometers on 0.1-1-µs215

time scale. Collapse of the bubbles formed close tothe free surface of the colloidal solution generateshydrodynamic flow, which can result in significantmaterial ejection in addition to any evaporativelosses.229 The bubble formation in the superheatingof colloids has also been exploited to set up a noveloptical switch.216 The scheme relies on the fact thatthe bubbles coalesce into a large one with a largerefractive index change from the surrounding me-dium.

The bubble formation results in a sharp acoustictransient,215 as detected by piezoelectric measure-ments and optically, whereas below the bubbleformation threshold, the detected pressure is muchweaker and due to the thermoelastic expansion of theparticles. Furthermore, the use of time-resolveddegenerate four-wave mixing has shown a largetransient change of the refractive index of colloidalsuspensions of carbon irradiated with nanosecondpulses.221,222 The amplitude oscillates on a time scalethat is much shorter than the laser pulse width. Thisbehavior has been associated with the acoustic wavegenerated in the fluid “coherently” by the explosiveboiling around the heated particles. Furthermore,this pressure generation may be responsible for the“giant” photoacoustic effect observed in the irradia-tion of carbon colloids,217 though other explanations,such as stable gas formation217 and/or formation ofmicroplasmas, had been advanced earlier. The bubbleformation around the heated particles has beentreated theoretically. In particular, a recent analyti-cal study stresses the importance of the pressure,besides the thermal factor, for bubble formation/growth.223 A dependence of the bubble dynamics(growth rate and size) on the laser pulse width ispredicted, even for times much shorter than thethermal relaxation time. MD simulations of thedamage effected to the absorbing particle itself athigher fluences have also been reported.224

Flaser )(Tboil - T0)

γ (43 RFgCpg+ 4FCpxDτpulse) +

4γ

FΛxDτpulse (17)


Absorbing particle suspensions have also beenemployed to effect laser ejection and ionization ofmacromolecules for mass spectroscopic examination.In fact, it was the method employed by Tanaka etal.3 in one of the first publications demonstrating thepotential of laser irradiation to effect “soft ejection”of macromolecules in the gas phase. To this end, thebiopolymers are dissolved in glycerol suspensions ofcobalt nanoparticles (30 nm in diameter) acting aschromophores. The term SALDI, for surface-assistedlaser desorption/ionization, has been introduced218 todifferentiate this technique from MALDI. The ad-vantages of this approach for analytical purposesinclude (a) the lack of matrix-related interferingpeaks in the mass spectra and (b) the nearly constantvalue of the particle suspension absorption over alarge wavelength range accessible by a variety oflasers, thereby overcoming the limitation of restrictedwavelength range absorption of matrices. The influ-ence of the particle size and nature on the ejectionefficiency of macromolecular analyte ions has beenstudied218-220 and has been largely explained by thedifferent degree of overheating attained for differentparticle sizes. Ejection of ions as high as ∼30 kDahas been demonstrated, but there seems to be alimitation for higher masses. It is likely that thislimitation is related mainly to the ionization steprather than the ejection process. The present evi-dence suggests that the ionization process is medi-ated by the liquid and not by the particles.

In all, the studies on superheating of liquidsadjacent to laser-absorbing substrates has provideddetailed information on the kinetics of nucleationphase change phenomena that would not have beenamenable through the conventional heating methods.The use of even shorter laser pulses will provide thepossibility of addressing in more detail the initialembryo formation and thus the validity of severaltenets of the classical thermodynamic theory ofmetastability. Furthermore, the method provides aconvenient experimental approach to study in detailproperties of the superheated liquids, particularlytheir photophysical and chemical properties (thusproviding a basis for understanding processes inother systems, e.g., van der Waals solids, sectionII.B.2), and in addition a convenient means of pro-ducing “controllably” a high-density ensemble ofstructures (i.e., bubbles) on the submicrometer/na-nometer level, with plausible applications in nano-technology.

B. Absorbing Solutions and Liquids

1. Photoinert Systems: Photomechanical Mechanism ofMaterial Ejection

Material ejection of absorbing liquids and solutionshas also been studied by techniques similar to thosedescribed in section III.A [Here the presentation islimited only to the studies of irradiation of the freesurface of the solutions. A large number of studies,e.g.,226,235,236 employing immersed optical fibers orirradiation of tissues within solutions, are more of amedical interest and are discussed by Vogel andVenugopalan in this issue]. For instance, time-

resolved transmission of a probing beam propagatingparallel to the free (irradiated) surface of the liquidand laser-flash photography (probing can be conve-niently performed in the visible range, where thesolutions are transparent) have been employed toprobe processes within the irradiated liquids. Usu-ally, aqueous solutions with dissolved (photostable)chromophores such as Orange G at 532 nm (τpulse ≈25 ns, R ) 50-150 cm-1),231,232 K2CrO4 at 355 and248 nm (τpulse ≈ 25 ns, R ) 1 × 102-1.7 × 103

cm-1),227,228,233,234 NaCl solutions at 193 nm,214 etc.have been studied (Figure 18). A noticeable drop inthe transmission through the solution of the probingbeam is observed above a specific fluence of the UV

Figure 18. Sequence of images within the liquid (1st and3rd row) and above its free surface (2nd and 4th row)obtained by laser flash photography in the 248-nm ablationof aqueous K2CrO4 solution (R ) 55 cm-1). For thephotographs in the left column, Flaser ) 1.2 J/cm2, and forthose in the right column, Flaser ) 1.85 J/cm2 (estimatedaverage temperature increase in the irradiated volume of15 and 26 K, respectively). Photograph a shows theinitiation and growth of cavitation bubbles in the surfacemillimeter layer. The black and white stripe at the bottomof the photograph is due to the refractive index changeinduced by the propagating bipolar pressure wave. Photo-graph c at 15 µs shows the fusion of bubbles into largerones (∼100-150 µm in diameter), while other bubbles havecollapsed (seen as empty spaces between the bubbles closeto the surface). Reprinted with permission from ref 227.Copyright 1995 American Institute of Physics.



beam.227,228,232-234 Flash photography at various timedelays relative to the ablating laser pulse demon-strates the drop to be due to the scattering of theprobing light by bubbles growing within the bulk ofthe liquid. Importantly, by either technique, bubbleformation at depths well below the optical penetra-tion depth is demonstrated (e.g., in the 355-nmirradiation of aqueous K2CrO4, R ) 55 cm-1, maximalbubble concentration is observed at twice the opticalpenetration depth, decreasing to zero only at twicethis length).227,228 The bubbles grow rapidly over thefirst microsecond, with the ones close to the surfacerupturing and resulting in material ejection, whilefurther material ejection occurs in a second stagefollowing the expansion, coalescence, and collapse ofbubbles (Figure 18).

The threshold for material ejection is more ac-curately determined from a change in the acoustictransients generated during laser irradiation (Figure19). The measurement relies on the fact that, without

mass ejection, the generated pressure pulse in themedium is bipolar (vide infra) and the time-inte-grated pressure is zero, whereas with material ejec-tion, the positive component of the pressure pulseexceeds the negative one due to the recoil momentum.On the basis of such measurements, the thresholdfor material ejection is indicated233,234 to be somewhathigher than the threshold for cavity formation. Atany rate, the temperature at the ablation threshold

is estimated (assuming a one-photon process) to bewell below the solvent boiling point. The absorbedenergy is about 1 order of magnitude lower than thespecific enthalpy of vaporization. Thermal explosionaround superheated micro-inhomogeneities cannot beinvoked to account for material ejection in thesesystems. In the case of the homogeneous aqueoussolutions, the average distance between chromophores(at least for reasonable concentrations) is in the rangeof a few nanometers, and thermal diffusion smoothsout any thermal heterogeneities within the typicalnanosecond laser pulse width. Furthermore, theemployed solutes (NaCl, K2CrO4, etc.) have beendemonstrated to be photochemically stable, even afterextensive irradiation. Thus, photochemical processessuch as those considered in section III.B.2 cannot beresponsible in the present case. Given the very smallthermal diffusion length into water (∼0.1 µm for t ≈1 µs), temperature changes below the optical pen-etration depth can be neglected (at least for moderatelaser fluences), and thus, they cannot be responsiblefor bubble formation at these depths. Bubble forma-tion and the effected material ejection have to beascribed to a photomechanical mechanism, i.e., to theeffect of stress-wave generation and propagation inthe bulk of the liquid at ambient temperature.

The photomechanical mechanism and its conse-quences are described in detail in the article in thisissue by Paltauf and Dyer; thus, only a brief descrip-tion is provided here. This mechanism relies on thefact that the rapid laser-induced thermal expansioncauses a pressure rise, given by the product of theGruneisen coefficient Γ and the absorbed energydensity:225

where â is the thermal expansion coefficient, cs thespeed of sound, CV the heat capacity at constantvolume, and κT the isothermal compressibility, andθ ) τpulse/tacoustic, where tacoustic is the time required foran acoustic wave to traverse the irradiated volume.The factor in the parentheses corrects for the reduc-tion in the stress amplitude due to the wave propa-gating out of the irradiated volume during the laserpulse (assumed to have a rectangular time profile).The stress entails a radially propagating cylindricalwave, which can be neglected for laser beam diam-eters substantially wider than the light penetrationdepth, and two plane waves counter-propagatingalong the laser beam axis (one toward the surfaceand the other into the sample). The plane wave thattravels toward the free surface (liquid/air surface)suffers a change of its amplitude sign upon reflectionfrom it, due to the acoustic impedance Fcs (F is thedensity and cs the speed of sound) of the irradiatedmedium being higher than that of air. Physically, thethermal expansion directed into the medium gener-ates a compression wave, whereas the outward ex-pansion generates negative stress (rarefaction wave).

The faster the heating, the higher the magnitudeof the stress in the medium, with the ultimate

Figure 19. (a) Temporal shape of the acoustic pulsesmeasured by a broadband piezoelectric transducer in the248-nm irradiation of K2CrO4 solutions (R ) 5.5 × 102 cm-1)for the indicated excimer laser fluences. Reprinted withpermission from ref 233. Copyright 1998 Elsevier ScienceB.V. (b) Flaser dependence of the integrated signal of theacoustic pulses. Above a specific fluence (ablation thresh-old), the integrated signal deviates from zero as a result ofthe recoil momentum of the ejected plume. Reprinted withpermission from ref 234. Copyright 1998 Springer-Verlag.

∆P ) ΓRFlaser(1 - e-θ

θ ) ) âFκΤcV

RFlaser (1 - e-θ

θ )(18)



efficiency of stress generation attained for heatingtimes much faster than the time required for stressto propagate through the irradiated volume. This isusually expressed as Rcsτpulse , 1 (R is the absorptioncoefficient, cs the speed of sound, and τpulse the laserpulse duration), where the dimensionless parameterrepresents the degree of temporal confinement of thelaser-induced stress in the irradiated volume. Fortypical excimer nanosecond pulses, this condition issatisfied for aqueous solutions for R e 103 cm-1.

The generated bipolar pressure wave propagatesinto the depth of the medium without significantattenuation. If the pressure of the tensile componentis lower than the saturation pressure of the liquidat a given temperature, the liquid is in a thermallymetastable state under tension, and the vapor-cavitynucleation is initiated. For amplitudes of the tensilestress exceeding the critical value balancing thesurface tension, cavity growth occurs, and the con-tinuity of the liquid is broken. Ablation can thereforebe interpreted in terms of a mechanical “rupture” ofthe liquid surface effected by large tensile stress. Thethreshold value of the tensile pressure for ablationfrom aqueous solutions is found227,231-234 to be ap-proximately 1.8-3.6 MPa (the exact value stronglydepending on the existing nucleation sitesstrappedgas bubblessin the liquid). Rupture of the bubblesgrown close to the liquid surface produces a plumeof vapor and microdroplets. The collapse of the initialcrater and of the large bubbles formed by coalescenceof smaller ones in the depth of the irradiated volumeresults in a complex hydrodynamic flow entailingfurther ejection of material in the form of liquidstreams (jets) (Figure 18).

Ablation creates sufficiently high recoil stressamplitudes that convert into so-called “weak shockwaves” upon propagation into the (acoustic nonlinear)medium. Low-amplitude (not exceeding several bars)acoustic waves propagate at the speed of sound (1.5km/s), but higher amplitudes propagate at supersonicvelocities that make the front of the pressure wavesteeper. This evolution is well monitored by flashphotography.227,228 In a medium with acoustic non-linearity, these shock waves develop after propaga-tion over a characteristic distance, Lsh:

where F is the medium density, cs is the speed ofsound, τac is the duration of a transient stress wave,and P is the pressure amplitude. The formation ofthese weak shocks is important for the optimizationof the various implementations of UV ablation, asthey propagate through the substrate and may causemechanical deformations178 at regions relatively faraway from the irradiation spot (e.g., injuries intissues, delaminations in painted artworks, etc.).

For irradiation at very high fluences, superheatingwill occur in parallel, and material ejection arisesfrom the combination of explosive boiling and me-chanical rupture. In this case, material ejection isindicated to occur in three distinct stages.233,234

During the nanosecond laser pulse irradiation, nosignificant surface deformation or plume ejection is

observed, despite the high surface temperature riseto the superheat limit. This delay can be related tothe time necessary for bubble growth. However,afterward a dense vapor-cavity zone is formed in thesubsurface region due to the explosive nucleation,accompanied by violent plume ejection. Thus, explo-sive vaporization is the dominant material ejectionmechanism in the early stage. The vapor-plumeejection is maintained for a few microseconds. Thesurface depression that is caused by the recoilmomentum of the ejected material activates subse-quently hydrodynamic motion, which results in up-ward flow and bulk liquid ejection. Thus, the finalstage is dominated by large-scale surface deforma-tions under the influence of gravity and liquid inertiaacquired by the impact of the ejected mass. This long-term hydrodynamic motion is not observed in thecase of the spallation-induced material ejection atlower fluences. In that case, no significant surfacedeformation is observed after ∼20 µs.

2. Photolabile Liquids: Photochemical Mechanism ofMaterial Ejection

In the previous studies, the absorbing analytes aswell as the solvents are photochemically inert. In theirradiation of photolabile organic compounds, a “pho-tochemical mechanism” may also contribute to mate-rial ejection. This mechanism was advanced11 alreadyin the very first studies on the UV ablation ofpolymers in order to account for the “clean” etchingeffected in this case. This result was suggested to berelated to the fact that UV light absorption resultsin bond dissociations and the fragments or the smallgaseous photoproducts they form were considered toexpand (requiring a larger volume than the initialmacromolecular structure), thereby causing a pres-sure rise and ejection of the material. Accordingly,ablation was suggested to occur above a certainfluence, at which the bond dissociation rate exceedsthe electronic energy relaxation time or the recom-bination rate of the produced radicals so as to achievea critical number of species (fragments and/or prod-ucts) (the need to consider rates is introduced in orderto account for the observation that for long enoughlaser pulses, volume ejection is not observed):

where q is the quantum yield, σabs the absorptioncross section, N the number density of the chro-mophores, R the rate of any “loss” mechanism, andFcr the critical number of radicals/photoproducts thatmust be formed to induce material ejection. Sincemost of the incident energy is used in bond dissocia-tions, this hypothesis seems to directly account forthe observed “clean” etching (i.e., little thermaldissipation and degradation, e.g., melting to thesurrounding area). It is clear that the validity of sucha mechanism would be of profound importance forthe optimization of the implementations of UV abla-tion. However, despite the large number of studies,it has proven difficult to obtain conclusive evidencefor this mechanism.

Lsh ≈ Fcs3τac/2πεP (19) F(t,x) ) ∫0

t [qσabsNIlaser(t ′,x)hν

- R(t ′,x)] dt ′ g Fcr

(20)


In the case of complex systems, it is very difficultto delineate between the various contributing mech-anisms (i.e., between photochemically induced reac-tions and thermal decomposition of the weakerbonds). To address the contribution of photochemicalmechanism in simpler systems, Srinivasan and Ghoshexamined photoproduct formation in the 248-nmablation of liquid C6H6.241 A number of differentspecies (naphthalene, biphenyl, carbon) were detectedin the post-irradiation analysis of the samples. Onthe basis of this result, a “pure” photochemicalmechanism was suggested for the ablation of thissystem. However, since reactivity may occur/continuewell after material ejection, product formation doesnot constitute sufficient evidence to establish thephotochemical nature of ejection. Furthermore, simi-lar products may be formed by different mechanisms.For instance, in the 193-nm irradiation of liquidbenzene at fluences higher than ∼10 mJ/cm2, prod-ucts similar to those reported by Srinivasan andGhosh were observed.242 However, in this case, theyare indicated to derive from the thermal decomposi-tion of the compound due to the much higher tem-peratures attained (the absorption coefficient of thecompound at this wavelength being orders of mag-nitude higher than that at 248 nm). In fact, most ofthese products are observed253,254 even in the laser-induced breakdown of the aromatic liquids, thoughtheir formation in this case is mediated via freeelectrons, and no electronically excited states aredetected.

Fukumura, Masuhara, and co-workers have re-ported several comparative studies on the KrF exci-mer laser (248 nm)-induced material ejection fromorganic photolabile liquids, namely neat samples ofC6H6 and its derivatives (C6H5X and C6H5CH2X,where X ) CH3, Cl), as well as solutions of thesecompounds in nonabsorbing solvents (alkanes andchloroalkanes). The threshold for material ejectionhas been established243-246 by photoacoustic spec-troscopy. The first sharp change in the generatedphotoacoustic signal as a function of laser fluence(Figure 20) is verified by shadowgraphic detec-

tion243-246 of the ejected plume to correspond to theonset of ablation (the second breakpoint in the figuremay be related to the enhanced scattering of theincident light by the plume). Most importantly, thedetermined ablation thresholds of the neat benzenederivatives and of their highly concentrated solutions

(0.6-1 M, R ≈ 250 cm-1) do not correlate with theboiling points of the liquids (Table 6). Accordingly,

the authors relate the 248-nm-induced materialejection from these systems to the photochemicalactivity (â-bond cleavage efficiency) of the com-pounds: C6H5CH3 with a high quantum photodisso-ciation efficiency having the lowest threshold, and thephotochemically inactive C6H6 having the highest.

The suggested radical formation has been con-firmed by time-resolved luminescence and absorptionspectroscopies.246 An absorption band around 320 nm,which can be ascribed to benzyl radical, is observedfor all examined systems. Chain reactions of theseradicals are likely responsible for the photoproductsobserved by Srinivasan and Ghosh.241 Time-resolvedabsorbance of the incident UV light is found to benearly constant during the laser pulse. It is thussuggested that one-photon absorption occurs mainlyby chromophores, which subsequently undergo ef-ficient fragmentation. The radical concentration isestimated to be ∼0.05 M at the threshold for all neatliquids and ∼0.006 M for the highly concentratedsolutions. In fact, it is somewhat smaller for alkyl-benzene derivatives and somewhat higher for thebenzyl compounds (the difference ascribed to thedifferent viscosity of the solvents). This indicates thata critical number of species/fragments must be pro-duced for material ejection to occur. The authorssuggest that the small radicals (H, Cl, CH3) producedby reaction 21 form gaseous products that induce avolume increase (estimated to be ∼20%), leading tomaterial ejection. The time necessary for the forma-tion of these products may be responsible for thedelayed material ejection in the UV laser irradiationof these liquids. Indeed, at the threshold, materialejection (plume development) was detectable by lasershadowgraphy nearly 100-500 ns after the laserpulse (for ambient pressure), and only at very highfluences is ejection observed to be initiated duringthe laser pulse.

In contrast, observations differ significantly forlow-concentration solutions (0.06-0.1 M, R ≈ 25cm-1) of the aromatic compounds. In this case, theablation thresholds are about twice the ones for thehigh-concentration solutions.248 This difference couldbe presumed to be simply due to the lower absorptioncoefficient of these solutions. However, this cannot

Figure 20. Photoacoustic signal amplitude as a functionof the laser fluence in the irradiation of benzyl chloride atλ ) 248 nm. Reprinted with permission from ref 243.Copyright 1994 American Chemical Society.

Table 6. Laser Ablation Thresholds (λ ) 248 nm, τpulse≈ 30 ns) of Aromatic Compounds (at AmbientConditions)a (Reprinted with permission from ref252. Copyright 1996 American Chemical Society.)

liquid Fthr (mJ/cm2) R (cm-1) Tthr (K) bp (K)

benzene 100 2100 410 353chlorobenzene 60 1900 370 405toluene 35 2400 350 383benzyl alcohol 60 ∼1900b ∼350 478benzyl chloride 30 ∼2200b ∼340 452

a Fthr, threshold laser fluence; R, absorption coefficient; Tthr,estimated surface temperature at threshold (Tthr ) Fthr/FCP,where F is density and CP specific heat); bp, boiling point.

C6H5CH2X 98hν

C6H5CH2 + X (21)



be the only factor, since the ablation threshold of thedilute solutions is found to scale with the boilingpoint of the solvent, whereas no such correlation isfound at high solute concentrations. Furthermore, thedelay in the jet-ejection onset with respect to theablating laser pulse, as established by laser shad-owgraphy, is shown245 to be related to the absorbedenergy density, ranging from ∼2 µs close to thethreshold down to a few tens of nanoseconds at muchhigher fluences. Similar dependences of the jet-ejection onset times on absorbed energy are observedfor each solvent independently of the employedabsorbing solute. Thus, ejection dynamics appears tobe regulated by the solvent and not by the solutes.In parallel, the luminescence spectra of the solutesare found245-247 to become unstructured and broaderwith increasing delay time between pump and prob-ing beams and with increasing pump fluence. Thesespectral changes are consistent with those expectedfor highly vibrationally excited/heated molecules. Thespectral broadening precedes the onset of materialejection, thereby providing strong evidence that theliquid is heated prior to the jet development. Materialejection is indicated to initiate at the ambient boilingtemperatures of the solvents (i.e., regular boiling),taking on the character of the explosive boiling athigher fluences (the exact estimation of the attainedtemperatures depends on the extent of multiphotonprocesses that take place in the irradiation of thesesystems, as described in section III.C).

On the basis of the previous results, the authors245

suggest the mechanism of material ejection from thesolutions of the benzene derivatives to switch from aphotochemical to a thermal one with decreasingconcentration of the photolabile solute. Evidently, itis assumed that the amount of radicals and of volatileproducts that is formed with decreasing photolabilesolute concentration becomes insufficient to result inmaterial ejection, and as a result, a thermal mech-anism takes over. Though this is a very attractivedelineation of mechanisms, it must be noted that thephotomechanical mechanism described in sectionIII.B has not been considered in detail. Given thestrong evidence for the importance of the laser-induced stress to effect “cold” ablation of solutions,the argument that estimated temperatures at thresh-old are lower than the boiling point (Table 6) isinconclusive for a photochemical mechanism. It isplausible that a combined contribution of the twomechanisms is involved. Examination of the laser-induced stresses and cavity formation in the irradia-tion of these systems would be useful in addressingthis point further.

Insight into the role of photochemical processes inlaser-induced material ejection has recently beenprovided by MD simulations157 (in fact, performed forvan der Waals films). These simulations have shownthat the energy liberated by exothermic reactions cancontribute to reaching the critical energy for materialejection (Ecr in eq 6). Additionally, the formation ofphotoproducts with weaker intermolecular interac-tions lowers the cohesive energy of the system,thereby facilitating further material ejection. In theframework of nucleation theory, it may be considered

that “volatile” photoproducts serve as nuclei to pro-mote heterogeneous bubble formation and growth(i.e., enhanced J in eq 15). In view of these results,it is important to examine in the photolabile organicliquids whether material ejection is due to photo-product-induced volume expansion (i.e., as usuallypresumed by simple photochemical models) or to theenergy released by exothermic reactions resulting inexplosive boiling (e.g., as suggested by the MDsimulations157).

It is interesting that, in the two cases that the samecompound has been studied in both liquid and solidphases, discrepancies are noted in relation with theobserved products. The naphthalene and biphenylspecies observed241 by Srinivasan and Ghosh in the248-nm ablation of liquid C6H6 have not been de-tected26 in the corresponding study on condensedC6H6 films. On the basis of this result, Buck and Hessquestioned the validity of the suggestion of Shrini-vasan and Ghosh241 for the photochemical nature ofUV ablation of C6H6 (section III.C.). In the ablationof condensed C6H5CH3 films, no products have beendetected47 (at least for moderate fluences above thethreshold), thus indicating a very low photolysisefficiency of condensed C6H5CH3.47 This sharplycontrasts with the efficient radical formation detectedin the irradiation of the liquid.245,252 Several plausiblereasons for these discrepancies can be advanced, butthere is no direct experimental evidence. A system-atic comparison of photoproduct formation accordingto the phase of the substrate is evidently required inorder to get further insight into the chemical pro-cesses in UV ablation. This comparative study is alsoneeded to assess the extent to which results onphotochemical processes in liquids are extrapolableto the ablation of polymers and tissues. [In the caseof polymers, an initial study has reported a numberof interesting differences in the material ejectiondynamics according to the state of the substrate.255]

Studies on the femtosecond-induced material ejec-tion from the above photolabile liquids have also beenreported. Hatanaka et al. have relied249,252 on time-resolved absorption spectroscopy to study processesin the irradiation of aromatic liquids with femtosec-ond pulses (248 nm, ∼300 fs). In the irradiation ofC6H5CH2Cl liquid at low laser fluences (<15 mJ/cm2),an absorption band ascribable to the benzyl radicalis already intense 10 ps after the pump pulse,suggesting a very fast predissociation of benzylchloride upon a one-photon excitation. Upon irradia-tion at higher pump laser intensities, a significantdrop in the transmittance is additionally observed attimes >1 ns. High-resolution surface-scattering im-aging250 reveals the liquid surface to remain nearlyflat in the first nanosecond, while enhanced surfaceroughness is indicated after this time. Assumingsimilar detection limits, the ablation threshold isindicated to be nearly equal for the nanosecond andfemtosecond laser pulses. This sharply contrasts withthe predictions of photomechanical models and ofmolecular dynamics simulations30 for a decrease ofthe threshold upon irradiation with femtosecondpulses. The authors suggest that the invariability ofthe threshold in the C6H5CH2Cl system indicates the


operation of a photochemical mechanism similar tothat suggested in the nanosecond case.242

On the other hand, in the corresponding femtosec-ond irradiation of liquid toluene, the ablation thresh-old is reported251 to be lower than that of C6H5CH2Cl,whereas the opposite trend is found in the corre-sponding nanosecond examination of the two com-pounds.242 Furthermore, the benzyl radical in thefemtosecond irradiation of C6H5CH3 liquid could notbe detected, even at the highest examined fluences,whereas the radical is efficiently produced in thecorresponding nanosecond process. To account forthis difference, the benzyl radical in the case oftoluene is suggested to be formed via a secondaryphoton absorption by the triplet state. This processis likely in the irradiation with nanosecond pulsesbut, of course, not in the case of femtosecond pulses.Concerning ablation of toluene, it is suggested thatablation with femtosecond pulses must be photother-mal in nature. However, clearly, there are manyaspects that must be examined before even a basisfor understanding femtosecond-induced material ejec-tion can be formulated.

C. Optical Processes in the Irradiation at HighIrradiances

Evidently, optical processes constitute the first steppreceding any of the material ejection processesdescribed in the previous sections. The reason forplacing this section last is the limited informationon the issue. Changes in the optical properties ofliquids upon superheating can be expected, sincetheir density decreases52,53 with increasing degree ofsuperheating. The consequent change in intermo-lecular interactions can affect strongly the electronicstates. However, as the following studies indicate,there are additional, dynamic processes that becomesignificant in the irradiation at high fluences.

Upon superheating at high energy densities (g600J/cm3 for R in the range of 825-11 500 cm-1), thesurface reflectivity of aqueous K2CrO4 solutions isobserved to decrease sharply in the later part of theincident nanosecond laser pulse (pulse “truncation”)(λ ) 248 nm).238 Pump-probe experiments employingtwo different wavelengths indicate that this decreaseis not due to scattering of the incident light by theejected plume. Thus, the pulse truncation must beascribed to a sharp decrease of the refractive indexof the liquid in the later part of the incident laserpulse. Nikiforov et al.238 suggest that the decreasereflects the highly reduced density of the overheatedH2O solvent. On the other hand, the 193-nm absorp-tion coefficient of superheated H2O is indicated237 tobe nearly 5 orders of magnitude larger than the room-temperature absorption value. In these experiments,a Q-switched Er:YAG laser is employed to heat H2O,while the absorption at 193 nm is probed 300 nsafterward. The increase has been ascribed to theblue-shifting of the UV absorption spectrum239 of theoverheated liquid due to the disruption of the hydro-gen bond network of H2O. In contrast, Longtin andTien240 ascribe the optical changes that are inducedin the irradiation of H2O at 266 nm (I ) 1.0 × 1010-2.0 × 1010 W/cm2) to the ejected electrons and the

OH radicals formed by the photoionization and thephotodissociation of H2O. The hydrated electronabsorbs strongly throughout the near-IR, visible, andUV regions (a 7-10 orders of magnitude increase ofthe absorption coefficient). Longtin and Tien sug-gested that these optical changes can be takenadvantage of in applications by coupling irradiationat 266 nm with wavelengths in the visible (e.g., 532nm).

Fukumura, Masuhara, and co-workers have re-ported particularly detailed studies on the issue ofelectronic excitation processes in the irradiation athigh UV laser irradiances. Time-resolved absorptionand luminescence spectroscopies have been used toprobe the dynamics of electronic excitation and de-excitation in the ablation process. In the irradiationof organic chromophores in nonabsorbing solvents(Table 5), they note245,247,248 that, for some of theemployed chromophores (biphenyl or phenanthrene),the internal conversion from the excited S1 state hasa very low quantum yield to account for the indicatedefficient heat generation. Furthermore, the totalfluorescence intensity as a function of laser fluencesaturates close to the threshold, indicating that a newchannel of the S1 state deactivation opens up. Inparallel, in the irradiation of the low-concentrationsolutions, the time-resolved absorbance is found toincrease during the excitation pulse.245 A plausibleexplanation was advanced on the basis of “a cyclicmutiphotonic absorption process”. According to this,the absorbing solutes and/or the benzyl radicalsproduced by the photolysis of the photolabile solutesabsorb further photons to higher electronic states.Alternatively, higher electronic states may be formedvia annihilation of S1 excited molecules. Deactivationof these A*(Sn) states occurs extremely fast (∼pico-second) and provides a way to rapidly convert theabsorbed energy into heat. Following deactivation,the recovered A*(S1) can participate in subsequentexcitation/de-excitation cycles.

Their results appear to be analogous to thosereported in the irradiation of van der Waals films,68-71

and also closely related to the previous changes inoptical properties noted upon superheating of aque-ous solutions.237-239 Indeed, the group has demon-strated that similar annihilation processes may bemost important also in the ablation of doped poly-mers.

Despite their limited number, these studies aremost important because they clearly illustrate thatphotophysical properties of highly heated liquids maydiffer significantly from those under equilibriumconditions, and they demonstrate the importance oftaking into account these changes for the quantita-tive understanding of the processes in the laser-induced material ejection processes. To this end,further studies must establish the exact conditionsunder which these processes become important.

A + hν f A*(S1)

A*(S1) + hν f A*(Sn) f A*(S1) + heat

A*(S1) + A*(S1) f A*(Sn) f A*(S1) + heat (22)


IV. Applications

In this section, applications of the ablation ofmolecular solids and liquids are presented. To avoidextensive overlap with other articles in this issue andother reviews, the discussion highlights a few specificdirections, rather than trying to provide an exhaus-tive list of the many different applications (describedin detail in refs 1, 5-7).

A. Cryogenic Films-Based TechniquesCryogenic films are finding increasing use in both

analytical and material-processing applications. Inthe first case, laser-induced material ejection fromfrozen aqueous solutions has been always of interestbecause of the widespread occurrence of H2O in theenvironment, in tissues, etc. However, the transpar-ency of H2O in the visible and UV regions has limitedwork on the ablation of frozen aqueous solutions. Onthe other hand, the high absorptivity of H2O in theIR and the recent availability of convenient andreliable lasers in this spectral range have resultedin renewed interest in these systems. Besides studieson the potential of MALDI,43,106,107,256-258 the IR-induced material ejection has been exploited for thecharacterization and monitoring of processes in icesin relationship with atmospheric studies.36,259,260

The ablation of frozen solvent/biopolymer systemshas also been demonstrated as a convenient tech-nique for the ejection and subsequent deposition ofnovel microstructure formation, as discussed byChrisey et al. in this issue. Leone19,22,112,113,261 firstadvanced the potential of using laser-ablated beamsto study collisional and reaction dynamics of hyper-thermal species. Harrison and Polanyi262 presenteda theoretical analysis of the feasibility of usingcrossed ablated beams to this end. The use of laser-ablated beams for such studies can meet a significantneed, since except for seeded beams and photolyticsources, there are no other convenient ways toproduce high-flux jets of hyperthermal neutral mol-ecules/atoms. An extra advantage of the use ofablated beams accrues from the capability to vary thedesorbate translational energy conveniently via thelaser irradiation parameters (laser fluence, forwardor backward irradiation, etc.). Experimentally, thispotential has been exploited to study the dependenceof the etching process of Si on this parameter.Hyperthermal Cl2 (Ekin up to ∼6 eV) is shown112 toetch Si >30 times more efficiently than thermal Cl2.For practical applications, etching may be furtherenhanced by using higher fluences to increase theCl fragment concentration and/or by applying anexternal electric field to enhance ionic formation inthe ejected plume.162 Enhanced oxidation of Si isalso demonstrated by ablated hyperthermal O3beams.279,280

The advantages of high-flux beams of radical/reactive species produced in the UV ablation ofphotolabile cryogenic films can be particularly usefulfor the processing of substrates. The use of laser-ejected reactive plumes for this purpose presentsversatility and can complement plasma techniques.The beams of high-purity reactive fragments that are

produced in the UV ablation of FPA/Ar condensedfilms (Table 5) have been used for the chemicalsurface modification of aromatic polyester, alkylthiol,poly(ethylene terephthalate), etc.281 The treatmentresults in the fluorination of the polymer surface,which thus becomes hydrophobic and chemicallyinert.281 Silylation of polymer surfaces has beeneffected with the plume produced in the ablation offrozen silanes.265 Surface nitridation of carbon andsilicon substrates is effected by using the reactivenitrogen plume produced in the ablation of N2films.274-276

Irradiation of frozen systems can also be used forthe deposition of new materials, thereby widening thechoice of target materials that can be used for pulsedlaser deposition. SiC and CN films have been pre-pared, respectively, by ablation of CH4 on Si and co-ablation of frozen CH4/N2 targets.276 Diamond-likecarbon films have been deposited by ablation offrozen CH4 and CO2 on graphitic carbon,275 ac-etone,271,272 or acetylene.270 Ablation of these com-pounds results in the efficient production of CHxspecies, which are known to play an important rolein diamond synthesis. Amorphous diamond-like filmsof a high electrical resistivity (>105 Ω‚m) with a∼40% sp3 fraction are produced. Some debris isobserved (for deposition at Tdep ≈ 200 °C), probablydue to droplets ejected from the frozen target (re-lated to the mechanisms discussed in sectionII.B.2). Deposition at higher substrate temperatures(Tdep ≈ 300 °C) results in the vaporization/disin-tegration of these droplets and in the formation of avery high quality film. No films could be producedin the irradiation of CH3OH films, probably be-cause the abundant oxygen radicals (detected opti-cally in the plume) react with the deposit. Pulsedlaser deposition has been used extensively for anumber of larger organic molecules (i.e., not cryo-genic films) with excellent results. For instance, thinfilms of pentacene deposited by ablation at 248 nmexhibit reduced surface roughness and increasedelectrical conductivity as compared to films grown byconventional thermal evaporation.285 Liquid crystalmixtures have also been transferred to a targetwhile retaining the original composition.277 Novelelectroluminescence molecules were ablated anddeposited for electroluminescence device prepara-tion.278 In this case, because of the photolabile natureof the end groups of the molecules, the fluence mustbe carefully optimized to avoid photodecomposition.Deposition methods and mechanisms are discussedin further detail in the article by Chrisey et al. inthis issue.

B. Liquid-Based Applications

1. “Steam Laser Cleaning” Technique

The studies on laser-induced superheating of liq-uids presented in section III.A have largely beenmotivated by the potential of so-called “steam laserremoval” of submicrometer particulates from sur-faces. Due to the ever-decreasing device size, cleaningmicrometer and submicrometer particles from sur-faces has become a strong need in lithography,


microelectronics, telecommunications, etc. Particu-lates of this size adhere with relatively strong force,van der Waals force being the dominant one forsubmicrometer particles [in which case the force isproportional to ARparticle/8πz2, where z represents theseparation between the particle and the surface, andA is a constant in the range of a few (1-10) electron-volts, depending on the nature of particle/substratematerial(s)], whereas electrostatic force becomesimportant for larger ones.202,287 Because the adhesionforce is proportional to the particle radius, whereasthe cleaning force is usually proportional to thesurface area or volume [i.e., inertial force m(d2x/dt 2)∝ R3(d2x/dt 2)], the acceleration necessary to effectejection scales as R-2. As a result, conventionalremoval techniques such as ultrasonic, plasma clean-ing, etc. become ineffective for micrometer and sub-micrometer particles.

By comparison to conventional techniques, laserirradiation has been shown to have high potentialas a solution to this problem.202 Laser-induced re-moval of the contaminants can be effected via eitherdirect absorption of the laser light by the particlesand/or the substrate (“dry laser cleaning”) or the priorapplication of a liquid film (“steam cleaning”). Theparticle removal can be followed by either opticalmicroscopic examination or the scattering via theparticles of an incident probing laser beam. Usually,“steam cleaning” is effected with strongly absorbingsubstrates and liquid films that are transparent atthe incident laser wavelength, though the feasibilityof CO2 laser radiation in combination with (absorb-ing) water films has also been examined.295 Typically,water mixed with 10-20% alcohol (to improve sub-strate wetting) is applied in a few tenths to severalmicrometers thickness (the optimal film thicknesshas not thus far been specified, though evidencesuggests that this parameter may be important forthe efficiency of the technique). A range of differentparticulates, such as those of Au, SiO2, Al2O3, Fe2O3,Si, polystyrene, debris, etc., are demonstrated to beefficiently removed from substrates such as Au, Mo,Si wafers and membranes, polyimide photoresistfilms, polymers, etc. (Figure 21) with 248-nm exci-mer-laser radiation.287-289,295,296 The important resultof these studies is that the laser cleaning efficienciesare much higher than those in dry cleaning. Thispermits the use of lower laser fluences, which isparticularly important for heat-sensitive substratesor for particles that can melt. It is noted also thatthe steam-laser technique differs distinctly from theusual liquid-removal/dissolution methods, since theapplied liquid film is nearly completely evaporatedupon laser heating and thus its penetration in thesubstrate is minimized.

The process has been studied for well-characterizedspherical polymer and silica particles of differentdiameters (∼ tens of micrometers to 100 nm) on Si(λ ) 532 nm, τpulse ≈ 7 ns) (Figure 21). The depen-dence of the cleaning efficiency on the number oflaser shots, Nlaser, can be approximated by H(Nlaser)≈ 1 - (1 - η)Nlaser, where η represents the cleaningefficiency for a single pulse (the formula assumesidentical isolated particles). In practice, deviations

are usually observed and may be ascribed to thedifferent cleaning threshold fluences for particles ofdifferent sizes, to the surface roughness of thesubstrate, etc. Applying typically 10-20 pulses, ef-ficiencies from 10-50% up to 40-90% are observeduntil finally they saturate with a higher number ofpulses. At least for submicrometer particles, a sharp,well-defined threshold for removal is demonstrated287

which is independent of the particle shape andmaterial, though for larger particles, removal ap-pears289,290 to require lower laser fluences. The par-ticle removal threshold is shown to correlate with theonset of vaporization.287-290 Thus, the particulatedetachment is related not to the high pressureassociated with the critical conditions of the liquidbut to the high-amplitude pressure wave which isgenerated by the fast-growing bubbles (section III.A).For micrometer-sized particles, the acoustic tran-sients result in large enough accelerations (108-109

m/s2) to effect the detachment of the particles, whichare subsequently carried away by the ejected high-speed jet stream. Theoretical models of the removalprocess have been reported,291,292 but several experi-mental observations remain to be accounted for in asatisfactory way. At any rate, the induced tempera-ture and acoustic transients are short enough thatno damage or deterioration is induced to the rela-tively robust solid substrates. For shorter laser pulses(30 ps, λ ) 583 nm), an even lower threshold is

Figure 21. Laser steam cleaning efficiency as a functionof laser fluence (λ ) 248 nm) for polystyrene spheres ofvarious sizes and for SiO2 and Al2O3 particles on Sisubstrate (20 cleaning steps have been applied). A filmconsisting of 90% H2O and 10% 2-propanol is applied in awell-controlled (200-400 nm) thickness. The results clearlyindicate a material and size-independent removal thresholdand a steep increase of the removal efficiency with increas-ing laser fluence. Reprinted with permission from ref 288.Copyright 2000 Springer-Verlag.



indicated.293 The decrease in the threshold has beententatively related to the lower conduction heat lossesin the metallic/Si wafer, and thus more efficientheating of the liquid.293

The use of “steam laser cleaning” has also beenexamined in relation with the removal of coatings.Oxide coatings from metallic and semiconductorsubstrates can be removed, with the removal ef-ficiency increased further via electrochemical poten-tial application.294 This enhancement appears toresult from an increase in the absorptivity of thecoating. Organic coatings can also be removed froma variety of substrates, including painted art-works. In this case, however, the use of liquid onthe highly sensitive organic/painted surfaces mayhave deleterious side effects, and we have advocatedthe potential of the “dry cleaning” process, in-stead.10,85,86

2. Liquid-Assisted Material Processing and NanostructureFormation

A different application concerns the laser irradia-tion of absorbing solutions for the etching/structuringof transparent or hard-to-process substrates such assilica,263,264,268 calcium fluoride,268 fluoropolymerfilms,267 etc. To this end, a thin layer of stronglyabsorbing solution is spread on the material to beprocessed, and irradiation is performed through thetransparent substrate (laser-induced backside wetetching, LIBWE).267,268,283 It is suggested that theliquid is highly overheated via cyclic multiphotonicprocesses (section III.C). The highly heated vapor ofhigh presure that is formed close to the interfaceattacks the softened surface of the substrate andresults in material removal.267,268 The importance ofthe liquid overheating has been directly shownrecently for the case of a (transparent) liquid layerapplied on an absorbing substrate.269 A substantialreduction (by 20-40%) of the ablation threshold andaugmentation of the ablation efficiency of the sub-strate were observed at fluences at which liquidexplosive vaporization is initiated. The rapid subse-quent cooling prevents damage to the bulk. Themethod overcomes most drawbacks of alternativelaser-processing schemes of such materials, presentsthe advantages of simplicity and of one-step process-ing in ambient conditions, and may thus be usefulfor mass production/processing.

There is an increasing number of reports on novelmaterial formation and, in particular, formation ofnanostructures via the laser heating of absorbingsurfaces immersed within liquids or of appropriatelyprepared absorbing solutions/colloids. Various formsof carbon, including amorphous diamonds films, havebeen deposited from simple aromatic compounds(benzene, toluene, cyclohexane, etc.) at a laser-heatedsolid-liquid interface.282,284 Cr2O3 has been depositedin the irradiation of aqueous CrO3. Au298 andAg296,299,300 nanoparticles have been prepared viaablation of corresponding targets in water by a Cuvapor laser (10-20 J/cm2). The metal nanoparticlesobtained upon solvent evaporation are disk-shaped,with a 20-60-nm diameter. The laser-prepared Agcolloids have been shown to be useful for surface-

enhanced Raman spectroscopy (SERS), and in facttheir activity is comparable or even superior to thatof chemically prepared systems.299,300 The approachovercomes the problems of particle contaminationand of limited stability that are encountered in theuse of chemically prepared colloids. Furthermore, thecolloids can be prepared in the presence of solutesand can be used directly for the SERS examina-tion.299,300 The reduction of cytochrome c by suchcolloids has also been examined.301 Cu particles havebeen prepared by ablation of CuO powder within2-propanol at 1064 and 532 nm.302 Most interestingly,the colloids obtained in the 1064-nm irradiation arestable under aerobic conditions even without the useof protective agents, whereas the corresponding col-loids prepared by chemical approaches are highlyunstable toward oxidation. The difference has beententatively ascribed to the somewhat higher size andmore perfect spherical shape of the particles producedby the laser method. Dispersed micrometer-sizedpowders of aromatic hydrocarbons and phthalocya-nines in poor solvents are fragmented upon excimerlaser irradiation, and as a result the initially turbidsolutions become transparent ones, composed of∼100-nm particles.303

A phenomenological model of the growth of thenanostructures has been described,297,298 but thefundamental mechanisms of their formation remainto be elucidated. It appears likely that the very fastvaporization of the liquid adjacent to the absorbingunits plays a critical role in the process.282-284 Thevery high transient pressures, temperatures, andvapor densities attained certainly affect crucially thedecomposition/reaction processes and the kinetics ofaggregation for the formation of the clusters. Thesubsequent fast cooling of the bubble products maybe responsible for the formation of the indicated novelmetastable forms. At any rate, despite the presentmechanistic uncertainties, several advantages oflaser ablation over other techniques for nanostruc-ture preparation are clear: a chemically “clean” prep-aration (i.e., reduced byproduct formation, simplerstarting materials, no need for catalyst, etc.), thepossibility of producing material forms that may notbe attainable by milder preparation methods, and ahigh degree of control over the size and properties ofthe produced structures via appropriate selection ofirradiation parameters (wavelength, fluence, repeti-tion rate, etc.). The degree of control afforded by thelaser method has been indicated in a number of cases.Particle size is generally observed to decrease withincreasing number of laser pulses.297,298,304 Controlover the average particle size and size distributionvia laser wavelength has also been indicated for Agnanoparticle formation in H2O (for instance, irradia-tion at 1064 nm results in particles of ∼30 nm meansize, whereas at 355 nm, their mean diameterdecreases to ∼10 nm).304 These effects seem to berelated to the extent of fragmentation of the nano-particles induced by self-absorption (i.e., absorptionof the incident laser light). Further control has beenobtained by employing solutes298,305 which may coator react with the surface of the clusters, therebylimiting their agglomeration.


The previous examples clearly indicate the poten-tial of irradiation of liquids/solutions at high laserirradiances for highly innovative processing and/ormaterial production schemes that may turn out tobe of very high and widespread impact in variousfields and, in particular, in nanotechnology.

V. AcknowledgmentsWe thank Dr. Akira Yabe and Dr. Hiroyuki Niino

(NIMC) for help and many suggestions in the firststages of the preparation of this manuscript. Thework was supported in part by the Ultraviolet LaserFacility operating at F.O.R.T.H. under the ImprovingHuman Potential (IHP)-Access to Research Infra-structures program (contract no. HPRI-CT-1999-00074), by the Training and Mobility of Researchers(TMR) program of the European Union (project no.ERBFMRX-CT98-0188), and by the PENED program(project no. 99E D 6) of the General Secretariat ofReseach and TechnologysMinistry of Development(Greece).

VI. References(1) Bauerle, D. Laser Processing and Chemistry; Springer-Verlag:

Berlin, 2000.(2) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299.(3) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida,

T. Rapid Commun. Mass Spectrom. 1988, 2, 151.(4) Hillenkamp, F.; Karas, M. Int. J. Mass Spectrom. 2000, 200, 71.(5) Boyd, I. A. Photochemical Processing of Electronic Materials;

Academic Press: London, 1992.(6) Miller, J. C., Haglund, R. F., Jr., Eds. Laser Ablation and

Desorption; Experimental Methods in the Physical Sciences 30;Academic Press: San Diego, CA, 1998.

(7) Azema, A. G., Laude, L. D., Eds. Excimer Lasers; KluwerAcademic Publishers: Dordrecht, The Netherlands, 1994; p 447.

(8) Tam, A. C.; Leung, W. P.; Zapka, W.; Ziemlich, W. J. Appl. Phys.1992, 71, 3515.

(9) Chrisey, D. B.; Hubler, G. K. Pulsed Laser Deposition of thinfilms; Wiley-Interscience: New York, 1994.

(10) Georgiou, S.; Zafiropulos, V.; Anglos, D.; Balas, C.; Tornari, V.;Fotakis, C. Appl. Surf. Sci. 1998, 127-129, 738.

(11) Srinivasan, R.; Braren, B. Chem. Rev. 1989, 89, 1303.(12) Zhou, X.-L.; Zhu, X.-Y.; White, J. M. Surf. Sci. Rep. 1991, 13,

73.(13) Domen, K.; Chuang, T. J. Phys. Rev. Lett. 1987, 59, 1484.(14) Domen, K.; Chuang, T. J. J. Chem. Phys. 1989, 90, 3318.(15) Domen, K.; Chuang, T. J. J. Chem. Phys. 1989, 90, 3332.(16) Chuang, T. J.; Domen, K. J. Vac. Sci. Technol. 1987, A5, 473.(17) Kutzner, J.; Lindeke, G.; Welge, K. H.; Feldman, D. J. Chem.

Phys. 1989, 90, 548.(18) Natzle, W. C.; Padowitz, D.; Sibener, S. J. J. Chem. Phys. 1988,

88, 7975.(19) Cousins, L. M.; Leone, S. R. J. Mater. Res. 1988, 3, 1158.(20) Cousins, L. M.; Leone, S. R. Chem. Phys. Lett. 1989, 155, 162.(21) Cousins, L. M.; Levis, R. J.; Leone, S. R. J. Phys. Chem. 1989,

93, 5325.(22) Cousins, L. M.; Levis, R. J.; Leone, S. R. J. Chem. Phys. 1989,

91, 573.(23) Nishi, N.; Shinohara, H.; Okuyama, T. J. Chem. Phys. 1984, 80,

3898.(24) Tabares, F. L.; March, E. P.; Bash, G. A.; Cowin, J. P. J. Chem.

Phys. 1986, 86, 738.(25) Braun, R.; Hess, P. J. Chem. Phys. 1993, 99, 8330.(26) Buck, M.; Hess, P. Appl. Surf. Sci. 1989, 43, 358.(27) Cho, C.-C.; Polanyi, J. C.; Stanners, C. D. J. Phys. Chem. 1988,

92, 6859.(28) Zhigilei, L. V.; Kodali, P. B. S.; Garrison, B. J. J. Phys. Chem. B

1997, 101, 2028.(29) Zhigilei, L. V.; Kodali, P. B. S.; Garrison, B. J. J. Phys. Chem. B

1998, 102, 2845.(30) Zhigilei, L. V.; Garrison, B. J. J. Appl. Phys. 2000, 88, 1281.(31) Mark, T. D.; Foltin, M.; Kolibar, M.; Lezius, M.; Schreiber, P.

Phys. Scr. T 1994, 53, 43.(32) Haskin, S. M.; John, P. J. Phys. Chem. B 1999, 103, 4566.(33) Srinivasan, R. Appl. Phys. A 1993, 56, 417.(34) Handschuh, M.; Nettesheim, S.; Zenobi, R. Appl. Surf. Sci. 1999,

137, 125.

(35) Han, C. C.; Han Y.-L. W.; Chen, Y. C. Int. J. Mass Spectrom.1999, 189, 157.

(36) Focsa, C.; Destombes, J. L. Chem. Phys. Lett. 2001, 347, 390.(37) Georgiou, S.; Koubenakis, A.; Kontoleta P.; Syrrou, M. Laser

Chem. 1997, 17, 73.(38) Georgiou, S.; Koubenakis, A.; Lassithiotaki, M.; Labrakis, J. J.

Chem. Phys. 1998, 109, 8591.(39) Koubenakis, A.; Elimioti T.; Georgiou, S. Appl. Phys. A: Mater.

Sci. Process. 1999, 69, 637.(40) Yingling, Y. G.; Zhigilei, L. V.; Garrison, B. J.; Koubenakis, A.;

Labrakis, J.; Georgiou, S. Appl. Phys. Lett. 2001, 78, 1631.(41) Koubenakis, A.; Labrakis, J.; Georgiou, S. Chem. Phys. Lett.

2001, 346, 54.(42) Duncan, M. A.; Knight, A. M.; Negishi, Y.; Nagao, S.; Nakamura,

Y.; Kato, A.; Nakajima, A.; Kaya, K. Chem. Phys. Lett. 1999,309, 49.

(43) Baltz-Knorr, M. L.; Scriver, K. E.; Haglund, R. F. Appl. Surf.Sci. 2002, 197-198, 110.

(44) Godovac-Zimmermann, J.; Brown, I. R. Mass Spectrom. Rev.2001, 20, 1 and references therein.

(45) Sadeghi, M.; Wu, X.; Vertes, A. J. Phys. Chem. A 2001, 105, 2578.(46) Robin, M. B. Higher Excited States of Polyatomic Molecules;

Academic Press: New York, 1974; (a) Vol. I, pp 155-178; (b)Vol. II, pp 230-232.

(47) Koubenakis, A. UV Laser Ablation of Molecular Cryogenic Films.Ph.D. Thesis, University of Crete, Heraklion, 2002.

(48) Kelly, R.; Miotello, A. Phys. Rev. E 1999, 60, 2616.(49) Miotello, A.; Kelly, R. Appl. Phys. Lett. 1995, 67, 3535.(50) Kelly, R.; Miotello, A. Appl. Surf. Sci. 1996, 96-98, 205.(51) Knight, C. J. AIAA J. 1982, 20, 519.(52) Debenedetti, P. Metastable Liquids: Concepts and Principles;

Princeton University Press: Princeton, NJ, 1996.(53) Skripov, V. P. Metastable Liquids; John Wiley & Sons Inc.: New

York, 1974.(54) Skripov, V. P.; Pavlov, P. A. Sov. Technol. Rev. B: Therm. Phys.

1989, 2, 171.(55) aus der Wiesche, S.; Rembe, C.; Hofer, E. P. Heat Mass Transfer

1999, 35, 25.(56) Batanov, V.; Fedorov, V. B. JETP Lett. 1973, 17, 247.(57) Martynyuk, M. M. Sov. Phys. Technol. Phys. 1974, 19, 793.(58) Martynyuk, M. M. Combust. Explos. Shock Waves 1977, 13, 178.(59) Porteous, W.; Blander M. AIChE J. 1975, 21 (3), 560.(60) Ready, J. F. Effects of High-Power Laser Radiation; Academic

Press: Orlando, 1971; p 433.(61) Oxtoby, D. W. Appl. Phys. B 1994, 58, 267.(62) Shen, V. K.; Debenedetti, P. G. J. Chem. Phys. 1999, 111, 3581.(63) Schilling, G.; Ernst, W. E.; Schwentner, N. Appl. Phys. B 1994,

58, 267.(64) Wu, P. K.; Ringeisen, B. R.; Bubb, D. M.; Auyeung, R. C. Y.;

Horwitz, J.; Houser, E. J.; Pique, A.; McGill, R. A.; Chrisey, D.B. J. Appl. Phys. 2001, 90, 3623.

(65) Westmacott, G.; Ens, W.; Hillenkamp, F.; Dreisewerd, K.;Schurenberg, M. Int. J. Mass Spectrom. Ion Processes 2002, 221,67.

(66) Dreisewerd, K.; Schurenberg, M.; Karas, M.; Hillenkamp, F. Int.J. Mass Spectrom. Ion Processes 1994, 141, 127.

(67) Fain, B.; Lin, S. H. J. Chem. Phys. 1989, 91, 2726.(68) Logan, D.; Wight, C. A.; Apkarian, V. A. Chem. Phys. 1997, 217,

99.(69) Ehring, J. H.; Sundqvist, B. U. R. J. Mass Spectrom. 1995, 30,

1303.(70) Ehring, J. H.; Sundqvist, B. U. R. Appl. Surf. Sci. 1996, 96-98,

577.(71) Knochenmuss, R.; Vertes, A. J. Phys. Chem. B 2000, 104, 8406.(72) Allwood, D. A.; Dyer, P. E.; Dreyfus, R. W. Rapid Commun. Mass

Spectrom. 1997, 11, 499.(73) Karas, M.; Gluckmann, M.; Schfer, J. J. Mass Spectrom. 2000,

35, 1.(74) Vertes, A.; Levine, R. D. Chem. Phys. Lett. 1990, 171, 284.(75) Vertes, A.; Gijbels, R. Scanning Microsc. 1991, 5, 317.(76) Vertes, A.; Gijbels R.; Levine, R. D. Rapid Commun. Mass

Spectrom. 1990, 4, 228.(77) Bencsura, A.; Navale, V.; Sadeghi, M.; Vertes, A. Rapid Com-

mun. Mass Spectrom. 1997, 11, 679.(78) Dutkiewicz, Ł.; Johnson, R. E.; Vertes, A.; Pedrys, R. J. Phys.

Chem. A 1999, 103, 2925.(79) Iwaki, L. K.; Deak, J. C.; Dlott, D. D. Chem. Phys. Lett. 1998,

293, 405.(80) Mashni, M.; Hess, P. Appl. Phys. B 1982 29, 205.(81) Antonov, V. S.; Lethokov, V. S.; Matveyets, Y. A.; Shibanov, A.

N. Laser Chem. 1982, 1, 37.(82) Chekalin, S. V.; Golovlev, V. V.; Kozlov, A. A.; Matveyets, Y. A.;

Yartsev, A. P.; Lethokov, V. S. Phys. Chem. 1988, 6855.(83) Demirev, P.; Westman, A.; Rimann, C. T.; Hakansson, P.;

Barofsky, D.; Sundqvist, B. U. R.; Chen, Y. D.; Seibt, W.;Siegbahn, K. Rapid Commun. Mass Spectrom. 1992, 6, 187.

(84) Dreisewerd, K.; Schurenberg, M.; Karas, M.; Hillenkamp, F. Int.J. Mass Spectrom. Ion Processes 1996, 154, 171.


(85) Lassithiotaki, M.; Athanassiou, A.; Anglos, D.; Georgiou, S.;Fotakis, C. Appl. Phys. A 1999, 69, 363.

(86) Athanassiou, A.; Andreou, E.; Anglos, D.; Georgiou, S.; Fotakis,C. Appl. Phys. A 1999, 86, 285.

(87) Arnolds, H.; Rehbein, C.; Roberts, G.; Levis, R. J.; King, D. A.J. Phys. Chem. B 2000, 104, 3375.

(88) Guo, H.; Saalfrank, P.; Seideman, T. Prog. Surf. Sci. 1999, 62,239.

(89) Niino, H.; Sato, T.; Yabe, A. Proc. SPIE 2001, 4274, 232.(90) Niino, H.; Sato, T.; Narazaki, A.; Kawaguchi, Y.; Yabe, A. Appl.

Surf. Sci. 2002, in press 7978.(91) Okoshi, M.; Higuchi, S.; Hanabusa, M. Appl. Surf. Sci. 2000,

154, 376.(92) Preuss, S.; Spath, M.; Zhang, Y.; Stuke, M. Appl. Phys. Lett.

1993, 62, 3049.(93) Baudach, S.; Kruger, J.; Kautek, W. Rev. Laser Eng. 2001, 29,

705.(94) Hosokawa, Y.; Yashiro, M.; Asahi, T.; Masuhara, H. J. Photo-

chem. Photobiol. A 2001, 142, 197.(95) Maechling, C. R.; Clemett, S. J.; Engelke, F.; Zare, R. N. J. Chem.

Phys. 1996, 104, 8768.(96) Handschuh, M.; Nettesheim, S.; Zenobi, R. J. Chem. Phys. 1997,

107, 2603.(97) Handschuh, M.; Nettesheim, S.; Zenobi, R. J. Chem. Phys. 1998,

108, 6548.(98) Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids; Oxford

University Press: Oxford, 1959.(99) Karaiskou, A.; Papazoglou, D.; Kafetzopoulos, D.; Zergioti, I.;

Fotakis, C. Appl. Surf. Sci., in press.(100) Ellegaard, O.; Schou, J. J. Appl. Phys. 1998, 83, 1078.(101) Jette, A. N.; Benson, R. C. J. Appl. Phys. 1994, 75, 3130.(102) Dou, Y.; Zhigilei, L. V.; Winograd, N.; Garrison, B. J. J. Phys.

Chem. A 2001, 105, 2748.(103) Dou, Y.; Zhigilei, L. V.; Postwa, Z.; Winograd, M.; Garrison, B.

J. Nucl. Instrum. Methods Phys. Res. B 2001, 180, 105.(104) Young, D.; Auyeung, R. C. Y.; Pique, A.; Chrisey, D. B.; Dlott,

D. D. Appl. Surf. Sci. 2002, 197-198, 181.(105) Young, D.; Auyeung, R. C. Y.; Pique, A.; Chrisey, D. B.; Dlott,

D. D. Appl. Phys. Lett. 2001, 78, 3169.(106) Nelson R. W.; Rainbow, M. J.; Lohr, D. E.; Williams, P. Science

1989, 246, 1585.(107) Williams, P. Int. J. Mass Spectrom. Ion Processes 1994, 131, 335.(108) Amster, I. J.; Speir, J. P. Anal. Chem. 1992, 64, 1041.(109) Koulikov, S. G.; Dlott, D. D. J. Photochem. Photobiol. A: Chem.

2001, 145, 183.(110) Herminghaus, S.; Leiderer, P. Appl. Phys. Lett. 1991, 58, 352.(111) Comsa, G.; David, R. Surf. Sci. Rep. 1985, 5, 145.(112) Campos, F. X.; Weaver, G. C.; Waltman, C. J.; Leone, S. R. J.

Vac. Sci. Technol. 1992, B10, 2217.(113) Campos, F. X.; Waltman, C. J.; Leone, S. R. Chem. Phys. Lett.

1993, 201, 399.(114) Kools, J. C. J.; Baller, T. S.; De Zwart, S. T.; Dieleman, J. J.

Appl. Phys. 1992, 71, 4547.(115) Elokhin, V. A.; Krutchinsky, A. N.; Ryabov, S. E. Chem. Phys.

Lett. 1990, 70, 193.(116) Zhang, J.-Y.; Nagra, D. S.; Li, L. Anal. Chem. 1993, 65, 2812.(117) de Vries M. S.; Hunziker H. E. J. Photochem. Photobiol. A:

Chem. 1997, 106, 31.(118) Heinrich, E. N.; Hunziker, E.; de Vries, M. S. Anal. Chem. 1999,

71, 1674.(119) Jones, A. C.; Dale, M. J.; Keenan, G. A.; Langridge-Smith, P. R.

R. Chem. Phys. Lett. 1994, 219, 174.(120) Kelly, R.; Dreyfus, R. W. Surf. Sci. 1988, 198, 263.(121) Kelly, R. J. Chem. Phys. 1990, 92, 5047.(122) Sibold, D.; Urbassek, H. M. Phys. Rev. A 1991, 43, 6722.(123) Sibold, D.; Urbassek, H. M. J. Appl. Phys. 1993, 73, 8544.(124) Anisimov, S. I.; Bauerle, D.; Luk’yanchuk, B. S. Phys. Rev. B

1993, 48, 12076.(125) Anisimov, S. I.; Luk’yanchuk, B. S.; Luches, A. Appl. Surf. Sci.

1996, 96-98, 24.(126) Kelly, R. Phys. Rev. A 1992, 46, 860.(127) Kelly, R.; Dreyfus, R. W. Nucl. Instrum. Methods B 1988, 32,

341.(128) Kelly R.; Miotello, A. Appl. Phys. B 1993, 57, 145.(129) Kelly, R.; Miotello, A. Nucl. Instrum. Methods B 1997, 122, 374.(130) NoorBatcha, I.; Lucchese, R. R. Phys. Rev. B 1987, 36, 4978.(131) NoorBatcha, I.; Lucchese, R. R.; Zeiri, Y. J. Chem. Phys. 1987,

86, 5816.(132) NoorBatcha, I.; Lucchese, R. R.; Zeiri, Y. J. Chem. Phys. 1988,

89, 5251.(133) Urbassek, H. M.; Sibold, D. Phys. Rev. Lett. 1993, 70, 1886.(134) Dreyfus, R. W.; Kelly, R.; Walkup, R. E. Appl. Phys. Lett. 1986,

49, 1478.(135) Itina, T. E.; Patrone, L.; Marine, W.; Autric, M. Appl. Phys. A

1999, 69, S59.(136) Vertes, A. Microbeam Anal. 1991, 1991, 25.(137) Vertes, A.; Irinyi, G.; Gibjels, R. Anal. Chem. 1993, 65, 2389.(138) Zhigilei, L. V.; Garrison, B. J. Appl. Phys. Lett. 1997, 71, 551.(139) Elam, J. W.; Levy, D. H. J. Chem. Phys. 1997, 106, 10368.

(140) Elam, J. W.; Levy, D. H. J. Phys. Chem. 1998, 102, 8113.(141) Georgiou, S.; Kosmidis, C.; Koubenakis, A.; Jia, W.; Ledinhgam,

K. W. D. Phys. Chem. Chem. Phys. 1999, 1, 5339.(142) Leboeuf, J. N.; Chen, K. R.; Liu, C. L.; Wood, R. F.; Geohegan,

D. B.; Puretzky, A. A.; Donato, J. M. Mater. Sci. Eng. B 1997,47, 70.

(143) Chen, K. R.; Leboeuf, J. N.; Wood, R. F.; Geohegan, D. B.; Donato,J. M.; Liu, C. L.; Puretzky, A. A. Phys. Rev. Lett. 1995, 75, 4706.

(144) Wood, R. F.; Chen, K. R.; Leboeuf, J. N.; Puretzky, A. A.;Geohegan, D. B. Phys. Rev. Lett. 1997, 79, 1571.

(145) Chen, K. R.; Leboeuf, J. N.; Wood, R. F.; Geohegan, D. B.; Donato,J. M.; Liu, C. L.; Puretzky, A. A. J. Vac. Sci. Technol. A 1996,14, 1111.

(146) Puretzky, A. A.; Geohegan, D. B.; Itina, T. E.; Marine, W.; Autric,M. Appl. Surf. Sci. 1998, 171, 127.

(147) Huth-Fehre, T.; Becker, C. H. Rapid Commun. Mass Spectrom.1991, 5, 378.

(148) Beavis, R. C.; Chait, B. T. Chem. Phys. Lett. 1991, 181, 479.(149) Westman, A.; Demirev, P.; Huth-Fehre, T.; Bielawski, J.; Sun-

dqvist, B. U. R. Int. J. Mass Spectrom. Ion Processes 1994, 130,107.

(150) Zhang, W.; Chait, B. T. Int. J. Mass Spectrom. Ion Processes1997, 160, 259.

(151) Fournier, I.; Beavis, R. C.; Blais, J. C.; Tabet, J. C.; Bolabch, G.Int. J. Mass Spectrom. Ion Processes 1997, 169/170, 19.

(152) Pan, Y.; Cotter, R. J. Org. Mass. Spectrom. 1992, 27, 3.(153) Puretzky, A. A.; Geohegan, D. B.; Hurst, G. B.; Buchanan, M.

V.; Luk’yanchuk, B. Phys. Rev. Lett. 1999, 83, 444.(154) Itina, T. E.; Tokarev, V. N.; Marine, W.; Autric, M. J. Chem.

Phys. 1997, 106, 8905.(155) Garrison, B. J.; Srinivasan, R. J. Appl. Phys. 1985, 57, 2909.(156) Yingling Y. G.; Zhigilei L. V.; Garrison B. J. Nucl. Instrum.

Methods B 2001, 180, 171.(157) Yingling Y. G.; Zhigilei L. V.; Garrison B. J. J. Photochem.

Photobiol. A 2001, 145, 173.(158) Sekine, S.; Ichimura, S.; Shimizu, H.; Hashizume, H. Jpn. J.

Appl. Phys. 1994, 33, L387.(159) Lippert, T.; Yabe, A.; Wokaun, A. Adv. Mater. 1997, 9, 105.(160) Weaver, G. C.; Leone, S. R. Surf. Sci. 1995, 328, 197.(161) Weaver, G. C.; Leone, S. R. J. Phys. Chem. 1996, 100, 4188.(162) Ventzek, P. L. G.; Suzuki, M.; Date, H.; Sakai, Y.; Tagashira,

H.; Kitamori, K. J. Appl. Phys. 1996, 80, 1146.(163) Kukreja, L. M.; Hess, P. Appl. Phys. Lett. 1993, 62, 105.(164) Georgiou, S.; Koubenakis, A.; Kontoleta, P.; Syrrou, M. Chem.

Phys. Lett. 1997, 270, 491.(165) Yingling, Y. G.; Garrison, B. J. Chem. Phys. Lett. 2002, 364, 237.(166) Nishiguchi, T.; Morikawa, Y.; Miyamoto, M.; Nonaka, H.;

Ichimura, S. Jpn. J. Appl. Phys. 2000, 39, L1200.(167) Georgiou, S.; Koubenakis, A.; Labrakis, J.; Lassithiotaki, M.

Appl. Surf. Sci. 1998, 127-129, 122.(168) Georgiou, S.; Koubenakis, A.; Kontoleta, P.; Syrrou, M. Chem.

Phys. Lett. 1996, 260, 166.(169) Niino, H.; Sato, T.; Yabe, A. Appl. Phys. A 1999, 69, 145.(170) Srinivasan, R.; Braren, B.; Casey, K. G. J. Appl. Phys. 1990,

68, 1842.(171) Dyer, P. E.; Karnakis, D. M. Appl. Phys. A 1994, 59, 275.(172) Koubenakis, A.; Labrakis, J.; Georgiou, S. J. Chem. Soc.,

Faraday Trans. 1998, 94, 3427.(173) Fournier, I.; Tabet, J. C.; Bolbach, G. Int. J. Mass Spectrom.

2002, 219, 515.(174) Krajnovich, D. J.; Vazquez, J. E. J. Appl. Phys. 1993, 73, 3001.(175) Hare, D. E.; Franken, J.; Dlott, D. D. J. Appl. Phys. 1995, 77,

5950.(176) Bubb, D. M.; Wu, P. K.; Horwitz, J. S.; Callahan, J. H.; Galicia,

M.; Vertes, A.; McGill, R. A.; Houser, E. J.; Ringeisen, B. R.;Chrisey, D. B. J. Appl. Phys. 2002, 91, 2055.

(177) Niino, H.; Sato, T.; Yabe, A. Appl. Phys. A 1999, 69, 605.(178) Athanassiou, A.; Andreou, E.; Bonarou, A.; Tornari, V.; Anglos,

D.; Georgiou, S.; Fotakis, C. Appl. Surf. Sci. 2002, 197-198, 757.(179) Athanassiou, A.; Andreou, E.; Fragouli, D.; Anglos, D.; Georgiou,

S.; Fotakis, C. J. Photochem. Photobiol. A: Chem. 2001, 145,229.

(180) Niino, H.; Koga, Y.; Yabe, A. J. Photochem. Photobiol. A: Chem.1997, 106, 9.

(181) Pimentel, G. C. Formation and Trapping of Free Radicals;Academic Press Inc.: New York, 1960.

(182) Song, Y.; Garder, P.; Conrad, H.; Bradshaw, A. M.; White, J. M.Surf. Sci. 1991, 248, L279.

(183) Siskin, M.; Katritzky, A. R.Chem. Rev. 2001, 101, 825.(184) Kawasaki, M.; Sato, H.; Nishi, N. J. Appl. Phys. 1989, 65, 792.(185) Venugopalan, V.; Nishioka, N. S.; Mikic, B. B. Trans. ASME J.

Biomech. Eng. 1994, 116, 62.(186) Venugopalan, V.; Nishioka, N. S.; Mikic, B. B. Biophys. J. 1995,

69, 1259.(187) Bityurin, N.; Arnold, N.; Luk’yanchuk, B.; Bauerle, D. Appl. Surf.

Sci. Phys. 1998, 127-129, 164.(188) Belov, M. E.; Alimpiev, S. S.; Mlynsky, V. V.; Nikiforov, S. M.;

Derrick, P. J. Rapid Commun. Mass Spectrom. 1995, 9, 1431.


(189) Dotter, R. N.; Smith, C. H.; Young, M. K.; Kelly, P. B.; Jones, A.D.; Mc Cauley, E. M.; Chang, D. P. Y Anal. Chem. 1996, 68,2319.

(190) Koubenakis, A.; Venturini, J.; Georgiou, S. Appl. Surf. Sci. 2002,197-198, 77.

(191) Claereboudt, J.; Claeys, M.; Geise, H.; Gijbels, R.; Vertes, A. J.Am. Soc. Mass Spectrom. 1993, 4, 798.

(192) Levis, J. Annu. Rev. Phys. Chem. 1994, 45, 4830.(193) Johnson, R. E. Int. J. Mass Spectrom. Ion Processes 1994, 139,

25.(194) Turro, N. J. Modern Molecular Photochemistry; University

Science Books: Mill Valley, CA, 1991.(195) Ehring, H.; Karas, M.; Hillenkamp, F. Org. Mass Spectrom. 1992,

27, 472.(196) Wen, X.; Tolbert, W. A.; Dlott, D. D. J. Chem. Phys. 1993, 99,

4140.(197) Root, R. G. Laser-Induced Plasmas and Applications; Marcel

Dekker: New York, 1989; p 69.(198) Krokhin, O. N. Generation of high-temperature vapors and

plasmas by laser radiation. In Laser Handbook; Arecchi, F. T.,Schulz-DuBois, E. O., Eds.; North-Holland Publishing Co.:Amsterdam, 1972; p 1371.

(199) Venugopalan, V.; Guerra, A.; Nahen, K.; Vogel, A. Phys. Rev.Lett. 2000, 88, 078103.

(200) Blander, M.; Katz, J. L. AIChE J. 1975, 21, 833.(201) Brennen, C. E. Cavitation and Bubble Dynamics; Oxford Uni-

versity Press: New York, 1995.(202) Tam, A. C.; Leung, W. P.; Zapka, W.; Ziemlich W. J. Appl. Phys.

1992, 71, 3515.(203) Leung, P. T.; Do, N.; Klees, L.; Leung, W. P.; Tong, F.; Lam, L.;

Zapka, W.; Tam, A. C. J. Appl. Phys. 1992, 72, 2256.(204) Do, N.; Klees, L.; Tam, A. C.; Leung, P. T.; Leung, W. P. J. Appl.

Phys. 1993, 74, 1534.(205) Yavas, O.; Leiderer, P.; Park, H. K.; Grigoropoulos, C. P.; Poon,

C. C.; Leung, W. P.; Do, N.; Tam, A. C. Phys. Rev. Lett. 1993,70, 1830.

(206) Park, H. K.; Grigoropoulos, C. P.; Poon, C. C.; Tam, A. C. Appl.Phys. Lett. 1996, 68, 596.

(207) Park, H. K.; Zang, X.; Grigoropoulos, C. P.; Poon, C. C.; Tam, A.C. ASME Trans. J. Heat Transfer 1995, 118, 702.

(208) Schilling, A.; Yavas, O.; Bischof, J.; Boneberg, J.; Leiderer, P.Appl. Phys. Lett. 1996, 69, 4159.

(209) Park, H. K.; Kim, D.; Grigoropoulos, C. P.; Tam, A. C. J. Appl.Phys. 1996, 80, 4072.

(210) Yavas, O.; Schilling, A.; Bischof, J.; Boneberg, J.; Leiderer, P.Appl. Phys. A 1997, 64, 331.

(211) Strauss, M.; Kaufman, Y.; Sapir, M.; Amendt, P. A.; London, R.A.; Clinsky, M. E. J. Appl. Phys. 2002, 91, 4720.

(212) Kim, D.; Park, H. K.; Grigoropoulos, C. P. Int. J. Heat MassTransfer 2001, 44, 3843.

(213) Yavas, O.; Leiderer, P.; Park, H. K.; Grigoropoulos, C. P.; Poon,C. C.; Tam, A. C. Phys. Rev. Lett. 1994, 72, 2021.

(214) Turovets, I.; Palanker, D.; Lewis, A. J. Photochem. Photobiol.1994, 60, 412.

(215) Lin, C. P.; Kelly, M. W. Appl. Phys. Lett. 1998, 72, 2800.(216) Lawson, C. M.; Euliss, G. W.; Michael, R. R. Appl. Phys. Lett.

1991, 58, 2195.(217) Chen, H.; Diebold, G. Science 1995, 270, 963.(218) Sunner, J.; Dratz, E.; Yu-Chie, C. Anal. Chem. 1995, 67, 4335.(219) Dale, M. J.; Knochenmuss, R.; Zenobi, R. Rapid Commun. Mass

Spectrom. 1997, 11, 136.(220) Schurenberg, M.; Dreisewed, K.; Hillenkamp, F. Anal. Chem.

1999, 71, 221.(221) McEwan, K. J.; Madden, P. A. J. Chem. Phys. 1992, 97, 8748.(222) Lowen, H.; Madden, P. A. J. Chem. Phys. 1992, 97, 8760.(223) Sun, J. M.; Gerstman, B. S.; Li, B. J. Appl. Phys. 2000, 88, 2352.(224) Zhigilei, L. V.; Garisson, B. J. Appl. Surf. Sci. 1998, 127, 142.(225) Gusev, V. E.; Karabutov, A. A. Laser Optoacoustics; American

Institute of Physics: New York, 1993; p 271.(226) Jansen, E. D.; van Leeuwen, T. G.; Motamedi, M.; Borst, C.,

Welch, A. J. J. Appl. Phys. 1995, 78, 564.(227) Oraevsky, A. A.; Jacques, S. L.; Tittel, F. K. J. Appl. Phys. 1995,

78, 1281(228) Oraevsky, A. A.; Jacques, S. L.; Esenaliev, R. O.; Tittel, F. K.

Lasers Surg. Med. 1996, 18, 231.(229) Esenaliev, R. O.; Karabutov, A. A.; Podymova, N. B.; Letokhov,

V. S. Appl. Phys. B 1994, 59, 73.(230) Thompson, C. R.; Gerstman, B. S.; Jacques, S. L.; Rogers, M. E.

Bull. Math. Biol. 1996, 58, 513.(231) Paltauf, G.; Schmidt-Kloiber, H. Appl. Phys. A 1996, 62, 303.(232) Paltauf, G.; Schmidt-Kloiber, H. J. Appl. Phys. 1997, 82, 1525.(233) Kim, D.; Grigoropoulos, C. P. Appl. Surf. Sci. 1998, 127-129,

53.(234) Kim, D.; Ye, M.; Grigoropoulos, C. P. Appl. Phys. A 1998, 67,

169.(235) Siano, S.; Pini, R.; Salimbeni, R.; Vannini, M. Appl. Phys. B 1996,

62, 503.(236) Frenz, M.; Konz, F.; Pratisto, H.; Weber, H. P.; Silenok, A. S.;

Konov, V. I. J. Appl. Phys. 1998, 84, 5905.

(237) Staveteig P. T.; Walsh J. T. Appl. Opt. 1996, 35, 3392.(238) Nikiforov, S. M.; Alimpiev, S. S.; George, M. W.; Sartakov, B.

G.; Simanovsky, Y. O. Opt. Commun. 2000, 182, 17.(239) Hale, G. M.; Querry, M. R.; Rusk, A. N.; Williams D. J. Opt.

Soc. Am. 1972, 62, 1103.(240) Longtin, P.; Tien C.-L. Int. J. Heat Mass Transfer 1997, 40, 951.(241) Srinivasan, R.; Ghosh, A. P. Chem. Phys. Lett. 1988, 143, 546.(242) Katoh, R.; Yokoi, H.; Usuba, S.; Kakudate, Y.; Fujiwara, S. Chem.

Phys. Lett. 1998, 291, 305.(243) Tsuboi, Y.; Hatanaka, K.; Fukumura, H.; Masuhara, H. J. Phys.

Chem. 1994, 98, 11237.(244) Tsuboi, Y.; Fukumura, H.; Masuhara, H. Appl. Phys. Lett. 1994,

64, 2745.(245) Hatanaka, K.; Kawao, N.; Tsuboi, Y.; Fukumura, H.; Masuhara,

H. J. Appl. Phys. 1997, 82, 5799.(246) Tsuboi, Y.; Hatanaka, K.; Fukumura, H.; Masuhara, H. J. Phys.

Chem. A 1998, 102, 1661.(247) Hirai, A.; Tsuboi, Y.; Itoh, K.; Ichioka, Y. Rev. Sci. Instrum. 1996,

67, 3222.(248) Tsuboi, Y.; Fukumura, H.; Masuhara, H. J. Phys. Chem. 1995,

99, 10305.(249) Hatanaka, K.; Itoh, T.; Asahi, T.; Ichinose, N.; Kawanishi, S.;

Sasuga, T.; Fukumura, H.; Masuhara, H. Chem. Phys. Lett.1999, 300, 727.

(250) Hatanaka, K.; Itoh, T.; Asahi, T.; Ichinose, N.; Kawanishi, S.;Sasuga, T.; Fukumura, H.; Masuhara, H. Appl. Phys. Lett. 1998,73, 3498.

(251) Hatanaka, K.; Itoh, T.; Asabi, T.; Ichinose, N.; Kawanishi, S.;Sasuga, T.; Fukumura, H.; Masuhara, H. J. Phys. Chem. A 1999,103, 11257.

(252) Hatanaka, K.; Tsuboi, Y.; Fukumura, H.; Masuhara, H. J. Phys.Chem. A 2002, 106, 3049.

(253) Toyota, K.; Nakashima, S.; Okada, T. Chem. Phys. Lett. 2000,323, 323.

(254) Toyota, K.; Tanaka, T.; Nishiwaki, S.; Nakashima, S.; Okada,T. J. Photochem. Photobiol. A: Chem. 2001, 141, 9.

(255) Hopp, B.; Smausz, T.; Wittmann, T.; Ignacz, F. Appl. Phys. A2000, 71, 315.

(256) Berkenkamp, S.; Karas, M.; Hillenkamp, F. Proc. Natl. Acad.Sci. U.S.A. 1996, 93, 7003.

(257) Schieltz, D. M.; Chou, C. W.; Luo, C. W.; Thomas, R. M.;Williams, P. Rapid Comm. in Mass Spectrom. 1992, 6, 631.

(258) Sheffer, J. D.; Murray, K. K. J. Mass Spectrom. 2000, 35, 95.(259) Livingston, F. E.; Smith, J. A.; George, S. M. Anal. Chem. 2000,

72, 5596.(260) Livingston, F. E.; George, S. M. J. Phys. Chem. A 2001, 105,

5155.(261) Levis, R. J.; Waltman, C. J.; Cousins, L. M.; Copeland, R. G.;

Leone, S. R. J. Vac. Sci. Technol. 1990, A8, 3118.(262) Harrison, I.; Polanyi, J. C. Z. Phys. D: Atoms, Molecules Clusters

1988, 10, 383.(263) Wang, J.; Niino, H.; Yabe, A. Appl. Phys. A 1999, 68, 111.(264) Wang, J.; Niino, H.; Yabe, A. Appl. Phys. A 1999, 69, S271.(265) Drinek, V.; Niino, H.; Pola, J.; Yabe, A. Appl. Phys. A 2001, 73,

527.(266) Yasui, Y.; Niino, H.; Kawaguchi, Y.; Yabe, A. Appl. Surf. Sci.

2002, 186, 552.(267) Wang, J.; Niino, H.; Yabe, A. Jpn. J. Appl. Phys. 1999, 38, L761.(268) Wang, J.; Niino, H.; Yabe, A. Appl. Surf. Sci. 2000, 154-155,

571.(269) Kim, D.; Lee, H. J. Appl. Phys. 2001, 89, 5703.(270) Hanabusa, M.; Tsujihara, K. J. Appl. Phys. 1995, 78, 4267.(271) Okoshi, M.; Higuchi, S.; Hanabusa, M. J. Appl. Phys. 1999, 86,

1768.(272) Okoshi, M.; Higuchi, S.; Hanabusa, M. Jpn. J. Appl. Phys. 2001,

40, 4926.(273) Hiroshima, Y.; Ishiguro, T.; Esaki, K. Jpn. J. Appl. Phys. 1997,

36, 798.(274) Hiroshima, Y.; Ishiguro, T.; Esaki, K. Jpn. J. Appl. Phys. 1997,

36, 2267.(275) Ishiguro, T.; Esaki, K.; Hiroshima, Y. Appl. Surf. Sci. 1998, 127-

129, 553.(276) Ishiguro, T.; Shoji, T.; Inada, H. Appl. Phys. 1999, A69, S149.(277) Allwood, D. A.; Dyer, P. E.; Gonzalo, J.; Snelling, H. V.; Hird,

M. Chem. Phys. Lett. 1999, 301, 91.(278) Farrar, S. R.; Contoret, A. E. A.; O’Neil, M.; Nicholls, J. E.;

Eastwood, A. J.; Richards, G. J.; Vlachos, P.; Helly, S. M. Appl.Surf. Sci. 2002, 186, 435.

(279) Nishiguchi, T.; Morikawa, Y.; Miyamoto, M.; Nonaka, H.;Ichimura, S. Appl. Phys. Lett. 2001, 79, 382.

(280) Nishiguchi, T.; Morikawa, Y.; Miyamoto, M.; Nonaka, H.;Ichimura, S. Jpn. J. Appl. Phys. 2001, 40, L897.

(281) Niino, H.; Yabe, A. Appl. Surf. Sci. 1996, 96-98, 572.(282) Simakin, V.; Voronov, V. V.; Shafeev, G. A.; Brayner, R.; Bozon-

Verduraz, F. Chem. Phys. Lett. 2001, 348, 182.(283) Bohme, R.; Braun, A.; Zimmer, K. Appl. Surf. Sci. 2001, 186,

276.


(284) Simakin, A. V.; Shafeev, G. A.; Loubnin E. N. Appl. Surf. Sci.2000, 154-155, 405.

(285) Salih, A. J.; Lau, S. P.; Marshall, J. M.; Maud, J. M.; Bowen, W.R.; Hilal, N.; Lovitt, R. W.; Williams, P. M. Appl. Phys. Lett.1996, 69, 2231.

(286) Rong, F.-X. Appl. Phys. Lett. 1995, 67, 1022.(287) Arnold, N. Appl. Surf. Sci. 2002, 197-198, 904.(288) Mosbacher, M.; Dobler, V.; Boneberg, J.; Leiderer, P. Appl. Phys.

A 2000, 70, 669.(289) Tam, A. C.; Park, H. K.; Grigoropoulos, C. P. Appl. Surf. Sci.

1998, 127-129, 721.(290) She, M.; Kim, D.; Grigoropoulos, C. P. J. Appl. Phys. 1999, 86,

6519.(291) Lu, Y. F.; Zhang, Y.; Wan, Y. H.; Song, W. D. Appl. Surf. Sci.

1999, 138-139, 140.(292) Wu, X.; Sacher, E.; Meunier, M. J. Appl. Phys. 2000, 87, 3618.(293) Mosbacher, M.; Chaoui, N.; Siegel, J.; Dobler, V.; Solis, J.;

Boneberg, J.; Afonso, C. N.; Leiderer, P. Appl. Phys. A 1999, 69,331.

(294) Pasquet, P.; del Coso, R.; Boneberg, J.; Leiderer, P.; Oltra, R.;Boquillon, J. P. Appl. Phys. A 1999, 69, S727.

(295) Heroux, J. B.; Boughaba, S.; Ressejak, I.; Meunier, M. J. Appl.Phys. 1996, 79, 2857.

(296) Lee, Y. P.; Lu, Y. F.; Chan, D. S. H.; Low, T. S.; Zhou, M. S.Jpn. J. Appl. Phys. 1998, 37, 2524.

(297) Mafune, F.; Kohno, J.-y.; Takeda, Y.; Kondow, T.; Sawabe, H.J. Phys. Chem. B 2000, 104, 9111.

(298) Mafune, F.; Kohno, J.-y.; Takeda, Y.; Kondow, T.; Sawabe, H.J. Phys. Chem. B 2001, 105, 5114.

(299) Prochazka, M.; Vlckova, B.; Stepanek, J.; Srnova, I.; Mal’y, P.J. Mol. Struct. 1997, 410-411, 213.

(300) Srnova, I.; Prochazka, M.; Vlckova, B.; Stepanek, J.; Mal’y, P.Langmuir 1998, 14, 4666.

(301) Sibbald, M. S.; Chumanov, G.; Cotton, T. M. J. Phys. Chem. 1996,100, 4672.

(302) Yeh, M.-S.; Yang, Y.-S.; Lee, Y.-P.; Lee, H.-F.; Yeh, Y.-H.; Yeh,C.-S. J. Phys. Chem. B 1999, 103, 6851.

(303) Tamaki, Y.; Asahi, T.; Masuhara, H. Appl. Surf. Sci. 2000, 168,85.

(304) Tsuji, T.; Iryo, K.; Watanabe, N.; Tsuji, M. Appl. Surf. Sci. 2002,202, 80.

(305) Bae, C. H.; Nam, S. H.; Park, S. M. Appl. Surf. Sci. 2002, 197-198, 628.

CR010429O


laser-induced material ejection from model molecular solids and liquids: mechanisms, implications,...

Documents