Hugo A.
López Peña
,
Derrick
Ampadu Boateng
,
Shane L.
McPherson
and
Katharine Moore
Tibbetts
*
Department of Chemistry, Virginia Commonwealth University, Richmond, VA 23284, USA. E-mail: kmtibbetts@vcu.edu
First published on 31st May 2021
The electronic potential energy surfaces of the nitrobenzene cation obtained from time-dependent density functional theory and coupled cluster calculations are used to predict the most efficient excitation wavelength for femtosecond time-resolved mass spectrometry measurements. Both levels of theory identify a strongly-coupled transition from the ground state of the nitrobenzene cation with a geometry-dependent oscillator strength, reaching a maximum at 90° C–C–N–O dihedral angle with a corresponding energy gap of ∼2 eV. These results are consistent with the experimental observation in the nitrobenzene cation of a coherent superposition of vibrational states: a vibrational wave packet. Time-resolved measurements using a probe wavelength of 650 nm, nearly resonant with the strong transition, result in enhanced ion yield oscillation amplitudes as compared to excitation with the nonresonant 800 nm wavelength. Analogous behavior is found for the closely related molecules 2- and 4-nitrotoluene. These results demonstrate that computational chemistry can predict the best choice of probe wavelength in time-resolved measurements of vibrational coherent states in molecular cations.
Intense femtosecond laser excitation coupled to mass spectrometry has been used for decades to study dissociation in radical cations because the short pulse duration induces rapid ionization prior to dissociative electronic excited state population, resulting in less molecular fragmentation.4 In particular, excitation with near-infrared wavelengths (∼1200–1600 nm) is widely observed to suppress fragmentation compared to excitation with the 800 nm Ti:Sapphire wavelength.5–11 The high yield of intact molecular ion is attributed to increased contribution of adiabatic electron tunneling to the ionization process, thereby enhancing population of the ground electronic state of the cation.6,9 However, extensive molecular fragmentation is still observed when the laser wavelength is resonant with an electronic transition between the ground electronic state and an excited state of the cation.7–15
Time-resolved pump–probe measurements are widely used to gain insight into how population of electronic excited states in radical cations induces dissociation.16–35 Many of these experiments have identified time-dependent oscillations in transient ion yields due to the preparation of a coherent superposition of vibrational states, i.e., a vibrational wave packet, by the pump pulse and its subsequent excitation by the time-delayed probe pulse.16–30 Complementary theoretical calculations of the participating electronic potential energy surfaces have uncovered mechanisms of dissociation via wave packet excitation. For instance, wave packet dynamics along the I–C–Br bending coordinate in CH2IBr+ create a dynamic resonance between the D0 and D3 electronic states that results in loss of neutral iodine to form CH2Br+.17,18 In acetophenone cation, wave packet dynamics along the torsional CH3CO–C6H5 twisting coordinate result in excitation from D0 to D2 at a dihedral angle of 90°, leading to loss of CH3 to form C6H5CO+.21,22
The known importance of electronic resonances to dissociation of radical cations has motivated the use of resonant probe wavelengths in time-resolved measurements,29–35 as is common practice in time-resolved studies of dynamics on neutral electronic excited states.36–41 Nevertheless, many time-resolved measurements on radical cations use the readily available 800 nm wavelength for probing cation dynamics.16–27 It would stand to reason that selecting the excitation wavelength to be resonant with electronic transitions in radical cations would enhance the amplitudes of the coherent oscillations resulting from vibrational wave packet dynamics. Computations of electronic potential energy surfaces can efficiently predict geometry-dependent electronic transitions in radical cations,17,18,21,22 making computational chemistry a potentially powerful tool for designing pump–probe control schemes to measure coherent vibrational dynamics.
In this work, we use time-dependent density functional theory (TDDFT) calculations of the electronic potential energy surfaces of the nitrobenzene (NB) cation to predict the most efficient excitation wavelength and verify this prediction with coupled cluster calculations and time-resolved mass spectrometric measurements. As a common model for nitroaromatic explosives such as 2,4,6-trinitrotoluene (TNT),42,43 NB has been subject to mass spectrometry studies using strong-field femtosecond laser excitation,42–45 although no time-resolved studies on NB have been reported to the best of our knowledge. We found a strongly-coupled transition in nitrobenzene cation at a C–NO2 dihedral angle of 90° with a ∼2 eV energy gap, making the transition nearly resonant with 650 nm excitation. Efficient geometry-dependent excitation at 650 nm is confirmed in pump–probe measurements, where oscillations in ion yields were enhanced by a factor of ∼2–5 as compared to excitation at 800 nm. Similar enhancement of oscillation amplitudes with 650 nm excitation is also observed in the cations of the related molecules 2-nitrotoluene and 4-nitrotoluene (2-NT and 4-NT). These results suggest that using computational chemistry to determine the best probe wavelength for time-resolved experiments can enhance the measurement of prepared vibrational coherent states, which is advantageous for coherent control and quantum information applications.
The calculated adiabatic and vertical ionization energies in units of eV, along with comparison to photoelectron spectroscopy experiments57,58 are given in Table 1. The numbers shown within parentheses are the signed errors of calculated values with respect to the corresponding experimental quantities, with negative numbers indicating underestimation and positive numbers indicating overestimation. Clearly, the BPW91 functional used for preliminary calculations performed poorly. Long-range-corrected hybrid GGA CAM-B3LYP and ωB97XD functionals overestimated adiabatic and vertical ionization energies in all cases with the combinations CAM-B3LYP/AUG-cc-pVDZ, ωB97XD/AUG-cc-pVDZ, and ωB97XD/Def2TZVPP giving deviations around 1 eV from the vertical experimental value. Estimations obtained with the M06 functional agree within 0.15 and 0.05 eV to the experimental adiabatic and vertical values respectively with the M06/Def2TZVPP level of theory giving the most accurate results. The agreement for B3LYP calculations falls within 0.09 and 0.04 eV for adiabatic and vertical quantities with the combinations with the CBSB7, AUG-cc-pVDZ, and Def2TZVPP basis sets performing accurate estimations of both experimental quantities. Particularly, the value obtained at the B3LYP/Def2TZVPP level of theory matches the vertical experimental measurement. Due to these reasons we decided to undergo further benchmarking considering the B3LYP and M06 functionals and their combination with the Def2TZVPP basis set.
Method | IEad | IEvert |
---|---|---|
a Ref. 57. b Ref. 58. | ||
B3LYP/6-311+G* | 9.69 (0.09) | 10.03 (0.04) |
B3LYP/CBSB7 | 9.61 (0.01) | 9.96 (−0.03) |
B3LYP/AUG-cc-pVDZ | 9.62 (0.02) | 9.97 (−0.02) |
B3LYP/Def2TZVPP | 9.64 (0.04) | 9.99 (0.00) |
M06/6-311+G* | 9.75 (0.15) | 10.04 (0.05) |
M06/CBSB7 | 9.70 (0.10) | 10.01 (0.02) |
M06/AUG-cc-pVDZ | 9.70 (0.10) | 9.98 (−0.01) |
M06/Def2TZVPP | 9.64 (0.04) | 9.95 (−0.04) |
CAM-B3LYP/6-311+G* | 9.89 (0.29) | 10.13 (0.14) |
CAM-B3LYP/CBSB7 | 9.81 (0.21) | 10.07 (0.08) |
CAM-B3LYP/AUG-cc-pVDZ | 9.81 (0.21) | 11.01 (1.02) |
CAM-B3LYP/Def2TZVPP | 9.83 (0.23) | 10.09 (0.10) |
ωB97XD/6-311+G* | 9.83 (0.23) | 10.09 (0.10) |
ωB97XD/CBSB7 | 9.78 (0.18) | 10.06 (0.07) |
ωB97XD/AUG-cc-pVDZ | 9.75 (0.15) | 10.95 (0.96) |
ωB97XD/Def2TZVPP | 9.75 (0.15) | 10.95 (0.96) |
BPW91/6-311G* | 9.26 (−0.34) | 9.61 (−0.38) |
BPW91/6-311+G* | 9.36 (−0.25) | 9.70 (−0.29) |
EOM-IP-CCSD/6-311+G* | 10.06 (0.07) | |
Expt. | 9.6a | 9.99b |
Additionally, we calculated the vertical ionization energy at the equation-of-motion ionization-potential coupled-cluster singles and doubles (EOM-IP-CCSD)59 level in conjunction with the 6-311+G* basis set considering NB optimized geometry at the B3LYP/Def2TZVPP level using Q-Chem 5.3.60 The optimized geometries of neutral NB and NB cation obtained at the B3LYP/Def2TZVPP level of theory are given in the ESI,† Table S1.
The frequencies and intensities of the vibrational modes in both the neutral molecule and the cation were calculated via normal mode analysis. In neutral NB, the values of the harmonic vibrational frequencies depended on whether the B3LYP or the M06 functional was used but, given a certain functional, the frequencies and intensities did not depend strongly on the basis set chosen. Fig. 1 shows the comparison between the predicted infrared spectra for the neutral molecule at the B3LYP/Def2TZVPP (red) and M06/Def2TZVPP (blue) levels to the experimental gas-phase spectrum obtained from NIST.61 The calculated harmonic frequencies match the experimental peaks reasonably well in the range of ∼500–2000 cm−1, but it is evident that the B3LYP/Def2TZVPP level of theory offers the best estimation. This agreement indicates the effectiveness of the method and suggests that the computed cation frequencies should be reasonably accurate. Fig. S1 and S2 of the ESI† show the comparison between the experimental infrared spectrum and the calculated spectra using the B3LYP and M06 functionals in combination with all the basis sets explored in this study. Full tabulated results of the harmonic vibrational frequencies and their corresponding intensities for the neutral molecule and its cation are presented in the ESI,† Tables S3 through S6.
Fig. 1 Experimental (black) and computed infrared spectra at the B3LYP/Def2TZVPP (red) and M06/Def2TZVPP (blue) levels of theory for NB. |
To determine the excited-state energies of the neutral and cationic NB, we performed time-dependent DFT (TDDFT)62 calculations using Gaussian 16. We calculated the first 10 singlet–singlet (for neutral NB) and doublet–doublet (for NB cation) transitions with the B3LYP and M06 functionals in combination with the above-mentioned four basis sets. We found an analogous behavior to that of simulated IR spectra: the dependence on the functional was stronger than that of the basis set. Fig. 2 shows the comparison of the experimental spectra for NB dissolved in methanol and in chloroform with the calculated ones at the B3LYP/Def2TZVPP and M06/Def2TZVPP levels. Experimental spectra in two solvents (see Section 2.2) are shown to illustrate any solvatochromic effects due to solvent polarity and ensure a reasonable comparison between experimental and calculated spectra. The comparison of the experimental UV-vis spectrum of NB dissolved in methanol with the spectra calculated using B3LYP and M06 functionals in combination with all the basis sets considered in this work can be seen in Fig. S3 of the ESI.† Full tabulated results of the excitation energies and their corresponding oscillator strengths for neutral and cationic NB are presented in Tables S7 through S10 of the ESI.† The better agreement of the computation at the B3LYP/Def2TZVPP level with the experiment follows in line with the previous bench-marking criteria. Unfortunately, the good performance of the B3LYP/Def2TZVPP level of theory in predicting the excitation spectrum of neutral NB cannot be extrapolated to NB radical cation. In Section 3.1 we will present TDDFT excited-state calculations for NB radical cation at the B3LYP/Def2TZVPP level but those calculations will be further supported with selected calculations at the equation-of-motion excitation-energies coupled-cluster singles and doubles (EOM-EE-CCSD)59 level. These EOM calculations were performed using Q-Chem 5.3.
Fig. 3 Geometrical structures for neutral and cationic NB optimized at the B3LYP/Def2TZVPP level of theory. Bond lengths are in Å and torsional angles in degrees. |
Previous femtosecond mass spectrometry measurements on NB have identified the most prominent ion signal as the C6H5+ fragment, obtained by NO2 loss from the parent NB+ ion.42–45 To determine the energy needed for C6H5+ generation, we performed a relaxed potential energy surface scan for the C–N bond in NB+ using the ModRedundant keyword in Gaussian at the B3LYP/Def2TZVPP level of theory (Fig. 4). The depth of the well obtained from this surface is around 1.58 eV. We estimated the zero point energy (ZPE) by adding the ZPE of the vibrational modes that heavily involve C–N bond stretching (modes 5, 10, 12, 21, 25, 27, 29, and 31, whose corresponding frequencies can be found in Table S5 of the ESI†). The result of this estimation was 0.55 eV (dashed black line in Fig. 4), which is reasonable since the total ZPE calculated by Gaussian, adding the ZPE for the 36 vibrational modes of NB, is 2.72 eV. Taking the ZPE into account it can be estimated that the minimum energy necessary for breaking the C6H5–NO2 bond is 1.03 eV (1.58–0.55 eV), less than the energy corresponding to 800 nm photons (1.55 eV) and very close to the energy of 1300 nm photons (0.95 eV) as can be seen in Fig. 4.
The ground and lowest 10 excited electronic potential energy surfaces (PESs) of NB cation along the C–NO2 torsional coordinate were computed to determine potential electronic excitation pathways in NB following ionization (Fig. 5). As seen in Fig. 5a, these PESs are distributed within two groups: the first comprised by D0 to D4 and the second by D5 to D10. It is evident that the D2 through D4 surfaces in the first group have a marked dependence on the dihedral angle, reaching their maximum energy value at 90°, whereas the surfaces within the second group have a less pronounced dependence on the dihedral angle. Fig. 5a also shows the energies corresponding to 1300, 800, and 650 nm photons for comparison (0.95, 1.55, and 1.91 eV respectively). Only the 650 nm photon energy is enough to promote the D0 → Dn, n = 1,…,4 transitions by means of single photon processes at any geometry. Moreover, 650 nm is nearly resonant with the D0 → D4 transition at a dihedral angle of 90°. For 800 nm photons, the D4 surface is out of reach from D0 at most geometries, the D3 surface is accessible at some geometries, and the D2 surface is reachable at any geometry. 1300 nm photons can only reach the D2 through D4 surfaces from D0 at dihedral angles near 0°, i.e., near the neutral NB geometry.
The importance of the dihedral angle for excitation probability is captured in the dependence of the oscillator strength on the dihedral angle (Fig. 5b). Some of the oscillator strengths are very close to zero for all values of the dihedral angle, but f02, f03, f04, f07, f09, and f010 have non-negligible values at different geometries. It is noteworthy that the oscillator strength for the D0 → D4 transition (f04) is particularly high and dependent on the geometry, reaching its maximum value at 90°. Moreover, this transition is the only one that has a substantial oscillator strength at 90° (Table 2). In order to benchmark our TDDFT calculations we also performed some computations using EOM-EE-CCSD in combination with the 6-311+G* basis set. The geometries considered were the ones obtained after optimization at the B3LYP/Def2TZVPP level. Table 2 contains the results of these calculations for NB cation with 90° C–NO2 torsional angle. Analogous EOM calculations for the optimized neutral and cation geometries are shown in Table S11 of ESI.† Comparison of results in Table 2 shows that EOM calculations confirm the presence of an excited state with a substantial oscillator strength at 90° dihedral angle and with excitation energy around 2 eV, in line with the TDDFT calculations, and very importantly, with the experimental results to be discussed in the following sections. However, the ordering of the excited states is not in agreement since EOM results point to D3 as the bright state while TDDFT computations point to D4. This situation will be briefly discussed in Section 4.
Transition | B3LYP/Def2TZVPP | EOM-EE-CCSD/6-311+G* | ||
---|---|---|---|---|
EE (eV) | f (a.u.) | EE (eV) | f (a.u.) | |
D0 → D1 | 0.73 | 0.0000 | 0.81 | 0.0000 |
D0 → D2 | 1.23 | 0.0010 | 2.00 | 0.0013 |
D0 → D3 | 1.70 | 0.0000 | 2.34 | 0.1156 |
D0 → D4 | 1.99 | 0.1039 | 2.49 | 0.0001 |
D0 → D5 | 3.03 | 0.0000 | 3.70 | 0.0013 |
D0 → D6 | 3.12 | 0.0000 | ||
D0 → D7 | 3.54 | 0.0004 | ||
D0 → D8 | 3.62 | 0.0002 | ||
D0 → D9 | 3.80 | 0.0000 | ||
D0 → D10 | 3.89 | 0.0000 |
Having a substantial oscillator strength for a transition from D0 to a specific bright state (D3 or D4) at 90° implies that, as the vibrational wave packet propagates along the D0 surface, it will have the highest probability of a one-photon excitation to the excited state surface at that particular dihedral angle. Hence, we can expect that pump–probe measurements using nearly resonant probe photons will most efficiently transfer ground-state NB+ to the bright state surface at pump–probe delays corresponding to the time needed for the wave packet to reach a 90° dihedral angle. To estimate this time we performed classical wave packet trajectory calculations21,65 using the D0 surface computed in fine increments of 2° (red dots, Fig. 6) and fit to a smoothing spline (solid red line, Fig. 6). The evolution of the dihedral angle of the wave packet centroid is determined by the equation of motion65
(1) |
Fig. 6 Calculated wave packet trajectories over the D0 PES (red dots and line) using 0 (bright green dashed line), 0.03 (turquoise solid line), and 0.24 eV (dark green dotted line) of excess energy. |
From Fig. 6 we can see a local minimum in the D0 surface at 0° dihedral angle; if we consider this surface as periodic, that local minimum constitutes a potential well. Hence, the wave packet will need at least Δ ≥ 0.002 eV excess energy to surpass the barrier at ∼9° dihedral angle. It is reasonable to assume that the source of that energy is some fraction of the excess energy after ionization, i.e., the difference between vertical and adiabatic ionization energies. At the B3LYP/Def2TZVPP level, this excess energy is 0.35 eV (Table 1), giving an upper limit of Δ = 0.24 eV in eqn (1). Fig. 6 shows the wave packet trajectories along the D0 PES (red line) obtained with Δ = 0.002 eV (light green dashed line), Δ = 0.03 eV (turquoise solid line), and Δ = 0.24 eV (dark green dotted line). As expected, the time taken for the wave packet to reach a 90° dihedral angle changes with the value of Δ, with times around 400, 200, and 100 fs for Δ = 0, 0.03, and 0.24 eV, respectively. As will be shown in Section 3.2 below, the pump–probe measurements indicate that the wave packet reaches the 90° torsional angle in ∼200 fs, suggesting a physically plausible value of Δ = 0.03 eV, or 9% of the available relaxation energy, to “kick-start” the wave packet.
Fig. 8 shows the transient signals of the ions highlighted in Fig. 7 as a function of pump–probe delay (τ) using 800 nm (a) and 650 nm (b) probe pulses at intensities 1013 W cm−2 (top) and 1012 W cm−2 (bottom). All transients are normalized to the yield of NB+ at τ < 0. Depletion in NB+ (red) and increase in C6H5+ (blue) signals at τ > 0 are observed under all conditions, with greater magnitude changes for high probe intensity (top panels) due to increased excitation probability. The yield of C4H3+ increases significantly for τ > 0 only at 1013 W cm−2, with greater enhancement for the 650 nm wavelength. The antiphase oscillations in the yields of NB+ and C6H5+ (inset in each panel of Fig. 8), with similar frequency to related aromatic molecules,19–23,29 indicate that electronic excitation of the NB+ vibrational wave packet along the C–NO2 torsional mode (Fig. 3 and 5) results in C6H5+ formation. The greater oscillation amplitude using the 650 nm wavelength, particularly at lower intensity (bottom panels), is consistent with the prediction based on the computed PESs in Fig. 5 that 650 nm photons are more efficient at exciting the NB+ vibrational wave packet.
(2) |
Under all conditions, the parent NB+ ion was fit to the complete eqn (2), C6H5+ fit to eqn (2) without the T3 component, and C4H3+ fit to the complete eqn (2) for 650 nm excitation at 1013 W cm−2 or without the oscillatory T1 component otherwise (ESI,† Tables S12–S15). To compare the oscillatory dynamics observed under the four experimental conditions in Fig. 8, the tabulated a, T1, t, and ϕ coefficients for the parent NB+ ion extracted from curve fitting are reported in Table 3. For all four conditions, the coherence lifetime T1, oscillation period t, and phase ϕ agree to within the reported errors (95% confidence intervals from curve fitting). This result confirms that the same initial wave packet was prepared by the ionizing pump pulse because the parameters T1, t, and ϕ are intrinsic to the wave packet dynamics. In contrast, the a coefficients associated with the oscillation amplitude are greater for 650 nm as compared to 800 nm excitation: the a value of 0.11 ± 0.02 obtained with 650 nm is twice the 0.05 ± 0.01 value obtained with 800 nm at the modest intensity of 1012 W cm−2. This result is consistent with the more prominent oscillations visible in Fig. 8b as compared to Fig. 8a. Greater values of a coefficients for the C6H5+ fragment are also observed with 650 nm probe as compared to 800 nm probe (ESI,† Tables S12–S15).
Eqn (2) | a | T 1 (fs) | t (fs) | ϕ (rad) |
---|---|---|---|---|
650 nm, I = 1013 | 0.14 ± 0.02 | 239 ± 15 | 430 ± 9 | −0.08 ± 0.09 |
650 nm, I = 1012 | 0.11 ± 0.02 | 242 ± 15 | 426 ± 10 | −0.04 ± 0.08 |
800 nm, I = 1013 | 0.08 ± 0.01 | 244 ± 24 | 439 ± 15 | −0.13 ± 0.19 |
800 nm, I = 1012 | 0.05 ± 0.01 | 225 ± 41 | 426 ± 10 | −0.05 ± 0.21 |
To further illustrate the enhancement of the oscillatory ion signals using 650 nm excitation, the residual oscillatory ion signals obtained after subtraction of the incoherent components (T2 and T3 terms) from eqn (2) are plotted in Fig. 9, left panel. This plot clearly shows the antiphase behavior of the oscillations in NB+ (red) and C6H5+ (blue), as indicated by the dotted lines at the minima of NB+ yield (200 and 630 fs). For 650 nm excitation at 1013 W cm−2, the C4H3+ (green) fragment also exhibits antiphase oscillations with respect to NB+. At both intensities, the larger-amplitude oscillations using 650 nm are apparent in the increased amplitude of the corresponding Fast Fourier Transform (FFT) peak at 80 cm−1 (dashed line) taken from the residual ion signals over the pump–probe delay range of 70–3000 fs (Fig. 9, right panel). In particular, the 80 cm−1 peak intensity for NB+ at 1012 W cm−2 is approximately five times higher for 650 nm as compared to 800 nm excitation. Hence, quantification of the oscillations indicate an amplitude enhancement by a factor of ∼2–5 using 650 nm for excitation as compared to 800 nm, consistent with the theoretical predictions from Fig. 5.
Both the 650 and 800 nm excitation wavelengths are capable of increasing the C6H5+ signal at the expense of the parent ion signal (Fig. 7), even though only the 650 nm probe is capable of promoting the D0 → D4 transition for any geometry according to our TDDFT calculations (Fig. 5a). This result is consistent with previous mass spectrometry studies using only one 800 nm pulse for ionization that show substantial C6H5+ ion signal and little parent ion,42–44 and suggests that ionization of NB at 800 nm can populate one or more electronic excited states of the cation. Because the 800 nm wavelength can access the D2 excited state at all geometries, excitation to D2 is likely sufficient to produce NO2 loss, at least at geometries where its energy is above the estimated dissociation threshold of 1.03 eV (Fig. 4). We cannot rule out the contribution of the D3 excited state to the dissociative event because 800 nm excitation is also able to promote the D0 → D3 transition for some geometries. The participation of the D0 → D1 transition can be excluded due to the negligible values of f01 (Fig. 5b) and more importantly because the energy of the D1 excited state at any geometry lies below the minimum energy to break the C–N bond. Hence, the key difference between 800 nm and 650 nm excitation is that only 650 nm gives full accessibility to the D4 surface, for which the coupling strength to D0 has a strong geometric dependence. This circumstance explains the enhanced oscillatory behavior for 650 nm excitation (Fig. 8 and 9).
The observation that the 650 nm excitation is more efficient at generating the C4H3+ fragment as compared to the 800 nm excitation (Fig. 7b) suggests that the excited state(s) involved should be at the reach of the 650 nm wavelength to a greater extent than for 800 nm. In addition, the finding of substantially enhanced C4H3+ yield at high probe intensity (1013 W cm−2, top panels of Fig. 8) as compared to modest probe intensity (1012 W cm−2, bottom panels of Fig. 8) supports the idea that the excited state(s) responsible for C4H3+ formation can only be reached by means of a two-photon process. These facts suggest that the two photon process(es) involved in the C4H3+ fragment production are more efficient for the 650 nm probe as compared to the 800 nm. Moreover, we can postulate the existence of a two-photon resonant process achievable only with the 650 nm excitation. After eliminating the incoherent part from the C4H3+ transient, the oscillatory nature of the transient is only observed with 650 nm probe, whereas the transient for the 800 nm probe only shows a feeble oscillatory behavior (Fig. 9). Because the D5 and D6 PESs do not depend on the dihedral angle while the D7 to D10 PESs show a modest dependence (Fig. 5a), we can postulate that the excited state(s) involved in the two photon process generating C4H3+ lie(s) between D7 and D10. These states span an excitation energy range from 3.54 to 3.89 eV (Table 2), which can be reached with two 650 nm photons (3.82 eV). The dependence on geometry for the D7 to D10 PESs could be associated with the oscillatory behavior observed for the 650 nm probe. On the other hand, the energy corresponding to two 800 nm photons (3.10 eV) cannot reach the D7 and higher states. As a last observation we want to point out that the residual ion signal for C4H3+ shows antiphase coupling between NB+ and C4H3+ (Fig. 9), suggesting that the C4H3+ fragment originates from the D0 surface of the parent ion. This result suggests that two-photon excitation of NB+ provides sufficient energy to further fragment initially formed C6H5+ into C4H3+, as observed in TPEPICO studies.66
Collectively, the results of this work both illustrate how quantum chemical computations can predict the outcome of pump–probe measurements on radical cations and highlight similarities in the observed dynamics, and thereby the similar electronic structures, of homologous radical cations. Hence, this work builds on previous studies wherein computed electronic PESs of cations including bromoiodomethane,17,18 acetophenone,21,22 and azobenzene27 are used to explain transient oscillatory dynamics of ion signals observed in pump–probe measurements. Moreover, this work demonstrates that computations of electronic PESs, even using the inexpensive TDDFT method, provide sufficiently accurate results to predict the best probe excitation wavelength in pump–probe measurements of nitrobenzene and closely related nitroaromatic compounds to maximize the visibility of vibrational wave packet dynamics. However, as the comparison of TDDFT and EOM results in Table 2 showed, our inexpensive TDDFT approach is not free of problems: TDDFT appears to misidentify the bright state as D4 instead of D3 and predicts excitation energy lower by 0.35 eV. Previous works have pointed out deficiencies on the TDDFT description of low-lying singlet excited states of various classes of molecules including polyacenes,67,68 thiophene and short thienoacenes,69 fused heteroaromatic rings,70 and naphthol isomers.71 Further work should be carried out to fully elucidate the nature of the excited states of related nitroaromatic radical cations and to rationalize and possibly mitigate the drawbacks of their TDDFT description. It is important to emphasize that any extension of this TDDFT approach to different groups of molecules should be carefully evaluated with the aid of higher level methodologies.
Additionally, our results complement previous studies showing that selection of near-infrared wavelengths for the ionizing pump also enhances the visibility of oscillatory ion dynamics by more selectively populating the ground electronic state of the cation.21,25 Finally, our results provide further evidence that similar coherent vibrational dynamics in radical cations are to be expected for families of homologous molecules.17,20,26 In particular, the ortho/para directing nature of the nitro group is confirmed in our pump–probe measurements due to the completely different dynamics of 3-NT as compared to NB, 2-NT, and 4-NT (Fig. 10).
Footnote |
† Electronic supplementary information (ESI) available: Supplemental figures of theory benchmarking and experimental curve fitting; tabulated theoretical results and curve fitting coefficients. See DOI: 10.1039/d1cp00360g |
This journal is © the Owner Societies 2021 |