Molecular Structure of Heterocycles : 6 . Solvent Effects on the 17 O NMR Chemical Shifts of 5-Trichloromethylisoxazoles

Com o objetivo de elucidar e quantificar os efeitos do solvente sobre os deslocamentos químicos de 17O de três 5-triclorometilisoxazóis [(1a) não-, (1b) 3-metile (1c) 4-metil-substituído] foi realizada uma análise de regressão multilinear, utilizando os parâmetros solvatocrômicos de KamletAbboud-Taft (KAT). Os deslocamentos químicos do átomo de oxigênio do anel, O1, dos compostos 1a-c mostraram dependências (em ppm) em função da polaridade-polarizabilidade do solvente de -4.8π*, -3.2π*, -8.9π*, em função da acidez do solvente (HBD) de 0.9α, -0.2α, -2.7α e em função da basicidade do solvente (HBA) de -0.4β, 1.9β, 0.9β, respectivamente. Os dados de carga líquida de O1 e de momento de dipolo, obtidos por cálculos de orbitais moleculares (AM1), são comparados com os parâmetros de efeitos do solvente determinados para os compostos 1a-c.


Introduction
Several papers have been devoted to the empirical and theoretical studies of solvent effect on the 17 O chemical shifts in different organic compounds 1,2 .Special attention has been devoted to the study of solvent effects in amides, where the 15 N and 17 O nuclei are observed 2 .Recently we applied a multilinear-regression analysis using the Kamlet-Abboud-Taft (KAT) 3 solvatochromic parameters in order to elucidate and quantify the solvent effects on the 17 O chemical shifts of 1,1,1-trichloro-4-methoxy-3-alken-2ones 4 and 5-hydroxy-4,5-dihydroisoxazoles 5a .According to the KAT formalism, the observed chemical shift of compound X at infinite dilution in solvent Y, δ X Y , would be given by the relationship 3 shown in Equation 1.
The solvent effects are described by the solvent parameters δ X CH , π* Y , δ Y , α Y and β Y .The π* Y scale is an index of solvent dipolarity/polarizability, which measures the ability of the solvent to stabilize a charge or a dipole due to its dielectric effect.The α Y scale of solvent hydrogenbond-donor (HBD) acidities describes the ability of the solvent to donate a proton in a solvent-to-solute hydrogen bond.The β Y scale of hydrogen-bond-acceptor (HBA) basicities measures the ability of the solvent to accept a proton (i.e., to donate an electron pair) in a solute-to-solvent hydrogen bond.The δ Y parameter is a polarizability correction term for polychlorinated (δ Y = 0.5) and aromatic (δ Y = 1.0) solvents.The coefficients s X , a X and b X in Equation 1 define the sensitivity of δ X Y to solvent dipolarity/ polarizability, acidity and basicity, respectively.The product of coefficients s X d X defines the sensitivity of δ X Y for the *e-mail: mmartins@base.ufsm.br; http://www.ufsm.br/nuquimhepolarizability correction term.The term δ X CH is the chemical shift of substrate X measured in cyclohexane since this reference solvent does not form hydrogen bond (α CH = β CH = 0) and was selected to define the origin of π* Y scale (π* CH = 0).The term s X (π* Y + d X δ Y ) accounts for the difference between the contribution to δ X Y in solvent Y and in cyclohexane from the solute-solvent interactions other than hydrogen bonding.The terms a X α Y and b X β Y represent the contributions from hydrogen bonds of substrate X with solvents HBD and HBA, respectively.
As a part of our research program we have studied the synthesis [6][7][8][9] , structure 5 and the multi-nuclear NMR chemical shifts 10 of 5-, 6-and 7-membered heterocycles.The aim of this work is to elucidate and quantify the solvent effects on the 17 O chemical shifts of 5-trichloromethylisoxazoles 1a-c using the Kamlet-Abboud-Taft (KAT) solvatochromic parameters 3 (Scheme).
All spectra were acquired in a 10mm tube, at natural abundance, in acetone, methanol, acetonitrile, dimethylsulfoxide, toluene, chloroform and dichloromethane as solvents.The concentration of the samples used in these experiments was 0.5, 1.0, 2.0, 3.0, 4.0 and 6.0 mol L -1 , and the signals were referenced to external H 2 O (in a capillary coaxial tube).

Semiempirical MO calculations
The MO calculations were carried out by the Austin Model 1 (AM1) semiempirical method 11 , implemented in the HyperChem 6.03 package (1999) 12 .Geometries were completely optimized without fixing any parameter, thus bringing all geometric variables to their equilibrium values.The energy minimization protocol employs the Polak-Ribiere algorithm, a conjugated gradient method 11,12 .Convergence to a local minimum is achieved when the energy gradient is < 0.01 kcal mol -1 .The calculations were performed on a PC Pentium IV 1.4 GHz computer equipped with a printer.

Results and Discussion
The 17 O chemical shifts of 5-trichloromethylisoxazoles 1a-c in various solvents are listed in Table 1.These values were determined by extrapolation to infinite dilution from spectral data obtained in several concentrations (0.5 to 6 mol L -1 ) relative to external water, at 323 K (see experimental).The Kamlet-Abboud-Taft (KAT) solvatochromic parameters (π* Y , α Y , β Y and δ Y ) used in the present work are also given in Table 1.Considering the 17 O NMR chemical shifts of the oxygen atom of the heterocyclic ring (O1) of compounds 1a-c and according to the KAT formalism, we can re-write Equation 1 as Equation 2(where X = O1).
Preliminary comparison shows that the response values of the oxygen chemical shifts to the solvent-solute dipolarity-polarizability (s O1 ) are of shielding effect for O1.The response to the solvent HBD acidities (a O1 ) is of deshielding for O1 of 1a and shielding for 1b and 1c.
Me H

Compounds
The synthesis of compounds 1a-c was developed in our laboratories 6a .

NMR spectroscopy
The 17 O NMR spectra were recorded on a Bruker DPX 400 at 54.25 MHz.The sample temperature was set at 323 ± 1 K.The instrumental settings were as follows: spectral widths 38 KHz (705 ppm), 8K data points, pulse width 12 µs (90 o ), acquisition time 54 ms, preacquisitions delay 10 ms, 16000-90000 scans, LB of 100 Hz, sample spinning 20 Hz.The spectra were recorded with a RIDE (RIng Down Eliminate) sequence 13 for suppression of acoustic ringing.The general reproducibility of chemical shift data is estimated to be better than ± 1.0 ppm (± 0.2 within the same series).The half-height widths were in the range 150-800 Hz.
The influence of the solvent hydrogen-bond-acceptor (HBA) basicities (b O1 ) is of shielding effect for 1a and deshielding for 1b and 1c.
The contributions (in ppm) to the 17 O chemical shifts of O1 for compounds 1a-c from the terms of Equation 2are listed in Table 3.The contribution of solvent-solute dipolarity-polarizability (s O1 π*) shows a shielding effect for chemical shift of O1 groups (1a-c) in all solvents, in the following order: dimethylsulfoxide > dichloromethane > chloroform > acetonitrile > acetone > methanol > toluene.The contribution of solvent HBD acidities (a O1 α) shows a negligible (< 1.0 ppm) deshielding (1a) or deshielding (1b and 1c) effect on the chemical shift of O1, except for methanol and chloroform in compound 1c.The contribution response to the solvent HBA basicities (b O1 β) show a negligible (< 1.0 ppm) shielding (1a) or deshielding (1b and 1c) effect for chemical shift of O1 oxygen atom, except for methanol and dimthylsulfoxide in compound 1b.
Considering that the terms s O1 , a O1 and b O1 are a measurement of the sensitivity of the studied compound to the solvent dipolarity/polarizability (π* Y ), the solvent hydrogen-bond-donor acidities (α Y ) and the solvent hydrogen-bond-acceptor basicities (β Y ), respectively, the differences between these parameters for compounds 1a-c must reflect the differences in some intramolecular properties of these molecules.In order to better understand the differences of the sensitivity of each compound to the solvent effects the MO calculations were performed.Selected data of the most stable molecular structure of compounds 1a-c were determined by energy minimization calculations using the AM1 semiempirical method 11,12 are listed in Table 4.
Although the solvent effect was obtained only for three compounds, and therefore we can not make a statistical treatment, when these data were compared with molecular data obtained by MO calculations, it was possible to observe some reasonable trends.
The solvent-solute dipolarity-polarizability (s O 1) does not show any trend when compared with the dipole moments of compounds 1a-c (Table 4).The other parameters show a tendency to decrease (a O1 ) and increase (b O1 ) with dipole moments of compounds 1a-c.All absolute values of solvent parameters (s O1 , a O1 , b O1 ) for O1 decrease with the decrease of the net charge in this atom (Table 4).

Conclusion
This work show the validaty to use the Kamlet-Abboud-Taft (KAT) model for complete evaluation of the solvent effects on the 17 O chemical shifts of compounds 1a-c.From a multilinear-regression analysis using the Kamlet-Abboud-Taft (KAT) solvatochromic parameters (π* Y , α Y , β Y and δ Y ) and the observed 17 O chemical shifts of compounds 1a-c, at infinite dilution, it was possible to determine the terms s O1 , a O1 and b O1 .These terms are a measurement of the sensitivity of the studied compounds to the solvent dipolarity/polarizability, the solvent hydrogen-bond-donor acidities and the solvent hydrogen-bond-acceptor basicities, respectively.

Table 1 .
17O NMR chemical shifts of compounds 1a-c at infinite dilution a and solvent parameters b used in Equation2.

Table 3 .
Contributions (in ppm) to the 17 O chemical shift of O1 of compounds 1a-c from terms of Equation 2.