Long Bonds and Short Barriers: Ionization and Isomerization of Alkyl

Dec 11, 2009 - Long Bonds and Short Barriers: Ionization and Isomerization of Alkyl Nitriles. Paul M. Mayer* ... Jaspreet Kaur , Bobby Antony. Journal...
0 downloads 0 Views 1MB Size
Download PDF
J. Phys. Chem. A 2010, 114, 867–878

867

Long Bonds and Short Barriers: Ionization and Isomerization of Alkyl Nitriles Paul M. Mayer,*,† Martyn F. Guest,‡ Louise Cooper,§ Larisa G. Shpinkova,| Emma E. Rennie,†,⊥ David M. P. Holland,# and David A. Shaw# Department of Chemistry, UniVersity of Ottawa, 10 Marie-Curie, Ottawa, Canada K1N 6N5, AdVanced Research Computing, Cardiff UniVersity, Redwood Building, King Edward VII AVenue, Cardiff CF10 3NB Wales, U.K., Department of Chemistry, Heriot-Watt UniVersity, Riccarton, Edinburgh EH14 4AS, U.K., Department of Nuclear Spectroscopy Methods, Institute of Nuclear Physics, Moscow State UniVersity, Moscow 119899, Russia, and Daresbury Laboratory, Daresbury, Warrington, Cheshire WA4 4AD, U.K. ReceiVed: October 8, 2009; ReVised Manuscript ReceiVed: NoVember 16, 2009

Ab initio molecular orbital calculations demonstrate that ionizing alkyl nitriles produces a dramatic geometry change involving lengthening of a C-CH2CN bond. The experimental determination of the adiabatic ionization energy of these species is thus very difficult. In addition, there are generally low barriers for 1,2-H shift reactions in the molecular ions leading to RCHCHN+• and RCHCNH+• isomers, which makes generating pure ionized alkyl nitrile in a mass spectrometer a challenge. Threshold photoelectron spectroscopy and threshold photoelecton photoion coincidence spectroscopy were employed to study the ionization and dissociation of two alkyl nitriles, in particular, pentanenitrile and 2,2-dimethylpropanenitrile. Threshold ionization is shown to result not in the respective molecular ions, but rather in isomeric forms, resulting in dissociation thresholds that lie below the calculated adiabatic ionization energies of the two molecules. Appearance energies for all observed fragment ions are reported and compared to available literature values. Charge separation in the dissociation of doubly ionized 2,2-dimethylpropanenitrile is observed as fragmention time-of-flight peak broadening at high photon energies. Introduction In the gas phase, acetonitrile and other nitriles are known constituents of, and major nitrogen carriers in, Titan’s atmosphere and interstellar space.1-3 When exposed to UV radiation, small ice-bound alkyl nitriles were shown to isomerize to ketenimine forms.4 Evidence for this phenomenon has also been found in interstellar clouds.5 This process occurs even in the dilute gas phase. It has previously been reported that it is very difficult to generate pure ionized acetonitrile CH3CN+• in the gas phase because of a low barrier for a 1,3-H shift to produce the ketenimine isomer, CH2CNH+•, lying 232 kJ mol-1 lower in energy.6 de Petris et al.6 demonstrated that the two isomers could not be distinguished based on their collision-induced dissociation (CID) mass spectra, as pure CH3CN+• could not be made by electron ionization (EI). They probed the relative populations of the two ions by charge- and proton-transfer experiments.6 These experimental results were consistent with a detailed computational treatment of the C2H3N+• potential energy surface by Choe.7 The 1,3-H shift barrier between the two isomers lies only 70 kJ mol-1 above CH3CN+• (G3 level of theory).6 In this work, we have extended the investigation to the higher alkyl nitrile homologues and computationally and experimentally (via threshold photoelectron spectroscopy) explored the generation of these ionic species. The effect of isomerization in the * Corresponding author. E-mail: [email protected]. Fax: (613) 5625170. † University of Ottawa. ‡ Cardiff University. § Heriot-Watt University. | Moscow State University. ⊥ Present address: Varian, Inc., 2700 Mitchell Drive, Walnut Creek, CA 94598. # Daresbury Laboratory.

molecular ions is demonstrated by threshold photoelectron photoion coincidence (TPEPICO) spectroscopy experiments on pentanenitrile (CH3CH2CH2CH2CN) and its isomer, 2,2-dimethylpropanenitrile [(CH3)3CCN]. Experimental Procedures The pulsed TPEPICO spectrometer8 and the 5-m normal incidence monochromator9 attached to the Daresbury Laboratory synchrotron radiation source have been described in detail previously. The coincidence spectrometer employs a pulsed extraction technique,10 the principal advantage of which is that threshold electrons can be detected with a high energy resolution while the associated ions can also be collected with a high mass resolution. In the present arrangement,8 a very low electric field is applied initially across the interaction region to extract threshold electrons. The detection of the electron triggers the application of a high-voltage (∼1 kV) pulse across the interaction region to draw the ion toward the drift tube, and initiates the time-of-flight (TOF) measurement. The time between the arrivals of the electron and the associated ion is measured electronically, with the summation of many events producing a TOF spectrum. With the present apparatus, the ion source residence time (time between the ionization event and the application of the ion extraction pulse) has been measured as 1.116 ( 0.050 µs, using the experimental procedure described in Holland et al.8 Threshold photoelectron spectra were recorded in the binding energy range from ∼10 to 25 eV (pentanenitrile) and from ∼10 to 24 eV (2,2-dimethylpropanenitrile) using the electron detection part of the coincidence spectrometer. The spectra were normalized to variations in the incident photon intensity using the signal from a photomultiplier that monitored the monochromated radiation after it impinged upon a sodium salicylate coated

10.1021/jp9096524  2010 American Chemical Society Published on Web 12/11/2009

868

J. Phys. Chem. A, Vol. 114, No. 2, 2010

Mayer et al.

screen. The threshold photoelectron spectra were measured at a photon resolution of 0.1 nm full width at half-maximium (fwhm) (∼18 meV at hν ) 15 eV). Lithium fluoride or indium filters could be inserted into the photon beam exiting the monochromator to help suppress higher-order radiation. The binding energy scale was calibrated by recording a threshold photoelectron spectrum of a gas mixture comprising sample, argon, and xenon. Pentanenitrile and 2,2-dimethylpropanenitrile were subjected to three freeze-pump-thaw cycles prior to their introduction into the spectrometer.

involves convoluting the theoretical k(E) curves with the parention thermal (rotational and vibrational) energy, the monochromator band-pass function (photon resolution ∼8 meV), and the threshold electron analyzer transmission function (width of ∼9 meV fwhm) derived from a threshold electron spectrum obtained from the photoionization of krypton in the region of the 2P1/2 ionization limit under the conditions used in the TPEPICO measurements. The activation energy, E0, and the transitionstate vibrational frequencies were varied in the RRKM calculation until a satisfactory fit to the experimental breakdown curves was obtained.

Computational Procedures Standard ab inito molecular orbital and density functional theory calculations11 were carried out with the Gaussian 98 suite of programs.12 Vibrational frequencies were calculated for all optimized geometries to verify that they corresponded to equilibrium structures, except for those optimized at the B3LYP/6-311++G(3df,2p) and QCISD/6-31+G(d,p) levels of theory. Vibrational frequency scaling factors were chosen from the Computational Chemistry Comparison and Benchmark DataBase.13 Single-point energy calculations at the geometries specified in the text were performed with the G3//B3-LYP14,15 and CBS-RAD(B3-LYP,B3,LYP)16 composite methods, employing B3-LYP/6-31+G(d,p) geometries and zero-point vibrational energies (ZPEs). Neutral and ion enthalpies of formation were derived using the atomization method outlined by Nicholaides and Radom.17 Thermal corrections incorporated standard H298 - H0 terms for the elements. For ionized acetonitrile, care was taken to explicity read in the calculated force constants for the optimized neutral molecule with C1 symmetry. Not doing this, or reading in those for the neutral optimized under C3V symmetry, resulted in the selfconsistent field (SCF) procedures in the QCISD(T), MP4, and MP2 calculations in the G3 protocol finding an excited, and not ground, state of the ion. Vertical ionization energies and their relative spectral intensities (pole strengths) of 2,2-dimethylpropanenitrile and pentanenitrile were obtained using the outer valence Green’s function (OVGF) approach.18 The calculations employed the GAMESS-UK19,20 suite of programs, with a variety of basis sets. For consistency, the geometries were fully optimized at the MP2 level using both cc-pVTZ and cc-pVDZ basis sets21-24 and resulted in symmetries of C3V for 2,2-dimethylpropanenitrile and Cs for pentanenitrile. The OVGF calculations were performed in Cs symmetry using both cc-pVDZ and cc-pVTZ basis sets. Therefore, a doubly degenerate (e) state in 2,2-dimethylpropanenitrile in C3V symmetry transforms into two states (a′ and a′′) in Cs symmetry, having the same energies. Six inner-shell orbitalssthe 1s functions on each C and Nswere frozen. The lowest-energy dissociation channels were modeled with the standard Rice-Ramsperger-Kassel-Marcus (RRKM) rate expression

k(E) )

‡ σ N (E - E0) h F(E)

(1)

where k(E) is the unimolecular rate constant at an ion internal energy of E, σ is the reaction degeneracy or symmetry number, h is Planck’s constant, E0 is the 0 K activation energy, F(E) is the reactant ion rovibrational density of states, and N‡(E - E0) is the transition-state rovibrational sum of states.25,26 The density and sum-of-states calculations employed the direct count algorithm of Beyer and Swinehart,27 using the calculated B3LYP/6-31+G(d,p) vibrational frequencies and rotational constants as outlined in the text. The next step in the analysis28-30

Results and Discussion Geometries of Ionized Nitriles: An Assessment of Theoretical Approaches. The structures of neutral and ionized acetonitrile (CH3CN), propanenitrile (CH3CH2CN), butanenitrile (CH3CH2CH2CN), pentanenitrile (CH3CH2CH2CH2CN), and 2,2-dimethylpropanenitrile [(CH3)3CCN] optimized at the B3LYP/6-31+G(d,p) level of theory are shown in Figure 1, along with selected geometrical parameters. For acetonitrile, the neutral and ion geometries are similar and result in similar vertical and adiabatic first ionization energies (see below). Starting with propanenitrile and continuing through the rest of the homologous series to pentanenitrile, the ions all exhibit a common feature: a long C-C bond between the CH2CN moiety and the rest of the alkyl chain. This is accompanied by an increased planarity of the two methylene groups involved in the long bond. Similarly, ionized 2,2-dimethylpropanenitrile also exhibits a long C-C bond (Figure 1). As a result, the optimized radical cations of all five alkyl nitriles exhibit significant spin contamination in their HartreeFock (HF) wave functions (〈S2〉 ) 0.9-1). The bond elongation allows the CR2CN (R ) H, CH3) moiety to more closely resemble a delocalized system with a partial CR2-CN double bond (evident from the bond lengths in Table 1). Such delocalized radical systems often suffer from significant spin contamination. For this reason, UMP2 was not considered suitable for geometry optimization, as it has been found to be particularly susceptible to errors in these cases.16,31-33 The vertical ion structures (i.e., ions having the same geometries as their corresponding neutral molecules) do not suffer from spin contamination, with HF wave functions exhibiting 〈S2〉 values of approximately 0.75. When spin contamination is significant, density functional theory has been found to be a more reliable approach to free-radical geometries.16,31-33 To ascertain whether this long bond is a computational artifact or is indeed real, the geometry of ionized propanenitrile was optimized at a variety of levels of theory. Table 1 lists selected geometric parameters (see Figure 2 for definitions) for this ion obtained with the B3-LYP density functional method employing a variety of basis sets to determine basis-set effects on the geometry. Moreover, to explore the method used, two other density functional methods, B3-PW91 and BHandH-LYP, and the QCISD ab initio method were used with the 6-31+G(d,p) basis set (Table 1). Even with spin contamination, QCISD has been found to be a reliable method for investigating radicals.16 It is apparent from the results in Table 1 that there is almost no difference in the geometry of ionized propanenitrile among the employed basis sets. The largest effect is a 0.01 Å bond length increase in the C3C2 bond in going from a double-split valence to a triple-split valence basis set [6-31+G(d,p) vs 6-311+G(d,p)]. The C1N and C2C1 bonds in the CH2CN moiety exhibit a decrease of ∼0.01 Å in going from small to large basis sets. There is a slightly more pronounced effect when the density

Ionization and Isomerization of Alkyl Nitriles

J. Phys. Chem. A, Vol. 114, No. 2, 2010 869

Figure 1. B3-LYP/6-31+G(d,p)-optimized geometries for neutral and ionized acetonitrile, propanenitrile, butanenitrile, pentanenitrile, and 2,2dimethylpropanenitrile, along with selected geometric parameters. Bond lengths are in angstroms; angles are in degrees.

TABLE 1: Comparison of Selected Optimized Geometric Parameters for Ionized Propanenitrilea r(C3H3)

∠(H1C2C1)

∠(H1 C2C1H1)

∠(H3C3H2H2)

B3-LYP/ 1.088 1.088 1.086 1.085 1.083

1.096 1.096 1.094 1.093 1.092

114.0 114.5 114.3 114.5 114.6

134.9 136.4 135.8 136.2 136.5

-132.0 -132.5 -132.9 -132.9 -132.9

6-31+G(d,p) 1.094 1.088 1.094 1.089 1.085 1.080 1.089 1.085

1.096 1.101 1.087 1.091

114.0 114.1 113.8 113.0

134.9 134.4 134.3 132.3

-132.0 -128.5 -128.7 -127.8

r(C1N)

r(C2C1)

r(C3C2)

r(C2H1)

6-31+G(d,p) 6-31+G(2df,p) 6-311+G(d,p) 6-311+G(2df,p) 6-311++G(3df,2p)

1.198 1.194 1.190 1.187 1.187

1.398 1.393 1.392 1.389 1.389

1.725 1.730 1.735 1.732 1.733

1.094 1.093 1.092 1.091 1.089

B3-LYP B3-PW91 BHandH-LYP QCISD

1.198 1.201 1.191 1.215

1.398 1.396 1.387 1.408

1.725 1.684 1.688 1.668

a

r(C3H2)

Bond lengths (r) in angstroms; angles (∠) in degrees. Refer to Figure 2 for definitions.

functional method is changed; the most significant change is a shortening of the C3C2 bond by 0.041 Å with B3-PW91 and 0.037 Å with BHandH-LYP (Table 1). The B3-PW91 geometry is the best approximation of the QCISD geometry. QCISD shortens the C3C2 bond further, to 1.668 Å, with the C1N and C2C1 bonds in the CH2CN moiety both elongating slightly. As a consequence of the similarity of the various optimized geometries, it is not surprising that the scaled ZPE values listed in Table 2 are all also virtually identical.

Why the Long Bond? What is the driving force for the elongation of the C-C bond in ionized nitriles? Figure 3 displays the spin densities and partial charges on the chemical groups in the equilibrium and vertical ion structures for propanenitrile, butanenitrile, and pentanenitrile. In general, the spin density is spread out slightly more over the alkyl side of the long C-C bond in the equilibrium ion structures than in the vertical ion structures. Threshold ionization will occur from the neutral highest occupied molecular orbital (HOMO), which

870

J. Phys. Chem. A, Vol. 114, No. 2, 2010

Mayer et al.

Figure 2. Atom definitions for propanenitrile used in Table 1.

TABLE 2: Comparison of Zero-Point Vibrational Energies (ZPEs) for Ionized Propanenitrile Obtained at a Variety of Levels of Theorya ZPE (hartrees)

ZPE (kJ mol-1)

scaled ZPE (kJ mol-1)

6-31+G(d,p) 6-31+G(2df,p) 6-311+G(d,p) 6-311+G(2df,p)

B3-LYP/ 0.070085 0.069841 0.069701 0.069742

184.0 183.4 183.0 183.1

177.4 177.3 176.6 177.4

B3-LYP B3PW91 BHandHLYP

6-31+G(d,p) 0.070085 184.0 0.070359 184.7 0.073014 191.7

177.4 177.3 178.1

For ease of comparison, scaled values employed the recommended scaling factors for vibrational frequencies listed in the Computational Chemistry Comparison and Benchmark DataBase: from top 0.964, 0.967, 0.965, 0.969, 0.964, 0.96, 0.929.13

Figure 3. Spin density values and partial charges (from Mullikin population analysis) for the chemical groups in ionized (equilibrium and vertical geometries) propanenitrile, butanenitrile, and pentanenitrile based on B3-LYP/6-31+G(d,p) geometries. Values might not sum to +1 spin and +1 charge because of rounding.

is a CN π bond. Stretching the C-C bond allows the alkyl chain to take on more radical character. It also allows the CR2CN moiety to become more like the CR2CN radical and cation, as the CN group is a net stabilizer of charge34 and spin33 in these species. Nitrile Ionization Energies and Enthalpies of Formation. Table 3 lists the G3//B3-LYP adiabatic ionization energies (IEa) for the five alkyl nitriles, along with known experimental values. The best experimental value is for acetonitrile, and as anticipated, the agreement between theory and experiment is excellent. For the other nitriles, the G3//B3-LYP values tend to lie between the photoelectron spectroscopy (PES) results of Ohno et al.35 and the PES values from Staley et al.36 and the photoionization values reported by Watanabe et al.37 For comparison, the IEa of propanenitrile was also calculated at the G3 level employing the BHandH-LYP/6-31+G(d,p) geometry and ZPE and the QCISD/6-31+G(d,p) geometry with the B3-LYP/6-31+G(d,p) ZPE, as well as the CBS-RAD(B3-LYP,B3-LYP) method with the B3-LYP/6-31+G(d,p) geometry and ZPE (Table 3). All of the results are very similar, indicating that there is no geometry effect on the calculated IEa. The difficulty in assigning the IEa values for these nitriles based on the observed onset of either the PES spectra or the PI onset of the molecular ions rests in the large geometry difference between the ion and neutral in these cases. This is demonstrated in the calculated differences between the vertical and adiabatic IE values for the five alkyl nitriles (Table 4). For acetonitrile, the difference is quite small (although the calculated difference is larger than the experimentally observed difference), and so, the experimental onset is easily deduced. For the other nitriles, the difference can range up to 1.46 eV, making the true onset to ionization almost impossible to determine in a photoionization or photoelectron spectroscopy experiment. It would likely require a chargetransfer equilibrium experiment, in which the ion has a chance of relaxing to its equilibrium structure in a collision complex, to experimentally access IEa values in these cases. A similar situation was recently demonstrated for hydrazine derivatives.38 Many thermochemical parameters of molecules and ions can be considered to be additive in nature. One example is the well-

known empirical trend in which plotting IEa values as a function of the reciprocal of molecule size (1/n, where n is the number of atoms in the molecule) yields a straight line for homologous series of compounds.39 The elongation of the C-C bond in the higher linear-nitrile homologues makes them distinct from, and not true homologues of, acetonitrile, as is evident from the discontinuity in the plot in Figure 4. Note that 2,2-dimethylpropanenitrile, by definition, is not a member of this homologous series. The new calculated adiabatic IE values yield new, more reliable values for the ∆fH of the alkyl nitriles (Table 5). The values obtained at the G3//B3-LYP level of theory for the neutral alkyl nitriles are included for comparison, and as expected, all are in very good agreement with the available experimental values. The CBS-RAD value for neutral and ionized propanenitrile is included because this approach it should perform better for spin-contaminated radicals.16 As can be seen from Table 5, the G3//B3-LYP and CBS-RAD values for the ion are in excellent agreement. There is a small difference between the two ∆fH values for the neutral, leading to the 0.03 eV difference in IEa seen in Table 3. Threshold Photoelectron Spectroscopy. The valence-shell threshold photoelectron spectra of 2,2-dimethylpropanenitrile and pentanenitrile are plotted in Figure 5, together with the OVGF ionization energies and relative spectral intensities. In 2,2-dimethylpropanenitrile, the predicted ionization energies for the outer valence orbitals agree remarkably well with the observed values, thereby enabling structure in the experimental spectrum to be correlated with ionization from specific molecular orbitals. Although the agreement between the calculated and experimental ionization energies is slightly less satisfactory for pentanenitrile (Figure 5b), the theoretical results still allow a reasonable interpretation to be proposed for the observed broad bands. The experimental vertical ionization energies, obtained through inspection of the threshold photoelectron spectra, are listed in Tables 6 and 7. In both molecules, the OVGF calculations demonstrate that the molecular orbital model of ionization40 holds for all but the two most tightly bound of the 17 valence orbitals.

a

Ionization and Isomerization of Alkyl Nitriles

J. Phys. Chem. A, Vol. 114, No. 2, 2010 871

TABLE 3: Comparison of Calculated and Experimentally Measured Adiabatic Ionization Energies (IEa) for Five Alkyl Nitriles IEa (eV) G3//B3-LYP CH3CN CH3CH2CN CH3CH2CH2CN CH3CH2CH2CH2CN (CH3)3CCN

G3//BHandH-LYP

G3//QCISD

CBS-RAD//B3-LYP

experiment a,c

12.20 11.79 11.34 11.13 11.07

11.79

11.79

11.76

12.20, 12.1b 11.84 ( 0.02,d 11.85,c 11.6b 11.67 ( 0.05,d 11.2b 11.85e

a

d

Photoelectron spectroscopy, Gochel-Dupuis et al.56 b Photoelectron spectroscopy, Ohno at el.35 c Photoelectron spectroscopy, Staley et al.36 Photoionization, Watanabe et al.37 e Electron ionization, Heerma and de Ridder.42

TABLE 4: Comparison of Calculated and Experimentally Measured Differences between Vertical (IEv) and Adiabatic (IEa) Ionization Energies for Ionized Alkyl Nitriles IEv - IEa (eV) CH3CN CH3CH2CN CH3CH2CH2CN CH3CH2CH2CH2CN (CH3)3CCN

G3//B3-LYP

experiment

0.23 0.40 0.74 0.69 1.46

0.05,a 0.09b 0.30a,b 0.54b

a Photoelectron spectroscopy, Staley et al.36 spectroscopy, Ohno at el.35

b

Photoelectron

Staley et al.36 have recorded the He I excited photoelectron spectrum of 2,2-dimethylpropanenitrile but did not attempt to assign the bands or provide ionization energies. A comparison between the He I excited spectrum36 and the present threshold photoelectron spectrum reveals several bands having a common origin, although the relative intensities of the corresponding bands might differ considerably. For example, in the He I excited spectrum the peak with the highest intensity occurs at a binding energy of ∼12.5 eV, whereas in the threshold photoelectron spectrum, the most prominent band appears at 14.55 eV. Such variations in relative intensities are typical of those observed in small polyatomic molecules.38,41 Experimentally, it has been found that the relative intensities of photoelectron bands associated with weakly bound orbitals tend to be greater in conventional spectra than those in threshold photoelectron spectra, whereas the converse holds for the more tightly bound orbitals. The enhancement in the latter bands can be attributed to resonant autoionization from numerous superexcited (Rydberg or valence) states. Resonant autoionization, an indirect twostep process, often influences threshold photoelectron spectra and can affect both the band profile and the relative intensity. In contrast, the signal in a He I excited photoelectron spectrum arises predominantly through direct photoionization. Within this context, it should be noted that the OVGF pole strengths correspond to the relative intensities for direct transitions between the initial neutral ground state and the final ionic states. Indirect processes, such as autoionization, are not taken into account. Therefore, the calculated pole strengths, plotted in Figure 5, might not correspond quantitatively with the measured threshold photoelectron band intensities. According to the OVGF calculations for 2,2-dimethylpropanenitrile (Table 6), ionization from the two least tightly bound orbitals (7a′′ and 16a′) gives rise to the broad band with a maximum at 11.78 eV and an onset at ∼10.95 eV, which is only slightly lower than the calculated value of 11.07 eV (Table 3). In addition to the large geometry change upon ionization (which normally makes the experimental IEa greater than the true value), the calculations also indicate that isomerization plays an important role in the threshold region (see below). As a result of these two effects, the onset observed at ∼10.95 eV might

not correspond to the true adiabatic ionization energy. On the other hand, the experimental value (11.78 eV) for the vertical ionization energy should be valid. The OVGF ionization energies of 12.68 and 12.90 eV for the 14a′ and the (15a′ and 6a′′) orbitals, respectively, correlate excellently with the doublet observed at 12.64 and 12.89 eV. Ionization from the next two orbitals (5a′′ and 12a′), with calculated energies of 13.79 and 13.96 eV gives rise to a broad band with a maximum at 13.79 eV, whereas the most prominent band in the threshold photoelectron spectrum, with a maximum at 14.55 eV, can be attributed to the 13a′ and 4a′′ orbitals with predicted ionization energies of 14.61 eV. The final two bands in the outer valence region, with observed vertical ionization energies of 15.79 and 17.01 eV, can be associated with the (11a′ and 3a′′) and 10a′ orbitals, respectively. In the inner valence region, the OVGF ionization energies are systematically higher than the experimental values. Nevertheless, the peak occurring at 20.4 eV evidently correlates with the 9a′ orbital, and the peak appearing at 22.6 eV originates from the 8a′ and 2a′′ orbitals. It is noticeable that the peaks associated with these inner valence orbitals are significantly less intense that those associated with the outer valence orbitals. Moreover, they are superimposed upon a substantial continuum. This indicates that the molecular orbital model of ionization40 no longer holds for these inner valence orbitals and that electron correlation leads to the intensity associated with a particular orbital being redistributed among numerous satellite states. The overlapping of these satellites results in the observed continuum. Neither a He I excited nor a threshold photoelectron spectrum of pentanenitrile has been reported previously. The present threshold photoelectron spectrum, shown in Figure 5b, displays a very broad but structured feature in the binding energy range ∼11-18 eV, due to many overlapping bands. The overall intensity distribution also suggests that resonant autoionization is strongly affecting the electron yield. This spectral congestion is confirmed by the theoretical results. According to the OVGF calculations, the ionization energies of the first outermost orbitals are separated by a little over 1 eV. It appears reasonable to associate the prominent peak at 12.84 eV with two, or possibly all three, of the 11a′, 4a′′, and 10a′ orbitals, whose predicted ionization energies lie at 12.56, 12.59, and 12.81 eV, respectively. The OVGF energies for the two outermost orbitals, 12a′ and 5a′′, are 11.72 and 11.93 eV, respectively, and fall in a region where the experimental intensity exhibits a gradual rise, with minor structure at 11.91 and 12.26 eV. It is not clear whether these two energies should be considered to correspond to the vertical ionization energies of the 12a′ and 5a′′ orbitals. As in the case of 2,2-dimethylpropanenitrile, a combination of the change in molecular geometry between the neutral and ionic states and potential isomerization near the ionization threshold (see below) means the observed onset of the threshold photoelectron signal at ∼10.80 eV is not associated with the adiabatic

872

J. Phys. Chem. A, Vol. 114, No. 2, 2010

Mayer et al.

Figure 4. Plot of G3//B3-LYP IEa vs 1/n for the four linear-chain alkyl nitriles. The equation for the line of best fit to the data for propanenitrile through pentanenitrile is shown.

TABLE 5: Comparison of Calculated and Experimentally Measured ∆fH° Values for the Five Alkyl Nitriles and Their Ionsa G3//B3-LYP

CBS-RAD

∆fH°0 ∆fH°298 ∆fH°0 ∆fH°298 CH3CN CH3CH2CN CH3CH2CH2CN CH3CH2CH2CH2CN (CH3)3CCN CH3CN+• CH3CH2CN+• +•

CH3CH2CH2CN

+•

CH3CH2CH2CH2CN (CH3)3CCN+• a

80 68 53 38 20 1257 1205

70 52 31 10 -8 1249 1191

1147

1127

1112 1063

1086 1088

-1

72

56

1207

1193

literatureb ∆fH°298 74.04 ( 0.37 51.5 31.2 11.1 -3.3 1251 1194 ( 2, 1195, 1171 1157 ( 5, 1112 1140

b

All values in kJ mol . Neutral ∆fH°298 values taken directly from the NIST database. Values for the ions were derived from these neutral ∆fH°298 values and the IE values quoted in Table 3.55

ionization energy. At higher binding energies, the broad doublet with maxima at 13.86 and 14.30 eV probably corresponds to ionization from the 9a′, 3a′′, and 8a′ orbitals, whereas another doublet, observed at 15.35 and 15.71 eV, arises from the 7a′ and 2a′′ orbitals, whose ionization energies are predicted to be almost identical. Finally, the two remaining outer valence orbitals, 1a′′ and 6a′, with calculated ionization energies of 17.12 and 17.21 eV, respectively, correlate with the experimental band at 17.19 eV. As with 2,2-dimethylpropanenitrile, although the predicted ionization energies for the inner valence orbitals of pentanenitrile lie consistently higher than the measured values, the theoretical results still allow an interpretation of the structure to be proposed. The calculations suggest that the peaks appearing at 19.7, 21.7, and 24.2 eV should be associated with the 5a′, 4a′, and 3a′ orbitals, respectively. These three inner valence peaks are particularly weak, thereby indicating that electron correlation is leading to a significant redistribution of main-line intensity. What Happens When an Alkyl Nitrile Is Ionized? We have also explored aspects of the rearrangement chemistry of the alkylnitriles. The results are summarized in the potential energy diagrams in Figure 6 obtained at the B3-LYP/6-31+G(d,p) level of theory. For ionized propanenitrile (1), three isomers were

Figure 5. Threshold photoelectron spectra of (a) 2,2-dimethylpropanentrile and (b) pentanenitrile, together with the theoretical results (Tables 6 and 7). The heights of the bars are proportional to the calculated relative spectral intensities (pole strengths).

calculated: the distonic isomer •CH2CH2CNH+ (1a), methylketenimine CH3CHCNH+• (1b), and the 1,2-H shift isomer CH3CHCHN+• (1c). The ketenimine isomer 1b is the lowestenergy species, lying 212 kJ mol-1 below 1. The distonic ion 1a is also much lower in energy, 151 kJ mol-1 below 1. Ion 1c lies only 19 kJ mol-1 below 1. The equivalent of ion 1c does not appear as an equilibrium structure on the C2H3N+• surface, but rather is the 1,3-H shift transition state between CH3CN+• and CH2CNH+•. In the case of ionized propanenitrile, the conversion of 1 to 1b occurs via two sequential 1,2-H shift reactions through 1c (Figure 6). Both of these transition states lie only ∼42 kJ mol-1 above 1, indicating a facile hydrogen-

Ionization and Isomerization of Alkyl Nitriles

J. Phys. Chem. A, Vol. 114, No. 2, 2010 873

TABLE 6: Experimental and OVGF/cc-pVTZ Vertical Ionization Energies of 2,2-Dimethylpropanenitrile

TABLE 7: Experimental and OVGF/cc-pVTZ Vertical Ionization Energies of Pentanenitrile

hopping mechanism across the partially delocalized CH2CN moiety. An independent 1,3-H shift transition state connecting 1 and 1b could not be located, but if it exists, it is unlikely to be lower in energy than the two 1,2-H shift reactions. Evident from this surface is that ionized propanenitrile will be even more difficult to generate in its pure state than ionized acetonitrile. RRKM calculations of the interconversion of 1 and 1b predict the rate constant to be 107 s-1 just 1 kJ mol-1 above the barriers. A similar situation exists for ionized butanenitrile (2) and pentanenitrile (3). Obviously, many more H-shift reactions can take place in these ions, and an extensive investigation of this chemistry is outside the scope of this study. Rather, the reactions analogous to those presented above for ionized propanenitrile are summarized in Figure 6. Isomers CH3•CHCH2CNH+ (2a), CH3CH2CHCNH+• (2b), and CH3CH2CHCHN+• (2c) and CH3CH2•CHCH2CNH+ (3a), CH3CH2CH2CHCNH+• (3b), and CH3CH2CH2CHCHN+• (3c) do not lie as low in energy relative to 2 and 3, respectively, as they do in the case of 1. This is due to the increasing stabilization of 2 and 3 from the longer alkyl chains. As was discussed above, the charge and spin densities in 1-3 become increasingly spread out over the alkyl group with increasing chain length (Figure 3). This lowers the energy of the nitrile ions relative to their isomers, in which the charge is largely localizedson the CH2CN moiety (∼70%) for isomers a, the CH group (∼65%) for isomers b, or the CHCH group (∼70%) for isomers csand thus are not greatly affected by extending alkyl substitution remote from the charge site. As the energies of the isomers increase relative to those of the nitrile ions, so do the barriers for interconversion, from ∼42 kJ mol-1 for 1 to 1b to 81-84 kJ mol-1 for 2 to 2b to 95-100 kJ mol-1

for 3 to 3b (Figure 6). So, it should be increasingly easier to generate pure ionized nitrile as the length of the alkyl chain increases. The RRKM rate constant for the isomerization of 3 to 3c, for instance, is much lower than it is for 1 to 1c, only reaching 74 kJ mol-1 above the transition state (3 to 3c). The internal energy distribution of the ions will also play a role here. If pentanenitrile ions are formed with a 298 K thermal internal energy distribution, the main portion of the distribution extends up to an internal energy of 50 kJ mol-1, which is 50% of the lowest isomerization barrier. If the ions are formed by electron ionization of the neutral, a much broader internal energy distribution will result, and some of these ions will undoubtedly be able to rearrange to 3b. Exploring the fragmentation of these alkyl nitrile ions with classical mass spectrometry techniques could be fraught with difficulty because any dissociation reaction will almost certainly occur above the threshold for the isomerization to the ketenimine forms 1b, 2b, and 3b. Ionized 2,2-dimethylpropanenitrile (4) exhibits a related chemistry. There are no neighboring H-atoms to undergo 1,2-H shift reactions across the nitrile functionality. Heerma and de Ridder42 proposed that fragmentation of this ion is initiated by H loss to form (CH3)2(CH2)CCN+•, followed by H-atom transfer to the nitrogen atom to form (CH3)(CH2)2CCNH+•, which then undergoes a methyl group shift to make CH3CNH+ (m/z 42) + C3H4. However, as demonstrated below, this three-step process requires much too high an energy to be competitive. A 1,4-H shift reaction could make 4a, (CH2)(CH3)2CCNH+•, but an equilibrium species with this connectivity could not be optimized (it hits a point where the SCF simply cannot converge). An accompanying 1,2-CH3 shift reaction can take place, though, to form the low-energy isomer ionized methyl ethyl ketenimine (CH3)(CH3CH2)CCNH+• (4b). There is also an intermediate (CH3)(CH3CH2)CCHN+• ion, 4c. It would appear that the

874

J. Phys. Chem. A, Vol. 114, No. 2, 2010

Mayer et al.

Figure 6. Relative energy diagrams [B3-LYP/6-31+G(d,p) level of theory] for common rearrangement reactions of ionized (- - -) propanenitrile, (s) butanenitrile, and (- - - -) pentanenitrile. Ionized nitrile refers to ions 1, 2, and 3. The corresponding isomers are isomer a (1a, 2a, 3a), isomer b (1b, 2b, 3b), and isomer c (1c, 2c, 3c). The relative energies for the four dissociation channels for ionized pentanenitrile (leading to m/z 55, 54, 43, and 41; see Table 10 below) are shown on the right.

TABLE 8: Experimental AE Values for Pentanenitrile AE (eV)

Figure 7. TPEPICO breakdown diagrams for (a) pentanenitrile and (b) 2,2-dimethylpropanenitrile.

rearrangement reactions in 4 that are analogous to those in 1-3 involve the interconversion of 4 to 4b, possibly via 4c. Threshold Photoelectron Photoion Coincidence Spectroscopy Study of Pentanenitrile and 2,2-Dimethylpropanenitrile. The dissociation of ionized pentanenitrile and 2,2dimethylpropanenitrile was explored with TPEPICO spectroscopy. The breakdown diagrams for the two molecules from 10.88 to 19 eV (for pentanenitrile) and from 10.89 to 40 eV (for 2,2dimethylpropanenitrile) are shown in Figure 7. For simplicity, only the most significant fragment ions are plotted. Appearance energies (AEs) for all fragment ions, derived from their onset energies, are listed in Tables 8 and 9. The listed uncertainty

m/z

this work

Heerma, de Ridder, and Dijkstra43

82 68 66 55 54 53 44 43 42 41 40 39 29 28 27 26 15