Experimental and theoretical study of the pyrrole cluster photochemistry: Closing the πσ[sup

Experimental and theoretical study of the pyrrole cluster photochemistry:Closing the ��* dissociation pathway by complexation

Viktoriya Poterya, Václav Profant, and Michal Fárníka�

J. Heyrovský Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Prague 8,Czech Republic

Petr Slavíčekb�

Department of Physical Chemistry, Institute of Chemical Technology, Technická 5, Prague 6,Czech Republic

Udo BuckMax-Planck Institut für Dynamik und Selbstorganization, Bunsenstrasse 10, D-37073 Göttingen, Germany

�Received 23 May 2007; accepted 12 June 2007; published online 13 August 2007�

Photolysis of size selected pyrrole clusters has been investigated and compared to the photolysis ofan isolated pyrrole molecule. Experimentally, size distributions of different mean cluster sizes�n̄=3 and n̄�5� have been prepared in supersonic expansions and the clusters were photolyzed at243 and 193 nm. The kinetic energy distributions of the H photofragments have been measured. Thedistributions exhibit a bimodal character with fast and slow H-fragment peaks similar to the spectraof the bare molecule. However, with increasing cluster size the slow component gains intensity withrespect to the fast one. A similar effect is observed with increasing the excitation energy from243 to 193 nm. Theoretical calculations at the CASSCF/CASPT2 level have been performed forbare and complexed pyrroles �pyrrole is complexed with an argon atom and with another pyrroleunit�. Combination of theoretical and experimental approaches leads to the conclusion that the directdissociative pathway along the ��* potential energy surface in the N–H stretch coordinate is closedby the presence of the solvent molecule. This pathway is an important channel leading to the fast Hatoms in the dissociation of the bare molecule. The solvent molecule influences significantly theelectronic structure in the Rydberg-type ��* state while it has little influence on the valence states.The slow channel is mostly populated by the out-of-plane deformation mode which is also notinfluenced by solvation. We have also studied other possible reaction channels in pyrrole clusters�hydrogen transfer, dimerization�. The present study shows that more insight into the bulk behaviorof biologically relevant molecules can be gained from cluster studies. © 2007 American Institute ofPhysics. �DOI: 10.1063/1.2754687�

I. INTRODUCTION

The pyrrole molecule C4H5N �Py� is one of the simplestheteroaromatic units with biological relevance. Pyrrole unitsare present in many larger biological structures, e.g., hemesor chlorophylls, and it is also structurally close to the basesof nucleic acids. A low luminescence quantum yield ex-plained by rapid internal conversion is typical for thesemolecules.1 This has an important biological consequencesince the photon energy dissipates and the molecule is notsusceptible to the potentially harmful reactions in the excitedstates.

For this reason pyrrole has attracted the attention of boththeoreticians and experimentalists, addressing the questionsof UV absorption spectrum interpretation 2–9 and the dynam-ics after the excitation.10–24 The characterization of the pyr-role absorption spectrum is still somewhat controversial con-

cerning the quantitative agreement between different abinitio methods. However, there is a general agreement thatthe spectrum starts with the S0→3s transition at about 5 eV.The population of this state is allowed by vibronic coupling.The two ��* states are expected to be about 1–1.5 eV abovethe 3s Rydberg state.

Despite the long lasting effort, the precise photodynam-ics of pyrrole molecules is still not completely understood.Successful mechanisms should account for a wide variety ofexperimental observations. In one of the earlier experimentsthe fragment distribution after photoexcitation and photo-fragment translational spectroscopy was studied.10 More re-cently Wei et al.11,12 employed photofragment velocity mapimaging to study the H-atom elimination mechanism at243 nm excitation. The results have recently been confirmedin a similar independent experiment.25 Ashfold andco-workers15,16 obtained photofragment translational energyspectra with vibrational state resolution using H-atom Ryd-berg tagging after photodissociation at various wavelengthsbetween 193 and 254 nm. In the same group in a differentexperiment, photoelectron imaging and time-of-flight massspectrometry were used to study multiphoton ionization and

a�Author to whom correspondence should be addressed. Electronic mail:[email protected]

b�Also at J. Heyrovský Institute of Physical Chemistry, Academy of Sciencesof the Czech Republic, Prague 8, Czech Republic. Electronic mail: [email protected]

THE JOURNAL OF CHEMICAL PHYSICS 127, 064307 �2007�

0021-9606/2007/127�6�/064307/12/$23.00 © 2007 American Institute of Physics127, 064307-1

Downloaded 21 Aug 2007 to 147.33.103.173. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp

http://dx.doi.org/10.1063/1.2754687

http://dx.doi.org/10.1063/1.2754687

http://dx.doi.org/10.1063/1.2754687

dissociation.14 Finally, time-resolved pump-probe femtosec-ond experiments have been performed.13 The major photo-fragment for wavelengths between 193 and 243 nm is thehydrogen atom. These hydrogen atoms are observed in twopeaks, one with a slow component and one with a fast com-ponent. At low excitation energies the fast component isdominant, while with increasing photon energy the slow hy-drogen component becomes more important.26 Experimentswith deuterated pyrrole indicate that at low excitation ener-gies, the N–H dissociation is strongly preferred, while athigher excitation energies, both the N–H and C–H dissocia-tions take place �especially for the slow hydrogen compo-nent�. Apart from hydrogen fragments, other molecular spe-cies are also formed in significant quantities. In particular,these are HCN for higher photon energies and CNH2 forlower photon energies.

All the above experiments can be interpreted in the fol-lowing way. At low photon energies, the system is promotedinto the ��* state. This state is of 3s Rydberg character andafter overcoming a small barrier it is repulsive with respectto the N–H coordinate and leads therefore to the productionof fast hydrogen atoms. Here we assume that the S0 groundstate is crossed diabatically. The role of the ��* state hasbeen pointed out by Sobolewski et al.17,18 based on the abinitio calculations. Wave packet calculations of the pyrrolephotodissociation on the ��* state have also beenperformed,19,20 showing the influence of the excitation wave-length and specific mode promotion on reaction products.Note that the excitation into the 1�-3s state is possible due tothe vibronic coupling with the ��* state.14

Several scenarios have been suggested to explain theslow hydrogen component. Lippert et al.13 have proposedthat the slow hydrogen atoms are produced after internalconversion from the ��* state to the ground state and sub-sequent energy randomization. Vallet et al.19 interpreted theslow hydrogen atoms as a result of adiabatic dissociation onthe S1 state, and Frank and Damianos24 suggested that theslow hydrogen atoms result from the relaxation in the excitedstate. All these processes might take place. It has, however,been suggested by Wei et al.12 and Cronin et al.15 and re-cently confirmed by theoretical calculations of Barbatti etal.23 that another reaction channel, connecting directly the��* and the ground state, exists. This channel, which ismainly important when exciting to the ��* state, competeswith the N–H stretching coordinate. The pyrrole photody-namics as presented by Barbatti et al. is sketched in theupper panel of Fig. 12 �see also Fig. 5 in Ref. 23�. The newreaction channel corresponds to the out-of-plane deformationof the pyrrole ring. This coordinate connects the ��* statedirectly to the ground S0 state. This internal conversion thenleads to the production of slow hydrogen atoms after theenergy is randomized in the S0 state. The out-of-plane modehas yet another significance: it connects the ��* state withthe ��* state. The ��* state can be thus populated evenwhen exciting the ��* state and vice versa. After excitationinto the ��* state the pyrrole molecule will most likely dia-batically proceed towards the intersection with the groundstate but some part of the wave packet can be trapped intothe ��* state, giving rise to the fast hydrogen atom compo-

nent. On the other hand, excitation into the ��* state can befollowed not only by direct dissociation via a small barrierbut also via transfer to the ��* state, leading to the deexci-tation without hydrogen abstraction. The ring deformationmode also connects the ��* state with a second ��* state.

All the studies quoted above deal with the isolated pyr-role molecule. Photochemical processes in nature, however,do not occur in the gas phase but in some environment, e.g.,in a solvent or in a geometry dictated by a phosphate back-bone of evolutionary selected proteins. The topography ofthe excited state potential energy surface can be significantlyaltered by the presence of a solvent or another environment�e.g., protein�.28,29 While in the ground state, the electronicstructure is influenced dominantly by long-range interactions�electrostatics�, short-range repulsion may deeply influencethe often delocalized excited state wave function.30 This isespecially true for states of Rydberg character. Since Ryd-berg states play an important role in the photochemistry ofpyrrole, profound effects of nonspecific complexation can beexpected. Solvent molecules can also react with the excitedsolute. In the case of pyrrole, hydrogen transfer to the sol-vating ammonia molecule31 or charge transfer to the solvat-ing water cluster32 was observed. These processes could alsooccur for clusters of aromatic molecules if proper hydrogenbonded arrangements are achieved, e.g., for aminopyridineclusters.33 Two aromatic units can also form an excimer.34

Our experiment and the accompanying theoretical treat-ment have been successfully applied to a series of differentsystems previously to study the influence of a molecule sol-vation on its photodissociation in a cluster environment.35

Effects of a rare gas cluster cage on hydrogen halide disso-ciation have been investigated.36–40 Here �i� the solute elec-tronic structure is not at all changed and �ii� the photochemi-cal product does not react with the solvent. Effects of thecomplexation in hydrogen bonded �HBr�n clusters have alsobeen studied.41,42 Here already interactions with the neighbormolecules were observed. Most recently the photolysis ofhydrogen halide molecule in large water clusters has beenexplored.43,44 Here the hydrogen definitely reacts with thesolvent.

In this contribution, we have combined the experimentaland theoretical approaches to study pyrrole clusters. On theexperimental side, pyrrole clusters of various �defined� sizeshave been prepared and subsequently photodissociated. Themean size n̄ of the cluster size distribution produced in asupersonic expansion has been determined in the scatteringexperiment with He atomic beam published independently.45

The results have been interpreted using high level ab initiocalculations. Pyrrole molecules act here as solvent. They in-fluence the electronic structure of the excited pyrrole unit byboth electrostatic interaction and Pauli repulsion. Under cer-tain conditions, the solvating pyrrole molecule can also reactwith the excited pyrrole. As we will argue later, the reactionchannel is determined by the ground state structure of pyr-role clusters, characterized by the N–H¯� arrangement.

In the following section we will describe the experimen-tal and theoretical methods. Then the results of the photolysisof pyrrole clusters of mean sizes n̄�3 and n̄�5 are pre-sented and compared to the photolysis of bare pyrrole mol-

064307-2 Poterya et al. J. Chem. Phys. 127, 064307 �2007�


ecule. Results obtained at the photodissociation wavelengthsof 243 and 193 nm are compared. Results of the theoreticalcalculations of pyrrole clustered with Ar atoms and with an-other pyrrole molecule are presented in the next section andthe experimental results are discussed in terms of these cal-culations. Finally, conclusions are drawn about the photodis-sociation dynamics of pyrrole in cluster environments.

II. EXPERIMENTAL AND THEORETICAL METHODS

A. Experimental apparatus

The schematic picture of the experimental apparatus isshown in Fig. 1. It was previously used at the Max-PlanckInstitute in Göttingen for both the scattering experiments46

and the photodissociation of molecules in clusters.35 Re-cently, the equipment has been newly installed at the J. Hey-rovský Institute of Physical Chemistry in Prague. The detailsof the experiment can be found in earlier publications fromGöttingen.35,42

The cluster beam was produced by a supersonic expan-sion of pyrrole vapor seeded in either He or Ar buffer gasthrough a conical nozzle of 60 �m diameter, 2 mm length,and 30° opening angle. The pyrrole �98%, Aldrich� was filledinto a reservoir outside the vacuum chamber, which was keptin a thermal bath at a constant temperature TR. He or Ar gascarried the pyrrole molecules to the nozzle kept at a tempera-ture T0. The expansion conditions are summarized in Table I.In a scattering experiment45 with the option to select andanalyze the neutral cluster sizes, we have explored the clustersize distributions for various expansion conditions. The He-expansion conditions used in the present experiment corre-spond to the �Py�n clusters of mean size n̄=3, while muchlarger mixed Py-Ar clusters n̄�5 are produced in the Arexpansion.47 For comparison the photodissociation experi-ment has also been performed under the conditions corre-sponding to the bare Py molecules in the beam, similar to theones found in the literature.11,12

After passing through a skimmer followed by two differ-entially pumped vacuum chambers, the clusters entered a

chamber hosting a transversally mounted time-of-flight as-sembly of the Willey-McLaren type �WMTOF�. Here theclusters were photolyzed, the H-atom photofragments wereionized with laser pulses, and the protons were detected. Thespectrometer was used to analyze the velocity distribution ofH atoms arising from the photodissociation of the moleculeson the clusters. For this purpose it was operated in the low-field mode with a small electric field of 3.5 V cm−1 appliedto extract the ions to the detector. With the small extractionfield, ions with initially zero velocity could be distinguishedfrom ions with a velocity vector pointing towards the detec-tor or in the opposite direction, and their velocities could beanalyzed. To suppress the background signal from dissoci-ated rest gas H atoms, the WMTOF was surrounded by acopper shield kept at a low temperature of 20 K by a highpressure helium compressor, which maintained a backgroundvacuum of approximately 2�10−9 Torr.

The tunable pulsed UV laser system was operated at awavelength of 243.07 nm, at which the H atoms were effi-ciently ionized by a �2+1� resonantly enhanced multiphotonionization �REMPI� process. This wavelength was generatedby mixing the fundamental 1064 nm of a neodymium dopedyttrium aluminum garnet �Nd:YAG� laser �Quanta RayGCR-5� with the frequency doubled output of a dye laser�LAS, LDL 20505� operated at 630.3 nm and pumped by thesecond harmonic �532 nm� of the Nd:YAG laser. The laserpulses were repeated at a frequency of 10 Hz with a pulseduration of 5 ns. The laser beam was focused into thevacuum chamber by a quartz lens onto the intersection pointof the cluster beam and the WMTOF axis. The polarizationof the linearly polarized laser light could be turned parallel�0°� or perpendicular �90°� with respect to the WMTOF axisby passing the laser beam through a series of prisms.

In the first experiment the pyrrole molecules were pho-tolyzed with the 243.07 nm laser pulse and the nascent Hfragments were ionized within the same laser pulse. In theother experiment the molecules were photolyzed with193 nm pulse synchronized with the following 243.07 nmpulse to ionize the H fragments. The 193 nm radiation wasproduced by an ArF/F2 excimer laser �Lambda: Compex102�. The repetition frequency of the laser pulses was 10 Hzwith a pulse length of 20 ns. The emitted light was focusedinto the chamber by a LiF lens onto the intersection point ofthe cluster beam and the WMTOF axis. The emitted lightcould be polarized using a thin film polarizer �TFP�. Therotation of TFP about the laser beam axis allowed to changethe laser light polarization from 0° to 90° polarization anglewith respect to the WMTOF axis.

The 193 nm laser beam, the cluster beam, and theWMTOF axis were mutually perpendicular. The 243 nm la-ser beam was introduced into the detector chamber at 17.5°angle to the dissociation beam, in the plane of the molecularbeam and the 193 nm laser beam �see Fig. 1�. The time syn-chronization of the two laser beams was achieved by trigger-ing the excimer laser by the Nd:YAG laser pulses via a pulsedelay generator. The dissociation pulses preceded the ioniza-tion by 5–20 ns. The time delay was adjusted on the inten-sity and the symmetric shape of time-of-flight �TOF� spectrameasured with �HBr�n system well studied with this experi-

FIG. 1. Schematic view of the cluster photodissociation experiment.

TABLE I. Experimental conditions.

Buffer gas He He Ar

Expansion pressure p0 �bar� 1.0 1.5 3.0Nozzle temperature T0 �°C� 30 9 9Pyrrole reservoir temperature TR �°C� −20 8 8Pyrrole concentration �%� 0.065 0.33 0.16Average cluster size n̄ 1 3 �5

064307-3 Pyrrole cluster photochemistry J. Chem. Phys. 127, 064307 �2007�


mental setup previously.42,48 The same procedure was used toachieve the spatial overlap of the two laser beams with theintersection point of the cluster beam and the WMTOF axis.

In the course of the present experiment the TOF �veloc-ity� distributions of the H-atom fragments after photodisso-ciation were measured by the WMTOF spectrometer. To ex-tract dynamical information about the photodissociationprocess, these spectra were converted to H-fragment kineticenergy distributions �KEDs�. The experimental points werefitted with a complete Monte Carlo �MC� simulation of theparticle trajectories in the photodissociation process, whichwas carried out considering the molecular beam data, thephotodissociation process parameters, the WMTOF geom-etry, the finite interaction volume, and the detector electronicresponse.48,49

B. Electronic structure calculations

The electronic structure of the isolated pyrrole moleculehas been studied by different groups, with interest either inthe absorption spectrum2–9 or in the photodynamics,17–24 us-ing various methods ranging from time-dependent densityfunctional theory to multireference configuration interaction�MRCI� level of description. Here we address the question ofenvironmental effects on the electronic structure and photo-dynamics of pyrrole molecule. Guided by the mechanismsuggested by Barbatti et al.,23 we have explored how thecluster environment influences the vertical excitation ener-gies, the ��* /S0 conical intersection, the ��* /S0 conicalintersection, and the barriers separating these intersectionswith the Franck-Condon point. We have chosen argon atomsas a structureless archetype of a solvent. To account for moresubtle solvent effects as well as for exploration of possiblereactivity of pyrrole molecules in the excited states, we havealso studied the pyrrole dimer.

The isolated pyrrole molecule has been optimized at theB3LYP/6–31g* level in the ground state. Five electronicstates of the pyrrole molecule have been investigated: theground state, two Rydberg states corresponding asymptoti-cally with �N–H� ��* states, and two ��* states. To describethese states we have employed a 6–31g* basis augmentedwith diffuse functions on nitrogen and on the hydrogenbound to a nitrogen atom. With this basis, complete activespace self-consistent field �CAS-SCF� calculation for eightactive electrons in seven orbitals averaged over five states ofinterest has been performed. To include dynamical correla-tion, the CASPT2 method has been then applied. This ap-proach provides results consistent with higher level MRCIcalculations of Barbatti et al.23 The above combination willbe referred to as CASPT2/SA-5-CAS�8/7� /6–31g*

+NHdif. The conical intersection between the ��* and S0

states has been obtained by prolonging the NH bond withoutfurther optimization. The conical intersection between the��* and S0 states has been optimized at theSA-3-CAS�6/6� /6–31g* level. Notice that the geometry ofthe latter intersection is robust with respect to the basis setand active space used. Diffuse functions have been omittedhere since they are not relevant for the crossing betweenthese two states. The reaction path connecting the Franck-

Condon point and the ��* /S0 crossing is simply the N–Hcoordinate, the reaction path connecting the Franck-Condonpoint and the ��* /S0 crossing has been approximated as alinear interpolation of internal coordinates.

The geometry of the pyrrole-argon complex has not beenoptimized. Instead, argon atom has been added to a pyrroleoptimized at the B3LYP/6–31g* level. The argon atom hasbeen placed in the pyrrole plane, 3.5 Å from the nitrogenatom in the NH axis. This is approximately the equilibriumdistance found in the pyrrole-argon complex.45 Note, how-ever, that this geometry does not correspond to a global mini-mum of this complex. The N–H¯Ar arrangement is chosento mimic the N–H¯� motive found in pyrrole dimer. Theinfluence of an argon atom in a fixed position on reactionpaths connecting the Franck-Condon point and ��* /S0 and��* /S0 intersections has been studied. Potential energycurves have been investigated at the CASPT2/SA-5-CAS�8/7� /6–31g*+NHdif level.

Finally, we have studied the photodynamical behavior ofthe pyrrole dimer. The structure of pyrrole clusters in theground state has recently been addressed by several groups,e.g., Ref. 50. Here we have optimized the dimer geometry atthe B3LYP/6–31g* level. For a pyrrole dimer, we have pri-marily addressed the question how the second �solvent� pyr-role molecule will influence the ��* /S0 intersections. Dif-fuse functions therefore were not included. Conicalintersections have been localized with the SA-3-CAS�6/5�method and recalculated at the SA-5-CAS�12/12� /6–31g*

level. This level of theory has also been used to calculate thereaction path between the Franck-Condon point and ��* /S0

intersections. The same approach has been used when thereactivity of the ��* system has been explored. To investi-gate possible hydrogen transfer reactions between two pyr-role molecules and hydrogen dissociation, diffuse functionshave been added to the corresponding nitrogen/hydrogen at-oms. All the calculations have been performed using theMOLPRO2006 program suite.51

III. RESULTS AND DISCUSSION

A. Experimental results

The size distributions of Py clusters produced undervarious expansion conditions were studied in a scatteringexperiment with He atoms. The obtained mean neutral clus-ter sizes are summarized in Table. I. Presentation of thewhole scattering experiment and its complicated analysis in-cluding the interesting results about the pyrrole cluster ion-ization and fragmentation upon the electron impact goes be-yond the scope of the present paper, which concentrates onphotodissociation of Py clusters and its dependence on clus-ter size. Therefore the scattering experiment is presented in aseparate publication.45 Here only the cluster mass spectraobtained with the quadrupole mass spectrometer at the end ofthe apparatus after the electron impact ionization are shownfor illustration in Fig. 2. Figure 2�a� shows the mass spec-trum of clusters produced in He expansions at p0=1.5 bar,T0=9 °C, and 0.33% concentration of Py in He. The meancluster size under these conditions was obtained from thecomplete scattering experiment to be n̄=3. Figure 2�b� shows



the mass spectrum of clusters produced in Ar expansions atp0=3.0 bar, T0=9 °C, and 0.16% concentration of Py in Ar,where the mean cluster size could not be determined exactlybut it is n̄�5 �see Ref. 45 for details�.

Figure 3 shows the H-fragment TOF distributions afterphotolysis of pyrrole clusters measured in He expansions forn̄=3 �a� and for n̄�5 �b�. The upper and lower spectra cor-respond to the laser polarizations of 0° and 90°, respectively.The solid lines show the MC simulation fits of the experi-mental points �circles�. These simulations were used to con-vert the measured TOF spectra into the KED of the H frag-ments shown in Figs. 4 and 5.

The spectra exhibit the bimodal character observed pre-

viously in photodissociation of pyrrole molecules, with anarrower intense peak of fast photofragments and a broaderlower intensity distribution of the slow ones.10–12,15 The nar-rower peaks at the sides of the spectra in Fig. 3 correspond tothe fast H atoms and the broader maxima of the slow atomsappear closer to the center, where there is a dip in the TOF

FIG. 2. �a� The mass spectrum of clusters produced in He expansions atp0=1.5 bar, T0=9 °C, and 0.33% concentration of Py in He. The meanneutral cluster size under these conditions was n̄=3. �b� The mass spectrumof clusters produced in Ar expansions at p0=3.0 bar, T0=9 °C, and 0.16%concentration of Py in Ar, where the mean cluster size could not be deter-mined exactly but it was n̄�5 �see Ref. 45 for details�. Gray bars corre-spond to Ark

+ clusters.

FIG. 3. Measured TOF distribution of H fragments from the photolysis ofpyrrole clusters produced �a� in He expansions corresponding to the meancluster size n̄=3 and �b� in Ar expansions corresponding to the mean clustersize n̄�5 for two different laser polarizations of 0° and 90° at 243 nm.

FIG. 4. H-fragment kinetic energy distributions from the photolysis of pyr-role clusters produced in He expansions for two different laser polarizationsof 0° and 90° at 243 nm. The KEDs were obtained by fitting the TOFspectra from Fig. 3�a� corresponding to the mean cluster size n̄=3.

FIG. 5. Comparison of H-fragment KEDs from the photolysis at 243 nm of�a� pyrrole molecule, �b� clusters with n̄=3 produced in He expansions, and�c� clusters with n̄�5 produced in Ar expansions.



spectra at the position of the zero kinetic energy fragments.The ratio of the fast versus slow fragment intensities is largerfor the 90° polarization. This is in qualitative agreement withthe results obtained for the molecule11,25 where a perpendicu-lar character ��=−0.33� of the excited optical transition lead-ing to the fast H atoms was observed, while the distributionof the slow fragments was isotropic ��0�.

A direct comparison of the present spectra with our spec-trum of a single molecule is shown in Fig. 5. Clearly theKED for the mean cluster size n̄=3 in Fig. 5�b� exhibits alarger contribution from the slow H atoms relative to the fastfragments. The slow component even increases when stilllarger clusters n̄�5 are generated in Ar expansions �Fig.5�c��.

Before discussing the trends in the KED with cluster sizea few notes should be made. First, it is worth mentioning thatspectra of different signal intensities are compared here.52

For illustration the measured TOF spectrum corresponding tothe KED in Fig. 5�c� is shown in Fig. 3�b�. The level ofsignal-to-noise ratio of the spectra in Fig. 3�b� is significantlylower than the one in Fig. 3�a�. This is reflected by the largererror bars in Fig. 5�c� compared to those in Fig. 5�b�. Simi-larly the molecular spectrum in Fig. 5�a� was measured withlower signals compared to KED for the clusters n̄=3.

Second, the monomer KED in Fig. 5�a� can be also com-pared to the previously measured spectra in Refs. 10–12, 15,and 25. Generally, the present spectrum exhibits a slightlyhigher slow component contribution. For example, the ratioof the fast versus slow maximum was 4.5 in Ref. 25, while itis 3.6 in the present KED at 0° polarization and 4.1 for the90° polarization �not shown�. This can still be regarded asgood agreement in view of the large error bars in our KEDfor single molecules. On the other hand, the spectra of Cro-nin et al.15 exhibit very little intensity for slow H fragmentsat the corresponding wavelengths of 244 nm. It ought to benoted that in the experiment of Cronin et al.15 neutralH-Rydberg atoms are detected and the slow fragments withvelocity component perpendicular to the TOF axis can es-cape detection more efficiently than the fast ones. On thecontrary, our TOF experiment with a low extraction fieldaccelerating the ionized H fragments towards the detector ismost sensitive to the H fragments with zero and low kineticenergies. In the conversion of the measured TOF spectra toKED, this high detection probability is accounted for by theMC simulation, yet the slow component contribution mightstill be enhanced. Nevertheless, our aim here is to observethe relative dependence of the KED on the cluster size,which is independent of the method used to obtain the spec-tra.

Also, a possible cluster contribution to the KED in Fig.5�a� cannot be excluded entirely, although the experimentalconditions for the present KED in Fig. 5�a� �see Table. I�were similar to the previous single moleculeexperiments11,12,14 such that presumably no clusters weregenerated in the expansion. In addition, the recorded quad-rupole mass spectra45 show no intensity of Pyn

+ ions for n�1 under these experimental conditions. Therefore the con-

tribution from pyrrole clusters can be assumed to be negli-gible �see the mass spectra and cluster size distribution inRef. 45�.

As for the larger clusters n̄�5 it should be mentionedthat there are some large mixed PykArm species as discussedin more detail in Ref. 45. The total number n=k+m of unitsin these mixed species is larger than the corresponding num-ber of Py molecules in the pure �Py�n clusters �k=4,m= 8 �.The presence of the mixed PynArm clusters, however, wouldnot influence significantly the discussion of our results, sincethey depend on the cluster size rather than on the complexityof the species with which the pyrrole molecule is clustered,as will be discussed below. Therefore, we refer to our “largeclusters” as n̄�5, keeping in mind that large mixed speciesare present in the beam. We note that in the case of the Heexpansions, pure Pyn, n̄=3, clusters are generated.

Finally, the last comment concerns the method of disen-tangling the fast and slow components in the KED. Weadopted a similar method as was used by Wei et al.,11,12

fitting the fast component with a Gaussian peak and using asimple polynomial to account for the broad background ofthe slow atoms. The fits are illustrated in Figs. 4 and 5 by thedashed �fast�, dotted �slow�, and dashed-dotted �sum� lines.Figure 6 shows the dependence of the ratio F/S between thefast and slow components obtained as the ratio of the inte-grated fits. It should be mentioned that the fitting procedureworks well for the molecular spectra dominated by the strongfast component peak but becomes questionable for the largerclusters with the strong slow component �as to what shapeshould be assumed under the fast component peak�. This canlargely influence the value of the F/S ratio. Therefore othermeans of evaluating this ratio were also adopted, rangingfrom simply taking the ratio of the maxima or integrating theslow component up to a fixed energy �0.55 eV� to fitting itwith functions of different shapes. Although the ratios differlargely depending on the integration method, the generaltrend of the F/S behavior with the cluster size is almost thesame independent of the method. The error bars in Fig. 6reflect the differences in this trend rather than the hugelydifferent ratio values obtained by the different integrationprocedures.

In what follows we shall discuss the trends observed inFig. 6. Clearly, the relative intensity of the fast componentdecreases with the cluster size for 0° polarization. This trendseems to stop or even slightly reverse for the 90° polarization

FIG. 6. Ratio of fast �F� and slow �S� channel fragments for different meancluster sizes and two laser polarizations.



between n̄=3 and n̄�5. The fast component stems prevail-ingly from a direct N–H bond fission which is a transition ofperpendicular character.11,12,15,25 In our experimental ar-rangement fragments of a perpendicular dissociation processcan be efficiently detected with 90° laser polarization, sincethey fly directly into the detector. In the larger clusters n�4 there are more unsaturated N–H bonds, which can un-dergo direct fission, than in the cyclic trimer.50 Thereforemore fast H fragments could be observed for n̄�5 than forn̄=3 with 90° polarization. With 0° polarization there can bealso more fast H fragments generated, but these are morelikely to escape our detection.

The experiment has been also performed at the wave-length of 193 nm for the single molecule case and for thesmall clusters n̄=3. Figure 7�a� shows an example of themeasured TOF spectra obtained with 0° polarization of the193 nm laser, and Fig. 7�b� shows the corresponding KED.Comparison with Fig. 5 shows a much higher contributionfrom the slow H fragments. The spectra obtained for singlemolecules and also the spectra at 90° polarization of the193 nm laser look very similar. More quantitatively, thesmall differences can be again expressed in terms of the F/Sratios, which are 0.18±0.04 and 0.22±0.04 for the singlemolecule 0° and 90° polarizations, respectively, and0.15±0.03 and 0.20±0.03 for the n̄=3 clusters 0° and 90°polarizations, respectively. First, these ratios at 193 nm aresignificantly smaller than the corresponding ratios at 243 nm.Clearly the channel leading to the slow H fragments is en-hanced at 193 nm, which is in agreement with the previousobservations for the bare molecule.15 Besides, going from thesingle molecule to the n̄=3 clusters similar trends �decreaseof F/S ratios� are observed at 193 nm as at 243 nm, althoughmuch less pronounced because of the lower significance ofthe fast channel at 193 nm.

It is interesting to note that, although there is about1.3 eV difference in the excitation energy deposited into themolecule with 193 and 243 nm, the position of the fast peakin the KED at about 0.8 eV remains the same. This suggeststhat the direct process releases the 0.8 eV into the fragmentkinetic energy independent of the excited initial state. On theother hand the higher excitation energy at 193 nm is reflectedin the high energy tail of the slow fragments, which nowseems to extend to somewhat higher energies compared tothe KED at 243 nm.

It ought to be mentioned that in our experiments with193 nm the contribution from the 243 nm laser used for theH-fragment ionization could not be avoided. We assume thatthe intense 193 nm pulse dissociates most of the species be-fore the ionizing pulse comes �some 10 ns later�; however,this pulse can still act on the undissociated species and con-tribute to the measured dissociation. Nevertheless, the differ-ent character of the spectra in Fig. 7 suggests that it is mainlya result of the 193 nm dissociation. The contribution from243 nm processes discussed above would make the F/S ratiolarger; therefore the present 193 nm spectra would representthe upper limits for the F/S ratios.

B. Calculations of photodynamics of bareand complexed pyrroles

The photochemical dynamics of a bare pyrrole moleculeis discussed in some detail in the Introduction. The discus-sion is based mostly on the work by Barbatti et al.23 Here weonly briefly review that upon photoexcitation of the pyrrole,the molecule can decay via two competing channels: �i� dis-sociation on the ��* state with a possible quenching into theground state �the direct dissociation process is mainlymarked by fast hydrogen components in the KED spectrumand a diabatic crossing of the S0 state� and �ii� ring deforma-tion conical intersection between the ��* and S0 states �inthis case slow hydrogens are likely to occur�. This mecha-nism is fully in accordance with the observed dependence ofthe kinetic energy distribution spectra on the excitationwavelength. It is also consistent with all other experimentalobservations. There is, however, one detail which still needsto be explained. Wei et al.12 have found that some fast hy-drogen atoms can be produced even from the pyrrole mol-ecules deuterated in the nitrogen position. This can be anexperimental artifact but it can also be explained by an in-tramolecular hydrogen transfer in the pyrrole molecule. Wehave calculated geometries and energetics of the Franck-Condon point, the ��* /S0 and ��* /S0 crossings, and theminimum with a hydrogen transferred from a nitrogen to acarbon atom �Fig. 8�. The ��* /S0 intersection can bereached by a ring deformation and the ��* /S0 intersectionby N–H stretching. The released hydrogen can, however, betrapped by a neighboring carbon atom forming a structurewith a transferred hydrogen. One of the two hydrogen atomscan later dissociate with the same energy as in the case of adirect dissociation. The intramolecular exchange process canthus be a possible explanation for the fast hydrogen atomsfound after dissociation of N-deuterated pyrrole. Note thatdirect dissociation from the C–H bond cannot lead to the fasthydrogen atoms.53

FIG. 7. Measured TOF distribution of H fragments from photolysis at193 nm of �a� pyrrole clusters produced in He expansions corresponding tothe mean cluster size n̄=3 and �b� the corresponding KED.



This mechanism is also a good basis for the discussionof the possible photochemical pathways of the pyrrole in aconfined environment, of which pyrrole clusters are an ex-ample. We have addressed the question how the excited statepotential energy surface of the pyrrole molecule changesupon the complexation. In more detail, two major questionsare considered: �i� How will the environment affect the ��*

state and the fast hydrogen reaction coordinate? �ii� How willthe complexation affect the ��* state and the slow hydrogencoordinate? �iii� Can excited pyrrole react with another pyr-role unit in a reasonable time? Answering these questionscan shed light on the observed size dependence of the kineticenergy distribution spectra and also on the photodynamics ofthe bare and complexed pyrrole molecules.

In the computational study we have focused our atten-tion to two systems: pyrrole complexed with an argon atomand the pyrrole dimer. Argon atoms have been chosen here asan archetype of a structureless solvent, the only effect ofwhich is the Pauli repulsion with another molecule. Notealso that it cannot be entirely ruled out that argon atomsactually solvate pyrrole molecules in larger clusters.45 Let usstart the discussion with the geometries of Py–Ar complex inthe Franck-Condon point, for the ��* /S0 crossing, and for adistant point along the N–H coordinate. The geometry of thepyrrole-argon complex has been chosen such that it re-sembles the structure of Pyn even though the character of theforces holding the particles together is different for the twocomplexes. The basic feature of the pyrrole complexes is thatthe N–H bond points toward another pyrrole unit. The C–Hbonds remain free for small pyrrole clusters.45,50 We havetherefore considered the argon atom to be placed 3.5 Å fromthe nitrogen atom, in the N–H axis, in the pyrrole plane.

Figure 9 shows how the electronic states are affected by thepresence of the solvent argon atom for the N–H dissociationreaction coordinate. The right-hand side panel corresponds tothe bare Py molecule, while the left-hand side shows thesituation for the Py–Ar complex. The first two excited statesare of Rydberg character, corresponding asymptotically tothe ��* states. By stretching the N–H bond we are able toreach a conical intersection between ��* and the groundstate. The Franck-Condon point is separated by only a smallbarrier of about 0.2 eV from this intersection. The excitedvalence states for the isolated molecule rapidly increase withthe N–H distance. Solvation with an argon atom has a pro-found influence on the process. The position of valence statesin the Franck-Condon region is virtually unchanged whilethe two Rydberg states are shifted up by 0.5 eV in energy.This shift is due to the Pauli interaction between extensiveRydberg orbitals and the argon atom. Upon the stretching ofthe N–H bond, valence states almost do not feel the argonatom while the ��* states feel strong repulsion from theatom. As a result the ��* /S0 intersection is not present forthis fixed cage geometry anymore. If we place the argonatom at a larger distance, the conical intersection will reap-pear. However, even in this case hydrogen atoms will notdissociate and instead a hot ground state will be produced.

Figure 10 shows how the solvation influences the secondpossible reaction coordinate connecting the Franck-Condonpoint with a ��* /S0 intersection. In this process, the nitro-gen atom is elevated from the pyrrole plane. The argon atomis kept in a fixed distance in the plane of the four carbonatoms. For an isolated pyrrole the first ��* state is separatedby only a small barrier �0.2 eV�, while the second ��* is

FIG. 8. Calculated geometries of the Franck-Condon point, the ��* /S0 and ��* /S0 crossings, and the H-transfer structure. Relative energies with respect tothe Py ground state are indicated. The energies of the lowest Rydberg state and the lowest valence state are shown for the Franck-Condon point. Energies werecalculated at the CASPT2/SA-5-CAS�8/7� /6-31g*+NHdif level �for H-transfer structure diffuse functions were placed on all atoms�.

FIG. 9. Potential energy curves along the rigid N–Hstretching coordinate for bare �right panel� and solvated�left panel� pyrrole molecules. Energies are calculatedat the CASPT2/SA-5-CAS�8/7� /6-31g*+NHdif level.



connected to the intersection without any barrier. This reac-tion coordinate is also important because of other reasons: itconnects the two ��* states one with the other and also withthe ��* states. After the ��* state is populated, this coordi-nate may lead not only to the quenching to the ground statebut also to the quenching to the ��* with a subsequent hy-drogen dissociation. Vice versa, ��* population can be fun-neled to the ground state via the ��* /S0 intersection withthis coordinate. Complexation by an argon solvent does notlead to qualitative change of the picture of the process.Rydberg states are higher in energy in the Franck-Condonregion and this results in a shifted ��* /��* intersectionwhile the valence states are almost unaffected. We can thusconclude that the out-of-plane reaction channel will remainopen even for the solvated pyrrole while the N–H dissocia-tion channel will be closed.

The effect of the argon atom on the N–H dissociationchannel comes from the Pauli repulsion between the argonwave function and the delocalized wave function of the ��*

state. This effect will certainly be dominant even for morecomplex solvents, including a second pyrrole atom. Indeed,we were not able to locate the conical intersection by pro-longing the N–H¯� bond while the intersection could belocalized for the free N–H bond in the pyrrole dimer. Thisarrangement is also typical for the larger complexes, e.g., intrimers and tetramers where there are no free N–H bonds.More subtle solvation effects can play a role for the out-of-plane reaction coordinate. We have therefore explored thephotodynamics of the pyrrole dimer by asking whether the��* /S0 intersections can be found for the dimer and whetherits relative energy position is significantly shifted by the in-fluence of the long-range electrostatic forces from the solventpyrrole molecule. Since the conformational space of a pyr-role dimer is richer than in the case of the monomer, we havefound several conical intersections. In all cases the intersec-tion can be viewed as an intersection for an isolated mol-ecule with a second pyrrole unit solvating this structure �seeFig. 11�. For isomers A and B, the out-of-plane motion islocalized to the hydrogen donor pyrrole unit while the hy-drogen acceptor unit is deformed for isomer C. The energiesof these conical intersections do not differ significantly. Thesame holds for the barriers separating the first ��* state and

the crossing: in all three cases the energy is about 0.1 eV andis also close to those found in bare pyrrole. Linear interpo-lation between the ground state and the conical intersectionB is also presented on Fig. 11. Note that the barrier is lowerthan the barrier found for the isolated molecule. This is, how-ever, caused mostly by a different choice of the active spacein the CAS-SCF.

So far the second pyrrole molecule has been only a spec-tator which has not been directly involved in the chemicalprocesses in the excited states. It can, however, not be ruledout that the pyrrole molecule in the excited state reacts withanother pyrrole molecule. The resulting bimolecular reactioncan then compete with the unimolecular reaction channelsoutlined above. We have considered two possible bimolecu-lar reactions. First, we have considered the option that thepyrrole molecule can donate a hydrogen atom to the nitrogenatom of another pyrrole unit. This process can occur on the��* potential energy surface. The corresponding structure isshown in Fig. 11. This process is analogical to the excitedstate hydrogen transfer observed in the pyrrole-ammoniaclusters.31 There is, however, an important difference. Whilethe pyrrole-ammonia complex in the ground state is a typicalhydrogen bond, with the N–H bond pointing towards theacceptor nitrogen, the pyrrole dimer represents a case of aN–H¯� bond. Here, the hydrogen atom points towards the� system of the pyrrole unit which cannot accept the hydro-gen. It is therefore necessary first to rearrange the heavyatom cage to allow for the transfer. Because of these entropi-cal reasons, it can be expected that this process will lastlonger than the internal conversion processes describedabove. Energetically, the hydrogen transfer process is pos-sible since the species with transferred hydrogen has some-what lower energy than the lowest excited state of the dimer.The barrier is estimated to be more than 1 eV with respect tothe lowest Rydberg state of the dimer �see Fig. 11�. Note,however, that this barrier results from a direct interpolationand paths with lower barriers exist. The structure with thehydrogen transferred from one pyrrole unit to the other is atthe same time not far from the conical intersection with theground state. It can be therefore expected that the system

FIG. 10. Linear interpolation between the ground state geometry and the ��* /S0 intersection for the bare pyrrole molecule �right panel� and the moleculesolvated by an argon atom �left panel�. Energies were calculated at the CASPT2/SA-5-CAS�8/7� /6-31g*+NHdif level. Interpolation index n=0 correspondsto the ground state geometry and interpolation index n=10 to the conical intersection.



quenches to the ground state and subsequently releases thehydrogen atom, similarly as in the case of the hydroniummolecule.

We have also considered possible reactions of pyrrole inthe ��* state. In this case two pyrrole units can dimerize. Wehave localized a conical intersection connecting the ��* andground states which leads to a dimer formation. This conicalintersection is energetically some 1.6 eV below the Franck-Condon point and it is thus energetically accessible. The di-rect interpolation path has a barrier of more than 1 eV �thecorresponding curve is shown in Fig. 11�. This is muchhigher than the barrier to reach the monomer out-of-planeintersection assisted by a second solvent pyrrole molecule. Amore complicated reaction path with a barrier of 0.5 eV hasalso been found. This is still more than the direct quenchingto the ground state via the ��* /S0 intersection; the bimo-lecular reaction is furthermore not preferred entropically.Note that after quenching to the ground state via this inter-section, the dimer stays bound and after some rearrangementit forms a structure which is unstable by some 0.7 eV withrespect to the van der Waals pyrrole dimer.

IV. CONCLUSIONS

Pyrrole clusters have been generated under well definedconditions which were controlled by the scattering analysisfrom helium beams published in an accompanying paper.45

In one data set the average cluster size is n̄=3; in the otherthe cluster size is n̄�5 and mixed Py-Ar clusters are gener-ated. These clusters are photodissociated by lasers at thewavelengths of 243 and 193 nm and two different polariza-tion angles of 0° and 90°. The outgoing H-atom products aredetected by �2+1� REMPI at 243.07 nm in a special time-of-flight mass spectrometer with low extraction fields whichallows us to measure the velocity of the H atoms up to zerovalues. The converted kinetic energy distributions exhibit thefollowing results, compared to the larger fast peak and thesmaller slow peak of the kinetic energy distribution of thephotodissociation of the bare molecule: �i� The clusters showan enhanced intensity of the fast peak at 90° polarization andthus the same tendency as was found for the bare moleculewith a preference for the perpendicular transition with �=−0.33 for the fast peak and an isotropic behavior with �=0for the slow peak. �ii� The intensity of the fast peak dropsappreciably at the expense of the slow peak with increasingcluster size for the fixed photolysis energy of 243 nm. �iii�The intensity of the fast peak drops in the same manner whenthe photolysis energy is increased to 193 nm for the samecluster size.

To explain these results we have extended the photo-chemical mechanism presented by Barbatti et al.23 in a the-oretical study to situations where the pyrrole molecule issolvated. These connections are presented pictorially in Fig.

FIG. 11. Important structures on the potential energy surface of the pyrrole dimer together with the respective linear interpolation potential energy curves. Thepresented relative energies are in eV and are calculated at the SA-5-CAS�12/12� /6-31g* level; for H-transfer structure the basis is further augmented withdiffuse functions on nitrogens and transferred hydrogen. Relative energies with respect to the Py2 ground state energy are shown. The energies of the lowestRydberg state and the lowest valence state are shown for the Franck-Condon structure. Linear interpolation between the ground state structure and out-of-plane��* /S0 intersection �structure B� is shown in panel �a�, interpolation between the ground state structure and chemically bound pyrrole dimer is shown in panel�b�, and interpolation between the ground state structure and hydrogen transfer structure is shown in panel �c�. Another reaction channel �not shown� is theN–H fission in the S1 state feasible for the free N–H bond but closed for the N–H¯� arrangement. Interpolation index n=0 corresponds to the ground stategeometry and interpolation index n=10 to the final structures �conical intersection, dimer, hydrogen transfer molecule�.



12. This includes the calculation of the ��* /S0 conical inter-section of the N–H stretch coordinate and the ��* /S0 and��* /��* conical intersections of the ring deformation. Theinfluence of the solvent was simulated by the presence of anAr atom at a fixed distance in the plane of the carbon atomswhich resembles the N–H¯� binding motive of the pyrroledimer. To account for other solvation effects than the Paulirepulsion, we have also explored the pyrrole dimer. The pres-ence of the Ar atom has a strong influence on the conicalinteraction in the N–H stretch coordinate �see the lower partof Fig. 12�. While the S0 and the excited valence states arevirtually unchanged, the delocalized ��* Rydberg states areshifted by 0.5 eV in the Franck-Condon region and the inter-action between the Rydberg state and argon gets strongerwith increasing N–H distance. As a result the ��* /S0 conicalintersection is no longer present and the channel for the fastH atoms is closed. The changes in the deformation coordi-nate upon solvation with Ar are on the other hand almostnegligible. The ��* /S0 intersection remains open and thepopulation of the slow peak is not hindered. This holds trueeven when the argon atom is replaced by another pyrrolesolvent molecule. Apart from unimolecular processes de-scribed above, we have also explored possible bimolecularreactions of pyrrole molecule in the excited state. Pyrrole inthe ��* state can transfer its hydrogen to the second pyrrolemolecule while pyrrole in the ��* state can form a C–Cbond with the solvating pyrrole unit. These processes areenergetically possible but are not favored neither energeti-cally nor entropically.

The results of our photodissociation experiments can befully understood in terms of this mechanism. We observemany more H atoms in the slow peak in clusters with respectto the molecule, because the route to the direct dissociationinto the fast peak is closed �furthermore, even when the mol-ecule reaches the ��* /S0 intersection, the solvent preventsdirect dissociation and the system quenches to the boundground state�. In addition, the out-of-plane mode, introducedby Barbatti et al., also operates in clusters and plays an im-portant role possibly even at 243 nm. The increase of theslow component intensity for 193 nm with respect to 243 nmis traced back to the direct excitation of the ��* state. The��* /S0 is in this case diabatically directly accessible and

this does not change with complexation. An important mes-sage which can be drawn out of these results is that conclu-sions about the photochemical behavior in a confined envi-ronment �e.g., in biological systems� which are solely basedon gas phase data can be misleading.

Despite the effort devoted to the theoretical study of pyr-role photodynamics, the picture is still incomplete. Ab initiocalculations have provided us with possible reaction chan-nels but dynamical calculations including all relevant statesand all relevant internal coordinates are needed to answer thequestion how the initial wave packet splits among states andreaction channels. The energetical possibility of a certain re-action channel does not guarantee that the reaction takesplace.54 As we have stressed here, it is imperative to considerenvironmental effects on the photodynamics. One can do thisvia continuum solvation models29,55 or via a mixed quantummechanics/molecular mechanics �QM/MM� potential energysurface.28 Since the pyrrole photochemistry is dictated by theshort-range Pauli repulsion, it is necessary to choose theproper treatment at either continuum solvation or theQM/MM level.30 Solvated pyrrole clusters can serve as animportant prototype for developing such methodology.

Very recently we have learned about new experiments inwhich photodissociation of pyrrole molecules clustered withxenon atoms has been observed.56 In this experiment,quenching of the N–H fission upon complexation has alsobeen found, fully in accordance with our calculations andexperiment with pyrrole clusters.

ACKNOWLEDGMENTS

The authors thank Vitaly Kresin for many valuable dis-cussions. Support by the special program “Nanotechnologyfor Society” of the Czech Academy of Sciences via GrantNo. KAN400400461 and by the Grant Agency of the CzechRepublic via Grant No. 203/06/1290 is gratefully acknowl-edged. One of the authors �P.S.� thanks the support of thepostdoctoral grant of the Grant agency of the Czech Repub-lic, 203/07/P449. Another author �V.P.� acknowledges Uni-versity Grant No. 113-10/257852.

1 C. E. Crespo-Hernandez, B. Cohen, P. M. Hare, and B. Kohler, Chem.Rev. �Washington, D.C.� 104, 1977 �2004�.

FIG. 12. Schematic figure showing the influence of thesolvent on the photochemistry of pyrrole. The upperpanel shows the situation for the bare Py molecule�partly adopted from Ref. 23� and the lower panel illus-trates the effect of solvation in the N–H bond stretchingand ring deformation coordinates.



2 L. Serranoandres, M. Merchan, I. Nebotgil, B. Roos, and M. Fulscher, J.Am. Chem. Soc. 115, 6184 �1993�.

3 P. Celani and H. Werner, J. Chem. Phys. 119, 5044 �2003�.4 O. Christiansen, J. Gauss, J. Stanton, and P. Jorgensen, J. Chem. Phys.

111, 525 �1999�.5 M. Palmer, I. Walker, and M. Guest, Chem. Phys. 238, 179 �1998�.6 M. Palmer and P. Wilson, Mol. Phys. 101, 2391 �2003�.7 B. Roos, P. Malmqvist, V. Molina, L. Serrano-Andres, and M. Merchan,J. Chem. Phys. 116, 7526 �2002�.

8 R. Burcl, R. Amos, and N. Handy, Chem. Phys. Lett. 355, 8 �2002�.9 J. Wan, J. Meller, M. Hada, M. Ehara, and H. Nakatsuji, J. Chem. Phys.

113, 7853 �2000�.10 D. A. Blank, S. W. North, and Y. T. Lee, Chem. Phys. 187, 35 �1994�.11 J. Wei, A. Kuczmann, J. Riedel, F. Renth, and F. Temps, Phys. Chem.

Chem. Phys. 5, 315 �2003�.12 J. Wei, J. Riedel, A. Kuczmann, F. Renth, and F. Temps, Faraday Discuss.

127, 267 �2004�.13 H. Lippert, H.-H. Ritzel, I. V. Hertel, and W. Radloff, ChemPhysChem 5,

1423 �2004�.14 A. J. van den Brom, M. Kapelios, T. N. Kitsopoulos, N. H. Nahler, B.

Cronin, and M. N. R. Ashfold, Phys. Chem. Chem. Phys. 7, 892 �2005�.15 B. Cronin, M. G. D. Nix, R. H. Qadiri, and M. N. R. Ashfold, Phys.

Chem. Chem. Phys. 6, 5031 �2004�.16 M. N. R. Ashfold, B. Cronin, A. L. Devine, R. N. Dixon, and M. G. D.

Nix, Science 312, 1637 �2006�.17 A. Sobolewski and W. Domcke, Chem. Phys. 259, 181 �2000�.18 A. Sobolewski, W. Domcke, C. Dedonder-Lardeux, and C. Jouvet, Phys.

Chem. Chem. Phys. 4, 1093 �2002�.19 V. Vallet, Z. Lan, S. Mahapatra, A. Sobolewski, and W. Domcke, Faraday

Discuss. 127, 283 �2004�.20 V. Vallet, Z. Lan, S. Mahapatra, A. Sobolewski, and W. Domcke, J.

Chem. Phys. 123, 144307 �2005�.21 A. Trofimov, H. Koppel, and J. Schirmer, J. Chem. Phys. 109, 1025

�1998�.22 H. Koppel, E. Gromov, and A. Trofimov, Chem. Phys. 304, 35 �2004�.23 M. Barbatti, M. Vazdar, A. J. A. Aquino, M. Eckert-Maksić, and H.

Lischka, J. Chem. Phys. 125, 164323 �2006�.24 I. Frank and K. Damianos, J. Phys. Chem. 126, 125105 �2007�.25 V. Papadakis and T. N. Kitsopoulos, Rev. Sci. Instrum. 77, 083101

�2006�.26 It is interesting to note that this behavior is just opposite to the recent

study of adenine photodissociation, where the fast hydrogens increase inintensity with the photon energy �Ref. 27�.

27 M. Nix, A. Devine, B. Cronin, and M. Ashfold, J. Chem. Phys. 126,124312 �2007�.

28 A. Toniolo, S. Olsen, L. Manohar, and T. Martinez, Faraday Discuss.127, 149 �2004�.

29 S. Yamazaki and S. Kato, J. Am. Chem. Soc. 129, 2901 �2007�.30 P. Slavicek and T. Martinez, J. Chem. Phys. 124, 084107 �2006�.31 O. David, C. Dedonder-Lardeux, C. Jouvet, H. Kang, S. Martrenchard, T.

Ebata, and A. Sobolewski, J. Chem. Phys. 120, 10101 �2004�.32 A. Sobolewski and W. Domcke, Chem. Phys. Lett. 321, 479 �2000�.33 E. Samoylova, V. Smith, H. Ritze, W. Radloff, M. Kabelac, and T.

Schultz, J. Am. Chem. Soc. 128, 15652 �2006�.34 G. Olaso-Gonzalez, D. Roca-Sanjuan, L. Serrano-Andres, and M. Mer-

chan, J. Chem. Phys. 125, 231102 �2006�.35 U. Buck, J. Phys. Chem. A 106, 10049 �2002�.36 P. Slavíček, P. Ždánská, P. Jungwirth, R. Baumfalk, and U. Buck, J. Phys.

Chem. A 104, 7793 �2000�.37 P. Slavíček, P. Jungwirth, M. Lewerenz, N. H. Nahler, M. Fárník, and U.

Buck, J. Phys. Chem. A 107, 7743 �2003�.38 P. Slavíček, P. Jungwirth, M. Lewerenz, N. H. Nahler, M. Fárník, and U.

Buck, J. Chem. Phys. 120, 4498 �2004�.39 N. H. Nahler, M. Fárník, U. Buck, H. Vach, and R. B. Gerber, J. Chem.

Phys. 121, 1293 �2004�.40 M. Fárník, N. H. Nahler, U. Buck, P. Slavíček, and P. Jungwirth, Chem.

Phys. 315, 161 �2005�.41 R. Baumfalk, U. Buck, C. Frischkorn, S. R. Gandhi, and C. Lauenstein,

Chem. Phys. Lett. 269, 321 �1997�.42 R. Baumfalk, U. Buck, C. Frischkorn, S. R. Gandhi, and C. Lauenstein,

Ber. Bunsenges. Phys. Chem. 101, 606 �1997�.43 V. Poterya, M. Fárník, P. Slavíček, U. Buck, and V. Kresin, J. Chem.

Phys. 126, 071101 �2007�.44 M. Fárník and U. Buck, Phys. Scr. 76, 1 �2007�.45 V. Profant, V. Poterya, M. Fárník, P. Slavíček, and U. Buck, J. Phys.

Chem. A �to be published�.46 P. Lohbrandt, R. Galonska, H. J. Kim, M. S. C. Lauenstein, and U. Buck,

in Atomic and Molecular Beams. The State of the Art 2000, edited by R.Campargue �Springer, Berlin, 2001�, p. 623.

47 A certain ambiguity exists, whether the clusters produced in Ar expan-sions are pure �Py�n clusters or mixed �Py�n ·Arm species of similar size�see Ref. 45 for more details�; however, this ambiguity has little influenceon the interpretation of the present experimental results, as discussed inthe text below.

48 R. Baumfalk, U. Buck, C. Frischkorn, N. H. Nahler, and L. Hüwel, J.Chem. Phys. 111, 2595 �1999�.

49 R. Baumfalk, N. H. Nahler, U. Buck, M. Y. Niv, and R. B. Gerber, J.Chem. Phys. 113, 329 �2000�.

50 A. Gómez-Zavaglia and R. Fausto, J. Phys. Chem. A 108, 6953 �2004�.51 H.-J. Werner, P. J. Knowles, R. Lindh et al., MOLPRO, version 2006.1, a

package of ab initio programs, 2006 �see http://www.molpro.net�.52 Since our conclusions do not depend on the absolute intensities of the

spectra, all the spectra are normalized to the same area.53 C. Barckholtz, T. Barckholtz, and C. Hadad, J. Am. Chem. Soc. 121, 491

�1999�.54 E. Muchová, P. Slavíček, A. Sobolewski, and P. Hobza, J. Phys. Chem. A

111, 5259 �2007�.55 R. Spezia, I. Burghardt, and J. Hynes, Mol. Phys. 104, 903 �2006�.56 L. Rubio-Lago, D. Zaouris, Y. Sakellariou et al., J. Chem. Phys. 127,

064306 �2007�, previous paper.



Experimental and theoretical study of the pyrrole cluster photochemistry: Closing the πσ[sup

Documents

Transcript of Experimental and theoretical study of the pyrrole cluster photochemistry: Closing the πσ[sup