Predicting Relative Stability of Conformers in Solution with COSMO-RS

Subscriber access provided by PEPPERDINE UNIV

Article

Predicting Relative Stability of Conformers in Solution with COSMO-RS Astrid Pung, and Ivo Leito J. Phys. Chem. A, Just Accepted Manuscript • DOI: 10.1021/acs.jpca.7b05197 • Publication Date (Web): 18 Aug 2017 Downloaded from http://pubs.acs.org on August 24, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry A is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Predicting Relative Stability of Conformers in Solution with COSMO-RS A. Pung, I. Leito* *Institute of Chemistry University of Tartu Ravila 14a, 50411 Tartu, Estonia corresponding author: Ivo Leito; e-mail: [email protected], Phone: +372-5-184-176

Abstract The ability of the COSMO-RS method to predict the relative stability of different conformers of the same species in solution was evaluated by comparing computational results with experimental conformer data for 101 molecules from 20 literature sources. The solvents CDCl3, CD2Cl2, benzene, toluene, ethyl ether, tetrahydrofuran, acetone, dimethylformamide, DMSO (and DMSO-d6), water and ethanol were used (dictated by the solvents used in the literature). In the case of 16 molecules the quantitative conformer abundances were also available and were compared with the data from computations. The results show that although COSMO-RS reproduces the conformer abundances only very approximately, the most stable conformer is determined correctly in 100 cases out of 101. This result validates the use of COSMO-RS in a number of applications (distribution coefficients, vapor pressures, solubilities, etc.) where the most stable conformer of a molecule can be different in different phases and accounting for this difference is important for obtaining meaningful results.

Introduction A conformation of a molecule – a specific spatial arrangement of the atoms in a molecule that arises from the rotation about the bonds linking such atoms – can vary. The relative stability of conformations – conformers – of a molecule depends besides the intrinsic properties of the molecule also on the environmental conditions, e.g. temperature and solvent. The ability to computationally predict the most stable conformers of a molecule under given conditions is much needed for a variety of applications. It helps understanding the interaction between molecules as well as the involved steric effects. This allows for better planning of chemical reactions, creating drugs for specific targets (e.g. predicting how enzymes would interact with different structural parts of a molecule) and gaining a better understanding of molecular properties in a given environment.1 This ability is critically important for such computational applications where the most stable conformer of a molecule can be different in different phases and accounting for this difference is important for obtaining meaningful results, e.g. distribution coefficients, vapor pressures and solubilities. Conformer analysis is also relevant for the field of host-guest chemistry for the creation of molecular receptors. Prior conformer analysis of the receptor and the host-guest complex geometries allows planning for several effects enhancing the binding affinity and selectivity, for example the preorganization and complementarity effects.1 1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 20

Multiple experimental methods can be used to gain some insight into most stable conformer geometries. NMR and IR spectroscopy are two of the most commonly used methods for conformational analysis.1,2 For conformer analysis with IR spectroscopy, the spectra of conformationally heterogeneous substances will usually display absorption bands due to individual conformers. The intensity of these absorption bands will be proportional to the conformer population. Generally, the intensities of bands are measured at two temperatures and the results used to calculate the ∆H values using the van't Hoff equation).1 For any spectroscopic measurements in determining conformers, it is important whether the time scale of the measurement is fast or slow compared to the time scale of rotation around single bonds. For NMR measurements, when the measurements are carried out at low temperatures, the conformational switching is slow enough so that conformer abundance can be directly monitored. If the measurement time scale is slower than the time scale of conformational change, the measured spectroscopic parameters are averages over all the stable conformers present in the solution.1 For systems where individual conformers can be detected by NMR measurements at low temperature, the relative conformer populations may be obtained from signal integration of a 13C or 1H NMR spectra and the structures can be assigned using the chemical shifts and coupling constant values measured for each conformer.1 For molecular systems where individual conformers cannot be detected, NMR can be used as a tool for conformational analysis by using the relation between the experimental NMR coupling constant and the coupling constants of individual conformers (Jobs = NA * JA + NB * JB, where Jobs is the experimental NMR coupling constant value and NA and NB and JA and JB the mole fractions and coupling constant values of the respective conformers A and B). This approach works when conformationally locked models are available for the molecules and the coupling constants of individual conformers can be found.1 For other molecular systems, for example small organic molecules, a common approach is using quantum mechanical methods to calculate specific NMR parameters, commonly the coupling constants and related parameters.2 A more detailed overview of using NMR for small molecule conformational analysis is found in Ref.2. The aim of this work was to check the ability of the COSMO-RS for the prediction of relative stability of different conformers of molecules in various solvents. First, a sufficiently large body of experimental data of relative abundance of conformers of molecules in different solvents has been collected from literature. COSMO-RS computations of the same species in the same solvents were carried out and the most stable conformers and, if available, their relative abundances, are compared. Experiments and computations yield quite different information on conformers. The results of computations are very detailed, giving the list of conformers together with their exact geometries, energies and abundances. In contrast, data obtained from experiments can often be limited to just a broad description of the most stable conformer, a la "chair conformation is

2 ACS Paragon Plus Environment

Page 3 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


significantly more abundant than boat conformation". For this reason, the criteria for comparison need to be elaborated. Having reviewed the literature sources, we have set two main criteria of determining the predictive ability of computations. Firstly, the most stable conformer found by computations should match the most stable conformer reported in the literature, as far as the literature geometry data are sufficient to identify it. Secondly, in case there is experimental information on several most stable conformers, the stability order of the two most stable conformers should be reproduced correctly. The possibility of applying these criteria was one of the conditions of using data from a particular literature source (see below).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 20

Theory and Computational Details COSMO-RS (Conductor-like Screening Model for Real Solvatation)3 is a method for computing the thermodynamic properties of fluids and solutions based on a combination of quantum mechanics and statistical thermodynamics. The COSMO-RS calculation consists of two steps. In the first step, DFT calculations are carried out for a CSM (continuum solvation model) method (COSMO), a molecular cavity is created and the screening charge distribution on the molecule's sigma surface is calculated imposing conductor-like boundary condition by setting the medium dielectric permittivity (epsilon) to infinity. The first calculation step results are the atomic coordinates of optimized geometry, the mapping of the polarization charge densities onto the cavity surface (sigma surface) and the total energy in the conductor. In the second step (RS), the molecules are "placed" into the liquid and methods of statistical thermodynamics are used to determine the different forces that affect the molecule- van der Waals forces, electrostatic interactions, hydrogen bonding interactions- through the interaction of surface segments of different molecules that are present in the liquid.4 The calculation finds the energy of a molecule in the chosen solvent, the value of which is converted to Gibbs free energy of the molecule in the predefined solvent composition through parametrization.3 These free energy values can eventually be used in the prediction of conformer abundance, solubilities, logP values, etc. in different mixtures.4 Some of the fields of application include chemical engineering thermodynamics, agricultural and pharmaceutical research.3 The key advantage of COSMO-RS compared to several alternative approaches of similar computational demand is that it can be used for computations in any solvents and solvent mixtures and at nonzero concentrations. The alternative methods are normally parametrized for one or few specific solvents and assume infinitely dilute solutions. An important step of COSMO-RS calculation is statistical weighting of the conformers of the calculated species X according to their energy in the solvent. Correct conformer weight prediction depends on the correctness of their relative energies and is vital to ensure the accuracy of further predictions of different data for species X from this method. This is especially important if the investigated phenomenon involves different phases where the relative stabilities of the conformers can differ. For example, in the case of predicting the partition between two phases, one conformer may be more stable in one phase while another conformer is more stable in the second phase. If the individual weights of the conformers were predicted incorrectly, it would be impossible to receive a correct final prediction for the partition coefficient. For COSMO-RS conformer analysis, the individual weight of a single conformer in a chosen solvent mixture is determined according to Boltzmann distribution between states of different free energies, expressed in eq 1.

 E X COSMO (i) + µ SX (i)   kT  

ω X (i) exp− π SX (i) =

 E X COSMO ( j ) + µ SX ( j )  ( j ) exp ω −  ∑j kT  

(1)

X

π SX (i) represents the population of conformer i of compound X in solvent mixture S. A compound X is represented by a set of COSMO files for the separate conformers. For each 4 ACS Paragon Plus Environment

Page 5 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


conformer, a multiplicity ωx(i) is assigned on the basis of geometrical degeneration aspects. Each conformer i has a respective COSMO energy E X COSMO (i ) . In COSMO-RS the interactions between all species in the mixture are taken into account, including interactions between different conformers of the same species. For this reason, the chemical potentials of conformers, µ SX (i) depend on the abundances of conformers. Therefore, in real calculations the eq 1 is iterated to self-consistency, starting from an initial population guess based on the condition µ SX (i) = 0.4

Selecting Experimental Data from Literature For the study, experimental conformer analysis data for 101 molecules from 20 sources5 – 25 were selected for computational analysis using the COSMO-RS method.3-4 The choice was made according to the following criteria: • •

•

The conformational analysis is carried out experimentally (although computations can assist) and the discussion on conformations is mostly based on the experiments. Reasonable size of the molecule or ion. This is important to ensure that information about their preferred conformations can be reliably gained from experimental analysis and that computational effort is acceptable. There needs to be sufficient information given about the conformers that meaningful comparison could be carried out with computational results. While experimental analysis cannot determine exact atom coordinates for the conformers, information can be gained about the orientation of substituents, basic conformation of cyclic molecules etc.

In case of some literature sources certain molecules were left out that would not have led to insightful information (either only a single conformer or insufficient information given on conformer distribution). The chosen molecules are grouped into four groups for further discussion. The first group consists of small to medium organic molecules from 5 literature sources.5-9 The second group consists of 12 molecules from 6 literature sources that were experimentally studied to determine their conformations related to intra- or intermolecular hydrogen bond forming abilities.10-15 The third group consists of 84 molecules from 7 literature sources. Larger sets of experimental data of similar molecules were compared with computational results to find if any discrepancies would arise in case of specific substituents or substituent positions in the molecules.16-22 In addition to the groups of molecules mentioned above, two literature sources23,24 were found where one molecule (ethylene glycol and cinchonidine respectively) were studied in various solvents, to determine the effect of solvent choice of the preferred conformation of the molecule. Finally, literature sources also determined the experimental conformer abundance of two most stable conformers in 17 cases. 5 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 20

Computational Parameters

Geometry optimization for this study was carried out using Turbomole v.6.525 with the following criteria: BP functional with TZVP basis set, wave function convergence criterion: max difference between SCF energies of two successive iterations 10-6 Hartree, geometry convergence: max gradient |dE/dxyz| 10-3 Hartree Bohr-1. For the conformer weight analysis, with the parametrization the program COSMOThermX17 v.C30_170126 BP_TZVP_C30_1701 was used. For the calculations, if deuterated solvents were used in the experiments, the solvents used in calculations were replaced with the corresponding nondeuterated version. The temperature was chosen according to the literature data. If room temperature was specified in the source then a temperature of 25º C was used. It was found very important to use a sufficient number of initial geometries for the geometry optimization. This is especially so with complex systems having high steric strain. Without giving a sufficient number of different geometries for the computational conformer stability predictions, important conformers may be missed, leading to incorrect conformer ratios. A number of initial geometries were generated manually to assess the prevalence of more likely conformations of the molecule. As the goal of the computations was to also assess the possibility of COSMO-RS method incorrectly predicting another conformer as the most stable, a suitably large number of potentially stable conformations had to be taken as the initial geometries for the work. The optimized geometries were used in the RS part of COSMO-RS calculation as a conformer set and it was checked that the computationally determined most stable conformer had the same characteristics (position of functional groups, conformation of cyclic part of molecule, etc.) as the experimental results from literature sources. Frequency calculations were carried out to ensure that the found conformers corresponded to true energy minima. Results and Discussion Qualitative prediction of the most stable conformer. The studied molecules are grouped into three groups and the results are presented in Tables 1-3. The structures of the compounds are available in Section I of the Supporting Information. The images of the most stable conformers of molecules studied in Table 1 and Table 2 are available in Section II of the Supporting Information. The first group consists of small to medium organic molecules from 5 literature sources (Table 1). In the case of 1 to 4 the experimental studies determined the prevalence of anti and gauche conformers at low temperature using Raman and FT-IR spectroscopy. The results varied from both forms being present to either anti or gauche form being dominant. The computational results agree in qualitative terms with experimental data in all cases. In the case of 5 the two rotamers with respect to the COOH group were evaluated.


Page 7 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1. Computational Prediction of Conformer Balance of Small Molecules (Ref. 5-9) Compounds Exp. Method

Solvent

silanes 1a-1b Silanes 2 Silanes 3 Cyclopropanes 4a-4b 5methoxyind ole-2carboxylic acid 5

FT/IR

Sample as a liquid Liquid Xe

FT/IR

Liquid Xe

Trans/gauche

FT/IR

Liquid Xe

Anti/gauche

NMR

DMSO

Orientation of COOH group in the molecule

Raman

Type of conformer studied Anti/gauche Anti/gauche

Information from experiment Both forms present Anti form more stable Trans form more stable Trans form more stable

Agreement? (Y/N)

Exp. Ref.

Yes

5

Yes

6,7

Yes

7

Yes

8

One possible orientation of COOH group is more stable

Yes

9

The second group (Table 2 and Scheme 1) consists of molecules from 6 literature sources capable of forming inter- or intramolecular hydrogen bonds. Intramolecular hydrogen bonds (HBs) are interesting to study for multiple reasons. The existence of an intramolecular HB will cause the conformer to be fixed in a set position. However, the existence and position of HBs is not always clear. In the case of some solvents, the solvent molecules may compete with forming their own hydrogen bonds with the hydrogen bond donor or hydrogen bond acceptor center of the molecule and therefore cause there to be no intramolecular HB in the most stable conformer. In addition, with larger molecules, there may be several different possibilities for intramolecular HB formation. In this case it would be important to correctly predict, which conformer and which HB is the more stable one. The experimental results confirmed the existence or lack of intramolecular hydrogen bonds in the most stable conformer for the chosen environment, computations confirmed the experimental findings in every case. The results show that in the case of simple molecules, the COSMO-RS method can correctly predict the existence or lack and the stability of hydrogen bonds and indicate, which of the possible alternative conformers is the most stable.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 20

Scheme 1. Most Stable Conformer Geometries for Molecules 6, 7b, 8b, 9a, 10a and 17b

Table 2. Computational Prediction of the Most Abundant Conformer in the Case of Molecules Capable of Forming Inter- or Intramolecular Hydrogen Bonds (Ref 10 - 15) Compound 2,5-dihydroxy-1,4benzoquinone 6 1,3-Bis(2-quinolyl)-2(p-chlorophenyl)-2propanol 7 pyridine-2,6dicarboxamide 8a 1,3benzenedicarboxamide 8b 3-diethylaminomethyl5-methoxy-salicylic

Exp. method FT-IR

Solvent CHCl3 CDCl3

NMR

CHCl3

NMR

DMSO-D6

NMR

DMSO-D6

FT-IR, NMR

CHCl3

Information from experiment 2 HBs between hydroxy and carbonyl groups OH⋯N HB rotates between the two nitrogen atoms Two intramolecular HBs Preferred conformation given in Scheme 1 OH⋯N intram. HB is dominant

Agreement? (Y/N) Yes

Exp. Ref 10

Yes

11

Yes

12

Yes

12

Yes

13


Page 9 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


aldehyde 9a 3-diethylaminomethyl5-bromo-salicylic aldehyde 9b 6 Thiosemicarbazones 10a-10f

FT-IR, NMR

CHCl3

OH⋯O=C intram. HB is dominant

Yes

13

NMR, IR, UV/Vis

CHCl3

One intramolecular HB exists in all molecules

Yes

14, 15

The third group consists of more complex molecules with various functional groups (Table 3). 84 molecules from 7 literature sources were studied to determine the geometry of the cyclic molecules (boat/chair/twist conformations) and/or gain information on the orientation of the substituent groups (axial and equatorial orientation). Sets of similar molecules were studied in each literature source to determine the difference in preferred conformations with the variation of substituent groups. The results displayed in Table 3 show that in 80 cases out of 80, the experimentally determined information about geometries of the most stable conformers is predicted correctly.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Page 10 of 20

Table 3. Computational Prediction of Conformer Balance for Miscellaneous Molecules from Ref. 16 - 22 Compound type

Exp. method 1H NMR; 13C NMR

Solvent

Information from experiment

Computations Exp. agree Ref 16 10/10

Chloroalkyl derivatives Anti-peri-planar form is predominant CDCl3 of naphtoquinones 11a11j a 17 2-oxo-1,3,2-dioxathianes NMR* DMSO Chair form most stable, axial vs equatorial orientation of 7/7 methyl substituents and SO group 12a-12k 18 Phosphorinanones 13a1H NMR; DMSO Chair form is most stable, C6H5 groups are axial or 6/ 6 13C NMR; equatorial depending on the molecule 13f 31P NMR 19 Phenyl cyclohexanone 1H NMR; CDCl3 Chair form with axial OH group and equatorial 4/4 oximes 14a-14db 13C NMR orientation of other substituents 20 Pyrimidinone 1H NMR; CDCl3 Half-chair form, methyl substituent orientation depends 22 / 22 (derivative)s 15a-15v 13C NMR on its position (6-axial, 7,8-equatorial, 9-mix of two) 21 Diazetidines 16a-16g H NMR; C Misc. Conformation equilibrium of the side groups and their 7/7 NMR rotation was studied 22 2-oxo-1,3-dioxane and H NMR; C CDCl3 Equatorial chair form is more stable, axial vs equatorial 24/24 derivatives 17a-17xa NMR orientation of methyl substituents a In case of molecule 11d, the exact conformation of the chloroethyl substituent was not given in the experimental data. b Some molecules studied in ref 19 have been omitted as not enough information is given in the source for meaningful comparison with computational geometries.

10

ACS Paragon Plus Environment

Page 11 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In Ref.17, 11 2-oxo-1,3,2-dioxathianes were studied in DMSO. Chair conformations were found to be the most stable with the orientation of SO group and methyl substituents depending on each particular molecule. Ref. 18 presents 6 phosphorinanones that were studied in DMSO. Three molecules were found to have axial orientation of C[2]-C6H5 bonds and the rest of the studied molecules should have equatorial orientation of the same bonds. The methyl substituents were found to have equatorial orientation. This was correctly predicted computationally. For the 4 phenyl cyclohexanone oximes studied in Ref. 19, a chair conformation was the most stable in CDCl3 with axial orientation of the OH group and equatorial organization of the other substituents. This was correctly predicted computationally. In Ref. 20, 2 pyrimidinones and 20 pyrimidinone derivatives were studied in CDCl3 by 1H NMR and 13C NMR. It was determined that the preferred conformation of the methyl substituents depends on their position: position 6 prefers axial orientation, positions 7 and 8 prefer equatorial orientation and position 9 takes a mixture of both. This was correctly predicted by the COSMO-RS method. In ref. 21, the conformational equilibrium of a selection of diazetidines was studied by low temperature NMR to determine the method of conformer inversion. Nitrogen inversion was determined to occur in all molecules and the preferred conformations for such conformer change were also correctly predicted by COSMO-RS. In Ref.22, the conformation of 2-oxo-1,3-dioxane and its 23 derivatives were studied in CDCl3 using 1H NMR and 13C NMR. Equatorial chair form was determined to be the preferred conformation and the axial or equatorial orientation of the methyl groups was also studied. Computational predictions fully agree with experimental data from literature. In refs 23 and 24, the stability of conformations of 2 molecules – ethylene glycol23 and cinchonidine (19),24 respectively – was studied in multiple solvents (D2O, CDCl3 and DMSOd6 for ethylene glycol; benzene, toluene, ethyl ether, tetrahydrofuran, acetone, dimethylformamide, dimethyl sulfoxide, water and ethanol for cinchonidine). In the case of ethylene glycol, it was determined that there is a strong preference for the gauche conformation.23 This was correctly predicted by our computations. For cinchonidine, four most stable conformers were found and it was determined that the prevalent conformer is not the same in all solvents.24 While some of the conformational information was determined computationally, experimental NMR measurements in d6-acetone solvent were also carried out for the study.24 We found computationally two most stable conformers for 19 (Scheme 2) and compared their abundances with experimental values in 10 different solvents. Conformer 19_01 is similar to the one determined to be the most stable conformer in ref. 24 (denoted as "Open 3" in ref 24). Conformer 19_02 is a conformer with an intramolecular hydrogen bond that was not mentioned in ref 24. According to our computations, it should exist and be the most stable conformer in the majority of cases. Comparison of computational and literature results for 19 is given in Table 4.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 20

Scheme 2. The Most Stable Conformer Geometries for Molecule 19 as found computationally.

Table 4. Conformation abundance of one of the most stable conformers (19_01 and 19_02) of cinchonidine. Computational Experimental abundance a abundance of of conformer 19_01 (ref 24) conformer 19_01 Benzene 0.59 0.08 Toluene 0.70 0.09 Ethyl ether 0.71 0.34 Tetrahydrofuran 0.62 0.44 Acetone 0.41 0.34 Dimethylformamide 0.34 0.49 Dimethyl sulfoxide 0.27 0.61 Water 0.33 0.05 Ethanol 0.78 0.19 a Abundance data were obtained from Figure 5 in ref 24. Solvent

Computational abundance of conformer 19_02 0.81 0.81 0.56 0.47 0.56 0.42 0.31 0.81 0.71

Table 4 reveals some discrepancies between the experimental and computational results. This may largely arise from the fact we did not computationally detect exactly the same two conformers to be the most stable ones, as were found experimentally. An important factor playing role in this is in the possibility of intramolecular hydrogen bond formation. It is possible that COSMO-RS overestimates the stabilizing effect of hydrogen bond formation in the molecule and underestimates the steric strain caused by the conformation that makes this hydrogen bond formation possible.

Quantitative prediction of conformer abundance. In few cases, literature sources also provide the relative abundances of conformers for some molecules. Expectedly, the computational 12 ACS Paragon Plus Environment

Page 13 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ratios between conformer abundances do not fully agree with experimentally found abundances. This conformer abundance data and the corresponding values determined computationally with the COSMO-RS method are shown in Table 5. Molecule 17n was omitted from the table as we could not reproduce a suitable second most stable conformer that would be optimized and pass the frequency calculations without imaginary frequencies.

Table 5. Computational Predictions for Conformer Abundance Data from Refs. 6, 17, 22. Molecule

Solvent Literature data Computational data ∆ECOSMO Abundance Ratio Abundance% Ratio (kcal/mol) (%) 1.76 Chloromethylcyclopropane Xe 88:12a 99.99:0.01 4a 0.14 0.00 2.20 Bromomethylcyclopropane Xe 92:8a 99.99:0.01 4b 0.09 0.00 80:20 0.83 Dioxane derivative 17c DMSO 76:24b 0.32 0.25 b 68:32 0.43 Dioxane derivative 17i DMSO 95:5 0.05 0.47 68:32 0.46 Dioxane derivative 17k DMSO 67:33b 0.49 0.47 91:9 1.35 Dioxathiane 12a CDCl3 80:20b 0.25 0.10 b 89:11 1.22 Dioxathiane 12b CDCl3 84:16 0.19 0.12 71:29 0.53 Dioxathiane 12c CDCl3 62:38b 0.61 0.41 89:11 1.24 Dioxathiane 12d CDCl3 85:15b 0.18 0.12 b 91:9 1.38 Dioxathiane 12e CDCl3 74:26 0.35 0.10 62:38 0.30 Dioxathiane 12f CDCl3 52:48b 0.92 0.61 b 73:27 84:16 0.96 Dioxathiane 12g CDCl3 0.37 0.19 b 91:9 1.37 Dioxathiane 12h CDCl3 79:21 0.27 0.10 93:7 1.56 Dioxathiane 12i CDCl3 68:32b 0.47 0.07 b 89:11 1.23 Dioxathiane 12j CDCl3 72:28 0.39 0.12 88:12 1.15 Dioxathiane 12k CDCl3 92:8b 0.09 0.14 a - cis-trans conformer balance, trans conformer more stable; b- balance between chair conformers.

These calculations were also carried out using the DFT BP-TZVPD-FINE computational parameters. The results can be seen from SI Table S1. Overall, there was no noteworthy improvement compared to the results of DFT BP-TZVP level calculations. Additionally, in the case of molecule 12f, the BP-TZVP-FINE level calculation actually predicted the order of the two most stable conformers incorrectly. A previous study in 2011 examined the effects of using different basis sets in the COSMO-RS method for the calculation of activity coefficients at infinite dilution. The study came to the conclusion that while at least a valence double-beta basis set is necessary, using larger basis sets shows no advantages.27 It can be seen from the results shown in Table 5 that while sometimes, the quantitative conformer abundance data can be approximately reproduced, it is not always the case and therefore the method should not be used for quantitative predictions of conformer weights. At the same time in all cases computations have predicted the most stable conformer correctly. The likely cause of this issue is simple – the ratio of conformers is very sensitive to the 13 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 20

difference of Gibbs' free energies between conformers. For example, the conformer ratio change from 1:3 to 1:1, i.e. three times, corresponds to only 0.6 kcal mol-1 change in the difference of their Gibbs' free energies, which for computations (especially in solution), is not much.


Page 15 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusions In general, while COSMO-RS can in the large majority of cases correctly predict the most stable conformer, it is generally not able to quantitatively accurately predict conformer abundance ratios. For the studied small molecules, the predictions are qualitatively correct for the studied experimental data for determining the most stable conformer. In all 12 cases, where conformations of molecules capable of intramolecular hydrogen bond formation were predicted, the predictions matched with literature data. Predictions were qualitatively accurate both in terms of the existence and finding the most stable intramolecular hydrogen bond conformer. For the determination of chair, boat and twist conformations and equatorial or axial orientation of substituents in cyclic molecules, the predictions were mostly qualitatively accurate. The orientation of substituents around the cyclic parts was predicted correctly. Questions arose in a few cases about the most stable conformation of flexible side groups of specific molecules. As the experimental data from literature gave no information, the computational predictions could not be checked with experiment. Supporting Information The supporting information for this manuscript includes the graphical representation of all studied molecules and the images of the most stable conformers for molecules 1a-10f. Acknowledgements This research was funded by the Ministry of Education and Science of Estonia (the institutional research grant No. IUT20-14).

References (1) (2)

(3)

(4) (5)

(6)

(7)

(8)

Eliel, E. L.; Wilen, S. H. Stereochemistry of Organic Compounds; Wiley, 1994. Tormena, C. F. Conformational Analysis of Small Molecules: NMR and Quantum Mechanics Calculations. Prog. Nucl. Magn. Reson. Spectrosc. 2016, 96, 73–88 DOI: 10.1016/j.pnmrs.2016.04.001. Klamt, A.; Eckert, F.; Arlt, W. COSMO-RS: An Alternative to Simulation for Calculating Thermodynamic Properties of Liquid Mixtures. Annu. Rev. Chem. Biomol. Eng. 2010, 1 , 101–122 DOI: 10.1146/annurev-chembioeng-073009-100903. Klamt, A. COSMO-RS: From Quantum Chemistry to Fluid Phase Thermodynamics and Drug Design; Elsevier, 2005. Powell, D. L.; Klaeboe, P.; Gruodis, A.; Nielsen, C. J.; Guirgis, G. A.; Durig, J. R.; Aleksa, V. Infrared and Raman Spectra, Conformations and Ab Initio Calculations of Ethyl Bromosilane and Ethyl Dibromosilane. J. Mol. Struct. 1999, 482, 391–396. Guirgis, G. A.; Nashed, Y. E.; Durig, J. R. Raman and Infrared Spectra, Conformational Stability, Ab Initio Calculations and Vibrational Assignments for 2Chloroethyl Silane. J. Mol. Struct. 1999, 510, 13-34. Guirgis, G. A.; Nashed, Y. E.; Klaeboe, P.; Durig, J. R. Conformational Stabilities, Ab Initio Calculations and Structural Parameters for 2-Chloroethyl Silane and 2Chloroethyl Trifluorosilane. J. Mol. Struct. 1999, 482, 385–389. Durig, J. R.; Shen, S.; Zhu, X.; Wurrey, C. J. Conformational and Structural Studies of Chloromethylcyclopropane and Bromomethylcyclopropane from Temperature 15 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(9)

(10) (11)

(12)

(13)

(14)

(15)

(16) (17)

(18)

(19)

(20)

(21)

(22)

Page 16 of 20

Dependent FT-IR Spectra of Xenon Solutions and Ab Initio Calculations. J. Mol. Struct. 1999, 485, 501–521. Cinar, M.; Karabacak, M.; Asiri, A. M. An Experimental and Density Functional Study on Conformational and Spectroscopic Analysis of 5-Methoxyindole-2-Carboxylic Acid. Spectrochim. Acta. A. Mol. Biomol. Spectrosc. 2015, 137, 670–676 DOI: 10.1016/j.saa.2014.08.090. Szabó, A.; Kovács, A. Hydrogen Bonding and Molecular Vibrations of 2, 5Dihydroxy-1, 4-Benzoquinone. J. Mol. Struct. 1999, 510, 215–225. Ośmiałowski, B.; Nissinen, M.; Kolehmainen, E.; Gawinecki, R. NMR Spectral and XRay Structural Investigation of 1, 3-Bis (2-Quinolyl)-2-(P-Chlorophenyl)-2-Propanol. J. Mol. Struct. 2000, 525, 241–245. Marlin, D. S.; Olmstead, M. M.; Mascharak, P. K. Extended Structures Controlled by Intramolecular and Intermolecular Hydrogen Bonding: A Case Study with Pyridine2,6-Dicarboxamide, 1,3-Benzenedicarboxamide and N,N`-Dimethyl-2,6Pyridinedicarboxamide. J. Mol. Struct. 2000, 554, 211–223. Szady-Chełmieniecka, A.; Rozwadowski, Z.; Dziembowska, T.; Grech, E.; Wojciechowski, G.; Brzezinski, B. Intramolecular Hydrogen Bonds in 3Diethylaminomethyl-5-R-Salicylic Aldehydes. J. Mol. Struct. 2000, 520, 39–43. Beraldo, H.; Lima, R.; Teixeira, L. R.; Moura, A. A.; West, D. X. Crystal Structures and IR, NMR and UV Spectra of 3-Formyl- and 3-Acetylpyridine N (4)Methylthiosemicarbazones. J. Mol. Struct. 2000, 553, 43–48. Beraldo, H.; Lima, R.; Teixeira, L. R.; Moura, A. A.; West, D. X. Crystal Structures and IR, NMR and UV Spectra of 4-Formyl-and 4-Acetylpyridine N (4)-Methyl-and N (4)-Ethylthiosemicarbazones. J. Mol. Struct. 2001, 559, 99–106. Prezhdo, V. V.; Ovsiankina, E. V.; Prezhdo, O. V. Conformational Analysis of Chloroalkyl Derivatives of 1,4-Naphthoquinone. J. Mol. Struct. 2000, 522, 71–77. Pihlaja, K.; Rossi, K.; Nikander, H. Conformational Analysis. 25. 13C NMR Chemical Shifts-Sensitive Detectors in Structure Determination. 3. The Proposal for Non-Chair Conformations in Methyl-Substituted 2-Oxo-1,3,2-Dioxathianes Challenged. J. Org. Chem. 1985, 50, 644–649. Rampal, J. B.; Berlin, K. D.; Edasery, J. P.; Satyamurthy, N.; Van der Helm, D. Carbon-Phosphorus Heterocycles. Synthesis and Conformational Analysis of AlkylSubstituted 1, 2, 6-Triphenyl-4-Phosphorinanones and Derivatives. J. Org. Chem. 1981, 46, 1166–1172. Balachander, R.; Sameera, S. A.; Mohan, R. T. S. Synthesis and 1H and 13C NMR Spectral Study of Some r(2),c(4)-Bis(isopropylcarbonyl)-c(5)-Hydroxy-t(5)-Methylt(3)-Substituted Phenyl, Cyclohexanones and Their Oximes. J. Mol. Struct. 2016, 1115, 63–69 DOI: 10.1016/j.molstruc.2016.02.086. Podányi, B.; Hermecz, I.; Vasvári-Debreczy, L.; Horváth, Á. Nitrogen Bridgehead Compounds. 62. Conformational Analysis of 6,7,8,9-Tetrahydro-4H-pyrido[1,2A]pyrimidin-4-Ones and Their Methyl Derivatives by NMR Spectroscopy. J. Org. Chem. 1986, 51, 395–399. Landis, M. E.; Mitchell, J. C.; Offen, D. L. Conformational Analysis of Fused-Ring 1, 2-Diazetidines by Carbon-13 Nuclear Magnetic Resonance Spectroscopy. J. Org. Chem. 1981, 46, 501–505. Pihlaja, K.; Rossi, K. Conformational Analysis. XXIII. 13C and 1H NMR Conformational Study of 2-Oxo-1,3-Dioxane and Its Methyl Derivatives. Acta Chem. Scand. 1983, 37, 289–296.


Page 17 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(23) Petterson, K.A.; Stein, R.S.; Drake, M.D; Roberts, J.D. An NMR Investigation of the Importance of Intramolecular Hydrogen Bonding in Determining the Conformational Equillibrium of Ethylene Glycol in Solution. Magn. Reson. Chem. 2005, 43, 225-230. (24) Bürgi, T.; Baiker, A. Conformational Behaviour of Cinchonidine in Different Solvents: A Combined NMR and Ab Initio Investigation. J. Am. Chem. Soc. 1998, 120, 1292012926. (25) TURBOMOLE, V6.5 2013; a Development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989–2007, TURBOMOLE GmbH, since 2007; Available from http://www.turbomole.com (accessed August 17, 2017). (26) Eckert, F.; Klamt, A.COSMOtherm, Version C30_1701; COSMOlogic GmbH & Co.: KG, Leverkusen, Germany, 2016. (27)

Franke, R.; Hannebauer, B. On the Influence of Basis Sets and Quantum Chemical Methods on the Prediction Accuracy of COSMO-RS. Phys.Chem.Chem.Phys. 2011, 13, 21344-21350. DOI: 10.1039/c1cp22317h

TOC Graph:



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1. Most Stable Conformer Geometries for Molecules 6, 7b, 8b, 9a, 10a and 17b 159x138mm (96 x 96 DPI)


Page 18 of 20

Page 19 of 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Scheme 2. The Most Stable Conformer Geometries for Molecule 19 as found computationally 440x203mm (96 x 96 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Toc Graph 174x81mm (96 x 96 DPI)


Page 20 of 20

Predicting Relative Stability of Conformers in Solution with COSMO-RS

Recommend Documents