Chemical Reactivity in Microscopic Systems - American Chemical

Aug 1, 1995 - S. Martrenchard-Barra,* C. Dedonder-Lardeux, C. Jouvet, U. Rockland2 and ... the NH3 molecule and the n system of the benzene ring leadi...
1 downloads 0 Views 2MB Size
J. Phys. Chem. 1995,99, 13716-13730

13716

Chemical Reactivity in Microscopic Systems: The Nucleophilic Substitution of Halogenated Benzene Ions by Ammonia Studied in Molecular Clusters S. Martrenchard-Barra,* C. Dedonder-Lardeux, C. Jouvet, U. Rockland2 and D. Solgadi Laboratoire de Photophysique Mol&culaire du CNRS, Batiment 213, 91405 Orsay, France Received: February 27, 1995; In Final Form: May 11, 1 9 9 9

The reactivity of small dihalogenated benzene ions clustered with ammonia has been investigated through mass-selected two-color two-photon ionization. Two reaction channels are evidenced: halogen elimination and hydrogen halide elimination. Appearance thresholds of the different reactive paths have been measured as well as the kinetics of the reactions. From these measurements, a new reactive scheme is proposed.

Introduction The study of chemical reactions within small molecular clusters provides a way to investigate the effect of solvation on bimolecular chemistry and thus to substantiate the comparison between dynamics in isolated molecules and in condensed phase. When a chemical reaction occurs in solution, several processes can happen simultaneously, and it may be difficult to discriminate between the reactive process itself and the environmental effects. A contrario, a size-selected cluster prepared in a supersonic expansion with a well-defined geometry, that is, without diffusion contribution, makes possible the study of a chemical reaction on the microscopic scale.' Binary reactions have been studied in a very detailed manner, in the case of excited state neutral reaction, starting from 1-1 van der Waals complexes excited with a laser to the reactive potential energy surface. A few examples have shown that the intermediate state as well as the entrance barrier2 of the reaction can be evidenced and that the real time evolution of such systems can be ~ b s e r v e d . ~ In complexes or clusters reacting in the ionic states, the parameters of the reaction are not so well-defined as for neutral reactions: the energy of the reactive system is not fully determined by the optical excitation, because the energy is shared between the ion and the outgoing electron. However, we show that barrier heights as well as time evolution of a chemical reaction can be measured for ion-molecule reactions. The ionic nucleophilic substitution has been extensively studied by Brutschy et al.4-7 and Mikami et on several halogenated benzene derivatives: chlorobenzene (ClB),5*63899 fluorobenzene (FB),4.5and fluorochlorobenzene (FClB)7 clustered with (NH3)n have been studied through one-color twophoton ionization coupled to mass spectrometry. Resonanceenhanced two-photon optical ionization presents the main advantage, as compared to electron impact ionization, that the spectroscopy of the SO SItransition is a fingerprint of a given cluster size. The comparison between spectra recorded at masses corresponding either to product ions or to parent ions enables the assignment of different products to their initial reactive cluster parent(s) ion(s). Very spectacular results have been obtained on halogenated benzene cations substituted by ammonia, where different reaction products are selectively observed, depending on the nature of the halogen atoms, their position (in the case of disubstituted benzene), and the number of ammonia molecules.' ~

1

.

~

9

~

-

'Institut fur Physikal und Theoritikal Chemie, Freie Universitat Berlin, Takustrasse 3, 14195 Berlin, Germany. @Abstractpublished in Advance ACS Absrracrs, August 1, 1995. 0022-365419512099- 13716$09.00/0

For fluorochlorobenzene, three reactive channels have been observed simultaneouslyor selectively, depending on the cluster size:

(a)chlorine elimination leading to the formation of fluoroanilinium: F-C6&-NHs+ F

F

CI

NH3'

Cp) HC1 elimination leading to the formation of fluoroaniline+: F-C6b-NH2+

0

+

+

NH3-@

CI

HCI

NHz

( y ) HF elimination leading to the formation of chloroaniline+: c1- c6&-N&+

CI

Ci

Fluoroanilinium (Fani1inium)-the a channel-is the only product observed in the gas phase experiments of Tholmann and Griitzmacher.'o,'' This reaction channel has been modeled by Shaik and P ~ O S S as'an ~ electron transfer mechanism between the NH3 molecule and the n system of the benzene ring leading, through an entrance barrier, to the formation of an intermediate addition complex: the (T complex.

F

The reaction ends with the rupture of the C-Cl bond in the intermediate (T complex to produce fluoroanilinium and C1'. In this model the barrier height is the limiting parameter to the reaction efficiency. This barrier height depends on several factors such as the difference in ionization potential between the aromatic molecule and the nucleophilic partner. For a series of position isomers, the barrier increases as the dipole moment 0 1995 American Chemical Society

J. Phys. Chem., Vol. 99, No. 37, 1995 13717

Chemical Reactivity in Microscopic Systems

TABLE 1: Summary of the Products Obtained in Jet Experiments for XYB-N&+ Complexes (X, Y = F, C1, or Br)' reactants products obsd in jet efficiency (gas phase) ( % ) b AH(cm-I) C1B+-NH3b anilinium + Cl' 13 - I925 FB+-NH3' nonreactive 7790 pFC1B+-NH3d p-fluoroanilinium + C1' 2.8 -750 p-fluoroaniline++ HC1 -7700 pFClB+-(NH3), p-fluoroaniline(NH3)x++ HCI + (n - x - 1)NH3 p-chloroaniline(NH3).+ HF (n - x - I)NH3 mFCIB+-NH3d m-fluoroanilinium+ C1' 10 -2500 m-fluoroaniline+ HC1 -8800 mFCIBt -(NH& m-chloroaniline(NH3). + HF + (n - x - 1)NH3 pFBrB+-NH3e p-fluoroanilinium + BI" 3.4 a Gas phase efficiencies are given for the a channel giving anilinium.'OJ' In the last column are listed the thermodynamical reaction enthalpies, when they are known.24 Reference 5. References 4 and 5 . Reference 7. e This work. f Reference 10.

+ +

+

of the halogenated benzene decreases in agreement with the experiments. Indeed, the transition state corresponds to an ionic NH3+ interacting with a neutral triplet aromatic with a chargedipole stabilizing interaction. If the mechanism for CY elimination seems to be well understood, the reactive mechanism leading to hydrogen halide elimination @ and y channels) is not so well documented. Several questions are still open: are there different intermediates for the different reaction paths, or is there a single intermediate leading to aniline+ and to anilinium? How can the different channels be simultaneously observed? In this work we have tried to gain deeper insight into these reaction mechanisms by studying the energetics and the dynamics of p-fluorochlorobenzene (pFC1B) and m-fluorochlorobenzene (mFClB) clustered with ammonia using two-color processes to obtain new results. Resonant two-photon two-color ionization experiments have been carried out in which a first photon hvl excites the complex SO S Ttransition, and the second photon hv2 of a different color ionizes it, allowing variation of the internal energy content in the reactive ion and the measurement of the appearance thresholds for the different reaction products. hmp/probe experiments have been set up to measure the reaction rates of various channels: one-color two-photon ionization of the cluster is followed by another color two-photon absorption, either of the parent ion or of the product ions, toward a dissociative ionic state leading to secondary fragments. Reaction rates are measured by looking at the competition between reactive channels and second laser absorption as a function of the delay between the two lasers. A few two-color experiments have also been performed to gain insight into the reactivity of 1-2 complex for p-fluorochlorobenzene (pFClB). Additional results have been obtained for the 1-1 p fluorobromobenzene (pFBrB)-ammonia complex, and some experiments complementary to those of Brutschy and Mikami's groups by MPI and those of Grover et al.I3 by synchrotron radiation have been undertaken on chlorobenzene (ClB)-NH3 complex. The reaction channels observed in this work and by Brutschy' as well as the reaction efficiencies measured in gas phase experimentsT0are presented in Table 1. The results of this work can be summarized as follows: 1. We have shown that the simultaneous appearance of the two reactive channels, X' and HX elimination, in the MPI experiments is due to energetic reasons: the ionization process leads to a large internal energy distribution within the ionic complex. 2. Using two-color ionization methods, the thresholds of the different reactive channels have been determined and it has been shown that the anilinef channel is observed only when the

-

"case 1"

-5000

F+iline*+HCl (-7700cm-')

-6000

reaction coordinate

Figure 1. Reactive potential energy surface proposed in the frame of model 1 (two disconnected channels giving aniline' and anilinium). The diagram is built with the threshold values measured for pFClBNH3 complexes. anilinium channel is energetically closed, which can be only achieved in complexes. For the fluorochlorobenzene derivatives, the anilinium channel (C1' elimination) opens at higher energies than the aniline+ channel (HC1 elimination),which then disappears, while for fluorobromobenzenethe anilinium channel (Br' elimination) is the only one observed. 3. For fluorochlorobenzene derivatives, long reaction times (microsecond time scale) have been measured on the Faniline+ channel, which opens at lower energies, whereas reaction times on the nanosecond time scale have been characterized on the Fanilinium channel, which is energetically higher. From these experimental results, two reaction mechanisms that can rationalize the data are discussed. One can sketch two potential energy surfaces as shown in Figures 1 and 2. Case 1 is represented in Figure 1 and assumes that the two reactions leading to anilinium X' (the a channel) and aniline+ HX (the /? channel), respectively, are disconnected. The starting point is the same precursorthe 1-1 complex ion-but the reactions proceed further under different paths. This point of view has been proposed by Brutschy et al. in ref 7. X' elimination after an entrance barrier proceeds through a bound u complex. HX elimination follows another route through another transition state such as the one suggested by Brutschy. As in the model proposed by Shaik and Pross for X' elimination,I2there is an electron transfer from NH3 to XB+. However, the accepting orbital is, in this case, the a* of the C-X bond instead of the n system of the benzene ring as in the u complex. Under this hypothesis, the transition state would be phenyl+X--NH3+. In a concerted manner, the reaction would end by (a) formation of a NH2' radical and H+ (which will be easier if the cluster involves more than one ammonia molecule) and

+

+

Martrenchard-Barra et al.

13718 J. Phys. Chem., Vol. 99, No. 37, I995 “case 2”

E h

.8 I

h

. d

2 -3000

-

ucomplex

-4000

I

100

a,

Faniline’tHCI (-7700cm-’)

I 20

:

140

MS

-5000

reaction coordinate Figure 2. Reactive potential energy surface proposed in the frame of model 2 (same u complex leading to the two products aniline’ and anilinium). The diagram is built with the threshold values measured for pFCIB-NH3 complexes.

100

attack of the phenyl+ moiety by this radical and (b) breaking of the C-X bond and formation of H+ -t X- = HX. This mechanism, which is not observed in the gas phase, is supposed to be less efficient than the simple halogen elimination. Case 2 is presented in Figure 2 and assumes that the two reactive channels proceed through the same intermediate 0 complex, and thus present the same entrance barrier. The separation of the final products takes place in the exit channel. In the anilinium case, the reaction ends with the departure of the chlorine or bromine atom through a simple rupture of the C-X bond. Below the anilinium thermodynamical threshold, the reaction proceeds through aniline+ formation (HX elimination) through a second late barrier. The experiments performed in this work were meant to measure the barriers and reaction times in order to try to discriminate between these two mechanisms.

Experimental Section The experimental setup consists of a skimmed supersonic jet coupled with a reflectron time of flight mass spectrometer. The supersonic jet is obtained by expanding a mixture of He or Ar (2 atm) with the benzene derivatives and NH3 through a pulsed (general valve, 200 pm) nozzle in a fist chamber pumped by a 2000 Us diffusion pump. The supersonic beam is skimmed before entering a second chamber where a 1 m reflectron time of flight mass spectrometer (Jordan Co.) is located. This chamber is pumped by a 700 L/s diffusion pump, and the time of flight is pumped by a 200 Us turbo pump. The axis of the time of flight is perpendicular to the beam axis, which allows us to get rid of thermal ions produced by the focused laser and particularly the oil coming from the diffusion pump. Under these conditions, horizontal deflection plates have to be used in order to make up for the horizontal velocity of the ions, which are traveling at 1700 m/s in helium, issued from the supersonic beam. The drawback is that for a given deflection voltage only a limited range of masses can be detected simultaneously (Am = 200 amu for the mass range of the complexes studied). In most of the experiments, a pulsed extraction field is used: the ionization takes place in a zero field region and the extracting pulse (180 V/cm with a 10 ns rise time) is applied after the ionization laser with a delay from a few nanoseconds up to a few microseconds. The ions extracted from the beam by this first voltage are accelerated between two other grids. Since all the ions get the same kinetic energy at the end of the acceleration stage at the entrance of a 1 m field free time of flight, they

120

110

110

120

130

140

150

mass (a.m.u) Figure 3. Mass spectra of pFClB-(NH3), clusters recorded through one-color two-photon ionization: hv is set on the 0; band of 1-1 complex, i.e. 36 405 cm-I. The effects of the delay (At) between the ionization laser and the extraction pulsed field are presented: Af = 0 ,us in the upper trace and Ar = 2.1 /AS in the lower trace. In the inserts, prompt (P) and metastable (MS) mass peaks are enlarged: the ratio between stable and metastable peaks increases with the delay. The

peaks are labeled as follows: CIA, p-chloroaniline+; FClB, p fluorochlorobenzene+,1- 1, [pFCIB-NH31f complex. For chlorinated compounds, the two mass peaks corresponding to the 35C1and j7Cl isotopes are indicated with arrows. The small peaks, shifted by Am = +1 amu from the main mass peaks, correspond to the I3C isotopes. reach the detector in “direct” mode at a time proportional to the square root of their mass. Using the reflectron mode, the ions are reflected by a second series of voltages and then detected after a second field-free time of flight. This reflectron mode enables us to improve the resolution. For example,-on the mass spectrum presented in Figure 3, the peaks at mass = 127, 129, 130 and 132 amu, corresponding to 35Claniline+, 37Claniline+, F35C1B, and F3’C1B, respectively, are easily resolved. Moreover, the use of a delayed extraction after the laser ionization combined with the reflectron mode enables measurement of the rate of ionic fragmentation (dissociation of a complex, chemical reaction, ...) when it occurs on a fragmicrosecond time scale. If a fragmentation (parent+ ment+) occurs, the flight time of the detected fragment+ ion depends on where and when the fragmentation happens with respect to extraction and acceleration times. If this dissociation occurs between the ionization of the neutral parent and the extraction pulse, ions are detected at the fragment’s mass (since the fragmentation occurs without any electric field). If it happens in the extraction-acceleration zone, where the kinetic energy increases, the time of flight of the fragment ion depends strongly on the exact fragmentation time. In our experimental conditions, this concerns a series of ions that fragment between 0 and about 2 ps after the extraction pulse. The observed mass peak is spread over a large range of arrival times from the “prompt” peak of the fragment to the metastable peak (see below), and the end of the signal will be lost in the background. In the field-free time of flight, that is, when the fragmentation time varies from =2 to 32 ps (for 1-1 complexes) after the extraction pulse, all the ions that fragment are detected at the

-

J. Phys. Chem., VoE. 99, No. 37, 1995 13719

Chemical Reactivity in Microscopic Systems same time of flight: they are accelerated with the parent’s mass and then reflected in the reflectron with their own mass. They form the metastable peak whose flight time lies between the time of flight of the fragment ion and that of the parent. The width of this peak is due to the dispersion of kinetic energies released in the dissociation. If the dissociation occurs after the reflection stages, the fragment ions are detected at the mass of their parents. When the dissociation time is in the order of a few microseconds, varying the delay between ionization and extraction changes the proportion of “prompt” and metastable peaks (see Figure 3). This represents a simple way to measure rate constants in a chemical reaction taking place on the microsecond time scale. In one-color experiments, the frequency-doubled dye laser, pumped by a BMI yag laser, is focused on the molecular beam between the extraction plates of the TOF mass spectrometer. In order to record the fluorescence excitation spectrum at the same time, a small part of this laser is sent with a quartz window into the first chamber to cross the supersonic beam a few millimeters from the nozzle; the emission is collected at right angle with a lens focusing on the entrance slit of a 20 cm monochromator. For excitation spectra, as the wavelength of the laser is scanned, the mass spectrum is recorded on a digital oscilloscope (Lecroy 4900). The oscilloscope trace is averaged for 15 or 30 laser shots and read by a PC computer, each mass peak being numerically extracted from the trace. Using two colors, two kinds of experiments have been performed: ionization threshold or appearance threshold measurements and pump-probe experiments. For ion threshold measurements, either a frequency-doubled YAG pumped dye laser or an excimer pumped dye laser (Lambda Physic) is used for the ionization. The first laser is set on a particular SO SI transition of a given cluster and kept at low intensity (104-l@ W/cm2) in order to minimize one-color two-photon ionization processes, while the second laser is scanned around the threshold, its power being 10 times larger than the first one. For a given cluster size, it has been carefully checked that, when the first laser is set off the SO SIresonance, the signal due to hvl hv2 disappears. This ensures that the contribution from larger clusters (n L 3), which are not selected through a resonant transition, is negligible. In the case of p - and m-fluorochlorobenzene, the lifetime of the S I state is about 1 ns, which is much longer than for chlorobenzene, 50 ps.I4 This lifetime is long enough to enable efficient ionization of the SI state by a nanosecond laser. However, the S I state lifetime of bromobenzene (5 ps) is 10 times shorter than that of chl~robenzene.’~ Even with a fluorine substituent, the lifetime of p-fluorobromobenzene is so short that only one-color experiment has been possible. Pump-probe experiments have been performed to measure the kinetics of the reactions. The principle is the following: reactive clusters are formed by ionization of the neutral precursors with the pump laser, and a second laser, delayed in time, is used to probe either the disappearance of the reactive cluster or the appearance of the product. In order to be able to probe an ion (no fluorescence can be detected), it must change its mass upon absorbing the probe laser. This kind of pumpprobe experiment in ions has been used to perform the spectroscopy of the Do D2 transition of aromatic cations. In the case of chlorobenzene,I6the Do state of the ion is populated by two-photon absorption of a f i s t UV laser; then a first photon of a visible laser excites the Do D:! transition of the cation,

-

-

+

-

-

this resonant absorption being followed by the absorption of a second visible photon to a dissociative state where the C-C1 D2 in an bond breaks. Indeed, the excited transition Do aromatic cation correspondsto a n x excitation (visible laser), whereas the neutral aromatic SO S I transition corresponds to x n* excitation, which lies in the ultraviolet spectral range (260 nm). By monitoring the ion current of the benzene’ fragment (C&+), one can record the absorption spectrum of the chlorobenzene cation. The same principle has been used here to probe both the reactants and the products of the chemical reaction for FClBNH3+ complexes. The aromatic chromophore is ionized in a one-color two-photon process by a frequency-doubled YAG pumped dye laser. Then, two photons issued from the excimer pumped dye laser (from 500 to 380 nm) promote the ions to dissociative states through the Do D2 absorption of aromatic cations. The delay between the two lasers is monitored with a BNC generator and varied by steps of 5 ns. The signal (depopulation of the parent or appearance of the fragment ion: FC6&+ and of ions issued from subsequent fragmentation) is averaged for 30 s for each step. The temporal overlap, that is, the temporal resolution, of the two lasers has been measured using a two color ionization procedure. The first photon is set on the 0; transition of the mFClB molecule (two photons at this energy cannot ionize the molecule) and the second W photon promotes the excited state to the ionization continuum. By detecting the ion signal as the delay is varied, one can obtain the laser’s temporal overlap, convoluted by the lifetime of the fvst excited state. The lifetime of this excited state has been measured in fluorescence and has been found in the order of 1 f 1 ns, that is, negligible as compared to the laser width. The observed overlap has been fitted with a Gaussian shape with a full width at half-height of 17 ns, that is, CJ = 10.2 ns.

--

-

-

-

Results

In this section, we present detailed results on both p - and m-fluorochlorobenzene,called pFClB and mFClB, respectively. At the end of the section, additional results on chlorobenzene (ClB) and p-fluorobromobenzene (pFBrB) will be given. A. p - and m-Fluorochlorobenzene-Ammonia (1-1) Complexes. Excitation Spectra/Mass Spectra. Excitation spectra of the 1-1 complex and of the reaction products have been recorded up to 1000 cm-I from the 0; band. In all the excitation spectra that will be presented, the energy origin is the 0; (SO S I ) transition of the bare molecule (36 275 cm-’ for pFClB and 37 027 cm-’ for mFClB7). These transitions have been detected by fluorescence because they cannot be observed in one-color REMPI, the ionization potential of these molecules being higher than 2hv(0:). p-Fluorochlorobenzene-NH3. For each vibration of the free molecule, two main bands of the 1-1 complex (labeled a and b in Figure 4) are due to different isomer^.'^ These bands are blue-shifted by 100 and 130 cm-’, respectively, from the bands of the free molecule.I8 When the 1-1 complex is excited on its 0; transition, three mass peaks have the same spectral signature and thus correspond to the same neutral precursor (see mass spectrum of Figure 3): the ionic complex of pFClB-NH3+ (147-149 amu depending on which isotope of the chlorine atom is considered), pFluoroaniline+ (1 11 amu), and a metastable peak (see Experimental Section). pFanilinium is not observed at this energy although the mass resolution is good enough to differentiate pFaniline+ (111 amu) and pFanilinium (1 12 amu).

-

13720 J. Phys. Chem., Vol. 99, No. 37, 1995

:z:

(x / ' b '[pFC;B-N&]':

I 4 7 ;a,

'

'

'

'

Martrenchard-Barra et al.

'

4000

t

90

Fanilinium

FClB

I

100

110

120

130

140

150

mass (amu)

cm-'

Figure 4. Excitation spectra of pFC1B-NH3 reactive complexes recorded through one-color two-photon ionization with the detection set on different mass peaks: (1 [pFCIB-NH31f, (2) pFanilinium, (3) metastablepeak, (4) pFaniline+,( 5 ) pFClB+. The energy is given with respect to the 0; band of the bare molecule: 36 275 ~ m - l .This ~ band is not ionized in a one-color two-photon process and is not seen in the lower trace. a and b point out the two isomers of the 1-1 complex.

The assignment of the metastable peak has been obtained by varying the delay between the ionization laser and the extraction pulse, allowing slow reactions to occur before the extraction. An increase of the pFaniline+ prompt peak and a decrease of the metastable peak are simultaneously observed as the delay between ionization and extraction increases, as represented in Figure 3: thus the metastable peak correspondsto the pFaniline+ product. Therefore, as already observed in ref 7, the metastable peak can be assigned to the j3 channel (HC1 elimination). Increasing the excitation energy in the SI state by pumping higher vibrational levels enables, in one-color experiments, an increase in the energy content of the reactive 1-1 complex ion. When a higher excitation energy is used (0; 340 cm-I), the Fanilinium mass peak appears as on Figure 4,and the Cl' elimination channel becomes the most important reaction channel. When the 1; vibration (+785 cm-I in the free pFC1B) is excited in the 1-1 complex, a signal is detected at the mass peak of the free molecule, resulting from the dissociation of the complex either in the excited state or in the ionic state. m-Fluorochlorobenzene-NH3. For each vibrational level of the bare molecule, three blue-shifted bands of the 1-1 complex are evidenced at +OS, +24, and +83 cm-l,I8 which are assigned to three isomers from hole burning experiments." Excitation spectra of mFClB -NH3 complexes have been recorded at the mass of the 1-1 complex ion mFClB-NH3+ and also at masses corresponding to reaction products: mFanilinium and mFaniline+ (prompt and metastable). The two channels, CI' elimination and HCl elimination, are observed as soon as the 0; level of the complex is excited (see mass spectrum on Figure 5a). It should be noticed that the metastable peak is much weaker than in the case of pFClB-NH3 complexes and that the ratio of reaction products vs nonreactive complex is greater. Finally, when the 1; vibration of the 1-1 complex is excited,

+

Figure 5. Mass spectra obtained with pump-probe experiments for mFClB-NH3 complexes: (a) One-color two-photon ionization. The pump laser hvl is set on the 0; band of the 1-1 complex, i.e., 37 105 cm-l. (b) Ion depletion and fragment appearance when the visible probe laser (hvz = 500 nm) is added to the pump laser with no delay. One can note the depopulation of the Fadine+ and Fanilinium peaks and the appearance of the FC&+ fragment. The increase of the fluorochlorobenzene (FClB) mass peak in the lower trace is due to evaporation of larger FCIB-(NH,), clusters. The peaks are labeled as follows: CIA, m-chloroaniline; 1-1, mFC1B-NH3; MS, metastable peak. For chlorinated compounds, the two mass peaks corresponding to the 35Cland 37Clisotopes are indicated with arrows. The small peaks, shifted by A m = + I amu from the main mass peaks, correspond to the I3C isotopes.

+

the dissociation in mFClB+ NH3 is less efficient than for pFClB -NH3+. As proved by hole-burning experiments" and also shown by calculations, several isomers are present in the expansion in mand pFC1B-ammonia complexes. One should expect the ammonia molecule to be located in the vicinity of the reactive chlorine site in some isomers and at more distance in others, which should change the reactivity. However, the proportion of each isomer is the same when the excitation spectrum is recorded at the mass of the complex and at the mass of the reaction products (see one-color spectra), indicating that the reactivity is independent of the initial structure. This can be understood on the basis (see below) of the energy difference between the adiabatic ionization threshold and the vertical one: the Franck-Condon ionization imparts more energy in the ionic state than necessary to cross over the isomerization barriers, and since the reaction time is long as compared to the vibrational redistribution, the different isomers are not expected to exhibit specific behaviors. Therefore, two-color experiments have been performed on the most abundant isomer obtained in the expansion. Ion Threshold Measurement. Threshold measurements in the ionic state have been performed using two-color experiments to vary the energy content in the ion, but without changing the intermediate vibrational level in the SIstate. p-Fluorochlorobenzene-NHj. Vertical ionization thresholds of the bare molecule and of the 1-1 complex have been measured under field free conditions. For the bare pFClB molecule, this threshold is 72 930 f 5 cm-I. This value may be slightly underestimated, even with pulsed field ionization, since long-lived high Rydberg states can be ionized with the high extraction voltage (180 V/cm).I9 The vertical ionization threshold of the 1-1 complex is found at 71 000 f 50 cm-l. From the vertical threshold up to 800 cm-I, scanning the second laser results in an increase of the ion signal at the 1-1 complex mass peak without any increase of the signal on the products peaks.

J. Phys. Chem., Vol. 99, No. 37, 1995 13721

Chemical Reactivity in Microscopic Systems

t

1

P‘

para F-anilinium

I

(para F-CI-Benzene..NH,]’

71600

72000

72200

72400

72600

,

‘ 1

Energy (cm-’) Figure 6. pFaniline’ appearance threshold recorded through two-color two-photon ionization. The first laser hvl is set on the 0; band of the 1-1 complex, i.e., 36 405 cm-I. (W) Appearance threshold recorded on the metastable peak; (0)appearance threshold recorded on the stable peak (1 11 amu). Ion signals are normalized to the signal recorded on 1- 1 complex mass peak.

73200

73400

73600

73800

+

The appearance thresholds of the two reaction products, pFaniline+ and pFanilinium, have also been measured. It has already been mentioned that when the 0; level of the 1-1 complex is excited, only the ,$ channel (pFaniline+) is observed, which already indicates that the appearance threshold for the pFaniline+ channel is lower than for pFanilinium. In two-color experiments,the pFaniline+ appearancethreshold has been measured by monitoring the ion signal both on the pFaniline+ “prompt” peak and on the metastable peak: the pFaniline+ metastable mass peak appears when the ionization energy reaches 800 cm-’ (7 1 800 & 50 cm-’) above the vertical ionization threshold, and the “prompt” peak appears 300 cm-’ higher in energy than the metastable one, as can be seen in Figure 6. The pFanilinium appearance threshold (a channel) can be measured by increasing the ionization energy again and is found at 73 300 f 50 cm-’. The ion currents recorded on the pFanilinium and 1- 1 complex mass peaks when the second laser is tuned around the pFanilinium appearance threshold are presented in Figure 7; it should be pointed out that above this threshold, only the pFanilinium signal increases while the two other ion currents stay unchanged. m-Fluorochlorobenzene-NH3. The vertical ionization threshold of the bare mFClB molecule is measured at 74450 & 5 cm-’ and is more abrupt (steplike) than that of pFC1B. For the 1-1 complex, the vertical ionization threshold is found at 71 950 f 50 cm-I. Appearance thresholds of reaction products have also been measured. In the case of mFCl-NH3+ reactive complexes, mFluoroaniline+and mFanilinium mass peaks are both observed when ionizing through the 0; level of the 1-1 complex. The appearance threshold for the ,8 channel, HCl elimination, has been measured at 72 630 & 50 cm-I. There is no significant energy difference between appearance thresholds for metastable and “prompt” mFaniline+ mass peaks, and the ratio of metastable peak vs the prompt one remains small even at the threshold, unlike what happens in the para case. The appearance threshold for the a channel, C1’ elimination, has been measured at higher energy: 73 470 f 50 cm-I (see Figure 8). As in the case of pFClB, the ion signal recorded at the mFaniline+ and 1-1 complex mass peaks remains constant, while the mFanilinium ion signal increases above this threshold. Long Lifetime Measurements. As explained in the Experimental Section, an ionic fragmentation time can be measured

hv,+hv2 (cm-’)

Figure 7. pFanilinium appearance threshold recorded through twocolor two-photon ionization. The first laser hvl is set on the 0; band of the 1-1 complex, Le., 36 405 cm-I. Upper trace: pFanilinium mass peak. Lower trace: 1-1’ complex mass peak. The signal remains nearly constant (the small ion signal increase as the photon energy increases is due to the higher power of the ionizing laser).

. 250 200 150 300

-3 5

100

1

meta F-anilinium

1

-

.

50-

$ 0 x 120 .-I VI

5

100

-6

80

I

.-

60 40

20

73200

73400

73600

73800

hv,+hv2 (cm-’)

Figure 8. mFanilinium appearance threshold recorded through twocolor two-photon ionization. The first laser hvl is set on the 0; band of the 1-1 complex, Le., 37 105 cm-’. The background signal observed is due to one-color two-photon hvl process. The plateau observed on the mFaniline+ mass peak (lower trace) corresponds to the mFanilinium

appearance threshold (upper trace).

if it takes place in the microsecond time scale. This is the case of the chemical reaction leading to Faniline+ ,$ channel, for which two mass peaks have been detected: a prompt one (reaction occumng during the delay between ionization and extraction) and a metastable one (reaction occurring in the fieldfree time of flight, that is, after the extraction-acceleration zone). In the following experiment a one-color two-photon process is used to ionize the clusters. p-Fluorochlorobenzene-NH3. The intensity variation of the pFaniline+ prompt and metastable peaks as a function of the

13722 J. Phys. Chem., Vol. 99, No. 37, 1995

Martrenchard-Barra et al.

delay between ionization and extraction is presented in Figure 9. These signals have been normalized to the signal recorded at the 1-1 complex mass peak. This procedure enables us to take into account the decrease in the collection efficiency when the delay increases (the ionization always takes place at the same point, but the position of the ions when they are extracted varies with the delay due to the velocity of the jet, decreasing slightly the collection efficiency). Assuming a first-order kinetic with a reaction rate t-l,one can fit the data to the equation

where t is the delay between the laser and the extraction pulse and tmaxan adjustable parameter that represents the time during which the ions that react in the extraction region can still be refocused by the reflectron at the product's mass: even at t = 0, the product ions formed in the first few hundreds of nanoseconds after ionization are refocused by the reflectron stage and are observed at the product mass spectra. The lifetime deduced for the /3 channel HCl elimination is t = 0.8 f 0.1 ps. Under the same assumptions, the variation of the metastable peak intensity with the delay is expected to decrease with the same lifetime, following eq 2, where Tdec-acc is the travel time

-.

*

3 0.15 v)

Q

3 - 0.10 c 2 Z 0.05 -

-

.d

i2

m

",.OO

of the ions in the extraction-acceleration region, that is, 1.84 ps for the 1-1 complex (147 amu). However, the signal on the metastable peak has been fitted with a different lifetime, t = 10 ,us (see Figure 9). This implies that at a given ionizing photon energy, different reaction times are observed for the same reactive channel. m-Fluorochlorobenzene-NH3. The same kinetic measurements have been performed by recording the mFaniline+ prompt mass peak intensity as a function of the ionization-extraction delay, when exciting the 0; level. The lifetime fitted with eq 1 is 400 & 100 ns. Here, the metastable peak intensity decreases rapidly with the delay and no long components ( e l 0 ps) are observed. Short Lifetime Measurements. We used pump-probe experiments to measure the shorter lifetimes observed for the Fanilinium channel. Clusters are ionized through a one-color 121 two-photon absorption, and a second laser (122 in the visible 500-390 nm spectral region) with a variable delay is used to probe the ions through a two-photon absorption to a dissociative state of the ionic chromophore. As can be seen on the mass spectra presented in Figure 5a, a and b, when A 2 is switched on, many fragments can be observed together with the decrease in intensity of the parent ion peaks. In order to get enough fragmentation and/or depopulation of the ions, a focused and powerful laser (0.1-1 GW/cm2) is used to probe the ions. Under these conditions, the Do DZ action spectrum (scanning A 2 and monitoring the fragment ions) is structureless between 500 and 390 nm. p-Fluorochlorobenzene-NH3. 121 is set on the first vibration in the SIstate of the 1-1 complex (0; 340 cm-I) in order to have enough energy in the ions to observe Cl' elimination (a channel): the ionization laser leads 200 cm-' above the measured threshold. The ion signal recorded at different mass peaks, pFanilinium, pFaniline', and 1-1 complex, as a function of the delay between

-

+

-

a

"

"

"

"

I '

"

'

delay ionization

"

/

"

I

"

"

"

'

a '

extraction (ns)

Figure 9. pFaniline' kinetics. Ion signals of prompt (a) and metastable (b) peaks are recorded as a function of the delay between ionization and pulsed extraction field. The experimental data (A) are fitted following the procedure explained in the text. The fitted lifetimes are (a) t = 0.8 ps (stable peak) and (b) t = 10 ps (metastable peak).

ionization by the first laser and probe of the ions by the second one, are presented in Figure 10. pFanilinium is the only peak on which a short kinetic depletion 30 f 5 ns is observed. For all the other mass peaks, the depopulation is nearly independent of the delay, reflecting the absence of measurable kinetics: the slow evolution with the delay is accounted for by the loss of some ions as the delay increases. One should notice also that varying the intermediate level from the 0; 340 cm-' band to the 1; band, that is, increasing the intemal energy in the ionic complex, does not lead to a measurable variation of the observed lifetime on the pFanilinium channel. m-Fluorochlorobenzene-NH3. When ill is set on the 0; vibronic state of the 1-1 complex, two-photon ionization leads 750 cm-' above the mFanilinium appearance threshold. The same pump-probe experiment has been performed, and the reaction time measured on the Fanilinum mass peak is nearly the same as for pFClB-NH3: 30 f 5 ns. Calculations. The binding energies of the complexes, which are not known experimentally, are necessary to scale the experimental thresholds with respect to the dissociation limit in FClB+ NH3. Accurate calculation^^^ have been performed on m- and p-fluorochlorobenzene-(NH& clusters by Millit et al. using Claverie perturbation exchange method*O and will be presented in a forthcoming paper, but the relevant values will be used in the next section. From the ionization threshold of the free molecules (pFC1B and mFClB), one can deduce the relative energies E (with respect to the dissociation limit) of the reactive processes occurring in the 1-1 complex by

+

+

J. Phys. Chem., Vol. 99, No. 37, 1995 13723

Chemical Reactivity in Microscopic Systems 150 100

50 anium

E.

(L,

-4000 *

I

para F-aniline+

0

i

100

50

150

an+

-5000

200

-4 -12300

mFan'

-4 -8800

pFan'

-4 -7700

-pFonium' -

150

Figure 11. Energy diagram for the ClB, mFClB, pFClB, and pFBrBNH3, 1 - 1 complexes. Experimental values of the appearance thresh-

0

3

100 r^

>ut

01

1

' I

'

0

1 "

"

I "

100 150 Probe-Pump delay (ns)

50

)

"

200

Figure 10. Pump-probe kinetic measurements in pFCIB-NH3 complexes. Ion currents are recorded on the pFanilinium, pFaniline+, and pFClB-NH3+ mass peaks as a function of the delay between pump laser hvl and probe laser hv2. hvl is set on the first band of the 1-1 complex which leads to pFanilinium: 0; + 340 cm-', i.e., 36 745 cm-', hv2 is a visible laser (500 nm). (a) The experimental Fanilinium depopulation is fitted using the procedure described in the text. The fitted lifetime is 30 k 5 ns. The dotted line is the temporal overlap of the two lasers. This temporal overlap is shorter than the observed depopulation. (b) pFaniline' and (c) pFClB-NH3' ion depopulation. This depopulation is independent of the pump-probe delay.

where E1.1 is the binding energy of the 1-1 complex in the ground state. p-Fluorochlorobenzene-NH3. For the 1-1 complex, two isomers are found in the ground state, and the more stable has a binding energy of 3.05 kcal/mol = 1070 cm-I. The binding energy of the ion is calculated to be 11.64 kcal/mol = 4070 cm-I. These calculated values will be used to build the energetical diagram of the substitution reaction. m-Fluorochlorobenzene-NH3. Several isomers with close binding energies have been found for mFClB-NH3 in the ground state. The binding energy of the most stable isomer is 950 cm-I, and the deepest well in the ionic potential energy surface is 4300 cm-I. The thresholds measured for the ionic reaction occumng in para and meta fluorochlorobenzene-NH3 complexes, scaled with respect to the dissociation limit (p- or mFClbenzene+ NH3), are summarized in Figure 11 and in Table 2. B. p-Fluorobromobenzene- Ammonia (1-1) Complex. The reaction of brominated benzene+ with ammonia has been investigated in gas phase experiments.I0 Brominated benzene+ clustered with ammonia presents the same kind of reactivity as chlorinated benzene: bromine (Bf) elimination (instead of CP) is observed giving anilinium ions. The main difference is the exothermicity of the reaction: 2000 cm-' for ClB and 6400 cm-' for BrB. These values may be compared to the average binding energy of halogenated benzene-ammonia ion: 40004500 cm-I. Thus in ClB+-NHs, Cl' elimination is endothermic at the ionization threshold, whereas in BrB+-NH3, B f elimination is exothermic. As mentioned in the Experimental Section, fluorine substitution of the benzene molecule increases the SI state lifetime; thus we have worked with p-fluorobromobenzene (pFBrB)NH3 complexes, where the reactivity is the same as for

+

olds for the a(ani1inium)and P(aniline+) channels (this work and ref 13 for ClB) and thermodynamical ex~thermicities~~ are scaled with respect to the dissociation limit XYB' + NH3. This is done using the calculated binding energies given by MilliC et aL1' in the case of pFClB-NH3 and mFClB-NH3 complexes (see Table 2). For C1BNH3 and pFBrB-NH3 (*), accurate values are not available, and the binding energies have been estimated as lo00 cm-' in the ground state and 4000 cm-I in the ionic state. For each compound, left full lines represent the appearance thresholds measured in complexes and right dashed lines the reaction exothermicity. The following abbreviations are used in this figure: diss. limit., dissociation limit in XYB+ NH3; aAT, a channel (anilinium) appearance threshold; PAT, /3 channel (aniline+)appearance threshold; vIT, vertical ionization threshold; aIT, adiabatic ionization threshold (calculated value); anium, calculated enthalpy of formation AH for the anilinium channel; an+, calculated enthalpy of formation AH for the aniline+ channel. In the case of pFBrB-NH3 complexes, the a channel appearance threshold (aAT**) has not been directly measured but estimated as explained in the text.

+

TABLE 2: Measured Appearance Thresholds (cm-') for pand mFClB-NH3 Complexee pFCIB -NH3 mFC1B-NH3 diss adiab diss adiab thresholds limit threshold limit threshold adiabatic ionizationb -4070 0 -4300 0 vertical ionization -3000 1070 -3400 900 channel P (Faniline') -2200 1870 -2770 1530 channel a (Fanilinium) -700 3370 -1920 2380 Adiabatic ionization thresholds are deduced from the calculated values of ref 17. Calculated values. bromobenzene-NH3 complexes, but where one-color twophoton ionization efficiency is strongly enhanced. However, two-color experiments remained difficult, especially near the ionization threshold where the ionization cross section is low. Thus only one-color experiments will be presented here. The free pFJ3rB molecule can be ionized when the 0; band (36 231 cm-I) is excited. This gives an upper limit of its ionization potential: 72 462 cm-' = 8.98 eV, a value slightly smaller than the vertical ionization potential measured by photoelectron spectroscopy: 9.02 f 0.02 eV.*' The excitation spectrum of the 1-1 pFBrB-NH3 complex in the vicinity of the 0; transition of the molecule has been recorded by monitoring the pFanilinium and the 1-1 complex mass peaks. Two close intense bands, +106 and +145 cm-I, are observed, as in the case of pFC1B-NH3 and are probably due to two isomers. Excitation of the 1; vibration in the 1-1 complex brings enough energy in the ion to induce ammonia evaporation, so that 1-1 excitation spectrum is observed on the Fanilinium and on the 1-0 mass peaks. The mass spectrum recorded with one-color ionization through excitation of the most intense band (+146 cm-I) is presented in Figure 12 for two delays between ionization and

13724 J. Phys. Chem., Vol. 99, No. 37, 1995

Martrenchard-Barra et al.

r-p-

anilinium

100

120

140 160 mass (a.m.u )

I

200

180

90

Figure 12. Mass spectrum of pFBrB-NH3 recorded through one-color two-photon ionization: hv is set on the 0; and of the 1-1 complex, Le., 36 377 cm-'. Two delays At between ionization and extraction are presented. The ratio between stable and metastable anilinium peak

increases with the delay: the metastable peak corresponds to slow reaction leading to Fanilinium. The splitting of the 1-1 complex mass peak corresponds to the 79Br and *'Br isotopes. extraction: 0 and 1 ps. When no delay is applied, three mass peaks issued from the same neutral precursor (1-1 complex) are observed: the pFanilinium peak, a metastable peak, and the 1-1 ionic complex peak. When the delay is increased, the metastable peak disappears but nothing appears at the Faniline+ mass. This clearly shows that the metastable peak corresponds to a slow reaction producing Fanilinium, that is, Br' elimination. The energetics deduced for pFBrB-NH3 complexes are displayed in the diagram of Figure 11. C. Chlorobenzene-Ammonia (1-1) Complex. The reactivity of chlorobenzene-NH3 complexes has been recently reinvestigated by single photon ionization (SPI) with synchrotron radiation by Grover et al.I3 The results are very similar to those we obtained on m- and p-fluorochlorobenzene-NH3 systems. Three thresholds are measured when increasing the energy: the vertical ionization threshold at 70 520 cm-I; the aniline+ appearance threshold at 71 370 cm-I; and the anilinium appearance threshold at 72 060 cm-I. Contrary to MPI experiments, the neutral precursor cannot be unambiguously assigned, since there is no SO S I selection, but one-photon ionization makes it possible to overcome the difficulties due to the very short lifetime of the S I state of chlorobenzene (50 ps). Using two-color two-photon ionization through excitation of the 0; band of the 1-1 complex, we have obtained results similar to those of Grover et al. Although our thresholds are overestimated because of the low two-color two-photon ionization efficiency, we have been able to see a difference between vertical ionization threshold, aniline appearance threshold and anilinium appearance threshold, whereas the aniline+ formation was not observed in the previous MPI experiment^.^,^.^.^ Figure 13 presents a mass spectrum recorded when the 0; band of the chlorobenzene-NH3 complex is excited. We have checked that the aniline' comes from the resonantly excited 1-1 complex and not from larger sized clusters. The aniline+ mass peak is broad as compared to the free molecule peak and broadens when the energy content of the ion is decreased, indicating a slow reaction (microsecond time scale). This is also validated by the observation of a small metastable peak. The energetics for ClB-NH3 complexes are also displayed in the diagram of Figure 11. D. p-Fluorochlorobenzene-(NH3)~Clusters (1-2 Cluster). The reactivity of m- and p-fl~orochlorobenzene-(NH3)~clusters

-

92

94 96 98 mass (amu)

100

Figure 13. Mass spectrum of CIB-NH3 recorded through one-color two-photon ionization: hv is set on the 0; band of the 1-1 complex, i.e., 37 140 cm-I. This signal is only observed when the laser is set on a transition of the 1-1 complex: anilinium and aniline' products are issued from the neutral 1-1 complex.

has been investigated by Brutschy and co-workers7 and is summarized in the Introduction. In the case of 1-2 clusters, the excitation spectra are structured so that the cluster size can be well selected. In pFClB-(NH&+ clusters, two reaction channels are observed in the ionic state: HC1 abstraction with Faniline+ formation, and HF abstraction with Chiline+ formation. Cl' elimination is no longer observed (and neither is the F elimination). Furthermore, pFClB -(NH3)2+ ions corning from direct ionization of the 1-2 neutral precursor are not observed in one-color experiments when exciting through the 0; band of this cluster. The pFClB(NH&+ ions observed come from larger pFClB -(NH3),>2+ clusters. We attempted to measure the appearance thresholds for the two reactive channels: HF and HC1 elimination through a twocolor experiment with the first laser set on the 0; band of the 1-2 cluster. But we have observed no difference between the vertical ionization threshold and the reaction thresholds, indicating that the reactions already occur at the vertical ionization threshold of 1-2 cluster, that is, 69 850 cm-'. The calculations give a binding energy of 2680 cm-' for the most stable pFClB-(NH3)2 isomer in the ground state and 7975 cm-' in the ionic state. The measured threshold corresponds to -5760 cm-l with respect to the dissociation limit in pFClB+ NH3 NH3. A final observation concerns the shape of the Fadine+ and Claniline+ mass peaks. These peaks are as narrow as those of the free molecule, indicating that the reaction proceeds in a time shorter than a 100 ns (see Experimental Section). This is true for one-color two-photon ionization on the 0; band (-2580 cm-' with respect to the dissociation limit using the calculated values) but also in the case of two-color experiments near the vertical ionization threshold.

+

+

Discussion In this last part, we interpret the energetics and the dynamics of the halogenated benzene-ammonia reactive complexes by discussing the simple models of reactive potential energy surfaces presented in the Introduction. The main part of the discussion will concern p - and m-fluorochlorobenzene-NH3 complexes, but a few points will be enlightened by comparison with the other systems. A. Internal Energy in the Ionic Complexes. Before going into the details of the reaction mechanisms, a first question has

Chemical Reactivity in Microscopic Systems to be clarified: how can several reactive processes, which do not have the same time scale, be simultaneously observed for a given photon energy (reactions giving Faniline+ and Fanilinium)? In the ionization process, the outgoing electrons can carry out some kinetic energy, leaving the ions with an internal energy distribution. The spread of this internal energy distribution is governed by the Franck-Condon principle, and since the difference of geometry between the neutral complex and the ion is large, this distribution can be very broad. Thus, depending on the internal energy of the ion, different channels can be observed for a given photon energy. This effect is clearly observed in two-color appearance threshold measurements: scanning the ionization laser through a threshold induces a new mass peak in the mass spectrum, indicating that a new channel is open, but the mass peaks corresponding to channels open at lower energies do not decrease. As shown in Figures 7 and 8, for p- and mFClB, the Fanilinium peak increases rapidly when the energy is increased above the threshold, whereas the Fanilhe+ and the 1-1 complex mass peaks remain constant, indicating the following: When the photon energy is set just above the threshold, the energy levels below this threshold are still populated. On increasing the photon energy, the ionization efficiency increases since more states are accessible in the ion. Most of the 1-1 complexes with an internal energy greater than the Fanilinium appearance threshold react to give Fanilinium and no longer Faniline+: for example, in Figure 8, the signal recorded on the mFaniline+ mass peak increases with the photon energy up to a limiting value and remains constant afterwards; this limiting value corresponds to the mFanilinium appearance threshold. Coming back to the reactivity of FClB-NH3 complexes, it is now easy to understand the simultaneous observation of the different channels for a given photon energy: The 1-1 ionic complexes observed at their own mass are nonreactive complexes; that is, their internal energy is lower than the lowest threshold of chemical reactions. The 1-1 ionic complexes that react to give Faniline+ have an internal energy that lies between the Faniline+ threshold and Fanilinium appearance threshold. The 1-1 ions with an energy higher than the Fanilinium appearance threshold react through the Cl' elimination channel (the a channel). This interpretation can be extended to one-color two-photon experiments: in the case of pFClB, for example, ionizing the 1-1 complex through its 0; level, pFaniline+ and the 1-1 ionic complex are simultaneously observed, but pFanilinium is not. As the laser is scanned to higher vibrational levels in the SI state, the energy content of the ions is also increased, and the opening of a new channel (pFanilinium) that becomes rapidly the most important is observed. However, whenever the intermediate vibrational level is changed, the Franck-Condon factors, that is, the accessible energy states of the ions, vary. Changing the excited vibrational level results in a translation of the internal energy distribution in the ions, and low internal energy ions are no longer produced: for example when the 1; vibration is excited, the 1-1 complex ion (at its right mass) is not observed, while the pFanilinium channel is the most important one. This assumption on an internal energy distribution responsible for the simultaneous observation of various channels for a given photon energy has been checked on a new experimental setup using the synchrotron radiation of superACO storage ring,22 where threshold photoelectron-photoion coincidences (TPEP-

J. Phys. Chem., Vol. 99, No. 37, 1995 13725 ICO) can be achieved. In these experiments, the internal energy of the 1-1 complex is exactly known (within 200 cm-') by selecting near zero kinetic energy electrons in coincidence with the reaction product ions (or with the non reactive 1-1 complex). For pFClB and mFClB-NH3, the following has been shown: Above the appearance threshold of Fanilinium, only Fanilinium is observed (no Faniline+). The mass peak is as narrow as that of the bare molecule, indicating that the reaction is fast. Below this threshold, Faniline+ is observed and the mass peak is broad. The enlargement of the mass peak is due to a slow reaction in the extraction-acceleration region of the TOF mass spectrometer. Decreasing again the photon energy, one detects only the 1-1 nonreactive complex, the vertical threshold being close to that measured in MPI. Fluorobromobenzene-NHj. In the case of pFBrB-NH3 complex, the internal energy distribution of the ions explains the simultaneous observation of 1-l+ ions and of the Fanilinium reaction products. This observation implies that the Fanilinium appearance threshold is higher than the vertical ionization threshold, and since the reaction is largely exothermic, there must be an energy barrier to B f elimination. We have not been able to measure directly the barrier through a two-color experiment because of the very short SI state lifetime, but it can be estimated through the intensity of the mass peaks (see below). It should be noticed that the thermodynamical values necessary to calculate the exothermicity of the reaction are only available for bromobenzene where AH = -6400 cm-I. To estimate the exothermicity of the bromine elimination channel for pFBrB-NH3 complexes, we have supposed that the exothermicity difference between pFBrB and BrB, on one hand, and pFClB and ClB, on the other hand, is the same, which leads to a value of AH = -5200 cm-I for pFanilinium formation from pFBrB-NH3 complexes. B. Lifetime Measurements. 1. C1' Elimination: the a Channel. In the two cases of mFClB and pFClB complexed with ammonia, Cl' elimination (the a channel) becomes dominant above its appearance threshold (the ions react to produce Fanilinium rather than Fanilinef). The reaction time is shorter than a few hundred nanoseconds, since no metastable peak corresponding to Fanilinium is detected, but is not very fast as shown by pump-probe experiments achieved on the nanosecond time scale. These experiments can be interpreted in the following way: 1. At the wavelengths (390-500 nm) used for the probe laser, only aromatic cations can absorb. The excited Do DZ transition in an aromatic cation corresponds to a n n excitation, whereas the neutral aromatic SO SI transition corresponds to n n* excitation which lies in the ultraviolet spectral range (260 nm). Since the ions are produced hot and since the laser fluence is high (0.1 - 1 GW/cm2),all the aromatic cations produced in the experiment (1-1+ complexes, Fanilhe+, free molecules, or large clusters) can be excited by the visible probe laser. Only Fanilinium cannot be excited with visible radiation because, the positive charge being localized on the NH3 group, its absorption spectrum is similar to that of a neutral aromatic molecule, around 260 nm.23 2. The absorption of the first probe photon is followed by the absorption of a second photon toward a dissociative state that leads to fragmentation. We did not observe any selectivity in the fragmentation of excited cations: FClB-NH3' complexes

-

-

--

13726 J. Phys. Chem., Vol. 99, No. 37, I995

Martrenchard-Barra et al.

and Faniline+ product give the same primary fragment: F-C&+ (by loss of CY), which fragments again by subsequent absorption. When the delay between the pump and probe lasers is varied, the different ion signals should reflect the dynamics of the ions produced by the pump laser: When an aromatic cation is observed, we should observe a depletion due to the direct excitation-fragmentation of this ion. When a reaction product is observed, we should see a depletion due to the excitation-fragmentation of the precursor ion. The dissociation fragments should show an increase. Since nonreactive complexes, reactive complexes, and Faniline+ all absorb the probe laser to give the same fragments, no kinetic information can be gained from these mass peaks. Fanilinium does not absorb the probe laser, so the depletion observed on the Fanilinium mass peak is due to the depletion of its precursor by the probe laser and is then a measure of the reaction time, since it is longer than the temporal overlap of the two lasers. Kinetic measurements have been performed on p- and mFClB-NH3 complexes with the pump laser set on different vibrations of the S I state in order to vary the excess energy in the ionic complexes, and thus the energy available for the reaction. Average reaction times have been extracted from pump-probe experiments by assuming a monoexponential decay. The reaction time measured is 30-40 ns for pFanilinium and for mFanilinium and does not show a clear dependence on the excess energy. The interpretation of this experiment is not straightforward because the reaction time likely varies with the internal energy of the complex. For a given photon energy, several components may then be included in the kinetics due to the internal energy distribution of the reacting ions. For a given energy E, the corresponding rate k(E) is fixed and by numerically integrating the rate equations, one can obtain the shape of the depopulation. The population No(t) of the 1-1 complex ion is

dNo/dr= -(k(E)

+ Z2 exp(-(t

+

- ZJ2/d))No(r)

NsdI2exp(-t2/2d)

(3)

where NSOis the ground state population; ZIand I2 are the pump laser ionization efficiency and the probe laser depopulation efficiency, respectively, which are parameters fitted on the experiment. The temporal shape of the lasers is taken as a Gaussian pulse of 10 ns width. T i s the delay between the two lasers. The Fanilinium population N&r) is given by

(4) and the observed population on the Fanilinium channel at a given delay T between the pump and the probe is

(5) where fobs is the maximum time for which the Fanilinium product issued from the complex can be observed at its right mass in the reflectron mass spectrometer: in our experimental conditions tobs = 300 ns. Simulations of the depopulation signal show that this kind of experiment is not sensitive to short reaction times: if the reaction is too fast, the reexcitation process by the probe laser becomes inefficient. As an example, the depopulation depth detected on the Fanilinium mass peak becomes 6 times weaker when the reaction time is varied from 40 ns to 0.5 ns.

Furthermore,monitoring Fanilinium prompt mass peak depletion does not make visible long reaction time components. Because of the integral energy distribution in the reactive ion, the experimental measurement must be considered as an average value, indicating that a nonnegligible part of the ions react in the order of a few tens of nanoseconds. The longer components are not preponderant (no metastable peak), but the shorter ones corresponding eventually to the larger internal energies in the reactive ions would not be detected. Fluorobromobenzene-NHj. Only Br' elimination (equivalent to the a channel for chlorinated compounds) is observed in the case of pFBrB-NH3 ionic complexes. As previously explained, the exothermicity of the reaction is here larger than the binding energy of the complex, which is not the case in p- and mFClBNH3 complexes. A broadening of the Fanilinium mass peak with respect to other nonreactive peaks, as well as the observation of a metastable peak, indicates that the ions with the lowest internal energy (in the vicinity of the top of the barrier) react at the microsecond time scale. 2. HCl Elimination: the /? Channel. Production of Faniline' HCl is a largely exothermic reactive process, the NH3 and exothermicity is AH = -7700 cm-' for pFClB AH = -8800 cm-' for mFClB NH3.24 Thus the Faniline+ appearance thresholds measured in this work correspond to barriers to the reaction. In the case of pFClB, this barrier is located -2200 cm-' below the dissociation limit p F C W NH3, assuming a binding energy of the 1-1 complex of 1070 cm-' in the ground state. This value has been determined by measuring the appearance of the pFaniline+ metastable peak as the energy content of the ion is increased (see Figure 6 ) . A prompt pFaniline+ peak is observed 300 cm-' higher in energy than this appearance threshold. This is indicative of a variation of the reaction rate constant with the internal energy of the ion. Just on threshold, the reaction begins to occur at the end of the first time of flight region (before the reflecting grids), which corresponds to a reaction time of a few tens of microseconds. The appearance of the prompt mass peak 300 cm-' above the threshold means that a significant part of the ions reacts between the ionization and the extraction pulse, that is, that the time scale of the reaction is 1 or 2 ps. Hence the reaction rate varies gradually with the intemal energy in the reactive complex. That is why, in onecolor experiment through the the 0; level (see Figure 9), it has not been possible to fit the intensity variation of the aniline prompt and metastable peaks as a function of the delay between ionization and extraction with only one rate constant. For mFClB, the barrier to mFaniline+ formation is located -2770 cm-l below the dissociation limit mFClB+ NH3, assuming a binding energy of 950 cm-' for the 1-1 complex in the ground state. But in this case, there is no significant energy difference between the appearance thresholds of metastable and prompt mFaniline+. It shows that the energy range where the reaction is "very slow" is narrower in the case of mFClB than in the case of pFClB. C. Reaction Mechanisms. 1. Summary of the Results. Let us first point out the main results brought up in these experiments, together with some results obtained on other systems in complexes or in gas phase collisions. (a) There is a banier on the reactive path leading to Faniline+, the /3 channel, -2200 cm-I for pFClB and -2770 cm-I for mFClB with respect to the dissociation limit. Above this barrier the reaction proceeds on a long time scale (microseconds), and the reaction time decreases as the energy increases. In the case

+

+

+

+

+

J. Phys. Chem., Vol. 99, No. 37, 1995 13727

Chemical Reactivity in Microscopic Systems of pFClB, time components of the order of 10 ,us have been evidenced that are not seen in mFClB. (b) At higher energy, a second threshold is observed corresponding to Fanilinium production, the a channel: the measured values are -700 cm-l for pFClB and -1920 cm-I for mFClB. It should be mentioned that these thresholds are very close (within the experimental uncertainty) to the thermodynamical ones: -750 and -2500 cm-I, respectively. Above this threshold, the reaction leading to Fanilinium proceeds on a nanosecond time scale, and the reaction leading to Fanilhe+ is no longer competitive. In the case of pFBrB-NH3, where the bromine elimination is more exothermic than chlorine elimination in the other cases, Fanilinium is the only observed product. There is an energy barrier in the formation mechanism of Fanilinium, whose height could not be directly measured but which will be estimated in the following. Just above this barrier, the reaction is slow: microsecond time scale. Other experiments on similar systems have been published and the relevant results are summarized below. (c) From gas phase measurements,I0where the only observed channel is the anilinium production (X' elimination), it has been shown that the reaction efficiency, which is the ratio between the reaction rate and the dissociation rate or in the collisions terminology the ratio between reactive and elastic cross sections, is governed not only by the exothermicity: for example, in the monohalogenated benzene derivatives, anilinium formation from iodobenzene is much more exothermic than anilinium formation from chlorobenzene (-9400 cm-I vs -1900 cm-I), but the efficiency is much lower (0.24% vs 13%). A mechanism involving the formation of an intermediate addition complex, the u complex, in which the kinetics is governed by the entrance barrier of formation of this intermediate has been proposed by Tholmann and Griitzmacher'o,'I following the model developed by Shaik and Ross.12Calculations have shown that the binding energy of the intermediate u complex is similar to that of the ionic 1-1 complex (for example, a value of 4500 cm-I has been found for the u complex in the case of mFClB NH3 by MNDO calculations and the ionic complex is bond by 4300 cm-' 17). (d) The reactivity of the ClB+-NH3 complex presents the same features as FClB+-NH3 ions. As recently evidenced by Grover et al.,I3 the appearance threshold for the anilinium channel is also close to the thermodynamical threshold. We have also shown that the reaction giving Faniline+ occurs on the microsecond time scale in the vicinity of the threshold. (e) For larger clusters, the reaction giving Fanilinium is not observed. The main reactive path is Claniline+ formation. This y channel is the only one observed for mFClB-(NH3)n>l, but the p channel (Faniline+) still subsists for pFClB-(NH3),>l. The appearance threshold of the Xaniline+ products for the pFClB-(NH& clusters is not different from the vertical ionization threshold: -5760 cm-I. But the observation of the 1-2+ peak at its own mass, coming from larger clusters, implies that there subsists a barrier to HX elimination. This barrier is small since it lies below the vertical ionization threshold. (0 A reaction equivalent to the fi channel, HF elimination, has been observed for fluoro aromatic derivatives with other solvents. Two molecules of ammonia or methan01~3~ have to be clustered to fluorobenzene to produce aniline+ or anisole+, respectively, and three molecules of water clustered with p-difluorobenzene induce the fluorophenol+ f0rmation.2~ 2. Energetics and Potential Energy Surface. One can now discuss the reactive potential energy surfaces (shown in Figures 1 and 2) that could rationalize the experimental data, for the

+

XB-NH3 complexes studied, X being the reactive halogen atom (Cl or Br). For X' elimination, we will follow the Shaik and Ross model,I2 assuming the existence of the intermediate u complex in the Fanilinium formation channel. Case 1 is represented in Figure 1 and assumes that the two reactions leading to anilinium f X' (the a channel) and aniline+ HX (the p channel), respectively, are disconnected. The starting point is the same precursor, the 1-1 complex ion, but the reactions proceed further under different paths. This point of view has been proposed by Brutschy et al. in ref 7 . X' elimination after an entrance barrier proceeds through a bound u complex. HX elimination follows another route through another transition state. The reaction ends up in a concerned manner by formation of a N H 2 + radical and H+ (which will be easier if the cluster involves more than one ammonia molecule) and attack of the phenyl+ moiety by this radical, and breaking of the C-X bond with formation of H+ f X = HX. Aniline' Formation: the fi Channel. In this case the threshold measured corresponds to the barrier of the reaction. The long reaction time (microseconds) implies that the crossover is difficult and may be limited by the proton transfer or rearrangement of the cluster ion. The mechanism proposed for the p channel enables to discuss the influence of the cluster size on the reactivity in larger FClbenzene-(NH3), clusters together with the reactivity of other clusters (see point f). The initial step of the reaction, the charge transfer from NH3 to FClB+, as in the Shaik and Ross model, will be favored if the ionization potential of the nucleophilic cluster decreases: the entrance barrier is thus lowered, either when the IP of the solvent decreases, or when the cluster size increases. The second step, formation of NH2' and rupture of the C-X bond with H+-X- formation, is assisted by a proton transfer to the rest of the cluster, whose efficiency is related to the proton affinity of the solvent cluster and will thus depend on the nature of the molecule, and on the cluster size: for example, in FClB (NH3), the barrier lowers as n increases and thus the reaction in larger clusters occurs at the vertical threshold. This model also lets us rationalize results on fluoro and chloro benzene reactions with ammonia, methanol, and water.4,5,7%25 The same arguments also hold for case 2. Anilinium Formation: the a Channel. In the case of chlorinated benzene, the two channels observed start from the same 1-1 complex but immediately follow two different paths. The simplest interpretation of the threshold observed for Fanilinium formation correspondsto the crossing of the entrance barrier to the u complex. According to the experimental thresholds and the calculated values for the binding energies in the ionic state, the barrier to the u complex would be 3370 cm-] above the calculated adiabatic potential of the 1-1 complex in pFClB-NH3 (-750 cm-I from the dissociation limit in pFClB+ f NH3), and +2380 cm-l for the mFClB+-NH3 complex (-1920 cm-I). However, these thresholds correspond, within the experimental uncertainties, to the thermodynamical values of the exothermicities of the reactions in chlorinated compounds. This is also the case for ClB-NH3 as measured by Grover et aLt3 Therefore, it seems that the barrier to u complex formation has not been directly measured and lies below these thermodynamical thresholds. This can be shown by comparison with brominated compounds. In the case of brominated derivatives, AH is larger than the binding energy of the 1-1 ion. The barrier to the u complex leading to pFanilinium formation is clearly evidenced in the

+

Martrenchard-Barra et al.

13728 J. Phys. Chem., Vol. 99, No. 37, I995 Number of populated levels

7 I I I , l - l + i Fanilinium II

II

I 1

I

I I

I

I

ionic

Figure 14. Scheme of the model used to estimate the reaction barrier in the case of pFBrB-NH3+ reactivity. The number ofpopulated levels is constant with the intemal energy of the ion. The following abbreviations are used in this figure: aIT, adiabatic ionization threshold;

vIT, vertical ionization threshold; Ebar, energy of the barrier to the reaction; 00,maximum excess energy in the pFBrB-NH3+ ion obtained through the excitation of the 0; band of the complex. When hv is set on the 0; band, all the ions with an internal energy between vIT and Ebar are detected on the 1-1+ (pFBrB-NH3)+ mass peak and all those with an internal energy between Ebar and 00 react and are seen at the Fanilinium mass peak. pFBrB-NH3+ ions (see above). Although not measured directly through two-color experiment, this barrier can be estimated in the following way: As pictured in Figure 14, the barrier height can be estimated by measuring the intensity of the 1-1+ complex mass peak as compared to the intensity of Fanilinium product peak on the mass spectrum obtained in exciting the 0; transition of the 1-1 complex. Indeed the intensity of the 1-1+ peak reflects the ion population produced in the ionization process with an internal energy lower than the barrier, while the intensity of the Fanilinium peak reflects the ion population produced with an energy greater than the barrier. To estimate this barrier some assumptions have to be made: 1. One has to know the intemal energy distribution in the ionic complex after the ionization process. Although the total density of vibrational state increases drastically with energy, due to Franck-Condon factors, all the levels are not populated. Indeed, the ion population at a given energy is quite constant between the energy of the vertical threshold Et and the energy of the photon: this can be deduced from the shape of the ionization cross section. At a given photon energy E , the ionization cross section varies as: a(@ = .@, dE,where e, is the density of vibrational levels really populated. From our experiment (Figure 7) or from the experiment of Grover et one can see that the ionization cross section increases linearly with the photon energy (at least within a few thousands of wavenumbers above the vertical threshold). Thus the density of levels populated (which is ep = du(E)/dE) stays nearly constant in the energy range of interest. 2. In order to scale the energies, the neutral ground state binding energy is evaluated as 1000 cm-I, similar to the calculated values for chlorinated benzene-NH3 complexes. 3. The vertical ionization threshold of pFBrB-NH3 has not been measured by two-color experiment. However, all the values measured for the other I-I+ complexes stay in the same range. The vertical and the adiabatic ionization thresholds of FBrB-NH3 are assumed to be similar to those measured on the other ClB-NH3 complexes and are taken as -3000 and -4000 cm-I, respectively (as for pFClB-NH3). In the mass spectrum of Figure 12 recorded with the laser set on the 0; transition of pFBrB-NH3, the proportion of Fanilinium and nonreactive complex are 75 f 5% and 25 f 5%, respectively. These values together with the above

assumptions enable us to estimate a barrier height of 1600 cm-' (the barrier is then located at -2400 cm-' from the dissociation limit). This estimation is of course crude due to the experimental uncertainties and the estimation of vertical and adiabatic thresholds.28 This 1600 cm-I barrier must be compared to the 3370 cm-I threshold measured in pFClB which is assumed to correspond to the barrier to u complex following the frame of the model 1. The reaction efficiencies measured in collision, 2.8% for pFClB and 3.4%for pFBrB, are very similar; one should then expect similar barriers to the reaction. Therefore, in chlorinated derivatives, one can conclude that we have measured the thermodynamical thresholds for the anilinium Cl' channel, the barrier to u complex formation being lower. It must be pointed out also that the barrier height deduced for pFBrB-NH3 producing anilinium is close to the barriers measured in FClB-NH3 complexes for the channel leading to Faniline+ HC1. Case 2 is presented in Figure 2 and assumes that the two reactive channels proceed through the same intermediate u complex. The separation of the final products takes place in the exit channel. The barrier to this intermediate u complex is the only one for the a channel-Cl' elimination. The Fanilinium appearance threshold, for chlorinated benzene, corresponds to the thermodynamical exothermicity as previously explained. The anilinium channel is open if the intemal energy of the reactive ion is greater than the exothermicity of the reaction. For brominated compounds for which AH is large, the reaction always ends by Br' abstraction. A contrario, in the chlorinated compounds AH is smaller than the binding energy of the 1-1 ion, and therefore below the anilinium thermodynamicalthreshold; and the reaction proceeds through Faniline+ formation through a barrier in the exit channel. Thus, along the Faniline+ reactive coordinate, the system crosses two barriers: the entrance barrier to the u complex formation, and an exit barrier of protonhydrogen transfer enabling the formation of HCl as leaving group. The appearance threshold for the Fanilhe+ channel thus correspondsto the highest of these two barriers. Then, the upper value for the barrier to the (5 complex formation is 1870 cm-' in pFClB and 1530 cm-' in mFClB (respectively, -2200 and -2770 cm-' below the dissociation in FClB+ NH3). As already mentioned, these barriers for HC1 elimination are surprisingly close to the barrier to the u complex estimated for pFBrB-NH3 producing Fanilinium Br, which supports the idea that the two reactions proceed through the same entrance channel. The geometry of the intermediate u complex seems also to be a good intermediate for the aniline+ formation. In this intermediate, the NH3 group is already attached to the carbon atom and the chlorine atom is only weakly bound to this carbon: only a few hundred wavenumbers are necessary to break the C-Cl bond (at the anilinium threshold). A mechanism involving a hydrogedproton transfer from NH3+ to C1, assisting the C-C1 bond cleavage by formation of HCl, can be proposed to account for the lowering of the energy necessary to break this bond in the u complex. It would correspond to a barrier in the exit channel for aniline+ formation, which lies below the thermodynamical threshold of Fanilinium formation. This process would be slower than the direct C-C1 bond rupture so that when the two channels are open, Fanilinium formation dominates. This model may also help to understand why, in the 1- 1+ complex, the reaction producing Faniline' is observed, while

+

+

+

+

Chemical Reactivity in Microscopic Systems the reaction leading to Claniline+ is not, although energetically allowed. From the Shaik and Ross model, the two u complexes on the chlorine and on the fluorine atom should present the same entrance barrier. The final step of the reaction leading to aniline+ products is governed by the “capture” by the halogen atom of an H (or proton) from the NH3+ group. This process is favored when the C-X bond is weakened, which is the case for the C-C1 bond, as discussed in the last paragraph. The C-F bond requires much more energy (about 10 000 cm-I). Therefore, in the FClB-NH3 complex the reactive site is the chlorine and not the fluorine atom. It should be stressed that the arguments used to explain the reactivity of larger clusters in the frame of case 1 are even more valid here. 1. The formation of the u complex is governed by the electron jump from the NH3 to the benzene; the lowering of ionization potential in solvent clusters due to size effect favors the u complex formation. As a matter of fact, experiments performed on pFClB-(NH&+ ion have shown that the barriers to the reaction lie below the vertical ionization threshold; that is, they are lowered with respect to the 1-1 complex. 2. In the exit channel of reaction /3, the proton transfer becomes easier with the help of solvent molecules and the stiffness of the C-X bond is no longer a limiting factor. This characteristic of assisting the proton transfer seems to be governed by sterical factors (formation of a molecule bridge between X and NH3 in the u complex) and also by the proton affhity, which depends on the nature of the solvent and increases with the cluster size. For example, two molecules of methanol have to be complexed to fluorobenzene to produce anisole+ and three molecules of water with p-difluorobenzene to get fluorophenol+, (H20)3 having the same proton affinity than the methanol dimer.26 Therefore, in large clusters, the proton transfer is no longer a limiting step, the exit barrier is considerably lowered, and the reaction is mainly governed by the exothermicity: 3. The reactions leading to the X’ abstraction (formation of Clanilinium or Fanilinium) are significantly less exothermic than the corresponding HX elimination and therefore are not observed. 4. The exothermicity for the HF elimination reaction is slightly higher than for HC1 elimination in the case of pFClB(NH3), (9240 vs 7740 cm-I) and thus becomes the dominant channel. For mFClB-(NH3), clusters, the difference in the exothermicities for reaction on the fluorine or chlorine atom becomes greater (11 690 vs 8780 cm-I) so that only HF elimination is observed. 3. Kinetics and Potential Energy Surface. For case 2, where the reaction proceeds through the same u complex, the measured Fanilhe+ appearance threshold is either the common entrance barrier for Cl’ elimination and HC1 elimination channels or the exit barrier for HCl elimination. In any case this barrier represents an upper value for the entrance barrier to u complex. If one assumes that the Faniline+ threshold measures the entrance barrier to the u complex, our kinetic measurements can be rationalized in the following manner: 1. The rate-limiting step for the two reaction paths is the crossing of the barrier to u complex formation. The reaction time is slow in the vicinity of the barrier: microsecond time scale and becomes faster (nanosecond time scale) for higher energies. Slow components have been observed on the Faniline+ channel for mFClB, pFClB, and C1B when the anilinium channel is energetically closed. In brominated compounds, where the exothermicity of B f elimination is large, slow components are also observed, this time on the Fanilinium

J. Phys. Chem., Vol. 99, No. 37, I995 13729 reaction product. Faster components have been recorded on the anilinium channel for the pFClB and mFClB-NH3 complexes through pump-probe experiments. 2. Kinetic arguments are also responsible for the competition between anilinium and aniline+ formation. Indeed, the aniline+ channel is open at lower energies than the anilinium one, but this channel is not competitive once the anilinium channel opens. In the model 2, chlorine abstraction is fast, whereas HCl formation requires the crossing of an exit proton (hydrogen) transfer barrier which slows down the process. 3. In the case of larger clusters, the reaction giving aniline+ is fast. In the measurements done for pFClB-(NH&+, no microsecond components are evidenced, even at the vertical ionization threshold (which corresponds to 2200 cm-I above the calculated bottom of the ionic well). This implies that the barriers are small. Comparison with Gas Phase Data. Shaik and Pross model predicts that the reaction efficiency in the gas phase (the ratio between the rate constants for anilinium formation and dissociation of the reactive pair) is governed by the height of the barrier to u complex formation, which depends on several parameters such as the dipole moment of the aromatic molecule. It was readily observed in collision that the reaction efficiency is not governed by the exothermicities and increases with the dipole moment of the aromatic molecule.lo-] Assuming that the thresholds measured on the aniline+ channel @ channel) correspond to the entrance barriers to the u complexes (case 2), one can check that these barriers are compatible with the collisional experiments. This can be done as follows: 1. We assume that the reaction is in competition with the dissociation in XYB+ NH3. The direct dissociation rate is constant for all the XYB-NH3 systems. 2. The reaction rate for the anilinium channel (a channel) is only governed by the entrance barrier to the u complex. 3. One can use statistical theory to predict the evolution of the reaction rate with the energy since we have measured that typical reaction times are slow as compared to vibrational redistribution times. 4. The RRK formalism is used: the reaction rate k(E) at a given excess energy ( E ) above the barrier (Et,) is given by the Kassel formula.27

+

where v is the vibrational frequency at the transition state and s the number of degrees of freedom involved in the reaction. 5 . Crossing over the entrance barrier mainly involves the intermolecular modes, as well as the C-C1 bond. The other intramolecular modes of the molecule are considered spectators and the number of degrees of freedom has been set as 9 (changing this number by 2 or 3 does not noticeably change the results). The results on the mFC1B-NH3 system have been used to calibrate the calculated rate constants from the barrier (-2770 cm-I) and the reaction efficiency (10%). The barrier heights for the other complexes have been then calculated with the same s and v and the known reaction efficiency. The results are summarized in Table 3. The agreement between the calculated and the measured barriers is good, although the model is very simple and we can conclude that (1) it is a good approximation to consider that the reaction efficiency observed for channel a is limited by the

13730 J. Phys. Chem., Vol. 99, No. 37, 1995

Martrenchard-Barra et al.

TABLE 3: Comparison between Measured Barriers and Barriers to the Anilinium Channel Calculated from Gas Phase Efficiencies Using the KasseP' Statistical Model reaction efficiency (%0)a mFClB -NH3 pFClB -NH3 pFBrB-NH3 CIB-NH3

10 2.8 3.4 13

calcd banier using measd Kassel formula (cm-I) bamer (cm-l) 1530 1800 1550 1300

1870 160Ob 12w

fl From Tholmann and Griitzmacher, ref 10. Estimated value. From Grover et a/.. ref 13.

u complex entrance barrier and (2) the barriers measured on the Faniline+ channel can be the entrance barrier to the u complex.

Conclusion In this work, we have investigated ion molecule reactions through the ionization of complexes or small clusters composed of dihalogenated benzene (XYB) derivatives and ammonia. In these ionic clusters several reactive channels are open: halogen substitution leading to monohalogenated anilinium (YBNH3+) and HX elimination leading to ionic monohalogenated aniline (YBNH*+). The results on the energetics and dynamics for these two reaction paths can be interpreted using a model potential where the two channels proceed through the same entrance barrier leading to the u addition complex. In fluorochlorobenzene, the barrier to the u complex probably corresponds to the threshold measured for fluoroaniline+ formation,'while the higher threshold measured on the anilinium channel corresponds to the thermodynamical exothermicity for this reaction. In fluorobromobenzene, the thermodynamical exothermicity for the anilinium channel is much lower than for fluorochloro compounds. This channel opens as soon as there is enough energy in the ionic cluster to cross the entrance barrier to the u complex. The observation of the anilinium product from the 1-1+ complex, despite its unfavorable exothermicity, is due to the exit barrier in the reaction path leading to aniline+ formation, which slows down the HC1 elimination as compared to the simple Cl' elimination.

Acknowledgment. We are indebted to Prof. B. Brutschy and Dr. P. MilliC for many helpful discussions and comments. We also thank Prof. J. R. Grover for providing us details on his work with the synchrotron radiation. Thanks also to S. Bouchonet. References and Notes (1) Jouvet, C.; Solgadi, D. Chemical Reaction in Clusters; Bemstein, E. R., Ed.; Oxford University Press: New York, in press.

(2) Jouvet, C.; Boivineau, M.; Duval, M. C.; Soep, B. J . Chem. Phys. 1987, 84, 5416. (3) Pdersen, S.; Herek, J. L.; Zewail, A. H. Science 1994, 266, 1359. Syage, J. A. Faraday Discuss. 1994, 97, 401. (4) Brutschy, B. J . Phys. Chem. 1990, 94, 8637. (5) Brutschy, B.; Eggert, J.; Janes, C.; Baumg2rtel. H. J . Phys. Chem. 1991, 95, 5041. (6) Awdiew, J.; Riehn, C.; Brutschy, B.; Baumgartel, H. Eer. Eunsenges. Phys. Chem. 1990, 94, 1353. (7) Riehn, C.; Lahmann, C.; Brutschy, B. J . Phys. Chem. 1992, 96, 3626 and references therein. ( 8 ) Maeyama, T.; Mikami, N. J . Am. Chem. Soc. 1988, 110, 7238. (9) Maeyama, T. N.; Mikami, N. J . Phys. Chem. 1990, 94, 6973. (10) Tholmann, D.; Griitzmacher, H. F. J . Am. Chem. Soc. 1991, 113, 328 1. (11) Tholmann, D.; Griitzmacher, H. F. Chem. Phys. Lett. 1989, 163(2-3), 225. (12) Shaik, S. S.; Pross, A. J . Am. Chem. Soc. 1989, I l l , 4306. (13) Grover, J. R.; Cheng, B. M.; Herron, W. J.; Coolbaugh, M. T.; Peifer, W. R.; Garvey, J. F. J . Phys. Chem. 1994, 98, 7479. (14) Ichimura, T.; Mori, Y. J . Chem. Phys. 1973, 58 (l), 288. (15) Dietz, T. G.; Duncan, M. A.; Liverman, M. G.; Smalley, R. E. J . Chem. Phys. 1980, 73 (10). 4816. (16) Ripoche, X.; Dimicoli, I.; Le Calve, J.; Piuzzi, F.; Botter, R. Chem. Phys. 1988, 124, 305. (17) Gaigeot, M. P.; De Pujo, P.; Millie, P.; Dedonder-Lardeux, C.; Jouvet, C.; Martrenchard, S.; Solgadi, D., to be published. (18) In p- and m-fluorochlorobenzene-NH3 complexes, there is a small disagreement on the position of the vibrational bands between our values and those of Brutschy et al. published in ref 7. We rely on our values since they have been calibrated with respect to the 0; band observed in fluorescence. Indeed, the 0; band could not be detected through one color two photon ionization (ionization threshold > 2hvw). (19) Rieger, D.; Reiser, G.;Muller-Dethlefs, K.; Schlag, E. W. J . Phys. Chem. 1992, 96, 12. (20) Claverie, P. In Intermolecular Interactions: From Diatomics to Biopolymers; Pullman, B., Ed.; Wiley: New York, 1978; Chapter 2. Hess, 0.; Caffarel, M.; Langlet, J.; Caillet, J.; Huiszoon, C.; Claverie, P. In Proceedings of the 44th International Meeting of Physical Chemistry on Modelling of Molecular Structures and Properties in Physical Chemistry and Biophysics, Nancy, France, 11-15, September 1989; Rivail, J. L., Ed.; Elsevier: Amsterdam, 1990. (21) Botter, R.; Gounelle, Y.; Jullien, J.; Men& F.; Solgadi, D. J . Chim. Phys. Fr. 1975, 10, 1094. (22) Dedonder-Lardeux, C.; Dimicoli, I.; Jouvet, C.; Martrenchard-Barra, S.; Richard-Viard, M.; Solgadi, D.; Vervloet, M. Chem. Phys. Lett., in press. (23) UV absorption spectrum of p- and m-fluoroanilinium have been recorded in solution by protonating Faniline with hydrochloric acid. (24) Pedley, J. B.; Rylance, J. Sussex-N.P.L. Computer Analysed Thermodynamical Data: Organic and Organometallic Compounds; University of Sussex: 1977. ( 2 5 ) Martrenchard, S.; Jouvet, C.; Lardeux-Dedonder, C.; Solgadi, D. J . Phys. Chem. 1991, 95, 9186. (26) Wanna, J.; Menapace, J. A,; Bemstein, E. R. J . Chem. Phys. 1986, 85, 1795. (27) Kassel, L. S. J . Phys. Chem. 1928, 32, 225. (28) Note added in proof recent results were obtained with synchrotron radiation and TPEPICO methods on the FBrB-NH3 complex: the Fanilinium appearance threshold has been measured around 1900 iz 300 cm-' above the adiabatic threshold, which is not so far from the estimation made here (to be published in Chem. Phys. Lett). JP950550X