Photodissociation dynamics of pyrrole: Evidence for mode specific dynamics from conical intersections

The H and D atom elimination mechanisms in the photodissociation of jet cooled pyrrole and pyrrole-d1 have been studied by photofragment velocity map imaging.

The molecules were excited to the 1 1A2 (πσ*) state at λ = 243 nm and to the 1 1B2 (ππ*) state at λ = 217 nm.

H/D atoms were detected by (2 + 1) resonance enhanced multiphoton ionization (REMPI) at λ = 243 nm.

The analysis of the images and the resulting translational energy distributions from the 1 1A2 state demonstrates the existence of two decay pathways, fast mode-specific cleavage of the NH bond in the excited state (channel A) and internal conversion (IC) to the electronic ground state (S0) followed by unimolecular decomposition of the vibrationally hot S0 molecules (channel B).

The angular distributions of the H/D atoms from the direct dissociation in the excited state are strongly anisotropic, whereas the decay of the S0 molecules leads to spatially isotropic distributions.

The results at λ = 217 nm indicate that the 1 1B2 state undergoes an ultrafast radiationless transition to 1 1A2 followed by the abovementioned direct mode-specific NH bond fission on the 1 1A2 potential energy surface (channel A′) or conversion to S0 and subsequent unimolecular decomposition (channel B′).

The latter pathway may also be initiated by a direct relaxation from 1 1B2 to S0.

The anisotropy parameter of β ≈ –1 for the direct NH bond fission at λ = 217 nm is in accordance with the expectations for a perpendicular electronic excitation and a dissociation lifetime that is short compared to the rotational period of the molecules.

The fast decay dynamics of both excited electronic states can be rationalized with reference to the theoretically predicted conical intersections between the ππ*, πσ*, and S0 potential energy surfaces and the antibonding nature of the πσ* potential energy surface with respect to the NH bond [A. L. Sobolewski, W. Domcke, C. Dedonder-Lardeux and C. Jouvet, Phys. Chem. Chem. Phys. 2002, 4, 1093].


Recent excited state ab initio calculations for H atom photodetachment processes from NH-containing heteroaromatic compounds, aromatic amines, or phenols have suggested a new paradigm for the photochemical dynamics in these systems.1

Accordingly, the fate of the photoexcited molecules is controlled by conical intersections between the usually strongly absorbing first excited 1ππ* state, a neighboring optically dark 1πσ* state, and the S0 electronic ground state.

The key role in the ensuing dynamics is played by the πσ* state which becomes repulsive along the NH or OH stretching coordinates and intersects the S0 state at long NH/OH bond lengths because the latter does not adiabatically correlate with the respective dissociation products in their electronic ground states.

Conical intersections between the ππ* and πσ* and the πσ* and S0 potential energy hypersurfaces thus open efficient pathways for radiationless electronic transitions leading to an ultrafast relaxation of the excited molecules to the electronic ground state and/or to a fast NH/OH bond fission and corresponding dissociation of the molecules.

These mode specific channels have been postulated to have great importance for the photochemical properties of the nucleic acid bases and the biologically relevant aromatic amino acids,2–6 for intramolecular H atom transfer reactions in excited electronic states (ESIHT),7 and for photo-induced intermolecular H atom transfer reactions in hydrogen-bonded molecular complexes.6,8–10

The pyrrole molecule is an almost ideal model system for which the photochemical dynamics can be studied in detail.

In addition, it is a building block of many biologically important compounds which may show related mechanisms.1,11

Last but not least, it plays a role in the synthesis of biologically active compounds, pesticides, organic polymers, and organometallic magnets.12

The UV absorption spectrum of pyrrole shows two strong bands, the first with an intensity maximum near λ = 210 nm, the second near 165 nm.

In addition, a very weak feature is observed with a maximum around 240 nm.13–16

The spectrum between 190 and 260 nm is reproduced in Fig. 1.

As can be seen, the absorption bands are very broad and unstructured and this has hampered their analysis for a long time.

Further work has been performed by electron energy-loss spectroscopy,15 photoelectron spectroscopy,17–19 and resonance enhanced multiphoton ionization (REMPI).20

In addition, the experimental studies were complemented by theoretical electronic structure investigations.15,21–23

The most recent quantum chemical calculations by Roos et al23. showed that the bulk of the intensity of the 210 nm band arises from the intra-valence transition to the 1 1B2 (ππ*) state.

A small contribution (≈15%) may also come from the neighboring 2 1A1 (ππ*) state.

The 165 nm band has been attributed to the excitation to the 1 1B1 and 2 1B2 states which belong to the 3p Rydberg series.

The analysis of the weak 240 nm band has, however, been controversial.

The respective transition has traditionally been assigned to the 1 1A2 (3s) Rydberg state.22,23

The observation of this electronically forbidden band can be explained by a vibronic coupling with the 1 1B2 (ππ*) state induced by out-of-plane vibrations (b1 symmetry).

One earlier paper has also assigned the 240 nm band to the excited 2 1A1 valence state.24

However, Sobolewski and Domcke1,11 recently demonstrated that the stretching of the NH bond in the 1 1A2 state leads to a Rydberg-to-valence transformation by which that bond becomes anti-bonding, i.e., the electron configuration becomes πσ*.

The Rydberg character remains evident by the diffuseness of the σ* orbital, but the πσ* character was postulated to be responsible for a fast mode-specific N–H bond dissociation and, through a conical intersection with the S0 state, for an efficient radiationless relaxation of the electronically excited molecules.

Moreover, a fast predissociation of the ππ* state by the πσ* state was suggested to explain the apparently extremely short lifetimes of the excited states of pyrrole and the resulting absence of observations of laser induced fluorescence.

Direct experimental evidence for these theoretically predicted processes had, however, been missing.

In a recent communication, we reported on first results of an investigation of the formation of H atoms from pyrrole and, for comparison, N-methylpyrrole in their lowest excited electronic states by photofragment velocity map imaging.25

Pyrrole was excited to the 1 1A2 (πσ*) state at λ = 243 nm, and the nascent H atoms were detected at the same wavelength by 2 + 1 REMPI.

The observed H atom velocity maps showed the existence of two dissociation channels.

The first was found to produce very fast H atoms and appeared to be due to a rapid direct NH bond cleavage in the excited electronic state, as predicted by the theoretical studies.1,11

The H atom kinetic energy distribution had a strong, narrow peak at high translational energies.

The less important second channel was observed to lead to much slower H atoms with a very broad kinetic energy distribution, consistent with a statistical formation via unimolecular decay reactions in the electronic ground state after internal conversion from the excited state.

This conclusion was supported by the results for N-methylpyrrole which showed only H atoms corresponding to the second channel.

The experimental results25 thus corroborated the proposal by Sobolewski and Domcke1,11 that the lowest excited electronic state of pyrrole has πσ* character and is anti-bonding with respect to the stretching of the N–H bond.

In this discussion paper, we present the results of an extended study of the photochemical dynamics of normal pyrrole (pyrrole-h1) and selectively deuterated pyrrole (pyrrole-d1) at two excitation wavelengths, λ = 243 nm and λ = 217 nm.

The reactions studied can be written as C4H4NH +  → H + C4H4NandC4H4ND +  → D + C4H4N → H + C4H3NDThe two excitation wavelengths (243 and 217 nm) will be indicated below by unprimed and primed labels, where appropriate.

In the first set of experiments (λ = 243 nm, one colour), we probed the H and D atom photodetachment from both isotopomers via the 1 1A2 (πσ*) state.

The obtained new D atom images from pyrrole-d1 (reaction (2d)) were seen to exhibit an efficient ultrafast mode-specific dissociation of the ND bond of the molecules on the πσ* potential energy surface, confirming the corresponding interpretation of our earlier experiments25 and providing direct evidence for the theoretical predictions of Sobolewski and Domcke.1,11

In the second set of experiments, we performed two-colour measurements, in which the molecules were excited to the 1 1B2 (ππ*) state at λ = 217 nm and the H atoms were again probed at λ = 243 nm.

The results of these experiments corroborate the postulate that the 1 1B2 (ππ*) state is predissociated by the repulsive πσ* state.

Experimental section

The photofragment velocity map imaging apparatus built in our laboratory will be described in some detail elsewhere.26

The basic set-up resembles that of Eppink and Parker.27

A gas mixture containing ≈0.1% of pyrrole in He was prepared by bubbling the seed gas through a sample of liquid pyrrole stored in a glass reservoir at –20 °C and 1 bar pressure.

The gases expanded into a differentially pumped stainless steel vacuum chamber through a solenoid actuated pulsed valve (General Valve #9) operated at 10 Hz repetition rate with ≈350 µs pulse–1 opening times.

A molecular beam entered the second vacuum stage through a 1 mm diameter electroformed conical skimmer (Beam Dynamics).

Based on our experience with other molecules under similar expansion conditions,28 the rotational temperature of the pyrrole in the molecular beam was estimated to be of the order of Trot ≈ (20 ± 10) K.

The photolysis/probe laser beams intersected the molecular beam at right angle halfway between the repeller and extractor plates of the Wiley-McLaren ion imaging electrode assembly.

A liquid nitrogen cryo-pump surrounding the arrangement minimized the residual hydrocarbon background.

The laser radiation at λ = 243 nm (≈0.05–1.4 mJ) was generated by frequency-doubling the output of a Nd:YAG pumped dye laser (Quanta-Ray GCR-3/Lambda Physik FL-3002) in a BBO I crystal and focused into the molecular beam with an f = 1 m lens.

Light at λ = 217 nm (≈0.1–2.0 mJ) was generated by frequency-doubling the output of a XeCl excimer pumped dye laser (Lambda Physik EMG 201/Lambda Physik FL-3002) in a BBO II crystal.

To avoid a photolysis of the pyrrole with the λ = 243 nm probe light in the two-colour experiments, the probe power was lowered until the 243 nm one-colour signal in the absence of the 217 nm photolysis beam had virtually disappeared.

The probe beam was time-delayed by ≈15 ns with respect to the counterpropagating photolysis beam.

The laser polarizations were oriented with the electric field vectors perpendicular to the plane defined by the molecular beam and the laser beams using Wollaston polarizers and Fresnel double rhombs.

All imaging measurements were made under velocity mapping conditions.27

The probe laser was periodically scanned by ±6 cm–1 resp. ±4 cm–1 around the line centers of the recoiling H and D atoms to cover their complete Doppler profiles.

The resulting H+/D+ ions were detected by a microchannel plate (MCP) detector coupled to a phosphorescence screen.

The obtained signals were recorded with a CCD camera (640 × 480 pixels, LaVision) and accumulated over up to 200 000 laser shots using single ion counting and centroiding29 to improve the detection sensitivity and spatial resolution and discriminate against noise.

Background images were taken under the same experimental conditions with the molecular beam valve closed.

Mass selectivity was achieved using a fast transistor switch (Behlke) to gate the MCP voltage.

Normal pyrrole (Aldrich, 98%) was used after repeated fractionated distillation until the liquid was clear.

Catalytic reactions in the stainless steel supply pipe were suppressed by maintaining a slow flow of the pyrrole/carrier gas mixture through a 1 mm id teflon tube extending in the supply line from the sample reservoir all the way to the molecular beam nozzle.

Pyrrole-d1 was prepared from the normal isotopomer by repeated stirring with excess D2O.

The final product was dried over Na2CO3.

The deuteration (>97%) was checked by NMR spectroscopy.

The gas supply lines of the apparatus were thoroughly purged overnight with D2O vapour before the measurements with the deuterated sample to prevent isotope exchange with H2O at the walls.


The two-dimensional (2D) velocity mapped H/D images provided by the experiments at λ = 243 nm and λ = 217 nm were inverse Abel transformed30 to obtain the respective three-dimensional (3D) velocity distributions using the basis-set expansion method of Dribinski et al.31

The required calibrations were performed by measuring H/D photofragment images from HBr/DBr and HI/DI, for which the dissociation energies are precisely known (D00(HBr) = 30 210 cm–1, D00(HI) = 24 630 cm–1).32–35

The results obtained at the two excitation wavelengths will be considered separately.

λ = 243 nm

H/D velocity maps

The H atom image from pyrrole-h1 and the D and H atom images from pyrrole-d1 recorded in the one-colour photodissociation experiments at λ = 243 nm are displayed in Fig. 2a–c in the upper row.

The respective 3D velocity distributions are given in the second row.

All images show a clearly visible, spatially anisotropic outer ring corresponding to very fast H/D atoms with a sharply peaked speed distribution and a diffuse, spatially isotropic inner region reflecting slower H/D atoms with a broad speed distribution.

With the data for the deuterated molecule, these distinctive dissociation channels can be identified with some confidence.

Reiterating first the conclusions for the case of pyrrole-h1 (reaction (1), Fig. 2a),25 the fast H atoms can be attributed to the direct NH dissociation in the πσ* state (referred to in our earlier paper and in the following as channel A) and the slower H atoms may be ascribed to unimolecular decay reactions after internal conversion (IC) to the S0 state (referred to as channel B).

However, the evidence for the assignment of the fast H atoms to the NH group had so far only been circumstantial.25

The present results provide the missing additional information: In particular, it can be seen from the image for reaction (2d) in Fig. 2b that the sharp ring of fast D atoms from pyrrole-d1, for which the site of the D atom is unambiguous, is even more striking than that of the H atoms from pyrrole-h1 (Fig. 2a).

At the same time, the slow D atom signal from pyrrole-d1 is the by far weakest in the series.

These results demonstrate clearly that the fast D atoms originate from the N position and not from any of the four C positions of the parent molecule.

In contrast, pyrrole-d1 shows the highest intensity of slow H atoms in the series of the three molecules (Fig. 2c, channel B), although there is also a small fraction of fast H atoms from pyrrole-d1.

The latter observation may indicate the existence of a third pathway (referred to in the following as channel C, see discussion).

The relative yields of D and H from pyrrole-d1 were determined by recording a REMPI spectrum (Fig. 3).

From the ratio of the integrated signals, the ND group accounts for ≈(74 ± 5)% of the observed atoms, while the four CH sites account for ≈(26 ± 5)%.

On a per bond basis, CH dissociation is therefore really only a minor process.

In the following, the attention will be focused mainly on the direct dissociation dynamics via channel A.

Translational energy distributions

The total center-of-mass (cm) translational energy distributions P(ET) of the photofragments which follow by integrating the 3D velocity distributions (Fig. 2) over the angular directions and applying linear momentum conservation for the two fragments are given in Fig. 4a–c.

The areas under the experimental curves in the three panels were normalized with respect to each other to reflect the respective branching ratios.

The clearly bimodal distributions nicely confirm the existence of the aforementioned distinctive dissociation channels that were apparent from the images.

In order to determine quantitative data for a comparison of the three isotopomers and their dynamics at the two photoexcitation wavelengths, the P(ET) profiles were fitted by using a sharply peaked Gaussian to represent the contributions of the fast atoms and a simple polynomial to account for the broad background of slow atoms.

As seen, the experimental profiles could be nicely described.

Alternatively, the fast atoms were also modeled using a three-parameter trial function of the form36P(ET) = Eavl[faT(1 – fT)b],where fT is the fraction of the available excess energy (Eavl) channeled into fragment translation.

This flexible fitting function has the advantage that Eavl, which is given by the photon energy minus the dissociation energy of the molecules, can be determined by the fitting.

Apart from this point, however, the differences between using this expression or the Gaussian were negligible.

The parameters from the fits are compiled in Table 1.

In addition, the table lists the relative yields (branching ratios) yi of the different dissociation channels given by the areas under the respective curves.

For reaction (2), the yi values account for the branching ratio between (2d) and (2h) from Fig. 3 so that the total yield is ∑iyi = 1.

Energy balance and determination of the NH/ND bond dissociation energy

The energy balance for the dissociation according to channel A is given by Ehν = D00 + ET + EPyVR – EPVR.Ehν is the photon energy, D00 the NH/ND bond dissociation energy, ET the total translational energy of the two fragments, EPyVR the vibration–rotation energy of the pyrrolyl co-product, and EPVR the internal energy of the parent molecules.

Considering the efficient cooling of the molecules in the molecular beam expansion, the latter term is small and may be neglected.

Thus, the available energy partitioned among the translational and internal (vibration–rotation) degrees of freedom of the fragments is Eavl = Ehν – D00 = ET + EPyVR.Since the fastest H/D atoms may be assumed to correspond to internally cold pyrrolyl radicals (EPyVR ≈ 0), Eavl can be determined from the sharp cut-off of the P(ET) distribution at the allowed maximal value (EmaxT) of ET , eqn. (I).

The data are given in Table 1.

The corresponding values for the bond dissociation energy according to eqn. (III) are

D00(NH) = (31 000 ± 500) cm–1,

D00(ND) = (31 500 ± 500) cm–1.

Furthermore, it can be seen from Table 1 that on average about 67% of the available energy is channeled into relative fragment translation, 33% is deposited as internal energy in the pyrrolyl radical.

Recoil anisotropies

The photofragment angular distributions P(θ) are obtained in principle by integrating the 3D velocity maps in Fig. 2 over the radii.

The resulting distributions can then be fitted using the standard expression37P(θ) = (4π)–1[1 + βP2(cos θ)].θ is the angle between the recoil direction and the laser polarization, P2(cos θ) is the second-order Legendre polynomial, and β is the anisotropy parameter.

In the limiting case of a single dissociation channel resulting from a pure parallel or perpendicular electronic transition, β is known to take the values +2 or –1, respectively, as long as the dissociation is fast compared to the rotation of the molecules.

In the present case, we have an anisotropic (βA < 0) distribution (channel A) above an isotropic (βB = 0) background (channel B).

In order to separate both contributions as well as possible, we integrated the 3D velocity maps by considering only the anisotropic outer shells with ET in the range 5000 cm–1 ≤ ET ≤ 8000 cm–1.

The resulting P(θ) distributions from reactions (1) and (2d) are presented in Fig. 5.

The values for the effective asymmetry parameters (βeff) which were fitted to these data are given in Table 1.

The anisotropy parameters for channel A (βA) listed in Table 1 in the last column were then found by correcting βeff for the relative fractions of the two channels in the integration range.

The data confirm the assumption of a preferentially perpendicular distribution.

λ = 217 nm

H/D velocity maps

The H/D velocity maps obtained after excitation of the 1 1B2 (ππ*) state at λ = 217 nm in the two-colour experiments and the resulting 3D velocity distributions are depicted in Fig. 6a–c.

In principle, the images look similar to those seen before at λ = 243 nm.

In particular, they exhibit again a sharp, anisotropic outer ring reflecting fast H/D atoms with a narrow speed distribution and a broad, structureless inner feature corresponding to much slower atoms.

The sharp rings in the H atom image from reaction (1′) in Fig. 6a and the D atom image from reaction (2d′) in Fig. 6b are again attributed to a fast dissociation of the NH/ND bond (referred to in the following as channel A′).

Since the excited 1 1B2 (ππ*) state does not correlate directly with H/D + pyrrolyl, this dissociation is assumed to proceed via the lower-lying πσ* state, in accordance with the theoretical predictions.1,11

The structureless inner signal is again ascribed to unimolecular decay processes in the S0 state (channel B′).

In addition, pyrrole-d1 also leads to H atoms (Fig. 6c, channels B′ and C′).

A REMPI scan (Fig. 7) gave relative yields of ≈(40 ± 5)% D and ≈(60 ± 5)% H (see discussion).

Translational energy distributions

The total translational energy distributions P(ET) at λ = 217 nm are given in Fig. 8.

The data were fitted in the same way as above; the results are collected in Table 2.

Considered on a per bond basis, the direct fission of the NH/ND bond (channel A′) can be seen to remain a major dissociation pathway.

However, compared to the results at λ = 243 nm, the data at λ = 217 nm show significantly higher yields for channel B′.

In line with the higher photon energy, the P(ET) distributions for both channels A′ and B′ extend to higher translational energies (i.e., higher EmaxT).

At the same time, the distributions are significantly broader.

However, the cut-off of channel A′ at EmaxT is not as well defined at λ = 217 nm than at 243 nm.

Indeed, the increase of the apparent EmaxT is smaller than the increase in the photon energies (ΔEhν = 4900 cm–1).

Furthermore, the distributions at the two wavelengths reach their peaks at practically the same energies (EpeakT).

These findings are reflected by a correspondingly smaller fraction of the available energy funneled into fragment translation (43% at λ = 217 nm compared to 67% at λ = 243 nm).

Thus, in summary, a higher fraction of the excess energy appears to be deposited in internal degrees of freedom of the pyrrolyl radical.

Recoil anisotropies

Despite the greater contributions of channel B′, the images at 217 nm show stronger spatial anisotropy than those at 243 nm.

The angular distributions P(θ) from reactions (1′) and (2d′) are presented in Fig. 9.

The effective asymmetry parameters βeff and the resulting βA′ values are listed in Table 2.

The average value of βA′ = –0.95 ± 0.05 is practically equal to the perpendicular limit of β = –1.


The photodissociation of pyrrole has first been studied by Blank et al. using photofragment translational spectroscopy (PTS).38

The authors performed experiments at two fixed wavelengths, λ = 248 nm and 193 nm.

The reported results for the formation of H atoms at both wavelengths are comparable to those obtained in the present work, although the underlying mechanisms could not be elucidated in detail at the time because of the lack of information on the excited electronic states of pyrrole.

However, the PTS results are of interest for this discussion because of the information on the branching ratios of the ensuing channels.

At λ = 248 nm, H + pyrrolyl were the only observed products.

At 193 nm, the authors also observed substantial amounts (≈50%) of products from ring opening reactions.

The experimental results presented in this paper demonstrate clearly that the photochemical mechanisms of pyrrole in the near UV can only be rationalized by strong interactions between two excited electronic states in the region, 1 1A2 and 1 1B2, and the electronic ground state S0.

The excited states have πσ* and ππ* character, respectively.

The electronic absorption is dominated by the oscillator strength of the 1 1B2 state (f = 0.209).23

The dipole moment for the excitation from S0 lies in the molecular plane and perpendicular to the NH bond.

The 2 1A1 (ππ*) state, which has been calculated only 0.14 eV below 1 1B2, also plays some role (f = 0.036).

In contrast, the calculated oscillator strength of the 1 1A2 state is negligible.

As mentioned, the 1 1A2 state is observed in the UV spectrum only very weakly as a result of vibronic coupling with the neighboring electronically allowed states.

A schematic diagram of the 1 1A2 (πσ*) and 1 1B2 (ππ*) potential energy surfaces is depicted in Fig. 10.

The energy profiles were plotted according to the information given by Sobolewski and Domcke.1,11

Note that the S0 electronic state itself does not correlate directly with H + pyrrolyl in their electronic ground states.

Dynamics of the 1 1A2 (πσ*) state

Direct NH/ND bond fission

The previously reported H atom velocity maps from the photodissociation of pyrrole and N-methylpyrrole at λ = 243 nm gave the first experimental evidence for the repulsive nature of the 1 1A2 excited electronic state of pyrrole with respect to the NH bond.25

The experimental observation of very fast H atoms with a narrow and spatially anisotropic velocity distribution (channel A) from pyrrole, but not from N-methylpyrrole, was consistent with the theoretical predictions1,11 that, as shown in Fig. 10, the NH bond in the first electronic state of pyrrole acquires πσ* character as the bond is stretched.

With the detection of very fast D atoms from the photodissociation of pyrrole-d1 with a similar narrow and spatially anisotropic velocity distribution, the present work provides unambiguous evidence for the proposed mechanism.

Indeed, the fast direct ND bond fission in pyrrole- d1 in the 1 1A2 (πσ*) excited electronic state accounts for 90% of the observed D atoms, or, when the H atoms from dissociation of the CH positions are taken into account, 67% of all channels.

The branching ratio for the direct H detachment in the excited state measured by Blank et al38. is somewhat smaller (47%) than ours (67%).

However, at λ = 248 nm, the molecules are excited just to the electronic origin of the 1 1A2 state.

The difference of the branching ratios at 248 and 243 nm can thus be explained by a trapping of the molecules with the lower excitation energy in the shallow potential energy well of the 1 1A2 state that has been found in the ab initio calculations.1,11

The shapes of the translational energy distributions of the fast H/D atoms in the present work and the study by Blank et al. are similar (see Table 1).

The mode-specific nature of the fast NH/ND dissociation process in the 1 1A2 state can be attributed to the unusual properties of the potential energy surface of the state (Fig. 10).

In particular, the large potential energy gradient along the NH stretching coordinate (πσ*) provides a strong driving force for a fast, direct cleavage of the NH/ND bond.

The conical intersection between the 1 1A2 and S0 states at rNH ≈ 1.85 Å then acts as a funnel, through which the H/D atoms are expelled with very high kinetic energies.

This occurs on a time scale before IC to S0 and intramolecular vibrational energy redistribution (IVR) may compete.

From the spatial anisotropy parameter (β ≈ –0.45, Table 1), we have to conclude in any case that the dissociation lifetime of the excited molecules is comparable to or even shorter than their rotational period.

The electronic excitation is seen to be preferentially perpendicular, as we would expect for a vibronic coupling of the 1 1A2 to the 1 1B2 state.

However, there may also be contribution from vibronic coupling to the 2 1A1 state, which can be reached from S0via a parallel transition.

Thus, the β value could arise from a mixture of the two states.

In the limiting case of a very fast dissociation with β1 = –1 for the perpendicular (1 1B2) and β2 = +2 for the parallel (2 1A1) branch, this would imply a ratio of 82∶18% for the two states.

Incidentally, this is close to the ratio of the oscillator strengths.

Obviously, some caution with these considerations is advisable because of a possible short trapping of the molecules in the 1 1A2 state.

We note nevertheless that in the case of the direct excitation of the 1 1B2 state at λ = 217 nm, we have β = –0.95 ± 0.05 (Table 2), practically equal to the perpendicular limit, even though in this case the excited molecules have to pass through another conical intersection (from 1 1B2 to the πσ* state) before the subsequent dissociation on the πσ* potential surface.

Unimolecular decay after conversion to S0

Considering the data for pyrrole-d1 (reaction (2)), a fraction of 33% of the H/D atoms appears to be produced via different other mechanisms.

IC from the 1 1A2 to the S0 electronic state is the first process that comes to mind.

This radiationless electronic transition is likely mediated by the same conical intersection that is responsible for the fast NH/ND bond dissociation (cf. Fig. 10).1,11

The intersection arises due to a vibrational mode of a2 symmetry, which corresponds to a screwing of the aromatic ring.

H/D atoms can then be produced from the resulting vibrationally hot S0 molecules via a variety of unimolecular decay processes (channel B).

The unimolecular dissociation of pyrrole in the electronic ground state has been studied experimentally by shock tube pyrolysis measurements39,40 and theoretically by quantum chemical techniques and chemical kinetic modeling.40–44

The major decomposition products were found to be HCN, methylacetylene, and acetylene, besides numerous other products appearing in smaller quantities.

The complete mechanism that has been developed includes ≈200 elementary reactions so that a detailed discussion exceeds the scope of the present paper.

The lowest energy dissociation channel was found to proceed via an H atom migration to pyrrolenine (2H-pyrrole) followed by decomposition to HCN and CH3CCH.39,40,43,44

From the measurements of Blank et al38. who probed the molecules at almost the same excitation energy (λ = 248 nm), we know that these products do not seem play a role here.

There is also a possibility for an isomerization to cis-crotonitrile or vinylacetonitrile.

H + pyrrolyl are formed preferentially by the dissociation of the CH2 group in the pyrrolenine intermediate rather than by direct NH bond fission.43

In this case, between one third and one half of the H atoms should originate from the NH position, the other two thirds to one half should originate from the CH positions, in line with the experimental results for reaction (2).

In addition, there are several other pathways to H atoms, including some that are connected with ring opening processes.

These atoms are expected to be produced with, essentially, statistical translational energy distributions.

Since the complex sequential unimolecular reactions of the molecules in the S0 state are not the main topic at this Discussion on non-adiabatic effects, we will not consider them in further detail.

CH dissociation in the 1 1A2 state

Surprisingly, the photodissociation of pyrrole-d1 also seems to produce a small amount (≈10%) of fast H atoms (referred to above as channel C) with a similar velocity distribution as the D atoms from channel A (Table 1).

While it is difficult to rule out an experimental artifact by a D/H exchange in the molecules in the gas supply line between the reservoir and the molecular beam valve, it is noted that the results were reproducible from day to day and did not seem to depend on the length of time the system was treated with D2O.

A similar channel has also been noted by Blank et al.38

Thus, we are tempted to postulate either a small contribution from a direct dissociation of one of the CH bonds in the excited 1 1A2 state or a fast partial intramolecular exchange in the excited state before the NH dissociation on the repulsive 1 1A2 potential energy surface.

Both processes should be investigated by additional theoretical calculations.

However, a fast D/H isotope exchange would also explain the observations.

NH dissociation energy

The NH bond dissociation energy of the pyrrole molecule is of fundamental importance for estimating the thermal stability of the molecule in combustion processes.

The present study can be considered as the most direct experimental determination to date.

The value of D00(NH) = (31 000 ± 500) cm–1 is in excellent agreement with that given by Blank et al.38

(D00 = 30 800 cm–1) and slightly lower that the ab initio quantum chemical estimate of 33 180 cm–1 reported by Bacskay et al.42

The discrepancy to our earlier report25 stems from the recalibration of the velocity map images based on the photodissociation HBr and HI.

The difference between the values of D00(NH) and D00(ND) is somewhat smaller than the expected isotope effect and not unreasonable considering the error limits.

The accuracy of the present data is limited essentially only by the resolution of the images.

Dynamics of the 1 1B2 (ππ*) state

NH/ND predissociation via the 1 1A2 state

The experimental results after excitation of the 1 1B2 (ππ*) state at λ = 217 nm show that a large fraction of the excited molecules decay again by a fast detachment of the H atom from the NH position, although the excited state is bound with respect to the NH coordinate (Fig. 10).

As shown in Fig. 10, the geometry-optimized 1 1B2 (ππ*) state lies slightly above the optimized 1 1A2 (πσ*) state.

In a multi-dimensional picture, both are connected by a surface of intersection.

Vibrational modes of b1 symmetry lift the degeneracies and lead to a conical intersection that can promote an efficient ultrafast radiationless transition from the 1 1B2 to the 1 1A2 surface.11,1

The subsequent dynamics is thus controlled by the πσ* potential surface.

In particular, the πσ* nature of the state leads to the same ultrafast NH bond dissociation and formation of H + pyrrolyl as observed before at λ = 243 nm.

Thus the experimental results for the observed channel A′ can be rationalized by a fast predissociation of the 1 1B2.

From the anisotropy parameter of β = –0.95, we have to conclude that this occurs on an ultrafast timescale (τ ≲ 300 fs).

The translational energy distributions resulting from the predissociation of the 1 1B2 state at λ = 217 nm are remarkably similar to those observed from the 1 1A2 state at λ = 243 nm.

In particular, the peak energies (EpeakT) are practically identical, despite of the higher photon energy at 217 nm, which increases Eavl by ≈4900 cm–1.

This picture is extended by the results at λ = 193 nm,38 where the addition of another 5700 cm–1 to Eavl raises EpeakT by just a few hundred cm–1.

The only effect of the increase of Eavl at the higher photon energies is a slight increase of the widths of the P(ET) distributions.

As suggested by Fig. 10, only the NH stretching vibration can lead to a high translational energy release.

Obviously, the extra energy cannot be released as translational energy and remains in the pyrrolyl radical as internal energy.

The almost identical EpeakT values at the different excitation wavelengths thus show that the coupling between the reaction coordinate (NH stretching motion) and the other nuclear coordinates during the fast decay on the 1 1A2 surface is almost negligible.

The geometries of the molecule at the conical intersection between the 1 1B2 and the 1 1A2 state are suggested to be very different compared to the equilibrium configuration.

Unimolecular decay after internal conversion to S0

As in the experiments at λ = 243 nm, the formation of the slow H/D atoms with a broad translational energy distribution (channel B′) is attributed to an electronic relaxation to the S0 state followed by unimolecular dissociation.

A comparison of Tables 1 and 2 shows that this pathway is more important at λ = 217 nm than at 243 nm.

The higher fraction of channel B′ can be rationalized by two decay processes, direct relaxation from the 1 1B2 to the S0 state and relaxation via the 1 1A2 state.

The additional internal energy from the higher photon energy may be responsible for the more efficient relaxation to S0.

It also leads to a faster unimolecular decomposition of the S0 state.

Products from ring opening reactions have not been detected in the present work.

At λ = 193 nm, however, Blank et al. observed ≈50% of HCN + C3H4.38

In addition, there is again a small fraction of fast H atoms from the photodissociation of pyrrole-d1 (see above).


The photodissociation of pyrrole in the 1 1A2 (πσ*) first excited electronic state at λ = 243 nm and in the higher lying 1 1B2 (ππ*) electronic state at λ = 217 nm has been studied by photofragment velocity map imaging.

The results show that the dynamics differ strikingly from the unimolecular decomposition of the molecules in the electronic ground state.

The 1 1A2 state decays mainly via fast mode-specific NH bond fission, which is driven by the repulsive nature of the πσ* potential energy surface along the NH stretching coordinate.

The H atoms are therefore eliminated with very high kinetic energy close to the maximum available energy.

In addition, the 1 1A2 state can undergo a fast IC to the electronic ground state followed by unimolecular decay of the vibrationally highly excited S0 molecules.

The 1 1B2 state is predissociated by the πσ* state.

Thus, it exhibits the same fast mode-specific NH bond fission channel.

However, IC to S0 and subsequent unimolecular decomposition plays a larger role.

The highly efficient radiationless electronic transitions from the ππ* to the πσ* state and from the πσ* to the S0 state are mediated by conical intersections between the respective potential energy surfaces.

From the strongly anisotropic angular H atom distributions, the lifetimes of the excited electronic states are short compared to the rotational period of the molecules.

Minor additional pathways which need further investigation may be a direct CH dissociation and/or a fast exchange between the NH and CH positions in the excited electronic state(s).

The radiationless decay mechanisms established in the present work for pyrrole are of substantial interest for the ongoing discussion of the photochemical dynamics of the nucleic acid bases.

In particlar, the purine bases are known to show a low-lying threshold, above which the excited states are abruptly quenched.2–6

It is currently a matter of controversy whether this quenching is due to a ππ*–nπ* coupling2,3 or to the suggested ultrafast NH/OH predissociation via the πσ* mechanism.1

The πσ* mechanism may also play a role in excited state intramolecular hydrogen transfer (ESIHT) processes.7

Furthermore, it has even been proposed to lead a fast NC bond fission in electronically excited N-methylpyrrole45.