You are on page 1of 5

PCCP

View Article Online


PAPER View Journal | View Issue

Accurate prediction of the optical rotation and


NMR properties for highly flexible chiral natural
Published on 09 August 2016. Downloaded by Pardubice University on 6/12/2019 7:32:13 AM.

Cite this: Phys. Chem. Chem. Phys.,


2016, 18, 24506 products†‡
Muhammad Ali Hashmi,a Sarah K. Andreassend,ab Robert A. Keyzersab and
Matthias Lein*ac

Despite advances in electronic structure theory the theoretical prediction of spectroscopic properties
remains a computational challenge. This is especially true for natural products that exhibit very large
conformational freedom and hence need to be sampled over many different accessible conformations.
Received 12th July 2016, We report a strategy, which is able to predict NMR chemical shifts and more elusive properties like the
Accepted 9th August 2016 optical rotation with great precision, through step-wise incremental increases of the conformational
DOI: 10.1039/c6cp04828e degrees of freedom. The application of this method is demonstrated for 3-epi-xestoaminol C, a chiral
natural compound with a long, linear alkyl chain of 14 carbon atoms. Experimental NMR and [a]D values
www.rsc.org/pccp are reported to validate the results of the density functional theory calculations.

1. Introduction
Recently a new 1-deoxysphingoid, 3-epi-xestoaminol C (11, see
Scheme 1) was isolated and identified.1 The relative configuration of
the two stereocentres in 11 was determined through derivatisation
and subsequent NMR analysis, while the absolute stereochemistry
was established using Mosher’s method.2 1-Deoxysphingoids are a
class of naturally occurring amino alcohols with long hydrocarbon
chains.3 Due to the length of the hydrocarbon chain, they possess
high structural flexibility and will adopt many conformations in
solution.4 To study the optical or spectroscopic properties of such Scheme 1 The 11 compounds that were considered in this investigation.
systems computationally, it is essential to consider all thermally 1 (2S,3S)-3-amino-2-butanol, 2 (2S,3S)-2-amino-3-pentanol etc. Note that
accessible conformers the molecule can adopt under experimental 11 is the naturally occurring 3-epi-xestoaminol C.
conditions according to their Boltzmann weighted contribution to
the desired property.5
Traditionally, conformational analysis is performed at the the identification of all contributing conformers, but because the
molecular mechanics (i.e. force field) level, e.g. through a Monte number of possible conformations increases greatly with increasing
Carlo search.6 In this context, empirical force fields are a molecular size,9 it is usually impossible to apply even at the
compromise between accuracy and computational cost.7 A molecular mechanics level, let alone using quantum mechanics
systematic search,8 i.e. one that samples the complete exhaustive (QM).10–12 For 11, the number of possible conformers exceeds
potential energy surface (PES), is the only method which guarantees 170 000. This demonstrates that a complete analysis is beyond
reach for even very modestly sized molecules.
a
School of Chemical and Physical Sciences, Victoria University of Wellington,
To avoid these problems, we present a new approach, in
Wellington, 6012, New Zealand which we combine the accuracy of QM calculations with an
b
Center for Biodiscovery, Victoria University of Wellington, Wellington, 6012, efficient search methodology that successfully identifies the
New Zealand most important contributions to a solution phase mixture of
c
Centre for Theoretical Chemistry and Physics, New Zealand Institute for Advanced
conformers in a relatively short time.
Study, Massey University Auckland, Private Bag 102904, North Shore MSC 0632
Auckland, New Zealand. E-mail: matthias.lein@vuw.ac.nz
In order to successively construct the conformational space
† Dedicated to Professor Gernot Frenking on the occasion of his 70th birthday. of 11, but restrict ourselves to the important contributions only,
‡ Electronic supplementary information (ESI) available. See DOI: 10.1039/c6cp04828e we started with the smallest possible analogue of 11 (structure 1,

24506 | Phys. Chem. Chem. Phys., 2016, 18, 24506--24510 This journal is © the Owner Societies 2016
View Article Online

Paper PCCP

see Scheme 1) and performed exhaustive PES mapping using


high level QM calculations.

2. Computational methods
All calculations were carried out with the Gaussian 09 suite of
programs (Revision D.01).13 All the calculations were carried
out using density functional theory (DFT) using Adamo’s
hybrid14 version of Perdew, Burke and Ernzerhof functional
(PBE0)15,16 which has been shown superior to a range of other
Published on 09 August 2016. Downloaded by Pardubice University on 6/12/2019 7:32:13 AM.

functionals especially for the prediction of optical rotations.17


Grimme’s empirical D3 correction with Becke–Johnson damping
(D3BJ)18–20 was applied. Alrich’s triplet z basis set Def2-TZVP21 was
applied in all the calculations with the polarizable continuum model
(PCM) using the integral equation formalism variant (IEFPCM).22–42
The SMD parameter set established by Cramer and Truhlar43 was
used to model the effects of the solvent (as implemented in
Gaussian 09).13 Dichloromethane was used as a solvent in all the
calculations unless otherwise specified. All the conformers were
confirmed to be minima on the potential energy surface at the
PBE0-D3BJ/def2-TZVP/SMDCH2Cl2 level through the calculation of
harmonic force constants. NMR calculations were performed
using the same level of theory.

3. Results and discussion


In our construction of the relevant conformational space of the
target molecule 11, we take advantage of the fact that the
conformational degrees of freedom that stem from the first
four carbon atoms (including the OH and NH2 functionality) is
well separated from the conformational degrees of freedom of
the linear alkyl chain that comprises the rest of the molecule.
Hence, we start by examining the PES of 1 through a series of relaxed
potential energy scans that map the energy landscape with respect to
rotations around all dihedral angles of 1 (see Fig. 1) in steps of 151.
This method identified 23 distinct conformers of 1, that were
subsequently confirmed to be minima on the PES (see Fig. 2) by Fig. 2 The 23 conformers of 1 and their relative Gibbs free energies
(DG in kcal mol 1) at the PBE0-D3BJ/def2-TZVP/SMDDCM level of theory:
1a 0.0; 1b 1.0; 1c 1.7; 1d 1.9; 1e 2.0; 1f 2.1; 1g 2.1; 1f 2.1; 1g 2.1; 1h 2.2;
1i 2.4; 1j 2.5; 1k 2.6; 1l 2.8; 1m 2.9; 1n 2.9; 1o 3.1; 1p 3.1; 1q 3.1; 1r 3.1;
1s 3.3; 1t 3.3; 1u 3.4 1v 3.8; 1w 4.7.

unconstrained relaxation. The relative DG value between the global


minimum and the highest energy structure is just 4.7 kcal mol 1.
In the next step, all possible low energy conformers of an
analogue of 1 with an additional carbon atom in its alkyl chain,
(2S,3S)-2-amino-3-pentanol (2), were constructed by substituting each
one of the hydrogen atoms on the terminal methyl group to generate
three new candidate structures for each of the 23 conformers of 1
(see Scheme 2). These 69 structures were then optimised with the
same QM methodology that was used previously. After the elimina-
tion of duplicates, 67 conformers of 2 were obtained (Table 1).
Fig. 1 Potential energy surfaces at the PBE0-D3BJ/def2-TZVP/SMDDCM level
Conformers of the compounds 3–11 were obtained in the
of theory that identify all 23 conformers of 1. Relative energies are in kcal mol 1. same fashion. From (2S,3S)-2-amino-3-octanol (5) onwards,
Optimisations were carried out for a total of 3750 individual structures. those that contribute to Z90.0% of the Boltzmann distribution

This journal is © the Owner Societies 2016 Phys. Chem. Chem. Phys., 2016, 18, 24506--24510 | 24507
View Article Online

PCCP Paper

energetically favourable interactions in conformations where the tail


of the molecule is in close proximity to the head of the molecule.
Indeed, we find that while there are descendants of low-
energy conformers of 1 that are relatively high in energy in 2,
there are very few descendants of high-energy conformers that
are low in energy. For example: if one divides the conformers of
1 into 12 ‘‘low-energy’’ and 11 ‘‘high-energy’’ conformers, it can
be seen that the first conformers of the ‘‘high-energy’’ group is
only the 23rd lowest energy conformer of 2 (which contributes
to 0.37% of the ensemble). This result is exacerbated when the
Published on 09 August 2016. Downloaded by Pardubice University on 6/12/2019 7:32:13 AM.

descendants of 2 are analysed in 3. Then the first descendant of


the ‘‘high-energy’’ conformers of 1, is only the 44th lowest
energy conformer of 3. To further quantify this relationship, we
Scheme 2 Construction of conformers through step-wise lengthening of analysed how the percentile rank (in terms of relative energy) of
the alkyl chain. The lowest energy isomer of 11 is shown at the bottom. the conformers of 2 is reflected in the percentile rank of their
descendants in 3. We find that there is an 83% value for the
correlation of the two. More importantly, once all conformers
Table 1 ncontr is the total number of distinct conformers considered (fully whose percentile rank in 3 has increased by at least 10 percentage
optimised at the PBE0-D3BJ/def2-TZVP/SMDDCM level of theory). nimp is the points with respect to their antecedent’s percentile rank
number of important conformers (i.e. the number of conformers that have to be
(i.e. those whose relative energy is markedly higher in 3 and
included to account for Z90.0% of the Boltzmann distribution at 298.15 K). c0 is
the percentage contribution of the lowest lying conformer at 298.15 K. P is the are hence of lower importance for the mixture of conformers)
total value of the partition function at 298.15 K. DG is the difference in Gibbs have been removed, the correlation improves to 95%. This
free energy between the energetically lowest conformer and the energeti- demonstrates that the descendants of high-energy conformers
cally highest lying conformer (of all ncontr conformers; in kcal mol 1). [a]D is become increasingly higher and higher in energy themselves
the predicted value of the optical rotation (based on all ncontr conformers)
and are thus of less and less importance to the observed
ncontr nimp c0 P DG [a]D mixture of conformations (details of this analysis can be found
1 23 7 67.0 1.49 4.7 +19.4
in the ESI‡). It is important, however, to note that while the
2 67 13 38.1 2.62 6.8 7.0 correlations we found are high, there are nevertheless several
3 188 27 27.1 3.69 9.4 20.5 exceptions to this trend (see ESI‡ for detail) that will affect the
4 190 50 14.6 6.84 6.2 21.3
5 240 80 11.3 8.83 6.0 22.4 reliability of the data obtained from the truncated set of con-
6 238 84 7.20 13.82 5.5 19.4 formers. This means that it is not guaranteed that no important
7 252 100 13.2 7.50 6.0 10.6 conformers are discarded, but merely that the likelihood of the
8 299 105 9.30 10.70 5.9 29.3
9 314 111 6.90 14.50 5.9 16.2 discarded conformers to give rise to important descendants is
10 332 113 4.90 20.50 6.1 19.1 relatively low. It is of course possible to increase the threshold
11 339 127 21.4 4.70 6.7 20.0 that is applied to the truncation if a higher accuracy (at the
expense of a larger conformational space) is desired.
were propagated to the next step (i.e. the 80–120 energetically We then further examined the fate of the descendants of the
lowest conformers). 10 highest energy conformers of 3–10 in the subsequent step
For 3-epi-xestoaminol C (or (2S,3S)-2-amino-3-tetradecanol, 11) (i.e. in conformers of 4–11). We found that all propagated high-
a total of 339 total conformers were optimised, the contributions of energy conformers resulted in equally high-energy conformers
the lowest 127 make up 490% of the total partition function in the next step. In fact, of the 30 conformers (of 4) descending
(see Table 1). from the 10 highest-energy conformers of 3, only 2 were low
In order to justify this systematic exclusion of high energy enough to be propagated themselves in the next step and they
conformers from propagating into the next step we examined contributed to only 0.11% and 0.12% of the ensemble. In each
the relative energies of all conformers of 1, 2 and 3 (where none step of the propagation, there were never more than 4 out of the
of the conformers had been eliminated). This is necessary to 30 descendants of the 10 highest energy conformers that
ensure that conformations that are high in energy in one themselves were low enough in relative energy to be propagated
compound and hence do not contribute much to the ensemble, and none of the descendants contributed more than 0.4% to
are not low in energy in another compound (and hence contribute the ensemble. Furthermore, it is noteworthy that all conformers
significantly to the ensemble in this other compound). Fortunately, of 9, 10 and 11 were descendants of conformers 1a and 1b (see
this is unlikely for systems with long alkyl chains such as the ones Fig. 2), i.e. the descendants of the conformers without the
in our investigation, since the energetics of the conformational stabilising hydrogen bond of either 1a or 1b were too high in
space is dominated by steric repulsion and the alkyl chain has no energy to feature. In other words, the sheer number of (relatively)
opportunity to interact with other parts of the molecule in a low-energy descendants of 1a and 1b ‘‘crowded out’’ the conformers
stabilising fashion. This is different for compounds that are able that descended from all other conformers of 1 after the 8th step of
to form e.g. intramolecular hydrogen bonds, halogen bonds or other our algorithm.

24508 | Phys. Chem. Chem. Phys., 2016, 18, 24506--24510 This journal is © the Owner Societies 2016
View Article Online

Paper PCCP

In contrast to a full systematic search of the configurational Table 2 Comparison of Experimental and theoretical NMR data for 11.
space of 11, which would have generated 177 147 conformers, C. no. represent the carbon number while dexp and dcalc represent the
experimental and theoretical chemical shifts (based on Boltzmann average of
the 339 optimised conformers found by us represent the most
all the conformers), respectively. dD is the difference between the experimental
important 0.2% of the configurational space that actually and calculated chemical shift at the PBE0-D3BJ/def2-TZVP/SMDDCM level
determines the observable behaviour of the system.
13 1
It should be noted that this new approach is likely only C NMR H NMR
applicable to systems with long alkyl chains that don’t interact C. no. dexp dcalc dD dexp dcalc dD
favourably with other parts of the molecule. Once functional groups 1 16.1 18.9 +2.80 1.27 0.90 0.37
are introduced that can interact with other parts of the molecule, 2 53.5 53.6 +0.10 3.09 2.21 0.88
the above statistical analysis will likely show that lengthening of the 3 73.2 75.3 +2.14 3.44 2.80 0.64
Published on 09 August 2016. Downloaded by Pardubice University on 6/12/2019 7:32:13 AM.

4 34.7 34.9 +0.19 1.40 1.47 0.07


alkyl chain can indeed lead to surprising new conformers that are 1.56 0.90 0.66
lower in energy than their antecedents would otherwise predict. 5 26.4 27.9 +1.47 1.39 1.57 +0.18
In order to test the validity of our approach against experiment, 1.54 1.01 0.53
6 30.8 31.4 +0.61 1.30 1.18 0.12
we measured the optical rotation of a pure sample of 11 in 1.30 1.07 0.23
dichloromethane and compared the result to the computationally 7 30.7 31.7 +0.96 1.30 1.19 0.11
obtained value. Experimentally, the optical rotation was measured 1.30 1.15 0.15
8 30.6 31.5 +0.94 1.30 1.14 0.16
to [a]D = 25.26 deg dm 1 cm3 g 1 (c 0.76 mg mL 1, in CH2Cl2 at 1.30 1.17 0.13
26.22 1C), which is in good agreement with the theoretical value of 9 30.5 31.5 +0.98 1.30 1.15 0.15
20.0 deg dm 1 cm3 g 1 that was obtained from the Boltzmann 1.30 1.14 0.16
10 30.5 31.6 +1.12 1.30 1.15 0.15
averaged result of 339 individual density functional calculations. 1.30 1.11 0.19
Because even close agreement between calculated and experi- 11 30.5 30.4 0.10 1.30 1.24 0.06
mental values of the optical rotation at a single wavelength isn’t 1.30 1.04 0.26
12 33.1 32.7 0.44 1.29 1.18 0.11
conclusive evidence, we have also calculated the corresponding 1.29 1.11 0.18
values for the optical rotation of 11 at several other wavelengths 13 23.8 23.3 0.46 1.31 1.19 0.12
that can be used in further experimental studies for verification. 1.31 1.15 0.16
14 14.5 11.8 2.73 0.91 0.77 0.14
In addition to the optical rotation of 11 in CH2Cl2 that is
reported in this work, we are also able to compare the compu-
tationally determined value of the optical rotation in chloroform Naturally, none of the agreements between theory and
(which was originally reported to be 6.19 deg dm 1 cm3 g 1). experiment are necessarily a reflection on the accuracy of the
Calculation, using the same set of previously obtained confor- calculations, as pointed out above. In addition, it is not always
mers, yields a theoretical value of 13.89 deg dm 1 cm3 g 1, clear if the theoretical setup really reproduces the experimental
which, while still in close agreement, comes with a greater error setup at all. Many compounds are known to form reverse
than the theoretical value in CH2Cl2. In addition to all sources of micelles, particularly in solvent like dichloromethane. Although
error that are mentioned above, the optical rotation in chloro- unlikely in the system under investigation here, it is conceivable
form of course suffers from the fact that its calculation is based that the ‘‘true’’ molecular structure of 11 in solution does indeed
on the ensemble of conformers that have been calculated with differ substantially from the one that was modelled here. We
CH2Cl2 as a solvent and could be expected to yield a better result would argue, however, that any large deviation is likely to
once all structures had been reoptimized. This last example produce a large discrepancy between experimental and compu-
should serve as a cautionary note, indicating that most molecular tational results.
properties are highly sensitive to the exact environment in which
they are measured or computed and while some might transfer
easily between different environments, most do not and need to 4. Conclusions
be recalculated in order to achieve the desired accuracy.
As a second test of our methodology we compared the We have systematically investigated the conformational space
calculated and experimentally obtained NMR chemical shifts for 3-epi-xestoaminol C with a new strategy using high level DFT
of 11. 13C-NMR chemical shifts are most sensitive to the calculations. Our approach began with relaxed surface scans of
chemical environment and are regularly used for the structural (2S,3S)-3-amino-2-butanol to locate all conformers. The obtained
assignment of complex molecules and natural products. Thus, energetic minima were then used to construct the energetically
the accuracy of calculated 13C-NMR chemical shifts is directly lowest conformers of analogues with successively longer alkyl
related to the quality of conformational space covered. chains. This procedure yielded the 339 most important conformers
Table 2 shows a comparison of experimentally and theoretically 3-epi-xestoaminol C. We were able to validate this method by
obtained NMR chemical shifts and their relative deviation (dD). The demonstrating excellent agreement between theoretically and
very close agreement is immediately obvious and closer examina- experimentally obtained values for the optical rotation and the
1
tion reveals that the mean absolute error (MAE) is 0.24 ppm for the H- and 13C-NMR chemical shifts of 11 and hence recommend
1
H-NMR chemical shifts and 1.00 ppm for the 13C-NMR chemical this method for the construction of the conformational space of
shifts respectively. natural product molecules with long alkyl chains.

This journal is © the Owner Societies 2016 Phys. Chem. Chem. Phys., 2016, 18, 24506--24510 | 24509
View Article Online

PCCP Paper

Acknowledgements J. Cioslowski and D. J. Fox, Gaussian Inc., Wallingford CT,


2010.
MAH acknowledges Victoria University of Wellington for a 14 C. Adamo and V. Barone, J. Chem. Phys., 1999, 110, 6158–6170.
Victoria Doctoral Scholarship. Computational facilities were 15 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,
provided by the Victoria University of Wellington High Performance 1996, 77, 3865–3868.
Computer SciFacHPC and Heisenberg. Lukas Hammerschmidt, 16 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,
Julia Schacht and Kevin Tuano are acknowledged for useful 1997, 78, 1396.
discussions. Additional computational time was provided through 17 M. Srebro, N. Govind, W. A. de Jong and J. Autschbach,
NeSI project # nesi00181. We would also like to acknowledge the J. Phys. Chem. A, 2011, 115, 10930–10949.
input from one of the reviewers for their valuable feedback on 18 S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799.
Published on 09 August 2016. Downloaded by Pardubice University on 6/12/2019 7:32:13 AM.

methodology and presentation that ultimately improved this work 19 S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys.,
immeasurably. 2010, 132, 154104.
20 S. Grimme, S. Ehrlich and L. Goerigk, J. Comput. Chem.,
2011, 32, 1456–1465.
Notes and references 21 F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005,
7, 3297–3305.
1 N. Dasyam, A. B. Munkacsi, N. H. Fadzilah, D. S. Senanayake, 22 S. Miertuš, E. Scrocco and J. Tomasi, Chem. Phys., 1981, 55,
R. F. O’Toole and R. A. Keyzers, J. Nat. Prod., 2014, 77, 117–129.
1519–1523. 23 S. Miertus̃ and J. Tomasi, Chem. Phys., 1982, 65, 239–245.
2 G. R. Sullivan, J. A. Dale and H. S. Mosher, J. Org. Chem., 24 J. L. Pascual-ahuir, E. Silla and I. Tuñon, J. Comput. Chem.,
1973, 38, 2143–2147. 1994, 15, 1127–1138.
3 P. Gangoiti, L. Camacho, L. Arana, A. Ouro, M. H. Granado, 25 M. Cossi, V. Barone, R. Cammi and J. Tomasi, Chem. Phys.
L. Brizuela, J. Casas, G. Fabriás, J. L. Abad, A. Delgado and Lett., 1996, 255, 327–335.
A. Gómez-Muñoz, Prog. Lipid Res., 2010, 49, 316–334. 26 V. Barone, M. Cossi and J. Tomasi, J. Chem. Phys., 1997, 107,
4 A. Bagno and G. Saielli, WIREs Comput. Mol. Sci., 2015, 5, 3210–3221.
228–240. 27 E. Cancès, B. Mennucci and J. Tomasi, J. Chem. Phys., 1997,
5 R. W. Hoffmann, Angew. Chem., Int. Ed., 2000, 39, 2054–2070. 107, 3032–3041.
6 N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller 28 B. Mennucci and J. Tomasi, J. Chem. Phys., 1997, 106,
and E. Teller, J. Chem. Phys., 1953, 21, 1087–1092. 5151–5158.
7 A. D. Mackerell, J. Comput. Chem., 2004, 25, 1584–1604. 29 B. Mennucci, E. Cancès and J. Tomasi, J. Phys. Chem. B,
8 R. E. Bruccoleri and M. Karplus, Biopolymers, 1987, 26, 1997, 101, 10506–10517.
137–168. 30 V. Barone and M. Cossi, J. Phys. Chem. A, 1998, 102, 1995–2001.
9 G. Barone, D. Duca, A. Silvestri, L. Gomez-Paloma, R. Riccio 31 M. Cossi, V. Barone, B. Mennucci and J. Tomasi, Chem. Phys.
and G. Bifulco, Chem. – Eur. J., 2002, 8, 3240–3245. Lett., 1998, 286, 253–260.
10 M. D. Beachy, D. Chasman, R. B. Murphy, T. A. Halgren and 32 V. Barone, M. Cossi and J. Tomasi, J. Comput. Chem., 1998,
R. A. Friesner, J. Am. Chem. Soc., 1997, 119, 5908–5920. 19, 404–417.
11 J. H. Jensen and M. S. Gordon, J. Am. Chem. Soc., 1991, 113, 33 R. Cammi, B. Mennucci and J. Tomasi, J. Phys. Chem. A,
7917–7924. 1999, 103, 9100–9108.
12 P. Palla, C. Petrongolo and J. Tomasi, J. Phys. Chem., 1980, 34 M. Cossi, V. Barone and M. A. Robb, J. Chem. Phys., 1999,
84, 435–442. 111, 5295–5302.
13 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, 35 J. Tomasi, B. Mennucci and E. Cancès, THEOCHEM, 1999,
M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, 464, 211–226.
B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, 36 R. Cammi, B. Mennucci and J. Tomasi, J. Phys. Chem. A,
X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, 2000, 104, 5631–5637.
J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, 37 M. Cossi and V. Barone, J. Chem. Phys., 2000, 112, 2427–2435.
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, 38 M. Cossi and V. Barone, J. Chem. Phys., 2001, 115, 4708–4717.
H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, 39 M. Cossi, N. Rega, G. Scalmani and V. Barone, J. Chem.
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, Phys., 2001, 114, 5691–5701.
V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, 40 M. Cossi, G. Scalmani, N. Rega and V. Barone, J. Chem.
A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, Phys., 2002, 117, 43–54.
N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, 41 M. Cossi, N. Rega, G. Scalmani and V. Barone, J. Comput.
V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Chem., 2003, 24, 669–681.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, 42 J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005,
J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, 105, 2999–3093.
G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, 43 A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys. Chem. B,
A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, 2009, 113, 6378–6396.

24510 | Phys. Chem. Chem. Phys., 2016, 18, 24506--24510 This journal is © the Owner Societies 2016

You might also like