Forschung / Jahresbericht 1996 / Feature-Artikel

Ultrafast Ionisation, Rearrangement and Fragmentation Processes in Ammonia Molecules and Clusters

W. Radloff and I.V. Hertel, V. Stert, Th. Freudenberg, K. Weyers, H.-H. Ritze

1. Introduction

The quest to fully understand the dynamics of chemical reactions in wave mechanical terms may  be considered the ultimate goal of all low energy gas phase physics (U. Fano). This is in particular true for all studies of free molecules and clusters. On the other hand, the selective making and breaking of molecular bonds and the control of chemical reactions by light may be seen as the ultimate technical goal of laser induced photochemistry. However, only in rare cases -if ever- 'conventional' laser chemistry has so far succeeded to initiate specific chemical reactions by selecting the wavelength so that a particular bond is activated (bond specificity). The reason is simply that the internal energy flow within the molecular reactants is extremely fast, typically much faster than the excitation rate and the duration of the laser pulses used to prepare the molecules for the reaction. Thus the photon energy initially deposited into specific bonds of the reactants is quickly randomised over all degrees of freedom so that only an unspecific heating of the molecules is the net result of the photoexcitation process - an effect which may be achieved by much cheaper means than laser technology. Here the potential of femtosecond technology (FST) in principle offers a completely new perspective by applying ultrafast laser pulses to trigger a chemical reaction. The various concepts of femtochemistry which have been discussed over the past few years [1] all involve a series of laser pulses (pump and probe, pump and control, coherent control etc.) with judiciously chosen parameters (pulselengths, time sequence, wavelengths, intensities) designed to control the ultrafast flow of energy and the particle motion within the reactants in real time so that specific reactive channels are opened before dissipation can occur.

A prerequisite for any successful approach to this kind of femtochemistry is, however, a detailed understanding of the photon induced dynamics in molecules and clusters, at least for a few prototype systems, which typically occurs on the pico- and femtosecond time scale. Molecular clusters, in particular, offer the possibility to 'construct' a relatively well defined microsystem with specific model properties for the study of simple intra- and intermolecular reactions. Ideally, to investigate the competition of particle rearrangement, reactive processes, photoionisation and fragmentation one needs a whole toolbox full of powerful, flexible femtosecond photon sources and measuring devices, ranging from the VUV (to excite higher lying states in realistic molecules or to probe the reactants by unambiguous single photon ionisation) via the visible (i.e. to rise the energy above a critical ionisation or fragmentation threshold) to the near and middle infrared (which allows one to characterise or excite specific vibrational motions). Such tools are being developed at the MBI, in particular in department A3, and are, at least in principle, available in the femtosecond laser application laboratories for internal and external users.

Ammonia molecules and clusters are specifically attractive in this context, being prototype polar solvent molecules with relatively well known structures and energetics. One line of research is based on seeding these species with a chromophore metal atom which gives access to ionisation with convenient laser sources in the visible and allows dynamical studies of metal solvation phenomena. We have reported last year about progress in this field. The present article is focused onto pure ammonia clusters and the ammonia molecule itself. The inclination of neutral ammonia clusters towards formation of hydrogen bonds and the well known preference of the ionic species for protonation as well as the accessibility of their excited states for multiphoton excitation with convenient laser sources has lead to a renewed world-wide interest in these systems when advanced femtosecond laser systems became available. We have chosen to attack the hitherto only poorly understood competition between ionisation, internal protonation and fragmentation which occurs in electronically excited states of NH₃ and (NH₃)_n.

Exploiting the UV and VUV femtosecond photon sources available at the MBI we are able to directly excite vibronic energy levels above the high lying electronic states states of the ammonia molecule (and their counterparts in the clusters) by a single photon. As illustrated in figure 1, we are able to use a reasonably straight forward two photon (sometimes three photon) pump-probe ionisation scheme to study the processes occurring in the electronically excited states in real time. As well known, the ionic mass spectra observed for the ammonia clusters are dominated by protonated species (NH₃)_n-2NH₄⁺ while the unprotonated mass peaks (NH₃)_n⁺ are only very weak. The protonated cluster ions are formed by NH₂ loss either from (NH₃)_n⁺ after the ionisation process (so called absorption-ionisation-dissociation mechanism, AID) or by fragmentation of the excited clusters (NH₃)_n* followed by ionisation (ADI mechanism). The pump-probe technique using fs laser pulses can, in principle, distinguish these two channels: At zero delay times only the AID process is possible whereas at delay times larger than the lifetime of the excited (NH₃)_n* clusters the ion signal observed arises from an ADI process.

Fig. 1: Energetics of electronically excited states of ammonia and ammonia clusters and schematics of the pump-probe scheme to study the ionisation and fragmentation dynamics.

Recent femtosecond pump-probe experiments [2,  3] of ammonia clusters excited to low vibrational levels of the à state have reported lifetimes for these excited states on the order of a few 100 fs. Unfortunately the pulse length of 350 fs and 290 fs, respectively, as well as the neglect of unprotonated ammonia clusters, limits the insight obtainable from these studies. We were able to use significantly shorter laser pulses of 160 fs and a more sophisticated model analysis of the time evolution of the ammonia molecule and the clusters excited to the à state. Additional information is gleaned from a systematic investigation of the deuterated species.

So far, no electronic states above the à state have been studied before by single photon excitation. In fact, the Castleman group [4] has reported a multi-photon experiment on the   state of (NH₃)_n which in the light of our present results very nicely illustrates that multiphoton excitation (in this case 6 red photons were needed) with femtosecond laser pulses is a very dangerous undertaking and one is easily lead to mistake the optical excitation dynamics as genuine molecular dynamics. In contrast, using fs laser pulses in the VUV spectral range we were able for the first time to excite ammonia molecules and clusters to the higher electronic state levels directly by one photon absorption (see fig. 1). Very interestingly, we have observed a quite different dynamical behaviour in these higher states in comparison to that seen when exciting the à state.

2. Experiment

The ammonia clusters were formed in a supersonic expansion of ammonia (5  % NH₃ or 5  % ND₃) seeded in He gas (2 bar) through a pulsed conical nozzle. After passing a skimmer (1 mm) the cluster beam travels through the ionisation region of a Wiley-Mc Laren time-of-flight mass spectrometer. Here the clusters interact with the fs laser pulses of two weakly focused copropagating laser beams. The clusters are excited with the pump pulse to energy levels corresponding to the different electronic states of the ammonia monomer and ionised with a probe pulse of a suitable photon energy. By varying the time delay t between pump and probe pulse the dynamics of the processes starting from the excited cluster states can be followed. The ion signals are detected by a micro-channel plate detector in the time-of-flight mass spectrometer, digitized by a fast digital oscilloscope and stored and processed by a PC.

The fourth harmonic (l₁  = 200 nm) of a commercial Ti:sapphire laser and amplifier system tuned to 800 nm was used to excite ammonia molecules and clusters to levels above the à state. A second part of the laser beam is frequency doubled (l₂  = 400 nm) or tripled (l₂  =  267 nm) and is used to probe the excited species. The pulse duration t_L of typical 160 fs for the pump and the probe beam is controlled permanently during the experiments. The energy densities in both beams are restricted to such low levels, that single photon absorption dominates for each laser beam.

In a second set of experiments, the states were excited by fs laser pulses with a wavelength of l_pu  = 155 nm (8 eV). The experimental scheme to generate these pulses by near resonant four-wave difference frequency

Fig. 2: Experimental scheme to prepare the VUV pulses by four wave difference frequency mixing [from ref. 5].

mixing in Ar (0.8 bar) in a gas cell is shown in Figure 2. In the gas cell two ultrashort pulses of wavelengths l_p  = 193.5 nm and l_i  = 258 nm are superposed co-linearly and yield a wavelength of l_s  =  155 nm, the photon energy being given by hn_s = 2  hn_p - hn_i [5]. The energy of the VUV pump pulses in the interaction zone of the TOF mass spectrometer is on the order of 100 nJ. The pulse length was estimated to be about 500 fs. As probe pulses usually the fundamental wave l_f =  774 nm (1.6 eV) with a pulse duration of about 80 fs is used (Fig. 2). In this case the ionisation of the ammonia molecule requires the absorption of two probe photons (Fig. 1). Thus, for comparison the third harmonic at l_pr  = 258 nm (4.81 eV) has been used as probe pulse in some experiments, too.

3. Results and discussion

3.1 Excitation to the à state

In Figure 3 a typical mass spectrum is shown as obtained in pump-probe experiments with 160 fs laser pulses (pump wavelength l₁  = 200 nm, 6.2 eV; probe wavelength l₂  = 400 nm, 3.1 eV). Generally, the unprotonated cluster signals are much smaller than the corresponding protonated ion signals. The increase of mass peaks from n  = 2 to n  = 5 reflects the increasing ionisation efficiency for the larger (NH₃)_n clusters as the ionisation potential drops significantly below the total photon energy of 9.3 eV. The characteristic decrease of the (NH₃)_n-1NH₄⁺ signals above n  = 5 reflects the well known closure of the first solvation shell where the NH₄⁺ cation is surrounded by four NH₃ molecules.

Fig. 3: Typical mass spectrum of ammonia clusters obtained in a pump-probe experiment via the à state with 160  fs laser pulses of l1  =  200  nm (pump) and l2  = 400  nm (probe).

The measured pump-probe delay scans are shown in Fig. 4 for the ammonia molecule and in Fig. 5 for the ammonia clusters. The theoretical fit to the monomer time evolution (Fig. 4) reveals a lifetime of about 40 fs assuming sech²(t) pulse shapes for both the pump and the probe pulse. The zero delay time is defined by the position of the curve maximum because the ammonia molecule was excited non resonantly at the given pump wavelength. For comparison the benzene molecule with approximately the same lifetime of the OS₂ state shows a significant shift of the curve maximum (see Fig. 4) because it is excited resonantly at the pump wavelength.

Fig. 4: Pump-probe delay scan for (upper part) NH3+ after non-resonant excitation of the NH3 Ã-state and (lower part) for the benzene ion after excitation of the OS2 state.

The zero delay time defined in Fig. 4 is identical to that value which is obtained by our model fits for the ammonia clusters excited resonantly (Fig. 5).

Fig. 5: Pump-probe delay scan for (NH3)n-2NH4+ (left) and for (NH3)n+ (right) after excitation of the (NH3)n Ã-state.

The details of the fit procedure are described in [6,  7]. The model is characterised schematically in Fig. 6 for the example of the ammonia trimer. The pump process from the ground state (1) into the à state (2) is described as a coherent interaction with the laser using optical Bloch equations and an exponential decay with the lifetime t₂₀. This decay rate is determined by the main loss channel X and rapid transitions to the rearranged state (3) as well as to the radical state (4) formed by NH₂ loss. Three rate equations (for states 3, 4, and 5) describe the rearrangement and fragmentation processes. The state (3) with the total lifetime t₃₀ may also fragment into the main loss channel X or evaporate an ammonia molecule to form the stabilised cluster state (5).

Fig. 6: Schematic illustration of the 5 state model used to fit the time evolution of the ammonia cluster ion signals as shown in [6,  7].

The lifetimes t₂₀ obtained by the fits for the protonated and the corresponding unprotonated ion signal curve are identical documenting that they arise from the same excited state. While t₂₀ increases slowly with cluster size as indicated in the figure, the lifetime t₃₀= 2 ps is found to be nearly constant for the clusters represented in Fig. 5. At long delay times (i. e. t  >= 6 ps) the cluster ion signals become nearly constant due to the long lifetime of the radicals (NH₃)_n-2NH₄ (left) and the stabilised 'internally protonated' fragments (NH₃)_n-2NH₂NH₄ in the state (5) (right). The protonated ion signals (NH₃)_n-2NH₄⁺ vanish at long delay times for n  =  3, 4 because the probe photon energy of 3.1 eV remains below the ionisation potentials of the corresponding radicals while for n  =  5 it exceeds the ionisation potential of (NH₃)₃NH₄. For the unprotonated ion signals a finite tail at long delay times was already found for (NH₃)₄⁺ which results from ionisation of the stabilised species (NH₃)₂NH₂NH₄ obtained by NH₃ evaporation of (NH₃)₃NH₂NH₄*.

The model used allows us to fit the measured delay curves for protonated ammonia cluster ions as well as for unprotonated ions consistently, demonstrating that the dynamics of the clusters excited to the à state is described reasonably well by this model. The lifetime of the primarily excited cluster states (which decay by ultrafast fragmentation and rearrangement) increases slowly with the cluster size starting with t₂₀  = 100 fs for (NH₃)₂ and t₂  = 200 fs for O(NH₃)₃ (cf. [6]).

3.2 Excitation to the B^~ and C^~' state

A surprisingly different dynamics was observed when ammonia molecules or clusters were excited to the electronic state. Figure 7 shows the time dependent ion signals for the two ammonia isotopes NH₃ and ND₃ excited by 8.05 eV photons. Note the different time scales for NH₃ and ND₃.

Fig. 7: Pump-probe delay scan for NH3+ (top) and for ND3+ (bottom) after excitation of the B~ or C~' state of NH3.

The decay of each ammonia isotope as a function of the delay time t can be fitted by a single exponential with a lifetime t_B  = 7.6 ps for NH₃ and t_B  = 71 ps for ND₃: a dramatic isotope effect. In comparison to the à state much longer lifetimes are obtained, thus, a completely different mechanism must be responsible for the decay of the electronically excited B^~ or C^~' state in ammonia. As suggested in recent studies (e.g. [8]) and supported by own preliminary theoretical estimates these lifetimes may be understood as being due to internal conversion from excited vibronic levels of the B^~ state to the à state followed by the ultrafast fragmentation of the à state discussed above. The difference between the lifetimes of NH₃ and ND₃ is then caused mainly by the higher reduced mass of ND₃ and its smaller vibrational frequencies.

Fig. 8: Pump-probe delay scan for (ND3)nND4+ after excitation of the B~ or C~' state of (ND3)n.

In contrast to the monomers the lifetimes of the clusters in vibronic levels of the B^~ or C^~' state are much shorter as can be seen in Fig. 8 for (ND₃)_n. The fit procedure requires the determination of the cross correlation function for the ultrafast pulses at t_pu  = 155 nm and t_pr  = 774 nm, which is non trivial in this wavelength region. We have used for this purpose the pump-probe delay scan of benzene ions which were obtained by resonant two-color two-photon ionisation of benzene molecules added to our gas mixture. Due to the very short lifetime of about 50 fs for the highly excited benzene molecule (at 8.05 eV in the region of the   G^~ state) the measured pump-probe delay curve represents the required cross correlation function in a very good approximation. The pulse durations obtained are t_L ~ 500 fs and t_L  = 80 fs for the pump and probe pulse, respectively (cf. section 2). Assuming sech²(t) shaped pulses with these half widths and using optical Bloch equations with coupled rate equations as described above we have fitted the measured curves in Fig. 8 (cf. [9]). In contrast to the à state, the lifetime t_B of the excited cluster states decreases with increasing (ND₃)_n cluster size from 570 fs for n = 3 to 400 fs for n  =  6 (with an uncertainty of about 50 fs). A quantitative interpretation of the size dependent lifetime based on the internal conversion from the initially excited state levels to the à state of the clusters is a rather complex matter. It would require detailed spectroscopic information on the clusters especially with respect to the vibrational modes of the highly excited cluster states. A qualitative approach to the problem may start by assuming a nearly resonant coupling from the excited state levels to highly excited vibrational levels of the à state which decay very fast with the rate t_A^-1. It is then plausible that the transfer rate from may increase with the lifetime t_A of the à state. Since the lifetime t_A was found to increase with the ammonia cluster size (see above) the coupling with the state increases and, hence, the lifetime t_B is expected to decrease with the cluster size just as observed (Fig. 8). Why the lifetimes for the monomers and the clusters are so vastly different certainly warrants a thorough theoretical analysis.

5. Conclusions

Ammonia clusters are interesting model systems for studying key reactions such as proton transfer in polar molecular systems. We have illustrated that the use of femtosecond laser pulses in the UV and VUV allows one to study the dynamics of the relevant highly excited electronic states in real time with great detail. Different electronic states are shown to behave very differently. These systems thus provide an interesting insight into different mechanisms of decay and rearrangement and will be a stimulating challenge for ab initio molecular dynamic studies.

Acknowledgement

We thank the Deutsche Forschungsgemeinschaft for support through Sonderforschungsbereich 337 'Energy and Charge Transfer in Molecular Aggregates' co-ordinated by the Freie Universität Berlin. We also wish to thank the laser group of the MBI, in particular Drs. Noack, Korn, Ringling and Güdde for continuous help by providing the laser sources over a wide spectral range.

References

[1]	See e.g. Femtochemistry, J. Manz and L. Wöste eds., VCH-Verlag, Berlin, 1993, as well as M. Chergui ed., Lausanne 1995, World Scient. Publ. Co, 1996
[2]	J. Purnell, S. Wei, S.A. Buzza, A.W. Castleman Jr., J. Phys. Chem. 97 (1993) 12530
[3]	K. Fuke, R. Takasu, Bull. Chem. Soc. Jpn. 68 (1995) 3309
[4]	S. Wei, J. Purnell, S.A. Buzza, R.J. Stanley, A.W. Castleman Jr., J. Chem. Phys. 97 (1992) 9480
[5]	O. Kittelmann, J. Ringling, G. Korn. A. Nazarkin, I.V. Hertel, Opt. Lett. 21 (1996) 1159
[6]	Th. Freudenberg, W. Radloff, H.-H. Ritze, V. Stert, K. Weyers, F. Noack, I.V. Hertel, Z. Phys. D 36 (1996) 349
[7]	Th. Freudenberg, W. Radloff, H.-H. Ritze, V. Stert, F. Noack, I.V. Hertel, Z. Phys. D, submitted (1997)
[8]	M.R. Dobber, W.J. Buma, C.A. de Lange, J. Phys. Chem. 99 (1995) 1671
[9]	Th. Freudenberg, V. Stert, W. Radloff, J. Ringling, J. Güdde, I.V. Hertel, Chem. Phys. Lett., submitted (1996)

Zu den Feature-Artikeln

Zum Jahresbericht 1996