Polynuclear Phosphoramidite Copper Complexes with Mixed Trigonal

Sep 30, 2014 - Emeline Rideau , Hengzhi You , Mireia Sidera , Timothy D. W. Claridge , and Stephen P. Fletcher. Journal of the American Chemical Socie...
1 downloads 0 Views 1MB Size
Note pubs.acs.org/Organometallics

Polynuclear Phosphoramidite Copper Complexes with Mixed Trigonal/Tetrahedral Coordination in THF Felicitas von Rekowski and Ruth M. Gschwind* Institute of Organic Chemistry, University Regensburg, D-93053 Regensburg, Germany S Supporting Information *

ABSTRACT: In the field of asymmetric copper-catalyzed 1,4addition reactions new applications were established using triorganoaluminum reagents in etheral solvents. This raised the question of whether phosphoramidite copper complexes have similar structures and aggregation levels in THF compared to noncoordinating solvents such as CD2Cl2. Therefore, NMR spectroscopic investigations of the precatalytic structure of phosphoramidite copper complexes in THF-d8 were performed. Combined low-temperature studies, spectra simulations, and a detailed integral analysis indicated [(CuIL)(CuIL2)4] as precatalyst in THF. This confirms the mixed trigonal/tetrahedral coordination of the copper atoms in phosphoramidite copper complexes as a more general motif. In addition, highly aggregated complexes might be more probable than expected even in THF.

A

Scheme 1. Proposed Precatalytic Structures of (a) Diphosphine Copper Complexes11 and (b) Phosphoramidite Copper Complexes13

symmetric copper-catalyzed 1,4-addition reactions are very often applied for the formation of enantioselective C−C bonds.1−4 In addition to the diorganozinc reagents typically used in these reactions recently also triorganoaluminum compounds have gained attention as transmetalation reagents.5,6 Based on the higher Lewis acidity of aluminum, an improved reactivity compared to zinc reagents is observed. As a result, sterically hindered substrates can be used, allowing for the formation of quaternary centers.6−9 Deviating from the typical experimental conditions of diorganozinc reagents, syntheses with triorganoaluminum reagents are performed in coordinating solvents, such as THF. In noncoordinating solvents, e.g., CH2Cl2, the biphenol backbone can be substituted by alkyl groups due to the high oxophilicity of aluminum.10 Despite the broad application of copper-catalytic systems, only a few investigations concerning the structures involved in these reactions have been published. The formation of a binuclear copper complex as a precatalytic species was observed by Feringa and co-workers while investigating the structure of copper complexes with chiral ferrocenyl-based diphosphine ligands in ethereal (Et2O) or halogenated (CH2Cl2) solvents.11 In more polar solvents (e.g., CH 3 CN or MeOH) a mononuclear copper complex was found (see Scheme 1a).11 In our working group previous NMR investigations revealed a binuclear phosphoramidite copper precatalyst with a mixed trigonal/tetrahedral coordination of the copper atoms in CD2Cl2 ([(CuXL)(CuXL2)]; see Scheme 1b).12−14 Recently, we identified also a transmetalation intermediate with an intact core structure of this mixed trigonal/tetrahedral-coordinated precatalyst in CD2Cl2.15 This retention of the precatalytic structure upon transmetalation emphasized the importance of © XXXX American Chemical Society

structural studies of precatalysts in the case of phosphoramidite copper complexes. The high sensitivity of copper-catalytic reactions toward the experimental conditions applied, e.g., the copper salt, solvent, ligand, and temperature, is well known.11−14,16 This raised the question of whether copper complexes with mixed trigonal/ tetrahedral coordination are a general structural feature for phosphoramidite copper complexes or a special motif in CD2Cl2. Second the general aggregation level of phosphoramidite copper complexes was an issue. Is the formation of binuclear species limited to CD2Cl2 and diethyl ether as previous studies suggested,5−7,9 or are higher aggregated complexes also possible in slightly better solvating solvents such as THF? Received: July 22, 2014

A

dx.doi.org/10.1021/om500751q | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Note

Figure 2a) to identify the main species present at a 2:1 ligand to salt ratio, which is typically used in synthesis. The spectra in

In a previous NMR investigation we already tested THF-d8 as solvent with CuCl as salt. However, the 31P NMR spectra showed very broad and overlapping signals of the copper complexes not appropriate for structure investigations.12,13 Due to the recent synthetic developments and applications of phosphoramidite copper complexes combined with triorganoaluminum reagents in ethereal solutions, we reinvestigated an NMR spectroscopic access to the precatalytic phosphoramidite copper complexes in THF-d8. Therefore, first, further Cu(I) salts were tested to find an appropriate model system with sharp and separated signals. Then, low-temperature studies, spectra simulations, and a detailed integral analysis were applied and revealed a polynuclear phosphoramidite copper complex with a mixed trigonal/tetrahedral coordination of the copper atoms also in THF-d8.



RESULTS AND DISCUSSION To find a suitable model system for the structure elucidation of precatalytic phosphoramidite copper complexes in THF-d8, ligand L was selected (see Scheme 1), because in previous investigations only complexes with L showed well-resolved 31P NMR spectra at low temperatures.12−14 To avoid severe NMR problems connected with paramagnetic Cu(II) salts, exclusively Cu(I) salts were tested. CuTC (TC = thiophenecarboxylate) and tetrakis(acetonitrile)copper(I) tetrafluoroborate ([Cu(CH3CN)4]BF4) were selected, because they are typically used in synthetic applications performed in coordinating solvents such THF or diethyl ether.6−9 In addition, CuI was chosen, which showed sharp and well-separated signals in previous investigations in CD2Cl2.13,17 All complexes were prepared according to the protocols applied in synthesis.6−9 In Figure 1 the 31P NMR spectra of 2:1 ligand to salt mixtures of L and CuI, CuTC, and [Cu(CH3CN)4]BF4 are

Figure 2. (a) 31P NMR spectra of L and mixtures of L with CuI (0.03 M) at varying ligand to salt ratios (1.3−3 equiv of L) at 230 K in THFd8; (b) 31P NMR spectra simulations of the signals at 130.5 and 128.4 ppm at 180 K; (c) proposed structures of the complex and corresponding signal splitting at low temperature.

Figure 2a showed that the three signals at 128.8, 121.8, and 118.6 ppm (also apparent in Figure 1a) are present at each ligand to salt ratio with varying integrals and that no additional complexes can be detected. The trends in the complex formation can be used for the assignment (see below), and the complex at 128.8 ppm is identified as the main species at a 2:1 ligand to salt ratio and can therefore be suggested as the precatalytic complex. The signal at 121.8 ppm at 230 K was assigned to [CuIL]3, because its absolute chemical shift, its signal pattern, and its titration trends (see Figure 2a) resemble the situation in CD2Cl2.12,13 The investigations in CD2Cl2 revealed also a correlation between the chemical shift and the average ligand to copper ratio of the complexes: the higher the amount of ligand in the complex, the higher the ppm value of the 31P signal.12,13 Therefore, the signal at 118.6 ppm hints either at a 1:1 complex [CuIL]n with a deviating aggregation number n or at a complex with a ligand to copper ratio lower than 1:1. In accordance with that chemical shift trend, for the complex at 128.8 ppm at 230 K a higher ligand to copper ratio is expected, which was previously assigned using low-temperature spectra.14 At 180 K line broadening appears for all three salts due to remaining ligand exchange contributions and high viscosity (Figure 1b).14 But again for CuI the spectra with the best signal resolution are observed. Especially, the main species at 128.8 ppm show a signal splitting indicating a mixed coordination of the copper atoms and allowing for the assignment of the complex constitution. Such a signal splitting was already observed in lowtemperature investigations in CD2Cl2 with an integral ratio close to 1:2, indicating a mixed trigonal/tetrahedral coordination of the copper atoms in the binuclear precatalytic complex [(CuIL)(CuIL2)] (for structure see Scheme 1b).12,14 Now, to determine the ratio of the CuIL and the CuIL2 units in the respective complex in THF, 31P NMR spectra with different integral ratios were simulated for the signals at 130.5 and 128.4 ppm and compared to the experimental spectrum at 180 K (see Figure 2b). Thereby an integral ratio of 1:8 (blue) was found to provide the best match with the experiment (black), indicating

Figure 1. 31P NMR spectra of 2:1 mixtures of L and three different CuX salts at (a) 230 K and (b) 180 K in THF-d8 (L 0.06 M, CuX 0.03 M).

presented at 230 K (a) and at 180 K (b). CuI provides the smallest line widths and well-separated signals for the free ligand as well as the different complex species. CuTC and [Cu(CH3CN)4]BF4 provided significantly broader line widths and/or reduced signal separation, hampering the structure elucidation. However, the spectra using all three copper salts showed a similar signal pattern at both temperatures, hinting at similar structures. Therefore, CuI was chosen for further investigations, providing the best spectroscopic properties. First the ligand to salt ratio was varied from 1.3:1 up to 3:1 (see B

dx.doi.org/10.1021/om500751q | Organometallics XXXX, XXX, XXX−XXX

Organometallics

Note

one CuIL and four CuIL2 units as a complex composition in THF-d8 (see Figure 2, b for simulations and c for structures). The simulation for 1:6 (one CuIL plus three CuIL2 units) shows a significant deviation. The simulation for 1:10 (one CuIL plus five CuIL2 units) also deviates from the experimental spectrum but is much closer than the simulation for 1:6. These data indicated the formation of a highly aggregated complex [(CuIL)(CuIL2)4]. In addition, the simulations suggested even a higher probability for [(CuIL)(CuIL2)5] than for [(CuIL)(CuIL2)3]. Considering the monomerization of ferrocenyl-based diphosphine copper complexes observed in polar solvents,11 higher aggregated copper complexes in THF compared to CD2Cl2 are quite surprising. Therefore, next, the temperature-dependent interconversion of the complexes (230−180 K, see Figure 3a) was addressed to prove or disprove this high aggregation in THF.18

Table 1. Integral Analysis of Copper Complexes (L 0.06 M, CuX 0.03 M)a [(CuIL)(CuIL2)n]b temp

n=1

n=2

n=3

n=4

n=5

230 K 220 K 210 K 190 K 180 K average

114.8 112.7 112.6 115.2 115.8 114.2

108.6 105.7 104.9 105.7 105.5 106.1

106.0 102.7 101.6 101.7 101.0 102.6

104.5 101.1 99.8 99.4 98.6 100.7

103.6 100.0 98.6 98.0 97.0 99.4

a

The right complex composition is indicated by 100%. calculations see the Supporting Information.

b

For

the composition of [(CuIL)(CuIL2)4]: The higher the temperature, the larger the ratio of CuIL compared to CuIL2 units. Similar temperature interconversion analyses with ligand to copper salt ratios of 1.3:1 and 2.5:1 corroborated these results (see Supporting Information). In summary, the here-presented study concerning the structure of phosphoramidite copper complexes in THF-d8 represents a further example of the sensitivity of these species toward the experimental conditions used. Spectra simulations of low-temperature 31P NMR spectra and detailed integral analysis of the temperature-dependent formation of copper complexes and their ligand distribution revealed a polynuclear complex [(CuIL)(CuIL2)4] in THF. This is a further example of the preferred formation of precatalytic phosphoramidite copper complexes with mixed trigonal/tetrahedral coordination of the copper atoms. Compared to the binuclear precatalyst [(CuIL)(CuIL2)] in CD2Cl2, surprisingly a higher aggregated complex is detected in THF. This reveals impressingly that the solvent interactions of these phosphoramidite copper complexes to THF are even less pronounced than to CH2Cl2 and that highly aggregated complexes might be more probable than expected.

Figure 3. (a) Temperature-dependent 31P NMR spectra of a 2:1 mixture of L and CuI-d8 (L 0.06 M, CuX 0.03 M); (b) corresponding ligand distribution normalized to 200.

Previous investigations in CD2Cl2 showed that the existence of the different complex species depends on the temperature. [CuIL]3 is observed only at temperatures above 200 K, whereas apart from [(CuIL)(CuIL2)] a second low-temperature species [CuIL2]2 appears below 180 K.14 In contrast to CD2Cl2 the formation of the thermodynamic stable complex [CuIL2]2 is not observed in THF-d8 (Figure 3a). Instead the signal of [(CuIL)(CuIL2)4] increases with lower temperatures accompanied by decreasing amounts of free ligand and [CuIL]3 and [CuIL]n (Figure 3b). Now, this temperature-dependent formation of the complexes enables a detailed analysis of the integrals on a single sample representing a second approach to reveal independently the composition of the copper complexes. Due to the ligand to copper ratio of 2:1, the sum of ligand integrals is equal to 200 (see Figure 3b) and the sum of copper is equal to 100 (see Table 1), provided that all copper atoms are incorporated in complexes19 and that the compositions of the complexes are assigned correctly. The smallest deviation from 100% is calculated for the complex [(CuIL)(CuIL2)4] (see Table 1), which corroborates the best fit of the 31P NMR spectra simulation (see Figure 2b). [(CuIL)(CuIL2)5] and [(CuIL)(CuIL2 ) 3 ] give average values of 99.4% and 102.6%, respectively, indicating again a larger deviation for smaller than for higher aggregated complexes. The overall small changes for all three complexes indicated that spectra simulations seem to be more reliable to assign the absolute stoichiometry of such complexes.20 Furthermore, the data in Table 1 show larger deviations from 100% with increasing temperature. This revealed a temperature-dependent trend in



EXPERIMENTAL SECTION



ASSOCIATED CONTENT

An argon-flushed Schlenk tube equipped with magnetic stirring bar and septum was charged with 2 equiv of ligand (0.06 M, 0.036 mmol, 15.81 mg) and 1 equiv of copper salt (0.03 M, 0.018 mmol, CuI: 3.43 mg, CuTC: 3.43 mg, [Cu(CH3CN)4]BF4: 5.66 mg), freshly distilled solvent THF-d8 (0.6 mL) was added, and the mixture was stirred for 1−2 h at room temperature until a clear solution was obtained. Subsequently the samples were transferred to an argon-flushed NMR tube. The samples were stored at −85 °C. For the ligand to salt dependent complex formation with CuI the same sample preparations were used, thereby the amount of CuI was held constantly (0.03 M), while the amount of ligand was varied from 0.023 mmol (0.04 M, 1.3 equiv) to 0.054 mmol (0.09 M, 3 equiv). NMR spectra were collected on a 600 MHz NMR spectrometer equipped with a 5 mm broadband triple resonance z-gradient probe (maximum strength 53.5 G/cm). 1H chemical shifts were referenced to the residual solvent signal of THF, and for the 31P chemical shifts the Ξ value was applied. 31P-spectra: relaxation delay = 2.5 s, acquisition time = 0.03−0.14 s, SW = 60 ppm, TD = 4k, NS = 512− 1k. The simulations of the spectra were performed using an iterative optimization of both line width and intensity. The final intensity ratios were obtained with line widths of 235 Hz for both signals. S Supporting Information *

Additional NMR data. This material is available free of charge via the Internet at http://pubs.acs.org. C

dx.doi.org/10.1021/om500751q | Organometallics XXXX, XXX, XXX−XXX

Organometallics



Note

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We would like to thank Carina Koch for the simulation of the low-temperature 31P NMR spectra. We gratefully acknowledge financial support from the DFG grant GS 13/1-2.



REFERENCES

(1) Jerphagnon, T.; Pizzuti, G. M.; Minnaard, A. J.; Feringa, B. L. Chem. Soc. Rev. 2009, 38, 1039. (2) Alexakis, A.; Backvall, J. E.; Krause, N.; Pamies, O.; Dieguez, M. Chem. Rev. 2008, 108, 2796. (3) Christoffers, J.; Koripelly, G.; Rosiak, A.; Roessle, M. Synthesis 2007, 1279. (4) Harutyunyan, S. R.; den Hartog, T.; Geurts, K.; Minnaard, A. J.; Feringa, B. L. Chem. Rev. 2008, 108, 2824. (5) Alexakis, A.; Krause, N.; Woodward, S. Copper-Catalyzed Asymmetric Synthesis; Wiley-VCH: Weinheim, 2014. (6) Alexakis, A.; Albrow, V.; Biswas, K.; d’Augustin, M.; Prieto, O.; Woodward, S. Chem. Commun. 2005, 2843. (7) d’Augustin, M.; Palais, L.; Alexakis, A. Angew. Chem., Int. Ed. 2005, 44, 1376. (8) Vuagnoux-d’Augustin, M.; Alexakis, A. Chem.Eur. J. 2007, 13, 9647. (9) Vuagnoux-d’Augustin, M.; Kehrli, S.; Alexakis, A. Synlett 2007, 2057. (10) Bournaud, C.; Falciola, C.; Lecourt, T.; Rosset, S.; Alexakis, A.; Micouin, L. Org. Lett. 2006, 8, 3581. (11) Harutyunyan, S. R.; Lopez, F.; Browne, W. R.; Correa, A.; Pena, D.; Badorrey, R.; Meetsma, A.; Minnaard, A.; Feringa, B. L. J. Am. Chem. Soc. 2006, 128, 9103. (12) Zhang, H.; Gschwind, R. M. Angew. Chem., Int. Ed. 2006, 45, 6391. (13) Zhang, H.; Gschwind, R. M. Chem.Eur. J. 2007, 13, 6691. (14) Schober, K.; Zhang, H.; Gschwind, R. M. J. Am. Chem. Soc. 2008, 130, 12310. (15) von Rekowski, F.; Koch, C.; Gschwind, R. M. J. Am. Chem. Soc. 2014, 136, 11389. (16) Arnold, L. A.; Imbos, R.; Mandoli, A.; De Vries, A. H. M.; Naasz, R.; Feringa, B. L. Tetrahedron 2000, 56, 2865. (17) CuI is not typically used in synthetic applications, but shows product formation in test reactions. (18) In principle a similar integral analysis can be performed using different ligand to salt ratios. However, in this case different samples have to be compared, providing higher experimental errors. (19) Previous elementary analysis in CH 2Cl2 revealed that precipitation and solubility are not affecting such kind of calculations.12 (20) Due to signal overlap in the 1H NMR spectra, DOSY measurements were not reliable; for 31P DOSY measurements the relaxation times are too short.

D

dx.doi.org/10.1021/om500751q | Organometallics XXXX, XXX, XXX−XXX