O–H Stretch Overtone Excitation of Ethyl Hydroperoxide Conformers

ARTICLE pubs.acs.org/JPCA

O
H Stretch Overtone Excitation of Ethyl Hydroperoxide Conformers Shizuka Hsieh,* Thida Thida, Margaret K. Nyamumbo, Kelly A. Smith, Noah Naamad, and Robert G. Linck Chemistry Department, Smith College, Northampton, Massachusetts 01063, United States

bS Supporting Information ABSTRACT: We present laser photoacoustic spectra of ethyl hydroperoxide (EHP) for 3
6 quanta of O
H stretch. Spectra are consistent with ab initio spectral simulations that take into account the influence of torsional motion about the O
O bond on O
H stretch overtone excitation. Experimental and simulated spectra share two trends. First, spectral features due to torsional excitation, including hot bands, become more prominent with increasing O
H stretch excitation, as has been shown previously for similar molecules such as methyl hydroperoxide. Second, contributions from the two different EHP conformers become clearly distinguishable at higher O
H stretch excitation, mainly due to the role of torsional motion. Results are consistent with a higher energy separation (330 cm
1) between the lower energy anti and the higher energy gauche conformers than predicted by electronic structure calculations (137 cm
1). Calculated absorption intensities are consistently higher than experimental values by ∼30% but within the experimental uncertainty. With each additional O
H stretch overtone, the dropoff in calculated integrated absorption intensities at room temperature becomes less extreme, with a 14fold dropoff from 3νOH to 4νOH and an 8-fold decrease from 5νOH to 6νOH.

1. INTRODUCTION Organic peroxides are intermediates in hydrocarbon oxidation, found both in the atmosphere1
4 and in combustion systems.5 Ethyl hydroperoxide (EHP, CH3CH2OOH) forms in the reaction of ethyl peroxy radicals (CH3CH2O2) with hydroperoxyl radicals (HO2).6,7 Along with methyl hydroperoxide (MHP, CH3OOH), hydroxymethyl hydroperoxide, peroxyacetic acid, bis-hydroxymethyl hydroperoxide, and isoprene hydroperoxide,8 EHP is one of the most abundant peroxides detected in the atmosphere.9 Inspiration for this work lies in demonstration10 of how O
H stretch overtone excitation can impact atmospheric chemistry and OH radical formation at dusk and dawn when UV radiation is scarce. Peroxides in particular have been of interest due to their relatively weak O
O bonds that enable photochemistry upon excitation to 5νOH11,12 and to the possibility of bimolecular reactions with oxygen upon excitation to 4νOH.13 Specifically for EHP, we demonstrated OH formation upon excitation to 5νOH.14 With the potential for overtone-induced chemistry in mind, we provide here experimental and calculated absorption wavelengths and intensities for a range of overtones (3
6νOH). Until now there has not been, to our knowledge, any experimental determination of the conformation of gas-phase EHP. Early computational work predicted the existence of two conformers15 but was inconclusive about which was lower in energy. The conformers differ by the O
C
C
H dihedral angle, here referred to as anti and gauche (Figure 1). More recent works5,16
18 have addressed the anti conformer. Our previous experimental work14 at 5νOH, however, indicated that both conformers are populated at room temperature, and calculations placed the gauche conformer only ∼100 cm
1 higher in energy than the anti. This work provides further experimental evidence for the population of both conformers. r 2011 American Chemical Society

2. METHODS 2.1. Experimental Details. Previous work from this laboratory19,20 describes the experimental setup for laser photoacoutsic spectroscopy. Gas-phase samples flow through a glass photoacoustic cell. Spectra are obtained by monitoring the signal from a Knowles EK-3132 microphone while scanning a 10 Hz-YAG-pumped dye laser (Sirah Cobra Stretch pumped by a Spectra Physics LAB150). Since the microphone signal has a linear dependence on laser power, all spectra are normalized for variation in power, as determined with a thermopile power meter (Ophir AP2). The wavelength scale is calibrated based on positions of lines in the photoacoustic spectrum of water. Aqueous CH3CH2OOH solutions are prepared in a manner analogous to syntheses described in the literature for CH3OOH (methyl hydroperoxide).21,22 Instead of extracting the hydroperoxide with ether, the samples are vacuum distilled, resulting in aqueous samples that range from 0.5 to 5.5 M CH3CH2OOH, as determined by titration.23 These aqueous samples also contain substantial ethanol, estimated to be up to 1 part in 10, by mole fraction. Argon bubbled through the aqueous mixture carries a mixture of CH3CH2OOH, ethanol, and water vapors through the photoacoustic cell for a total pressure around 100 Torr. To subtract contributions from ethanol and water, spectra are also collected for ethanol
water mixtures. The basis for determining experimental cross sections for CH3CH2OOH is the assumption that the power-normalized photoacoustic signal is proportional to the absorber’s absorption cross section and its relative abundance in the gas mixture. Signal Received: September 2, 2011 Revised: October 29, 2011 Published: October 31, 2011 14040

dx.doi.org/10.1021/jp208467f | J. Phys. Chem. A 2011, 115, 14040–14044

The Journal of Physical Chemistry A

ARTICLE

Figure 2. Calculated O
H stretch frequencies (ω) and anharmonicities (ωx) for the anti (left) and gauche (right) conformers.

Figure 1. Optimized geometries of the anti (top) and gauche (bottom) conformers of ethyl hydroperoxide, determined at the B3LYP/aug-ccPVTZ level.

comparisons are conducted back to back, so that differences in parameters such as microphone gain can be assumed negligible. Known absorption cross sections24,25 and comparison of their photoacoustic signals yield relative abundances of water and ethanol in the gas mixture. Rather than assuming Henry’s law as in our previous work19,20 to estimate the partial pressure of the hydroperoxide in the vapor phase, this work relies on direct analysis23 of condensate collected in a trap, as the gas mixture flows out of the cell. Given the relative abundances of the hydroperoxide and water, cross sections are determined from comparison of their photoacoustic signals. Wavelength regions for monitoring water, ethanol, and CH3CH2OOH are selected to avoid overlap of molecular signatures. Water lines are scanned at laser step sizes small enough to collect several dozen points across each feature. The method above yields cross sections at 4νOH, from an average from 11 separate runs with different aqueous solutions. For 3νOH and 5νOH cross sections, we take advantage of interchangeable sets of dye circulators that allow for quick changes between 532 nm pumped dyes and back to back comparison of signals from the same sample but at different levels of O
H stretch excitation. Cross sections at 3νOH and 5νOH come from their relative signal magnitude compared to the 4νOH region. No cross sections are determined for the 6νOH region, due to the low signal level and use of a 355 nm pumped dye that prohibits direct comparison to 4νOH. 2.2. Computational Details. Spectral simulations follow our previous work.14 The model combines O
H stretch vibration and torsional motion about the O
O bond and assumes adiabatic separation between them. Transitions considered here

involve the first eight torsional energy levels within each O
H stretch vibrational level, including transitions between torsional states with opposite wave function symmetry, the importance of which has been noted for a similar system, CH3OOH.26 For the simulations, calculated transition intensities are weighted by the Boltzmann population of the initial state at 298 K and convoluted with Gaussian functions to approximate the rotational envelope. Gaussian peak widths are chosen for suitable agreement with experimental data, as determined by the eye. Dipole moment derivatives, O
H stretch frequencies, and torsional potentials are based on calculations at the B3LYP/ aug-cc-PVTZ level using the Gaussian03 program.27 For each conformer, optimized geometries, energies, frequencies, and dipole components are obtained at C
O
O
H dihedral angles every 10
, from 0
to 360
. At each torsional angle,there are also calculations with compressed and extended O
H bond lengths in 0.5 Å steps around the equilibrium value, with six points shorter and eight longer. These later calculations hold other degrees of freedom fixed at the optimized value for the particular torsional angle. 2.2.1. Energies and Wave Functions. The O
H stretch vibrational energies and wave functions follow those expected for a Morse potential V ðqÞ ¼ Dð1
e
aq Þ 2

ð1Þ

where q is the displacement from the lowest energy O
H bond length. Frequencies (ω) and anharmonicities (ωx) for the O
H stretch are determined at each torsional angle by a nonlinear leastsquares fit of D and a28 to the energy/distance pairs with equal weighting of the points.29 The least-squares method yields better fits than the method used previously,14,30 as the sum of the squared deviations are smaller by a factor of ∼30. The torsional energies and wave functions at each vibrational level come from solving the Schr€odinger equation using a 20term basis set of sines and cosines. The kinetic energy operator is obtained identically to previous work,14 with parameters that depend on geometries in the ground and vibrationally excited states as described for HOOH.31 Along with the Morse O
H stretch energy contribution at each vibrational level, each torsional potential takes into account electronic and zero-point energies from Gaussian03, where zero-point energies do not include the C
O
O
H torsional and O
H stretch frequencies and are 14041

dx.doi.org/10.1021/jp208467f |J. Phys. Chem. A 2011, 115, 14040–14044

The Journal of Physical Chemistry A

Figure 3. Calculated torisonal potentials and energy levels for the anti (left) and gauche (right) conformers for ground and excited O
H stretch vibrational states.

scaled by a factor of 0.9674. Potentials are fit to a Fourier series consisting of 11 cosine and 3 sine terms. The sine terms are necessary for describing the gauche conformer potentials and are treated as a perturbation to a potential symmetric about 180
. 2.2.2. Transition Intensities. We use numerical methods to evaluate the necessary integrals,28 as shown for the ith dipole (μ) derivative ! i 1 d μ a jv00 n00 æ ð2Þ Æv0 n0 j qi i! dqi q¼0

where v designates O
H stretch vibrational quantum number, n designates torsional quantum number, and a designates the x, y, or z principal axis, with i = 1
6. Dipole derivatives at each torsional angle come from seven-term polynomial fits to the dipole components as a function of O
H bond length. Components in each integral that are independent of the torsional wave functions are evaluated at 10
increments, from 0
to 360
, and fit to a Fourier series with 10 cosine and 10 sine terms, so that the integration is for the product of three Fourier series.

3. RESULTS AND DISCUSSION 3.1. Computational Results. Optimized geometries for the anti and gauche conformers are shown in Figure 1. B3LYP/aug-ccPVTZ level calculations place the anti conformer lower in electronic energy than the gauche conformer by 98 cm
1. Including zeropoint energies, as described in section 2.2.1 and including O
H stretch and torsional motion, increases the energy difference to 137 cm
1, which is 21 cm
1 higher than determined14 at the B3LYP/6-311++G(3d,2p) level using Gaussian98. The conformers have Morse O
H stretch frequencies (ω) and anharmonicities (ωx) that differ slightly (3835.5 and 99.4 cm
1 for the anti

ARTICLE

Figure 4. Photoacoustic (top) and simulated spectra (below) at 3νOH. For comparison with experimental data, simulations are shifted 30 cm
1 to the red. Calculated line intensities for 298 K are convoluted with a 25 cm
1 Gaussian to simulate the rotational envelope, including any line broadening (middle panel, solid for anti and dotted for gauche). A 3:1 weighted sum of anti:gauche contributions (bottom) mimics the experimental spectrum.

conformer; 3837.6 and 99.5 cm
1 for the gauche conformer). Upon torsional motion about the O
O bond, the gauche conformer has a second local minimum, higher in electronic energy by 46 cm
1. These two gauche conformers differ in the relationship of the OH hydrogen atom and the methyl group; in one, these groups are on the same side of the plane defined by O
O
C(H2), whereas they are on opposite sides in the other. Due to differences at its two local minima, asymmetry about the H
O
O
C torsional angle of 180
results for the gauche conformer, as shown in Figures 2 and 3. Figure 2 shows the variation in ω and ωx with torsional angle, and Figure 3 shows the torsional potentials and energy levels. Torsional energies and the parameters used to determine them are listed in the Supporting Information, as are dipole moment derivatives. 3.2. O
H Stretch Overtone Spectra. Figures 4
7 show experimental photoacoustic (top panels) and simulated absorption spectra (middle and bottom panels). They include, to our knowledge, the first 3νOH and 6νOH experimental spectra for EHP and improved resolution of features compared to previous work for 4νOH and 5νOH.14,19 Both data and simulations show patterns demonstrated for CH3OOH.20,26 Each spectrum features a main peak that involves vibrational excitation, without any change in torsional level, while peaks to the red and blue correspond to accompanying changes in torsional quantum number. With increasing O
H stretch vibration, relative intensities of the main peak decrease while peripheral features due to torsional excitation become more prominent. The trend has been attributed to the changes in the shape of the torsional potential with increasing O
H stretch (Figure 3); specifically, the shift in minimum torsional angle at high overtones results in greater wave function overlap for transitions with changes in torsional quantum number.20 A trend specific to CH3CH2OOH is the increased energy separation of anti and gauche conformer peaks with increasing 14042

dx.doi.org/10.1021/jp208467f |J. Phys. Chem. A 2011, 115, 14040–14044

The Journal of Physical Chemistry A

Figure 5. Photoacoustic (top) and simulated spectra (below) at 4νOH. Calculated line intensities for 298 K are convoluted with a 35 cm
1 Gaussian to simulate the rotational envelope, including any line broadening (middle panel, solid for anti and dotted for gauche). A 3:1 weighted sum of anti:gauche contributions (bottom) mimics the experimental spectrum.

Figure 6. Photoacoustic (top) and simulated spectra (below) at 5νOH. For comparison with experimental data, simulations are shifted 40 cm
1 to the blue. Calculated line intensities for 298 K are convoluted with a 45 cm
1 Gaussian to simulate the rotational envelope, including any line broadening (middle panel, solid for anti and dotted for gauche). A 3:1 weighted sum of anti:gauche contributions (bottom) mimics the experimental spectrum.

O
H stretch vibration. Experimentally, shoulders due to the higher energy gauche conformer become distinct at higher overtones (top panels in Figures 4
7). Differences in O
H stretch overtone energies account only for a fraction of the increased wavenumber separation between conformer peaks; rather, differences in the torsional energy levels are primarily responsible for

ARTICLE

Figure 7. Photoacoustic (top) and simulated spectra (below) at 6νOH. For comparison with experimental data, simulations are shifted 90 cm
1 to the blue. Calculated line intensities for 298 K are convoluted with a 55 cm
1 Gaussian to simulate the rotational envelope, including any line broadening (middle panel, solid for anti and dotted for gauche). A 3:1 weighted sum of anti:gauche contributions (bottom) mimics the experimental spectrum.

the ability to distinguish conformers. For example, the observed energy separation between conformer main peaks at 6νOH is 86 cm
1, but the difference expected based on O
H stretch frequency (ω and ωx) alone would be 11 cm
1. Weighted sums (bottom panels in Figures 4
7) of simulations for the two conformers mimic the experimentally observed shoulders when assuming a 3:1 ratio of anti:gauche population. While simulations at 3νOH are relatively insensitive to changes in weighting, those at 5νOH and 6νOH have visibly better agreement with experiment assuming this ratio, compared to either 5:3 or 4:1. Assuming a 2-fold degeneracy for the gauche conformer, to account for the two possible O
C
C
H dihedrals, the 3:1 ratio corresponds to an energy separation of 330 cm
1 between conformers at 298 K. The range of energies bracketed by the 5:3 and 4:1 ratios is 200
390 cm
1. The Gaussian peak widths, which we empirically choose to mimic the rotational envelope in our simulations, become wider with increased O
H stretch overtone (middle panels in Figures 4
7). Because this current study is limited, we cannot interpret the physical causes. Higher resolution studies, such as jet-cooled double-resonance studies, however, can address questions of line broadening and dynamics. Similarly, more sophisticated simulations that include rotational band types for the two conformers might account for the trend. While limited, compared to our previous work for 5νOH,14 simulations here have better agreement with experiment in two ways. First, the new cross sections are closer to the experimental values. Calculated integrated cross sections (middle panel in Figure 6) in this work are two and three times lower than previous for the anti and gauche conformers, respectively. Previous simulations for 5νOH had peak heights doubling experimental values. In contrast, calculated cross sections here enable simulations for 3νOH, 4νOH, and 5νOH (bottom panels in Figures 4
6) that are 14043

dx.doi.org/10.1021/jp208467f |J. Phys. Chem. A 2011, 115, 14040–14044

The Journal of Physical Chemistry A consistently higher than experimental values by ∼30% but within the experimental uncertainty. Second, simulation peak positions better match the experimental data, requiring smaller shifts for alignment. For example, shifts in this work range from no shift to 90 cm
1 as opposed to 507 cm
1 previously.14 We anticipate that improvements in our calculations demonstrated here will support future predictions of O
H stretch overtone spectra. Such calculations, for example, contribute to our understanding about the extent to which absorption intensities drop with increased overtone excitation.25,32 For a series of alcohols and acids, Phillips and co-workers demonstrated dropoffs in absorption intensities ranging from 10 to 15 for 4νOH compared to 3νOH.32 Calculations here for EHP fall within this range, with a corresponding 14-fold decrease for 298 K. The predicted dropoff becomes less steep with increasing overtone, with an 11-fold decrease from 4νOH to 5νOH and an 8-fold decrease from 5νOH to 6νOH.

4. CONCLUSIONS We present experimental photoacoustic spectra for EHP, from 3νOH to 6νOH, and account for their features and intensities with a model that includes torsional motion about the O
O bond. The torsional motion plays an important role in elucidating the existence of two conformers of EHP at room temperature, with contributions from the higher energy gauche conformer becoming distinct from the anti conformer at higher overtone excitation. Experimental and calculated absorption intensities are similar to that of CH3OOH20,26 and low enough that like CH3OOH26 overtone-initiated photochemistry can also be expected to have a small impact on atmospheric OH formation. ’ ASSOCIATED CONTENT

bS

Supporting Information. Kinetic energy parameters and potential energy curves for determining torsional energies and wave functions, torsional energy levels, and dipole moment derivatives. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work was supported by a Henry Dreyfus Teacher-Scholar Award and the NSF (CHE-0910617). We thank Lauren E. Owen and Katherine L. Meek for assistance with sample preparation and data collection and Hannah T. Hitchner for computational contributions. ’ REFERENCES (1) Jackson, A. V. Crit. Rev. Environ. Sci. Technol. 1999, 29, 175–228. (2) Hewitt, C. N.; Kok, G. L. J. Atmos. Chem. 1991, 12, 181–194. (3) Hua, W.; Chen, Z. M.; Jie, C. Y.; Kondo, Y.; Hofzumahaus, A.; Takegawa, N.; Chang, C. C.; Lu, K. D.; Miyazaki, Y.; Kita, K.; Wang, H. L.; Zhang, Y. H.; Hu, M. Atmos. Chem. Phys. 2008, 8, 6755–6773. (4) Lee, M.; Heikes, B. G.; O’Sullivan, D. W. Atmos. Environ. 2000, 34, 3475–3494. (5) Simmie, J. M.; Black, G.; Curran, H. J.; Hinde, J. P. J. Phys. Chem. A 2008, 112, 5010–5016.

ARTICLE

(6) Hasson, A.; Tyndall, G.; Orlando, J. J. Phys. Chem. A 2004, 108, 5979–5989. (7) Wallington, T.; Japar, S. Chem. Phys. Lett. 1990, 166, 495. (8) Warneke, C.; Holzinger, R.; Hansel, A.; Jordan, A.; Lindinger, W.; Poschl, U.; Williams, J.; Hoor, P.; Fischer, H.; Crutzen, P.; Scheeren, H.; Lelieveld, J. J. Atmos. Chem. 2001, 38, 167–185. (9) Baasandorj, M.; Papanastasiou, D. K.; Talukdar, R. K.; Hasson, A. S.; Burkholder, J. B. Phys. Chem. Chem. Phys. 2010, 12, 12101–12111. (10) Donaldson, D. J.; Tuck, A. F.; Vaida, V. Chem. Rev. 2003, 103, 4717–4729. (11) Donaldson, D. J.; Frost, G. J.; Rosenlof, K. H.; Tuck, A. F.; Vaida, V. Geophys. Res. Lett. 1997, 24, 2651–2654. (12) Brown, S. S.; Wilson, R. W.; Ravishankara, A. R. J. Phys. Chem. A 2000, 104, 4976–4983. (13) Donaldson, D. J.; Francisco, J. S. Phys. Chem. Chem. Phys. 2003, 5, 3183–3187. (14) Closser, K. D.; Vogelhuber, K. M.; Hsieh, S. J. Phys. Chem. A 2008, 112, 1238–1244. (15) Lay, T. H.; Krasnoperov, L. N.; Venanzi, C. A.; Bozzelli, J. W.; Shokhirev, N. V. J. Phys. Chem. 1996, 100, 8240–8249. (16) Maciel, G. S.; Bitencourt, C. P.; Ragni, M.; Aquilanti, V. Int. J. Quantum Chem. 2007, 107, 2697–2707. (17) Chen, D.; Jin, H.; Wang, Z.; Zhang, L.; Qi, F. J. Phys. Chem. A 2011, 115, 602–611. (18) Hou, H.; Li, J.; Wang, B. J. Phys. Chem. A 2005, 109, 11206–11212. (19) Homitsky, S. C.; Dragulin, S. M.; Haynes, L. M.; Hsieh, S. J. Phys. Chem. A 2004, 108, 9492–9499. (20) Haynes, L. M.; Vogelhuber, K. M.; Pippen, J. L.; Hsieh, S. J. Chem. Phys. 2005, 123, 234306. (21) Vaghjiani, G. L.; Ravishankara, A. R. J. Chem. Phys. 1990, 92, 996–1003. (22) Davies, D. M.; Deary, M. E. J. Chem. Soc., Perkin Trans. 2 1992, 559–562. (23) Behrman, E. J.; Biallas, M. J.; Brass, H. J.; Edwards, J. O.; Isaks, M. J. Org. Chem. 1970, 35, 3069–3075. (24) Rothman, L. S.; et al. J. Quantum Spectrosc. Radiat. Trans. 2003, 82, 5–5
44. (25) Phillips, J. A.; Orlando, J. J.; Tyndall, G. S.; Vaida, V. Chem. Phys. Lett. 1998, 296, 377–377
383. (26) Matthews, J.; Martinez-Aviles, M.; Francisco, J. S.; Sinha, A. J. Chem. Phys. 2008, 129, 074316. (27) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (28) Atkins, P. W. Molecular Quantum Mechanics; Oxford University Press: Oxford, 1983. (29) Moore, R. H.; Ziegler, R. K. Los Alamos Sci. Lab. Report LA, 1960, LA-2367; Los Alamos Scientific Laboratory: Los Alamos, NM. (30) Howard, D.; Jorgensen, P.; Kjaergaard, H. J. Am. Chem. Soc. 2005, 127, 17096–17103. (31) D€ubal, H.-R.; Crim, F. F. J. Chem. Phys. 1985, 83, 3863–3872. (32) Lange, K. R.; Wells, N. P.; Plegge, K. S.; Phillips, J. A. J. Phys. Chem. A 2001, 105, 3481–3486. 14044

dx.doi.org/10.1021/jp208467f |J. Phys. Chem. A 2011, 115, 14040–14044