Computer-Aided Prediction of 125 Te and 13 C NMR Chemical Shifts of Diorgano Tellurides

Esse trabalho descreve um método para previsão dos deslocamentos químicos de RMN de 125Te e 13C de diorgano-teluretos baseado em uma abordagem tridimensional. Para tanto, 150 substâncias foram selecionadas da literatura e tiveram suas respectivas geometrias otimizadas através do método semi-empírico PM3. A partir desses dados, as estruturas foram codificadas através do programa FOCOS, que descreve o ambiente químico de cada átomo presente na estrutura, num total de 1411 focos para as substâncias. O método descrito foi testado com dez diorgano teluretos não inseridos na base de dados e mostrou uma exatidão maior na previsão de deslocamentos químicos de RMN de 125Te e 13C do que os dados previstos por um programa comercial. Os dados e os parâmetros estatísticos obtidos nesse estudo demonstram que o método aplicado foi capaz de prever com sucesso os deslocamentos químicos de RMN de 125Te e 13C dos diorgano teluretos.


Introduction
0][11][12] In view of this interest in the organic compounds of this element, analytical methods for their rapid identification are welcome.Nuclear magnetic resonance (NMR) spectroscopy in nowadays is the routine technique of choice for this end.There are eight naturally occurring isotopes of tellurium.However, only 123 Te and 125 Te are active in NMR spectroscopy.The latter isotope is significantly more naturally abundant (7%) than the former one (1%).The 125 Te isotope is 12.5 times more receptive than 13 C. 13 In addition, 125 Te NMR presents a spectral width of about 7000 ppm. 14Consequently, minor structural changes in a molecule cause sensitive chemical shift responses in its 125 Te NMR spectrum of compounds.
Considering the increasing importance of the organic tellurium compounds and the magnetic properties of 125 Te commented above, in this work we developed a method to predict the 125 Te and 13 C NMR chemical shifts of diorganotellurides based in a three-dimensional approach.

Theoretical approach
The structure determination of a new organic compound is currently made by coupling information derived from several spectroscopic methods with those obtained from 2D NMR spectroscopy.For this end, tables of chemical shifts, 15,16 are many times insufficient for the validation of complex and hypothetical structures.In order to help in the spectra analysis, a number of methods for 13 C and 1 H NMR chemical shift prediction have been developed.8][19][20][21][22][23][24][25][26][27] However, to the best of our knowledge, the methods up to now developed are unable to predict 125 Te chemical shifts.On the other hand, the 13 C NMR prediction for organotellurium compounds, by these methods are scarce and show large error ranges.

The three-dimensional database
A total of 150 diorgano tellurides were selected for this work.The 13 C and 125 Te NMR chemical shift data assigned in the literature were used.The structures of the selected compounds were optimized at the PM3 semiempirical level 28 employing the HyperChem program. 29After the optimization process of the organo tellurides, their structures were coded by the FOCOS program, 30 which uses data derived from the optimized geometry and the reported 13 C and 125 Te NMR chemical shifts.

Atom codification
The following codes were computed for each carbon and tellurium atom using the geometrical and connectivity information obtained: At i -atom type code, that encodes the nature of the atom i, its hybridization state and hydrogenation pattern.Thirty three atom types (Table 1) were considered in the codification of the FOCOS program.
Sc i -stereochemical code of tertiary and quaternary carbon atoms.For the chiral centers the codes 01 and 02 were attributed if the atom presents the R or the S configuration, respectively.For the achiral centers it was assigned the code 00.
The chemical environments were coded for the structure (all atoms) and designated as foco.The foco contains the atom type description (At i ) according to Table 1 for each element present in the structure, its respective stereochemical code (Sc i ) and all the neighboring atoms coded according to Table 1 and correlated with their distance from the centralized atom.The neighboring atoms are organized according to the increasing distance value from the focused atom.The maximum distance for the codification process can be selected by the user; in this study, the focos were constructed with the maximum distance of 5 Å.For each codified foco, the chemical shift value was stored together with the chemical environment.
All atoms i having the same foco were grouped into a cluster.Two focos atoms were considered as being equal if the distance tolerance is not larger than 0.25 Å.At the end of the clustering process, the mean chemical shift of the atoms belonging to the cluster was stored.The database elaborated contains 1411 focos for the 150 diorgano tellurides.

Results and Discussion
In order to predict the chemical shifts of the tellurium and carbon atoms, the three-dimensional model of the hypothetical structure was first constructed following the same method used for the database elaboration.Then the parameters of each atom of the structure were generated using the same variables as described above.2][33][34] The structures of these compounds are shown in Figure 1.
The 125 Te and 13 C NMR chemical shifts predicted by the FOCOS program for these ten tellurides are shown in Table 2.The chemical formula were also submitted to the commercial ACD program, 35 which predicted their 13 C NMR chemical shifts for the compounds (Table 2).For the compound 7 the 125 Te NMR data was not displayed in Table 2 because this data was not available in the original reference.
Analyzing the results generated by both programs one can observe that for 13 C NMR chemical shift prediction, the FOCOS program shows an average error of 0.67 ppm and a maximum error of 6.40 ppm, whereas the ACD program exhibits an average error of 2.11 ppm and a maximum error of 27.77 ppm.Linear regression of predicted against observed 13 C NMR chemical shifts for the FOCOS and ACD programs are displayed in Figures 2 and 3, respectively.The coefficients obtained from the regressions are displayed in the equations below.The main errors on the 13 C NMR chemical shift prediction were observed in both programs for the carbon atoms directly bonded to the tellurium atom (α-carbons), once these atoms show the typical high field resonance.
Equation for 13 C NMR data predicted from the FOCOS program: Experimental = +0.99680(±0.00464)Focos +0.03896(±0.403839)(n=098; r=1.000; s=1.266;F=182507.818;)Equation for 13 C NMR data predicted from the ACD program: Experimental = +1.01349(±0.01845)ACD -2.06478(±1.608659)(n=098; r=0.996; s=4.936;F=11907.938;) The data prediction carried out by the FOCOS program exhibits a higher accuracy than those predicted by the ACD program, because in the former the database was addressed for diorgano tellurides.This specific database shows atoms with similar chemical environments of the test compounds, resulting in smaller errors in the 13 C NMR chemical shift prediction.However, it was not possible to compare the 125 Te NMR data predicted by the FOCOS program with other commercial software because they do not present 125 Te NMR predictions.
Using the results obtained by the FOCOS program for the tellurium atom, a linear regression of the data predicted against the experimental chemical shifts was plotted (Figure 4).The coefficients obtained from the regression for the 125     The parameters obtained from the regression methods are statistically significant and demonstrate that despite of the experimental data show the large variation due to the higher sensibility to external factors, the method here described was able to predict successfully the 13 C and 125 Te NMR chemical shifts.

Conclusions
The method described in this paper allows the prediction of 125 Te and 13 C NMR chemical shifts of diorgano tellurides, employing a database that contains 150 substances.The three-dimensional code described is detailed enough to consider all atoms that can influence the magnetic behavior of the resonant atom.This approach lead to accurate results in the restricted structural domain of the diorgano tellurides, but could be extended to other heavy resonant atoms.

Figure 1 .
Figure 1.Tellurides used to test the programs.

Figure 2 .Figure 3 .Figure 4 .
Figure 2. 13 C NMR chemical shifts (ppm) experimental vs. predicted by the FOCOS program.-40 -20 0 20 40 60 80 100 120 140 160 Focos chemical shifts are dependent upon solvent, temperature and concentration of the sample.In this way, the prediction of125  Te NMR chemical shifts is a hard task due to the number of variables involved.However, the data shown indicate that the values calculated by the methods described in this work agree satisfactorily with the experimental data.

Table 1 .
Atomic groups and respective codes used in the molecular codification by the FOCOS program T: triple bond representation; *aromatic atom representation.
*This data was not available in the original reference.