Advanced
Elimination Reactions of Aryl Furylacetates Promoted by R<sub>2</sub>NH-R<sub>2</sub>NH<sub>2</sub> <sup>+</sup> in 70 mol% MeCN(aq). Effects of β-Aryl on the Ketene-Forming Transition-State
Elimination Reactions of Aryl Furylacetates Promoted by R2NH-R2NH2 + in 70 mol% MeCN(aq). Effects of β-Aryl on the Ketene-Forming Transition-State
Bulletin of the Korean Chemical Society. 2014. Jul, 35(7): 2143-2147
Copyright © 2014, Korea Chemical Society
  • Received : March 26, 2014
  • Accepted : March 31, 2014
  • Published : July 20, 2014
Download
PDF
e-PUB
PubReader
PPT
Export by style
Article
Author
Metrics
Cited by
TagCloud
About the Authors
Sang Yong Pyun
Ju Chang Kim
Bong Rae Cho
Department of Chemistry, Korea University, Seoul 136-713, Korear

Abstract
Ketene-forming elimination from 2-X-4-nitrophenyl furylacetates ( 1a-d ) promoted by R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq) has been studied kinetically. When X = Cl and NO 2 , the reactions exhibited second-order kinetics as well as Brönsted β = 0.37-0.54 and |β lg | = 0.31-0.45. The Brönsted β decreased with a poorer leaving group and |β lg | increased with a weaker base. The results are consistent with an E2 mechanism. When the leaving group was changed to a poorer one [X= H ( 1a ) and OCH 3 ( 1b )], the reaction mechanism changed to the competing E2 and E1cb mechanisms. A further change to the E1cb mechanism was realized for the reaction of 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + in 70 mol % MeCN-30 mol % D 2 O. By comparing the kinetic results in this study with the existing data for ArCH 2 C(O)OC 6 H 3 -2-X-4-NO 2 , the effect of the β-aryl group on the ketene-forming elimination was assessed.
Keywords
Introduction
Extensive studies of the structure-reactivity relationships in the ketene-forming elimination reactions have led to the qualitative understanding of the relationship between the reactant structure and the mechanism. 1-11 Earlier, we reported that the ketene-forming elimination reactions from O -aryl phenylacetates ( 3 ) promoted by R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq) proceeded by an E2 mechanism, 11a while the reactions of O -aryl thienylacetates ( 2a-c ) proceeded by a competing E2 and E1cb mechanism under the same condition. 11g The results have been attributed to the higher acidity of the C β -H bond in the latter, 12 which would stabilize the carbanion intermediate such that the E1cb mechanism can compete with the E2. Alternatively, the smaller aromatic resonance energy of the thiophene than that of benzene may also have contributed to the stability of the carbanion. 13 To assess the relative importance of C β -H bond acidity and the aromatic resonance energy of the β-aryl group on the keteneforming eliminations, we have now studied the reactions of aryl furylacetates ( 1 ) with R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq) (eq 1). Since the p K a values of 2-methylfuran and toluene are identical, 12 whereas the aromatic resonance energy of furan is much lower than that of benzene, 13 this seemed to be an ideal system for such a purpose. We have investigated the reaction mechanism under various conditions. By comparing with the existing data for eliminations of ArCH 2 C(O)OC 6 H 3 -2-X-4-NO 2 [Ar = thiophene ( 2 ), phenyl ( 3 )] with R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq), the effect of β-aryl group was assessed.
PPT Slide
Lager Image
Results
Aryl furylacetates 1a-d were synthesized by the reaction between 2-furylacetic acid, substituted phenols, 2-chloro-1-methylpyridinium iodide, and Et 3 N in CH 2 Cl 2 as reported previously. 11a,14 The spectral and analytical data for the new compounds were consistent with the proposed structures.
For the reactions of 1a-d with R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq), the yields of aryloxides determined by comparing the UV absorption of the infinity samples of the kinetic runs with those of the authentic aryloxides were in the range of 94-98%. The rates of elimination reactions were followed by monitoring the increase in the absorption at the λ max for the aryloxides in the range of 400-434 nm. Excellent pseudofirst order kinetics plots, which covered at least three halflives were obtained. The rate constants are summarized in Tables S1-S4 in Supporting Information. For the reactions of 1c and 1d with R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq), the plots of k obs versus base concentration were straight lines passing through the origin, indicating that the reactions are second-order, first order to the substrate and first order to the base ( Figures 1 , S8 and S9). The slopes are the overall second-order rate constants k 2 E . Values k 2 E for eliminations from 1c and 1d were summarized in Table 1 .
PPT Slide
Lager Image
Plots of kobs versus base concentration for eliminations from 2-chloro-4-nitrophenyl furylacetate (1c,■) and 2,4-dinitrophenyl furylacetate (1d,●) promoted by R2NH-R2NH2+ in 70 mol % MeCN(aq) at 25.0 ℃, [i-Pr2NH]/[i-Pr2NH2+] = 1.0, μ = 0.10 M (Bu4N+Br-).
Rate constants for ketene-forming elimination from ArCH2C(O)OC6H3-2-X-4-NO2(Ar = furyl)apromoted by R2NH-R2NH2+in 70 mol % MeCN(aq) at 25.0 ℃
PPT Slide
Lager Image
a[Substrate] = 5.0 × 10-5 M. b[R2NH] = 8.0 × 10-4- 5.0 × 10-2 M. cReference 15. dAverage of three or more rate constants. eEstimated uncertainty, ± 3%. fcis-2,6-Dimethylpiperidine.
In contrast to the reactions of 1c and 1d , the plots of k obs versus base concentration for the reactions of 1a and 1b were curves at low base concentration and straight lines at higher base concentration ( Figures 2 and S1-S7). The data were analyzed with a nonlinear regression analysis program by assuming that the reaction proceeds by concurrent E2 and E1cb mechanisms (eq. 2). 11a,11d By utilizing the computer program, the k 2 E , k 1 , and k -1 / k 2 values that best fit with Eq. (2) were calculated, and the plots were dissected into the E2 and E1cb components. In all cases, the correlations between the calculated and the experimental data were excellent. The calculated k 2 E values are included in Table 1 and k 1 and k -1 / k 2 values are summarized in Table 2 . The k 1 increased and k -1 / k 2 decreased with a stronger base. For a given base, k 1 decreased and k -1 / k 2 ratio increased as the leaving group was changed to a poorer one.
PPT Slide
Lager Image
PPT Slide
Lager Image
Plots of kobs versus base concentration for eliminations from 4-nitrophenyl furylacetate (1a) promoted by R2NH-R2NH2+ in 70 mol % MeCN(aq) (●) and 70 mol % MeCN-30 mol % D2O (■) at 25.0 ℃, [i-Pr2NH]/[i-Pr2NH2+] = 1.0, μ = 0.10 M (Bu4N+Br-). The closed circles (●) and squares (■) are the experimental data, and the solid line shows the fitted curve by using eq. 2 (see text). The curve for the reaction in 70 mol % MeCN(aq) is dissected into the E2 and E1cb reaction components (dashed lines). Calculated values of k2E, k1, and k-1/k2 for 1a and 1b are summarized in Tables 1 and 2.
Thek1andk-1/k2values for elimination from ArCH2C(O)-OC6H3-2-X-4-NO2(Ar = furyl,1a-d)apromoted by R2NH-R2NH2+in 70 mol % MeCN(aq) at 25.0 ℃
PPT Slide
Lager Image
a[Substrate] = 5.0 × 10-5 M. b[R2NH] = 8.0 × 10-4-5.0 × 10-2 M. cCalculated from kobs by using eq. 2. dThe slopes of the plots of log k1 versus pKa of the base for 1a and 1b are 0.38 ± 0.04 and 0.29 ± 0.12, respectively. eThe slopes of the plots of log k-1/k2 versus pKa of the base for 1a and 1b are -0.17 ± 0.12 and -0.27 ± 0.07, respectively. fSolvent was 70 mol % MeCN-30 mol % D2O. gcis-2,6-Dimethylpiperidine.
The k 2 E values for 1a-d showed excellent correlations with the p K a values of the promoting bases on the Brönsted plots ( Figure 3 ). The β values are in the range of 0.37-0.54 and decrease with a better leaving group ( Table 3 ).
PPT Slide
Lager Image
Brönsted plots for the elimination from ArCH2C(O)-OC6H3-2-X-4-NO2 (Ar = furyl) promoted by R2NH-R2NH2+ in 70 mol % MeCN(aq) at 25.0 ℃ at 25.0 ℃ [X = H (1a,●), OMe (1b, ■), Cl (1c,▲), NO2 (1d,▼)].
Brönsted β values for elimination from ArCH2C(O)-OC6H3-2-X-4-NO2(Ar = furyl) promoted by R2NH-R2NH2+in 70 mol % MeCN(aq) at 25.0 ℃
PPT Slide
Lager Image
aReference 16.
When the reaction of 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + was conducted in 70 mol % MeCN-30 mol % D 2 O, the k obs values were larger than those measured in 70 mol % MeCN(aq) ( Figure 2 and Table S2), and the rate data could be fitted with the E1cb mechanism, that is, the 2nd term in Eq. (2). Calculated values of k 1 and k -1 / k 2 for this reaction are summarized in Table 2 .
The plots of the log k 2 E versus the leaving group p K lg values are depicted in Figure 4 . As reported for 2a and 2b , 11g the data for 1a and 1b showed large negative deviation from the straight lines. Therefore, the β lg values were calculated without the date for 1a and 1b . The β lg values are in the range 0.31-0.45 and decrease as the p K a value of the base increase ( Table 4 ).
PPT Slide
Lager Image
Plots log k2 versus pKlg values of the leaving group for the elimination from ArCH2C(O)OC6H3-2-X-4-NO2 (Ar = furyl, 1a-d) promoted by R2NH-R2NH2+ in 70 mol % MeCN(aq) at 25.0 ℃. [R2NH = Bn(i-Pr)NH (■), i-Bu2NH (▲), i-Pr2NH (●), 2,6-DMP (▼)].
Brönsted βlgvalues for Eelimination from ArCH2C(O)-OC6H3-2-X-4-NO2(Ar = furyl) promoted by R2NH-R2NH2+in 70 mol % MeCN(aq) at 25.0 ℃
PPT Slide
Lager Image
aReference 15. bcis-2,6-Dimethylpiperidine.
To provide additional evidence for the reaction mechanism, the H-D exchange experiment was conducted by mixing 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + in 70 mol % MeCN-30 mol % D 2 O at −10 ℃. The reactant was recovered immediately after mixing. The NMR spectrum indicated that approximately 35% of furfuryl C-H bond of 1a was converted to the C-D bond.
Discussion
Mechanism of Eliminations from 1. Results of kinetic investigations and product studies reveal that the reaction of 1c and 1d with R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq) proceed by an E2 mechanism. The reactions produced elimination product and exhibited second-order kinetics, thereby ruling out all but bimolecular pathway. The (E1cb) R , (E1cb) ip , and internal return mechanisms, which would exhibit either a specific base catalysis or Brönsted β values near unity, 16,17 were negated by the observed general base catalysis with the Brönsted β ranging from 0.37 to 0.54. Moreover, the β values increased with a poorer leaving group. This effect corresponds to a positive interaction coefficient, i.e ., pxy = ∂β/∂p K lg , = ∂β lg /∂p K BH > 0, which describes the interaction between the base catalysis and the leaving group. 17,18 The positive interaction coefficient provides additional support for an E2 mechanism in which the C β -H and C α -OAr bonds are partially cleaved in the transition state.
When the leaving group was changed to a poorer one ( 1a and 1b ), the plots of k obs versus the base concentration were curves ( Figures 2 and S1-S7). The result could be explained with a competing E2 and E1cb mechanism, as explained below. First, the rate date fitted well with Eq. (2) and the shapes of dissected lines were typical for an E2 and E1cb mechanisms. Second, the calculated values of k 2 E , k 1 , and k -1 / k 2 are consistent with the competing mechanism; (i) The calculated k 2 E values for 1a and 1b and the experimentally measured k 2 E values for 1c and 1d show excellent correlation on the Brönsted plots ( Figure 3 ), as expected for the E2 mechanism. (ii) The increase in the k 1 value with a stronger base is in good agreement with the deprotonation step of the E1cb mechanism. (iii) The decrease in the k -1 / k 2 ratio with a stronger base can be attributed to the decreased acidity of the R 2 NH 2 + which would reduce k -1 .
It is interesting to note that the k 2 E values for 1a and 1b showed large negative deviation in Figure 4 . This outcome could be explained, if the extent of the C α -OAr bond cleavage is smaller than those for 1c and 1d . Since the β values for 1a and 1b are larger than those for 1c and 1d , the transition state of the E2 pathway would be skewed toward the E1cb-like with greater extend of proton transfer and smaller degree of cleavage of the C α -OAr bond. This would destabilize the transition state as a result of the smaller extent of double bond formation due to the inefficient charge transfer from the β- to α-carbon and retard the rate. This result underlines the unusual sensitivity of the ketene-forming transition state to the reactant structure in the borderline between the E2 and E1cb.
The mechanistic diversity of the ketene-forming elimination was further demonstrated in the reaction of 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + in 70 mol % MeCN-30 mol % D 2 O. There is convincing evidence in support of the mechanism change from the competing E2 and E1cb to E1cb under this condition. First, the k obs values measured in the deuterated solvent could be fitted to the 2 nd term of Eq. (2), indicating an exclusive E1cb mechanism ( Figure 2 ). Second, the larger k 1 value in the deuterated solvent is also consistent because the basicity of R 2 NH would increase in the poorer solvating D 2 O. 19 Third, since the k 2 value should be almost the same in the two solvents, the 3-fold smaller k -1 / k 2 value in the deuterated solvent predicts that k -1 H2O / k -1 D2O ≈ 3.0 ( Table 2 ). Since all of the R 2 NH 2 + should be converted to R 2 ND 2 + in the deuterated solvent, the large primary isotope effect is not unexpected. Fourth, the smaller k -1 D2O would increase the 2 nd term of Eq. (2) more in the deuterated than in undeuterated solvent to make the k obs values larger. If it becomes the predominant term, a change of the mechanism would occur. Fifth, the partial H-D exchange of the furfuryl C-H bond in the recovered starting material from the reaction of 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + in 70 mol % MeCN-30 mol % D 2 O provides additional support for the E1cb mechanism.
Effect of β-Aryl Group on the Ketene-Forming Transition State. Table 5 compares the rates and transition state parameters for the eliminations from ArCH 2 C(O)OC 6 H 3 -2-X-4-NO 2 . The relative rate of E2 reaction changed from 1.0 to 10 to 1.2 as the β-aryl group was varied from furyl ( 1a ) to thienyl ( 2 ) to phenyl ( 3 ). Since the p K a values of 2-methyl furan, 2-methylthiophene, and toluene in DMSO are 43, 42, and 43, respectively, 12 the faster rate observed for 2 can be attributed to the higher acidity of the C β -H bond. However, b and |β lg | values of 1a , 2 , and 3 are almost the same within experimental errors, indicating that structure of the E2 transition states are not significantly influenced by the β-aryl group. In addition, the k 1 value is larger for 2 than that for 1a is consistent with the higher acidity of the C β -H bond. The larger k -1 / k 2 value for 2 is as expected for the E1cb mechanism. Although the lower basicity of the conjugate base of 2 would reduce both k -1 and k 2 , the effect should be more pronounced for the slower step ( k 2 ). This would predict a larger k -1 / k 2 . The most interesting result from this study is that 1a and 2 , but not 3 , showed mechanistic changes from E2 to a competing E2 and E1cb. While the change of the mechanism for 2 can be attributed to the higher acidity of the C β -H bond of 2 (see above), the difference between 1a and 3 cannot be explained similarly because the acidities of their C β -H bonds are the same. Alternatively, the negative charge density developed at the β-carbon can better be stabilized by the furyl than phenyl group because of the smaller aromatic resonance energy. 13 This would stabilize the carbanion intermediate for 1a such that the E1cb mechanism can compete with E2. A similar explanation can be applied to the competing E2 and E1cb mechanisms for 2 , although the higher acidity of the C β -H bonds would undoubtedly contribute to the carbanion stability.
Transition state Parameters for ketene-forming eliminations from ArCH2C(O)OC6H4-4-NO2Promoted by R2NH-R2NH2+buffers in 70 mol % MeCN(aq) at 25.0 ℃
PPT Slide
Lager Image
aReference 11g. bReference 11a. cR2NH = Bn(i-PrNH).
In conclusion, we have studied the elimination reactions of aryl furylacetates ( 1a-d ) promoted by R 2 NH-R 2 NH 2 + in 70 mol % MeCN(aq). The reactions of 1c and 1d proceeded by a concerted E2 mechanism. The mechanism changed to the competing E2 and E1cb mechanisms as the leaving group was changed to a poorer one [X = H (1a) and OCH 3 ( 1b )]. For elimination from 1a with i -Pr 2 NH/ i -Pr 2 NH 2 + , a further change to the E1cb mechanism was realized in 70 mol % MeCN-30 mol % D 2 O. When the β-aryl groups was changed from phenyl to thienyl to furyl, the structure of the E2 transition state remained almost the same although k 2 E value changed according to the acidity of the C β -H bond. Noteworthy is the mechanistic diversity observed in the keteneforming eliminations from 1 by the changes in the leaving group and the base-solvent system.
Experimental Section
Materials. Aryl furylacetates 1a-d were prepared from 2-furanacetic acid and substituted phenols in the presence of Et 3 N and 2-chloro-1-methylpyridinium iodide in CH 2 Cl 2 as previously reported. 11a,14 The yield (%), IR (KBr, C=O, cm -1 ), 1 H NMR (400 MHz, CDCl 3 , J values are in Hz), and 13 C NMR (100 MHz), and mass spectral data for the new compounds are as follows.
2-(4-Nitrophenyl)furylacetate (1a): Yield 75%; IR 1760 cm -1 ; 1 H NMR δ 3.98 (s, 2H), 6.37 (m, 2H), 7.29 (d, J = 9.24, 2H), 7.41 (s, 1H), 8.25 (d, J = 9.24, 2H); 13 C NMR δ 34.1, 108.8, 110.8, 122.4, 125.2, 142.6, 145.5, 146.2 ,155.2, 167.0; LRMS (EI); m/z 247 [M + ] (64), 109 (12), 108 (79), 81 (100) ,82 (34), 63 (6), 53 (46), 50 (5).
2-(2-Methoxy-4-nitrophenyl)furylacetate (1b): Yield 78 %; IR 1774 cm -1 ; 1 H NMR δ 3.90 (s, 3H), 3.99 (s, 2H), 6.38 (m, 2H), 7.20 (d, J = 8.56, 1H), 7.41 (d, J = 1.72, 1H), 7.83 (d, J = 2.4, 1H), 7.88 (dd, J = 2.4, 8.56, 1H); 13 C NMR δ 33.7, 56.5, 107.7, 108.7, 110.7, 116.4, 123.1, 142.5, 144.9, 146.4, 146.5, 151.6, 166.5; LRMS (EI); m/z 277 [M + ] (14), 108 (99), 81 (100), 53 (27).
2-(2-Chloro-4-nitrophenyl)furylacetate (1c): Yield 58 %; IR 1780 cm -1 ; 1 H NMR δ 4.03 (s, 2H), 6.37 (m, 2H), 7.35 (d, J = 8.88, 1H), 7.42 (d, 1.72, 1H), 8.15 (dd, J = 2.72, 8.88, 1H), 8.33 (d, J = 2.72, 1H); 13 C NMR δ 33.7, 109.0, 110.8, 123.2, 124.3, 125.9, 128.2, 142.7, 145.9, 151.9, 166.1; LRMS (EI); m/z 281 [M + ] (42), 109 (14), 108 (29), 82 (64), 81 (100), 63 (11), 53 (53).
2-(2,4-Dinitrophenyl)furylacetate (1d): Yield 49%; IR 1745 cm -1 ; 1 H NMR δ 4.08 (s, 2H), 6.39 (m, 1H), 7.42 (m, 1H), 7.50 (d, J = 9.06, 1H), 8.51 (dd, J = 2.72, 9.06, 1H), 8.96 (d, J = 2.72, 1H); 13 C NMR δ 33.6, 109.3, 110.8, 121.3, 121.8, 126.7, 129.1, 131.6, 142.8, 145.4, 148.4, 166.3; LRMS (EI); m/z 292 [M + ] (23), 125 (9), 108 (27), 82 (40), 81 (100), 53 (38).
Reagent grade acetonitrile and secondary amine were fractionally distilled from CaH 2 . The base-solvent solutions were prepared by dissolving equivalent amount of R 2 NH and R 2 NH 2 + in 70 mol % MeCN(aq). In all cases, the ionic strength was maintained to 0.1 M with Bu 4 N + Br - .
Kinetic Studies. Reactions of 1a-d with R 2 NH/R 2 NH 2 + in 70 mol % MeCN(aq) were followed by monitoring the increase in the absorbance of aryloxides at 400-434 nm with a UV-vis spectrophotometer as described. 11a-11e
Product Studies. The products of eliminations from 1a-d promoted by R 2 NH/R 2 NH 2 + in 70 mol % MeCN(aq) were identified as reported. 11a The yields of aryloxides determined by comparing the UV absorptions of the infinity samples with those for the authentic aryloxides were in the range of 94-98%.
H-D Exchange Experiment. To determine whether the β-furfuryl protons of 1a undergo H-D exchange, 1a (0.047 g, 0.19 mmol) was reacted with i -Pr 2 NH/ i -Pr 2 NH 2 + (0.05 M, 5 mL) in 70 mol % MeCN-30 mol % D 2 O at −10 ℃. The reaction was quenched by adding dilute HCl(aq) immediately after mixing. The recovered reactant was isolated by extracting the products with ethyl acetate followed by the silca gel column chromatography using ethyl acetate/hexane = 1/4 as eluent. The proton NMR spectrum of the recovered reactant from 1a was identical to that of the substrate except that the integration at δ 3.98 decreased by approximately 35%.
Control Experiments. The stabilities of 1a-d were determined as reported. 11b-11d The solutions of 1 in MeCN were stable for at least two weeks when stored in the refrigerator.
Acknowledgements
Support Information Available.Observed rate constants for elimination from1a-dpromoted by R2NH/R2NH2+in 70 mol % MeCN(aq), plots ofkobsversusbase concentration, and spectroscopic and analytical data for the new compounds are available on request from the correspondence author (20 pages).e-mail:sypyun@pknu.ac.kr.
References
Pratt R. F. , Bruice T. C. 1970 J. Am. Chem. Soc. 92 5956 - 5964    DOI : 10.1021/ja00723a024
Tagaki W. , Kobayashi S. , Kurihara K. , Kurashima K. , Yoshida Y. , Yano J. 1976 J. Chem. Soc. Chem. Commun. 843 - 845
Broxton T. J. , Duddy N. W. 1981 J. Org. Chem. 46 1186 - 1191    DOI : 10.1021/jo00319a028
Chandrasekar R. , Venkatasubramanian N. 1982 J. Chem. Soc., Perkin Trans. 2 1625 - 1631
Douglas K. T. , Alborz M. , Rullo G. R. , Yaggi N. F. 1982 J. Chem. Soc. Chem. Commun. 242 - 246
Isaac N. S. , Najem T. S. 1988 J. Chem. Soc., Perkin Trans. 2 557 - 562
Chung S. Y. , Yoh S. D. , Choi J. H. , Shim K. T. 1992 J. Korean Chem. Soc. 36 446 - 452
Bordwell F. G. 1988 Acc. Chem. Res. 21 456 -    DOI : 10.1021/ar00156a004
Bernardi F. 1988 J. Mol. Struct. 163 173 - 177    DOI : 10.1016/0166-1280(88)80389-0
Saigo K. , Usui M. , Kikuchi K. , Shimada E. , Mukaiyama T. 1977 Bull. Chem. Soc. Jpn. 50 1863 - 1866    DOI : 10.1246/bcsj.50.1863
Cho B. R. , Lee S. J. , Kim Y. K. 1995 J. Org. Chem. 60 2072 - 2076    DOI : 10.1021/jo00112a030
Coetzee J. F. 1965 Prog. Phys. Org. Chem. 4 45 - 92
Gandler J. R. , Patai S. 1989 The Chemistry of Double Bonded Functional Groups, part 1 John Wiley and Sons Chichester 734 - 797
Lowry T. H. , Richardson K. S. 1987 Mechanism and Theory in Organic Chemistry Harper and Row New York (a) pp 214 - 218. (b) pp 591-616. (c) pp 640-644
Gold V. 1969 Adv. Phys. Org. Chem. 7 259 - 267