Violation of frozen shell approximation in dissociative electron capture ...

RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

Violation of frozen shell approximation in dissociative electron capture by halogenated anthraquinones N. L. Asfandiarov1*, V. S. Fal’ko1, V. G. Lukin1, E. P. Nafikova1, S. A. Pshenichnyuk1, A. I. Fokin1, G. S. Lomakin1 and Yu. V. Chizhov2 1

Institute of Physics of Molecules and Crystals, Ufa, 450075, Russia St. Petersburg State University, St. Petersburg, Russia

2

Received 4 June 2001; Revised 3 August 2001; Accepted 4 August 2001

A series of halogenated anthraquinone (AQ) derivatives has been studied by means of electron capture negative ion (NI) mass spectrometry (ECNI-MS). 1Cl-AQ and 2Br-AQ display dramatically steep positive temperature dependencies of Hal ion abundance in the low electron energy region. Molecular NI intensity decreases rapidly with increasing temperature in the case of 1I-AQ. In the case of 2Br-AQ, a metastable NI peak (m/z 22.9) corresponding to the process BrAQ → Br ‡ AQ0 was recorded. This means that the characteristic dissociation lifetime of the molecular NI Br-AQ is at least 25 ms at the energy 0.67 eV in the low-temperature spectrum (T 80 °C), and at the energy 0.13 eV in the hot spectrum (T 290 °C). Together with the observed temperature dependence of the 2Br-AQ curves of effective yield (CEY), this proves that this anion dissociates according to Coulson's model. The same halogen anion behavior is observed in the case of 1Cl-AQ. There are three consecutive stages in the process of molecular NI dissociation of Cl- and Br-substituted AQ, namely, electron capture into the empty p-orbital by means of the shape resonance mechanism, followed by a radiationless transition into the ground electronic p-state of the anion, as predicted by Compton in the case of the parabenzoquinone molecule, and, finally, a fluctuative dissociation of the molecular NI accompanied by the transition from the p-term into the s-term, so-called predissociation. Calculations show reasonable agreement with the experimental data. In the case of 1I-AQ, an effect of inversion of empty levels in the process of electron capture by the molecule takes place, a violation of the so-called frozen shell approximation. The phenomenon found may be of significance not only in the case of ECNI-MS, but also in other experimental investigations using low-energy electronmolecule and ion-molecule collisions. Copyright # 2001 John Wiley & Sons, Ltd.

The problem of symmetry conservation in the processes of negative ion (NI) formation has a 35-year-old history.1 Coulson showed that the symmetry of the molecular NI electron shell defines the pathway of its dissociation.1 Really, the symmetry conservation law in quantum mechanics corresponds to the law of angular momentum conservation in classical mechanics. However, there are enormous difficulties in the description of molecular NI decay in the case of the ClC6H5 molecule. Thus, a decay with nonradiative transition from the p-term (p7 state in the notation of Coulson1) to the s-term (p6 state) in the neighborhood of the crossing point (see Fig. 1) implies both temperature and pressure dependencies of Cl ion abundance.1 Neither has been reliably observed by experiment.1±4 There have been some attempts to explain this paradox5,6 by an out-of-plane vibration of the Cl atom. Unfortunately, these considerations involve only qualitative reasoning without any quantitative estimation of the effect expected. Meanwhile, the probability *Correspondence to: N. L. Asfandiarov, Institute of Physics of Molecules and Crystals, Ufa, 450075, Russia. E-mail: [email protected] Contract/grant sponsor: Russian Foundation for Basic Research; Contract/grant number: 00-02-16578. DOI:10.1002/rcm.446

Figure 1. Schematic terms representation for the chlorobenzene dissociation.1 1A1 denotes the molecular term, 2B1 and 2A1 denote the anion terms. I denotes the term of the molecule, II that of the p-term of the anion, and III that of the dissociative s-term of the anion. Copyright # 2001 John Wiley & Sons, Ltd.

1870 N. L. Asfandiarov et al.

of these processes has been evaluated.4,7 It has been shown that, in the case of planar (thus forbidding dissociation) benzotiadiazole derivatives7 and substituted anthraquinones,4 the population of out-of-plane conformations of the target molecule (allowing dissociation) at the temperature of typical ECNI-MS experiments is about 5±10%. This means that the relative intensity of symmetry-forbidden channels of dissociation may only be up to some tens of percent of that of the allowed channels. Therefore, the so-called `Freeman4,7 model' seems to be unrealistic in the case of C6H5Cl. In addition, the Freeman model implies the existence of a temperature dependence of Cl ion formation according to the Boltzmann distribution of out-of-plane conformation populations. Present-day theoretical investigations of electronic structures of molecular negative ions are carried out within the framework of Koopmans' theorem,8 see, for example, Refs 9±11. A typical procedure is as follows. Virtual orbital energies calculated at the B3LYP/D95//B3LYP/D95 level of theory, or with another quantum chemical method with geometry optimization in the ground neutral state of a molecule, are compared with the experimental electron attachment energies obtained from electron transmission spectroscopy. It is known that there is a strong linear correlation between these parameters for a large number of aromatic and unsaturated hydrocarbons: " = (x ‡ a)/b, where " denotes the experimental attachment energy, x is the calculated orbital energy, and a and b are empirical constants. This kind of consideration of molecular NI electronic structure is very convenient and simple to use. But, as we will show below, one should be very careful in some situations. As the mechanism of Cl ion formation by means of p-s non-radiative transition is invalid in the case of ClC6H5,1 there must be other mechanisms for this dissociation channel. Beland et al.12 have found that, in the case of some chlorinated benzenes, an effect of inversion of empty levels in the process of anion formation takes place, namely, the lowest unoccupied orbital is a p* orbital for 1-, 1,3-di-, 1,4di-, 1,2,4-tri-, 1,2,3,5-tetra-, 1,2,4,5-tetra-, and 1,2,3,4,5-pentachlorobenzenes according to CNDO/2 calculations,12 whereas open shell calculations for the radical anions of the chlorobenzenes show that the highest occupied MO is a s* orbital in all molecules under investigation.12 Unfortunately, this important result went unnoticed by specialists in dissociative electron capture NI mass spectrometry. The latest investigations of halogenated benzenes4 confirmed the existence of an empty levels inversion effect, or, in other words, the violation of the frozen shell approximation in the processes of dissociative electron capture by molecules. The order of empty levels in ClC6H5 is as follows: ‡0.063 eV p* b1 (LUMO); ‡0.161 eV p* a2; ‡1.46 eV s*C Cl a1, according to semiempirical PM3 calculations. Open shell PM3 calculations for the anion ClC6H5 gives the following result: HOMO is a 0.275 eV s* a1. An appropriate scheme of the terms in the process of anion dissociation is shown in Fig. 2. It is easy to show that there is no symmetry ban for the dissociation of the ground electronic state of the chlorobenzene anion to Cl ion.4 The same result has been obtained for the cases of p-Cl,NO2C6H4 and p-I,NO2C6H4.4 Obviously, it Copyright # 2001 John Wiley & Sons, Ltd.

Figure 2. Schematic representation of the molecular and anion terms of chlorobenzene. rC Cl denotes the reaction coordinate. Predicted Emax(Cl ) = 0.55 eV, the experimental value is 0.74 eV.4

Figure 3. NI CEYs for 1F-AQ measured at a series of ion source temperatures. Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

Dissociative electron capture by halogenated anthraquinones

1871

Figure 4. NI CEYs for 1Cl-AQ measured at a series of ion source temperatures.

is necessary to obtain more examples of the violation of the frozen shell approximation on other classes of molecules to prove that fact convincingly.

EXPERIMENTAL Negative ion mass spectra were obtained using a modified13 MI-1201 mass spectrometer under the following conditions: accelerating voltage 4 kV, electron trap current 1 mA, FWHM of electron energy distribution DE1/2 = 0.35 eV, Eel varies in the range 0±12 eV. Vaporization temperatures of compounds under investigation were in the range 300± 350 K. Temperature dependences of the curves of NI effective yield (CEYs) were measured in the range 100± 300 °C. The details of ECNI-MS experiments, using a method Copyright # 2001 John Wiley & Sons, Ltd.

developed for static mass spectrometers,13 have been described previously.4,7 Drift time for the SF6 ion (m/z 146) from the moment of formation in the ion source, through the mass separator system to the moment of detection by the secondary electron multiplier, is 25 ms.

RESULTS AND DISCUSSION Negative ion CEYs as a function of electron energy, for each of the molecules under investigation, are shown in Figs 3±6. NI mass spectra of these molecules are presented in Table 1. All data listed in Table 1 correspond to the lowest temperature at which spectra were recorded, see Figs 3±6. There is one principal difference between 1F-AQ NI spectra and all the others: the molecular 1F-AQ NI has at least three Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

1872 N. L. Asfandiarov et al.

Figure 5. NI CEYs for 2Br-AQ measured at a series of ion source temperatures.

long-lived resonant states, while the others have only two (Cl- and Br-substituted) or only one resonant state (Iderivative). This observation may be interpreted in terms of the competition of the molecular NI dissociation process with halogen anion formation. This proposal will be confirmed by the temperature-dependence considerations described below. The ECNI mass spectrum of AQ (C14H8O2) was measured previously.14 Resonant states interpretation is given in Table 1, where RS0 denotes the nuclear-excited Feshbach resonance,15 SR the first shape resonance, FR the electron-excited Feshbach resonance,15 and RS1, RS2, etc., denote 1st, 2nd, etc., inter-shell resonances.16 The energy of the shape Copyright # 2001 John Wiley & Sons, Ltd.

resonance varies according to the electronic properties of the halogen atom in the series of molecules under investigation. A method of quantitative evaluation of the energy of a shape resonance was proposed in Refs 17 and 18, and calculations performed on this basis show reasonable agreement with the experimental data (see Table 3). The third resonance at the energy of about 1.7 eV is related to the electron-excited state of the target molecule, namely a triplet n → p* transition. This transition energy is essentially constant in the series of substituted molecules (Table 2).19 Only Tnp* and Snp* transition energies (practically constant) correlate with the energy of the Feshbach resonance in the ECNI mass spectra. Substitution has a dramatic influence on Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

Dissociative electron capture by halogenated anthraquinones

1873

Figure 6. NI CEYs for 1I-AQ measured at a series of ion source temperatures.

the energy of the triplet and singlet p → p* transition in the series of anthraquinone derivatives.19 Therefore, we can assume that, in the AQ derivatives, the lowest unoccupied molecular orbital (LUMO) is the p* orbital. Semiempirical PM3 calculations of electron structures of the molecules under investigation confirmed that in all cases the LUMO was found to be p*. Consider now the mechanisms of long-lived molecular NI formation in ECNI mass spectra of anthraquinone derivatives at non-thermal electron energy. There are two generally accepted explanations of the formation of long-lived molecular NI of para-benzoquinone in the high-energy region 1.8 eV.2 The first involves the direct attachment of an incident electron into an unoccupied orbital followed by internal conversion to the ground state of the molecular negative ion (MNI).2 The second proposal invokes simultaneous excitation of a core electron into a higher orbital (electron-excited Feshbach resonance20,21). In the latter case, the MNI is stable relative to autodetachment because the term of the MNI lies below an appropriate molecular term, and the autodetachment channel is closed.15 A resonance of this kind is called `two-particle one-hole resonance' (2p, 1h). In a characteristic time (about a typical time of a radiationless transition), this excited-electron state relaxes into the ground electron state, and changes into a `conventional' nuclearexcited Feshbach resonance. We believe that, in the case of anthraquinone derivatives, the resonance at the energy 0.5 eV forms by the first mechanism, and the high-energy Copyright # 2001 John Wiley & Sons, Ltd.

resonance (1.8 eV) forms by means of the second one. Proof of the validity of this resonance states interpretation has been offered in our previous work.14 Figures 3±6 show that there is an obvious temperature dependence of the NI curves of effective yield. This temperature dependence is weaker in the case of 1F-AQ, for which the molecular NI has three long-lived resonant states, namely, the nuclear-excited resonance at 0.0 eV (low energy), the shape resonance at 0.38 eV (middle energy), and the core-excited Feshbach resonance at 1.76 eV (high energy). The vibrational energy of the MNI increases with temperature, which leads to a decrease in the molecular NI lifetime and, as a result, the maxima of the CEY curves for 1F-AQ drift toward lower energy. Moreover, the relative intensity of the high-energy peak of the MNI, relative to that at middle energy (0.38 eV), decreases with temperature. This may be explained by shortening of the MNI lifetime, a well-known effect in ECNI-MS.13 On the whole, the shape of the 1F-AQ MNI CEY is the same as that for AQ.10 We did not detect F ions in the ECNI mass spectrum of 1F-AQ. The 1F-AQ MNI lifetime relative to the autodetachment process is very long (>10 3 s) in the low- and middle-energy regions, and linearly decreases from 50 ms at 1.4 eV to 4 ms at 2.36 eV. There were no fragment negative ions observed in the spectrum of 1F-AQ. In the case of 1Cl-AQ the situation is more complicated. The molecular NI has a very long lifetime, about 300 ms, relative to autodetachment (Table 1), but the strong Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

1874 N. L. Asfandiarov et al. Table 1. ECNI mass spectra of anthraquinone derivatives at the lowest recording temperature. ta means MNI lifetime relative to autodetachment I. AQ* Emax (eV) 0.0 0.44 1.7 5.91sh*** 6.8 8.4

ta (ms) 1400 880 19

Int. (%) 59 100 22 0.1 0.2 0.07

Resonant state RS0** SR FR RS1 RS2 RS3

Structure M

Emax (eV) 0.0 0.38 1.76

ta (ms) >1000 >1000 40

Int. (%) 60 100 19

Resonant state RS0 SR FR

III. 1Cl-AQ m/z 242

Structure M

ta (ms) 400 130

35

Cl

Emax (eV) 0.0 0.41 1.93 3.62

Int. (%) 54 100 55.2 11.2

Resonant state RS0 SR FR RS1

IV. 2Br-AQ m/z 286

Structure M

ta (ms) 1000 900

79

Br

Emax (eV) 0.0 0.3 0.67 1.92 3.6

Int. (%) 66 100 4.1 39.0 8.3

Resonant state RS0 SR SR FR RS1

V. 1I-AQ m/z 334 127

Structure M I

Emax (eV) 0.0 0.0 0.1 1.88

ta (ms) 9

Int. (%) 2.3 92.8 100 12.45

Resonant state RS0 RS0 SR FR

m/z 208

Structure M

207

[M

II. 1F-AQ m/z 226

H]

* Data from Ref. 10. ** RS0 means nuclear excited Feshbach resonance, SR means the first shape resonance, FR means electron excited Feshbach resonance, RS1, RS2, etc., mean 1st, 2nd, etc., intershell resonances (see text). *** sh means shoulder.

Table 2. Spectral properties of the substituted anthraquinones.15 FR means Feshbach resonance Molecule AQ 2OH-AQ 1NH2-AQ 2-NH2-AQ 1F-AQ 1Cl-AQ 2Br-AQ 1I-AQ

Tnp*, eV

Snp*, eV

Tpp*, eV

Spp*, eV

2.71 2.73 2.80 2.81 2.62 2.70 2.71 2.71

2.96 2.96 3.03 3.06 2.94 2.95 2.95 2.94

3.11 1.67 1.73 1.62 2.77 2.74 2.79 2.75

3.86 2.41 2.47 2.37 3.51 3.48 3.53 3.44

FR, eV 1.7 1.76 1.73 1.6 1.76 1.93 1.92 1.88

DE(Tnp*-FR), eV 1.01 0.97 1.07 1.21 0.86 0.77 0.79 0.83

Table 3. Shape resonance lifetimes calculated in the united anion approximation.14 EA stands for electron affinity, U stands for spherical potential well depth, r denotes the radius of the well, L the orbital momentum of the captured electron, Eexp the experimental energy of the resonance, Ecalc the calculated value, and tSR means the calculated lifetime

I II III IV V VI VII VIII

Molecule

EA (eV)

U (eV)

Ê) r (A

L

Eexp (eV)

Ecalc (eV)

AQ 2OH-AQ 2NH2-AQ 1NH2-AQ 1F-AQ 1Cl-AQ 2Br-AQ 1-I-AQ

1.59 1.64 1.49 1.46 1.52* 1.71 1.81* 1.55*

8.27 8.07 7.80 7.90 7.95 8.22 8.04 6.82

3.72 3.79 3.82 3.78 3.78 3.77 3.86 4.2

3 3 3 3 3 3 3 3

0.44 0.34 0.40 0.56 0.38 0.41 0.3 0.1

0.54 0.46 0.54 0.60 0.57 0.43 0.27 0.21

tSR (s.10

15

)

110 170 100 80 90 200 600 1200

* Means PM3-calculated vertical electron affinity. Copyright # 2001 John Wiley & Sons, Ltd.

Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

Dissociative electron capture by halogenated anthraquinones

1875

Figure 7. Schematic terms representation for the 1Cl-AQ molecule, p- and sanion states. Short-lived term of the electron-excited shape resonance is not shown.

competition with dissociation leads to disappearance of the MNI CEY at an energy of about 1 eV (Fig. 4). The shape resonance has two observed channels, M ion at the energy 0.3 eV, and Cl ion at the energy 0.5 eV. Relative intensities of both channels change with the experimental temperature (Fig. 4). Consider now electron state evolution in the case of Cl ion formation in the middle-energy resonance. According to the model of Compton,2 formation of the MNI 1Cl-AQ at the energy 0.5 eV proceeds through the shape resonance mechanism. An incident electron is captured into the empty level of positive energy of about 0.5 eV, evidently of psymmetry. PM3 semiempirical calculations predict the existence of a p* 5 molecular orbital with the energy ‡1.18 eV (index 5 means the fifth empty orbital of the target molecule). Our evaluation of such a shape resonance lifetime gives a value about ySR = 2  10 13 s. This time is long enough for nuclear motion in the temporary MNI to get started. Thus it is possible that the ion geometry will be changed in such a way that the SR term will lie lower than the term of the molecule,2 thus closing the autodetachment channel. Radiationless transition is possible from the SR electron configuration to the ground electron state of MNI (p* 5 → p* 1). Such a transition takes about a thousand vibration periods.22 Theoretical evaluation predicts the probability of radiationless transition of about P = 10 1± 10 2 in the case of the CO molecule, and of about P = 10 6 for the aromatic molecules.22 The value P 10 3 seems to be a reasonable estimate in the case of the MNI of anthraquinone derivatives. So, in the case of the middle-energy resonance, we can propose the schematic term diagram shown in Fig. 7. After a radiationless transition, the MNI is placed on the pterm (dotted line in Fig. 7), and a direct dissociation with Cl ion formation is forbidden by symmetry conservation.1 In the neighborhood of term intersection, the next radiationless transition to the dissociative s-term (dashed line) is possible. In that case, the MNI energy excess is equal to the electron Copyright # 2001 John Wiley & Sons, Ltd.

affinity of the molecule plus the kinetic energy of the incident electron. The total time elapsed from the moment of MNI formation to the relaxation to the ground electronic state is of the order of 10 11±10 10 s (tvib 10 14±10 13 s, P 10 3), which is quite enough for the dissipation of the excess energy into all internal degrees of freedom. Therefore, the C-Cl bond, whose length plays the role of reaction coordinate in the process of dissociation, has an average vibrational energy DE(T, Eel) = kT ‡ (EA ‡ Eel)/(3N 6), where T denotes the temperature, k is the Boltzmann constant, and N = 24 is the number of atoms in the molecule of AQ. For the case of 1Cl-AQ at the lowest temperature (100 °C) and Eel = 0.5 eV, the excess energy is DE 0.066 eV per degree of freedom. At the same time, the dissociation energy for the anion is about ED = 1.16 eV, see Fig. 7. The terms in this scheme were built proceeding from the values of EA(1Cl-AQ) = 1.71 eV,19 EA(Cl) = 3.62 eV,23 C-Cl bond dissociation energy DE(C-Cl) = 3.18 eV;24 equilibrium internuclear distances were calculated by means of PM3 Ê for the semiempirical calculations, to be r0C Cl = 1.682 A Ê molecule, and r C Cl = 1.7 A for the anion p-state. It is easy to see that Cl ion formation is allowed energetically in the first (zero-energy) resonance. However, because the average vibrational energy of the C-Cl bond is only about DE0 = kT ‡ EA/(3N 6) 0.058 eV, it takes a very long mean time (>1400 ms, see Table 4) for enough energy to concentrate in this bond. We used a many-particle statistical approximation for this time evaluation: td ˆ tvib =10

3

Emax Z

Elim

2 p e pT 3

" p T "d";

…1†

13

s, Elim denotes the anion dissociation where tvib = 10 energy, Emax denotes the maximum available vibration energy for each vibration mode, and the coefficient 10 3 accounts for the probability of transition from the ground state p-term to the dissociative s-term. Some comments on Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

1876 N. L. Asfandiarov et al. Table 4. Evaluation of fluctuative dissociation lifetime td (ms) of Hal-AQ MNI for different temperatures in the low- and middleenergy resonances. MNI drift time means the time from the ion formation in the ion source to detection at the electron multiplier 1Cl-AQ. MNI drift time 32.19 ms T °C\Eel

0 eV

80 110 170 230 290 310

1430 ms 526 ms 128 ms 38.6 ms 12.8 ms 10.2 ms

0.4 eV

1.5 eV

204 ms 104 ms 32 ms 11.8 ms 4.98 ms 3.84 ms

6.24 ms 4.2 ms 2.0 ms 1.1 ms 0.6 ms 0.5 ms

2Br-AQ. MNI drift time 35.11 ms T °C\Eel 80 110 170 225 290

0 eV

0.3 eV

411 ms 198 ms 54.9 ms 20.3 ms 7.5 ms

119 ms 63.4 ms 21.1 ms 8.9 ms 3.7 ms

Figure 8. Schematic terms representation for the 2Br-AQ molecule, p- and s-anion states.

1I-AQ. MNI drift time 37.81 ms T °C\Eel 105 150 200 250 270

0 eV

0.1 eV

19.0 ms 3.17 ms 0.83 ms 0.27 ms 0.18 ms

9.52 ms 1.90 ms 0.544 ms 0.189 ms 0.129 ms

the evaluation of Emax evaluation are appropriate: it is reasonable22 to assume that only about 30±40% of the degrees of freedom can contribute to the energy concentration needed for the fluctuation dissociation. In other words, only about 35% of the degrees of freedom are active in the statistical fluctuation of vibrational energy, and therefore: Emax ˆ 0:35
…kT…3N

6† ‡ EA ‡ Eel †:

…2†

Thus, we see (Table 4) that there is a reasonable agreement of our evaluation of the average time of dissociation with the experimental data for the case of 1Cl-AQ (Fig. 4). Table 4 (column 0.4 eV for 1Cl-AQ, td = 32 ms) shows that it is possible to detect the metastable peak m* (m/z 5.06) corresponding to the process ClAQ → Cl ‡ AQ at a temperature of about 170 °C, at the drift time of the 1Cl-AQ ion 32 ms. In fact, this metastable peak is present in the spectrum for the appropriate temperature with maximum at the energy 0.44 eV. Calculation for the zero energy resonance predicts that Cl can be detected in the spectrum only at a high measurement temperature, T 290 °C, when td (290 °C) = 12.8 ms. Figure 4 confirms this conclusion, as the lower energy part of the CEY of m* overlaps the CEY of the molecular NI only at the temperature 290 °C. It is important to point out that, in the first approximation, the position of the crossing point does not influence the value of dissociation probability. In fact, if the transition between p- and s-terms takes place (with appropriate probability), we have a new pathway from the anion to the dissociation products, and only the energy difference Copyright # 2001 John Wiley & Sons, Ltd.

between them is important. The most important parameter in our evaluation of anion dissociation lifetime is the value of dissociation energy Elim. More precise evaluations should take into account the speed of movement of the nuclei when passing a crossing point.22 However, our estimates show that, even with our limitations, there is reasonable agreement between calculations by Eqn. (1) and the experimental data. Meanwhile, it is necessary to remember that the schemes of terms in Figs 7±9 are only a convenient method of visualization of very complicated multidimensional hyperplanes of potential energy, and it is not reasonable to expect it to be a complete description of reality. Nonetheless, Coulson pointed out that such a kind of representation is sufficiently correct in the case of chlorobenzene and similar molecules1. The same calculations were made for the case of 2Br-AQ, see Table 4 and Fig. 5. The behavior of the terms for the 2BrAQ molecule and of its anion states is shown in Fig. 8. The term of an anion formed in the ground electronic state p crosses the dissociative s-term, and with a probability of about 10 3 an anion may dissociate with Br ion formation. Calculated dissociation lifetimes are listed in Table 4. Parameters used are as follows: EA(2Br-AQ) = 1.81 eV, DE(C-Br) = 3.08 eV, EA(Br) = 3.36 eV, Elim = 1.71 eV. Equilibrium C-Br distances for the molecule and p-state of the Ê , and r C Br = 1.882 A Ê, anion are equal to r0C Br = 1.866 A according to PM3 calculations. Br ion formation is allowed energetically in the thermal electron energy region, but, in the low-temperature spectra, this process is not observed. This result is obvious from comparison of the CEY for Br and M . Evidently, the low-energy tails of these CEYs correspond to the Maxwellian distribution of electron energy at the working temperature of the electron gun cathode.9 Therefore, if the appearance energy of the Br ion is higher than that of M which formed by nuclear-excited resonance at zero energy, then it is clear that Br may not be produced at zero energy. The Br CEY overlaps the M CEY only at a Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

Dissociative electron capture by halogenated anthraquinones

Figure 9. Schematic terms representation for the 1I-AQ molecule, p- and s-anion states.

temperature of about 290 °C, see Fig. 5. Our calculations predict this effect at a temperature of 225 °C, see Table 4. A temporary NI m* (m/z 22.9), corresponding to the process BrAQ → Br ‡ AQ0, was recorded, see Fig. 5. This means that the characteristic dissociation lifetime of the molecular NI Br-AQ , td, is 20 ms at the energy 0 eV and the temperature 225 °C, and about 21 ms at the energy 0.3 eV and the temperature 170 °C. Clearly, the calculated temperature dependence is somewhat higher than the experimental one, which may be explained if we suppose that the number of active degrees of freedom in the case of 2Br-AQ is lower than for 1Cl-AQ. Such an assumption is reasonable because the typical vibration period of the C-Br bond is longer than that of the C-Cl bond, so there are fewer other vibrational modes that are able to contribute to the energy concentration in the reaction coordinate C-Hal. Thus, we have a reason to decrease the coefficient 0.35 in Eqn. (2) in the case of 2Br-AQ. However, the data in Table 4 were obtained using a coefficient value of 0.35; nevertheless calculated results are in reasonable agreement with experiment. The maximum of the CEY for the metastable peak m* clearly shifts to the lower energy region with increasing temperature, as expected. A quite different situation is apparent in the case of 1I-AQ, see Fig. 6 and Tables 1 and 4. The dominant channel of molecular NI dissociation is the channel of I formation. This peak is in the low-energy region and evidently forms at zero electron energy in the `cool' spectra. The temperature dependence exhibits a dramatic decrease of the M ion intensity with increasing temperature. Moreover, the CEY of the MNI has the same shape as that for SF6 /SF6. This means that the dissociation process is very rapid in the epithermal energy region. Semiempirical PM3 calculations predict the terms' behavior in the process of I-atom elimination, as shown in Fig. 9. The dissociative s-term is the ground electronic state of the anion, and the dissociation is allowed both energetically and by symmetry. Naturally, semiempirical calculation may not be strict proof of the inversion of empty levels in the process of electron capture by the 1I-AQ Copyright # 2001 John Wiley & Sons, Ltd.

1877

molecule, but we can use indirect evidence. If there is no symmetry ban for the dissociation, and therefore there is no radiationless transition involved in the process of dissociation, then the first coefficient accounting for this event in Eqn. (1) is equal to 1, in contrast to the cases of Cl- and Brsubstituted AQ derivatives. The results of the dissociation lifetime calculations are listed in Table 4. There is reasonable agreement between the lifetime evaluations obtained and the results observed. It is easy to see that with a coefficient of 10 3 the results would strongly disagree with the experiment. We believe this result is strong evidence for the violation of the frozen shell approximation in the case of 1IAQ ECNI-MS. There are many qualitative explanations for the effect of inversion of empty p* and s* levels in the process of electron capture by an aromatic molecule. For example, we can suppose that the Coulomb repulsion of the electron on the s*-orbital of C-Hal bond from the core p-electrons is weaker that in the case of p*-location, by analogy with the wellknown effect of destabilization of n → p* transitions with respect to p → p* transitions in absorption spectroscopy.19 Alternatively, it is possible to speculate about a polarization effect of an incident electron on the electronic shell of the target molecule. It is evident also that the conditions that possibly lead to the violation of the frozen shell approximation may be realized in other molecular processes, e.g. ionmolecule reactions, orbital-controlled chemical reactions, catalysis, etc. It is clear that the frozen shell approximation is only a convenient and simple method of description of certain cases of the electron-molecule interaction, but is not a strict and fundamental rule. Therefore, a discussion of the status of the frozen shell approximation is pertinent and useful. The final solution to this problem would involve direct experiments on molecular NIs (e.g. angular distribution of photoelectrons in laser photoelectron spectroscopy of negative ions25), and sophisticated quantum-chemistry investigations of MNI electron shells.

CONCLUSIONS The ECNI-MS data obtained were compared with the results of statistical calculations of dissociation lifetimes of molecular NIs of halogenated anthraquinones. It was shown that, in the cases of 1Cl-AQ and 2Br-AQ, molecular NIs dissociate according to Coulson's model1 through the radiationless transition between p- and s-terms. It has been demonstrated that, in the case of 1I-AQ, there is convincing evidence for the effect of violation of the frozen shell approximation in the process of electron capture by the molecule. It is shown that the MNI of 1I-AQ forms in a low-energy region in the ground electronic state of the s-type, and dissociates without symmetry ban violation. Thus, Koopmans' theorem has a limited applicability in the case of NI formation by electron capture.

Acknowledgement

The present work was sponsored by the Russian Foundation for Basic Research (grant # 00-02-16578). Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878

1878 N. L. Asfandiarov et al.

REFERENCES 1. Clarke DD, Coulson CA. J. Chem. Soc. A 1969; 1: 169. 2. Cooper CD, Naff WT, Compton RN. J. Chem. Phys. 1975; 63: 2752. 3. Landau LD, Lifshitz EM. Quantum Mechanics. Nauka: Moscow, 1989; 767. 4. Asfandiarov NL, Fal'ko VS, Fokin AI, Khvostenko OG, Lomakin GS, Lukin VG, Na®kova EP. Rapid Commun. Mass Spectrom. 2000; 14: 274. 5. Freeman PK, Srinivasa R, Campbell JA, Deinzer ML. J. Am. Chem. Soc. 1986; 108: 5531. 6. LarameÁe JA, Mazurkiewich P, Berkout V, Deinzer ML. Mass Spectrom. Rev. 1996; 15: 15. 7. Asfandiarov NL, Fal'ko VS, Fokin AI, Lomakin GS, Pozdeev NM, Podkopaeva OY, Chizhov YV. Rapid Commun. Mass Spectrom. 1998; 12: 595. 8. Koopmans T. Physica 1934; 1: 104. 9. Mazurkiewicz PH, Deinzer ML. J. Am. Chem. Soc. 1999; 121: 3421. 10. Berkout VD, Mazurkievicz P, Deinzer ML. J. Am. Chem. Soc. 1999; 121: 2561. 11. Staley SW, Strnad JT. J. Phys. Chem. 1994; 98: 116. 12. Beland FD, Farwell SO, Callis PR, Geer RD. J. Electroanal. Chem. 1977; 78: 145. 13. Khvostenko VI. Negative Ion Mass Spectrometry in Organic Chemistry. Nauka: Moscow, 1981; 159. 14. Asfandiarov NL, Fokin AI, Lukin VG, Na®kova EP, Lomakin GS, Fal'ko VS, Chizhov YV. Rapid Commun. Mass Spectrom. 1999; 13: 1116.

Copyright # 2001 John Wiley & Sons, Ltd.

15. Christophorou LG, Grant MW, McCorkle DL. Adv. Chem. Phys. 1977; 36: 413. 16. Khvostenko VI, Vorob'ev AS, Khvostenko OG. J. Phys. B: Atom. Mol. Phys. 1990; 23: 1975. 17. Fokin AI, Na®kova EP, Lomakin GS, Asfandiarov NL. Electronic J. ªInvestigated in Russiaº, 029/000325, 412 (2000). Available: http://zhurnal.ape.relarn.ru/articles/2000/029. pdf 18. Na®kova EP, Fal'ko VS, Fokin AI, Lomakin GS, Lukin VG, Pshenichnyuk SA, Asfandiarov NL. Proceedings of the Intern. Symposium on Electron-Molecule Collisions and Swarms. July 14±16, 2001, Lincoln, Nebraska USA. (in press). 19. Strokach NS, Shigorin DN, Scheglova NA. Electron-Vibration Spectra of Polyatomic Molecules. Nauka: Moscow, 1982; 144. 20. Christophorou LG, Carter JG, Christodoulides AA. Chem. Phys. Lett. 1969; 3: 237. 21. Collins PM, Christophorou LG, Chaney EL, Carter JG. Chem. Phys. Lett. 1970; 4: 646. 22. Nikitin EE. Theory of Elementary Atomic-Molecular Processes in the Gas Phase. Khimiya: Moscow, 1970; 455. 23. NIST Chemistry WebBook. Available: http://webbook.nist. gov/chemistry/. 24. Gurevich LV, Karachevtsov GV, Kondratev VN, Lebedev YA, Medvedev VA, Potapov VK, Hodeev YS. Chemical Bonds Break Energies. Ionization Potentials and Electron Af®nity. Nauka: Moscow, 1974; 351. 25. Ervin KM, Ho J, Lineberger WC. J. Phys. Chem. 1988; 92: 5405.

Rapid Commun. Mass Spectrom. 2001; 15: 1869±1878