’Li, 23Na and 9Be Nuclear Magnetic Resonance Investigations of the Influence of N-Substitution on the Solvation Interaction of Amides with Alkali and Alkaline Earth Metal Ions BY BERND M. RODE,* THOMAS PONTANI Innrain 52 a, A 6020 Innsbruck, Austria Institut fur Anorganische und Analytische Chemie der Universitat Innsbruck, AND GERNOT HECKMANN Institut fur Anorganische Chemie der Universitat Stuttgart, Pfaffenwaldring 55, D 7000 Stuttgart, Germany Received 25th March, 1977 ’Li, 9Be and 23Na chemical shifts of metal salt solutions in formamide and its mono- and di-N- substituted derivates have been investigated in order to obtain information about the influence of N-methylation and N-ethylation on the interaction of the amide group with alkali and alkaline earth metal ions. Concerning the interaction with Li+ and Na+ ions, methyl and ethyl substitution were found to have an opposite influence on the shifts, whereas in the case of BeZ+ ions all kinds of substitution shift the resonance signal towards higher field.Quantum chemical calculations with minimal Gaussian basis sets were employed in order to obtain some additional information about the background of the substituent effects and the exceptional line broadening observed for the metal ion solvates with diethylformamide. The results are discussed with respect to reported relations between metal resonance shifts and the donor abilities of the solvents. The interaction of amides, peptides and proteins with metal ions has interest for inorganic, theoretical and biological chemists and has been, therefore, the subject of numerous investigati0ns.l Considering the strongly differing solvating properties of various N-substituted amides,2 questions concerning the specificity of peptide binding sites for alkali and alkaline earth metal ions in biosystems and the affinity of such ions to certain parts of protein membrane layers, it seemed desirable to begin some systematic experimental and theoretical studies on the influence of substituents at the nitrogen atom of the peptide group on its ion binding properties.In the first part of these studies, presented here, this influence has been studied investigating the effect of methyl and ethyl groups at nitrogen on the amide’s ligand properties with respect to Li+, Na+ and Be2+ ions. As an experimental tool for these investigations we chose 7Li, 23Na and 9Be n.m.r.measurements, which reflect, by means of chemical shifts and line widths, the ligand induced changes in the electronic environment of the nuclei of the solvated metal ions. These experiments have been supplemented by ab initio calculations with minimal basis sets for the 1 : 1 complexes of the amides with the cations. EXPERIMENTAL METHOD It has been shown by several workers, that the alkali metal n.m.r. technique, particularly 23Na and 7Li n.m.r., is a very sensitive tool for investigation of the immediate chemical 7172 N.M.R. OF SOLVATION INTERACTION environment of alkali metal ions. It has been used extensively for studies on ion interactions with non-aqueous solvents as well as for determination of preferential ~olvation.~-~ Magnitude and direction of chemical shifts have been related to basic chemical properties like the basicity or donor abilities of the solvent^.^^ Using this technique we could expect, therefore, to obtain some reliable experimental information about the influence of N-substitution on the interaction of the amides (and the peptide group, respectively) with these ions. Only very little work has been done using 'Be n.m.r.' The reported data indicate, however, that the method should be as suitable for our investigations as 7Li- and 23Na n.m.r.spectroscopy. Thus we included the Be2+ ion in our investigations in order to obtain some general information about the principal differences between the singly charged alkali and the doubly charged alkaline earth metal ions in their interaction with amides.Theoretical calculations on ion complexes with dimethylformamide, by means of a mixed electrostatic- quantum chemical model,1° have indicated the existance of characteristic, different behaviour of the alkali and alkaline earth ions, respectively, in their influence on the ligand's electronic structure. Having measured the shifts for the metals complexed by the various substituted amides one can determine the influence of the substituent, regarding the values for the unsubstituted formamide as a reference for the other compounds. Since the differences are not expected to be too big, an accurate determination of chemical shifts was necessary. Determination of the magnetic susceptibilities of the solvents was, therefore, inevitable, since there are no data available in literature, except for formamide itself.For the solutions used in the experiments (and, of course, for the aqueous standard solutions) corrections to the susceptibility due to the salt influence were made using the tabulated values for these salts.ll An estimation of the accuracy in the determination of the chemical shifts, based on the line widths of the resonance lines for the nuclei, led to the values kO.2 Hz for Li, and +4 Hz for Na and Be. According to the results for anion and concentration influence on the chemical shift '1 l2 we chose the perchlorates of Lif and Na", and the sulphate of Be2+ for our measurements, which should guarantee the best conditions for the determination of shifts.N.M.R. MEASUREMENTS All spectra were recorded at 25°C with a HFX-90 n.m.r. spectrometer (BRUKER-Physik AG) with Fourier unit. 7Li spectra were recorded at 33.55 MHz, 23Na and 9Be spectra at 22.62 and 9.12 MHz, respectively. The magnetic field strength was adjusted correspondingly to these frequencies at 20.28,20.08 and 15.24 kG, respectively. The samples were measured in rotating 10 mm 4 tubes with 3 mm 4 reference capillaries. In the tables, a positive sign denotes a shift to higher field, a negative sign a shift to lower field strengths. For the Li measurements, no n.m.r. lock was necessary, since the use of the pulse Fourier technique with an acquisition time of 4.27 s allowed the field to drift at a value of +O.l Hz during the recording of one spectrum.The 23Na spectra were recorded with frequency sweep, the field being kept constant by a homolock to the reference signal. For the 'Be pulse Fourier spectra, an accumulation of 100 interferograms was necessary. During the total recording time of 42.5 s, a maximum field drift of 1 Hz could be maintained, so that the use of an n.m.r. lock could be avoided. As reference solutions, we used saturated aqueous solutions of reagent grade LiCl, NaCl and BeCl,. The relative methodical errors within these experimental conditions have been pointed out in the previous chapter. PREPARATION OF THE SAMPLES A N D SUSCEPTIBILITY MEASUREMENTS Anhydrous reagent grade salts, LiC104 (Alfa Ventron), NaC104 (Merck) and BeS04 (Fluka), were dried for 48 h at 150°C before use. The amides (Fluka) were dried over a molecular sieve (4 A).0.5 mol dm-3 solutions of the salts in the amides were prepared under nitrogen atmosphere. The susceptibility of the amides was determined after the Gouy method l 1 with a 3-MB-6B . M. RODE, T . PONTANI AND G. HECKMANN 73 (BRUKER Physik AG) magnetic balance, using n-butylbenzoate as reference. Taking into account the salt contributions to the diamagnetic susceptibility," the corrections to the chemical shifts for the solutions were performed according to the formula of Zimmermann and Forster.13 QUANTUM CHEMICAL CALCULATIONS In order to facilitate the discussion of the results, it seemed desirable to have some theoretical information about ligands and solvated cations. For this purpose, we performed ab initio calculations with minimal I4 GLO basis sets for these systems. The basis set has already been successfully used in several works on ion influence on amides and could, therefore, be expected to give reliable information at a qualitative level.For the amides, we maintained the experimental geometry, the cation positions were optimized with respect to total energy. The calculations were performed in part on the CDC 3300 computer of the University of Innsbruck and in part on the CDC Cyber 74 computer of the technical University of Vienna, using the program MOLPRO by Meyer and Pulay. RESULTS AND DISCUSSION SUSCEPTIBILITY MEASUREMENTS OF THE AMIDES I n table 1 we have listed the volume susceptibilities of the amides, which were investigated during this work. These values show that &substitution may signifi- cantly alter the magnetic behaviour of amides and that correction of the chemical shifts was actually necessary for a reliable interpretation of the n.m.r.spectra. The calculated l1 susceptibilities of the reference solutions are also listed in table I. TABLE 1 .-VOLUME-SUSCEPTIBILITIES OF AMIDES AND STANDARD SOLUTIONS formamide (FA) -0.550~ low6 LiCl a : - 1.087~ N-methylformamide (NMF) - 0 . 5 1 3 x NaCl a : - 1.000~ N, N-dimethylformamide (DMF) BeCI, a : - 0.889 x N, N-diethylformamide (DEF) - 0 . 5 4 4 x - 0.668 x N-ethylformamide (NEF) - 0 . 6 4 2 x a Saturated aqueous solutions N.M.R. MEASUREMENTS In table 2, the non-corrected chemical shifts for Li, Na and Be in the five amides are presented and compared with the susceptibility corrected shifts.TABLE 2.-cHEMICAL SHIFTS, RELATED TO STANDARDS (a) UNCORRECTED (6) BULK SUSCEPTIBILITY CORRECTED (IN HZ) ion a b a b a b a b a b FA NMF DMF NEF DEF Lif 1 1 . 6 -24.9 5.0 - 3 4 . 1 7.0 -30.0 5.1 - 2 4 . 9 5.5 -22.5 Na+ 118 98 1 2 6 1 0 4 126 105 1 1 2 96 104 89 Be2+ -7 - 1 3 -2 -9 8 2 -4 - 8 4 0 Since the influence of the ligand molecule on the electronic environment of the cations can be assumed to increase approximately linearly with the number of ligands in the first coordination sphere, we have to take into account the average coordination numbers n in this sphere for the evaluation of the " specific '' substituent influence. These average coordination numbers iz were obtained from recent vapour pressure measurements and are listed in table 3, together with the chemical shifts, JFA174 N.M.R.OF SOLVATION INTERACTION related now to formarnide as the unsubstituted “standard” amide. The final values, which allow us to estimate the substitution effect on the complexation of the cation by the amide’s carbonyl group, are collected in table 3, namely the coordination number corrected shifts, a,, related to the coordination number of the Li, Na and Be amide solutions, and the line widths Av of the n.m.r. signals. One can first observe the opposite direction of the substituent induced shifts d,, for Li and Na, respectively. Methyl groups shift the Li signal to lower field strengths, ethyl groups to higher field. The Na signals show the opposite behaviour. The different electronic structure of these ions may provide a reason for this opposite effect.TABLE 3.-CHEMICAL SHIFTS &A, RELATED TO FORMAMIDE, AVERAGE COORDINATION NUMBERS, n, COORDINATION NUMBER CORRECTED SHIFTS a,, AND LINE WIDTHS Av (IN Hz) ion FA NMF DMF NEF DEF - 9.2 4.9 5.6 - 8.1 8 .O 8.5 6 4.1 4.8 +5 45 60 4 (6) (6) +4 30 30 - - - - - - - 5.1 2 . 4 - 10.4 7.0 7 2.2 + 13 50 15 4.0 + 23 24 0 4.8 0 9.5 -2 4.0 -2 150 5 (6) +5 30 + 2.4 3 .O + 3.9 19.0 -9 2.0 - 19 130 13 (4) a + 20 60 a Values in parentheses are estimated coordination numbers Whereas lithium with its ls2 electron configuration should be influenced mainly by changes in the diamagnetic shielding, a dominating paramagnetic shielding term is expected in the case of sodium. This is confirmed by the results of 23Na and 7Li shift measurements in various nonaqueous solvents with strongly differing donor abilities,15~ l6 where Li and Na also show opposite behaviour with respect to their n.m.r. absorption frequencies.It is diEcult to find a reasonable explanation, however, for the behaviour of CH,-groups at nitrogen as “ electron acceptors ” for Li+ and C2H,-groups as “ donors ” without further information, as, for example, could be provided by quantum chemical calculations. For Be, we found a shielding effect for all kinds of substitution investigated ; disubstitution increases the shielding effect strongly in the case of methyl as well as ethyl groups. Obviously, a character- istic difference exists between the substitution effects with respect to the interaction with either mono- or di-valent ions.A second interesting fact, observed in the results of our n.m.r. experiment is the strong line broadening of the resonance signals of all ions upon diethylation of the amide’s nitrogen, indicating that this kind of substitution causes a strong non- symmetry in the electronic environment of the nucleus. This effect will be discussed in more detail in the following section. QUANTUM CHEMICAL CALCULATIONS Calculations were carried out for the five amides and most of their 1 : 1 complexes with Li+, Be2+ and Na+. For comparison with the n.m.r. results, special attention was paid to the population analysis, which was performed after the MullikenB . M. RODE, T . PONTANI AND G . HECKMANN 75 procedure, and to the orbital densities at the metal nucleus, which can both represent a helpful tool in the discussion of chemical shifts.Owing to the use of a rather unsophisticated wave function, i.e. the minimal basis sets, one cannot formulate more than a qualitative discussion on the basis of these calculations. For this reason, an analysis of the total electron density function of the systems was not expected to supply more information than the atomic net charges and orbital densities at the nucleus to a discussion of chemical shifts. On the other hand, the larger systems did not allow an extension of the basis set without reaching unreasonable computing times. Semi-empirical procedures are not suitable for the ion-amide complexes for reasons of method,lP l4 and because the description of the metal ion in a valence basis set only cannot lead to reliable information about the electronic environment of the metal nucleus.TABLE 4.-cALCULATED ATOMIC NET CHARGES OF LIGAND ATOMS AND METAL IONS, 4, AND ORBITAL DENSITIES AT THE METAL NUCLEI, Q system FA/Li NMF/Li DMF/Li NEF/Li DEF/Li FA/Be NMF/Be DMF/Be NEF/Be DEF/Be FA/Na NMF/Na DMF/Na NEF/Na FA NMF DMF NEF DEF 4c 0.553 0.594 0.618 0.579 0.597 0.706 0.727 0.748 0.708 0.723 0.529 0.571 0.591 0.551 0.436 0.478 0.505 0.465 0.490 4 0 - 0.528 - 0.535 - 0.541 - 0.526 -0.517 - 0.985 -0.951 - 0.957 - 0.956 - 0.922 - 0.464 - 0.470 - 0.472 - 0.462 - 0.250 -0.251 - 0.257 - 0.247 - 0.246 4-n + - 0.012 - 0.012 - 0.012 -0.012 - 0.012 - 0.003 - 0.003 - 0.003 - 0.003 - 0.003 - 0.027 - 0.028 - 0.028 - 0.028 - - - - - Q 2.765 24 2.765 45 2.765 49 2.765 41 2.765 74 9.487 71 9.476 76 9.476 88 9.476 80 9.475 97 248.717 23 248.717 81 248.712 46 248.712 25 - - - - - In table 4, the net atomic charges of the ligand’s carbonyl group qc, qo and the metal ions’ qMe are presented, together with the calculated orbital density Q at the metal nucleus.In the last part of this paper we will try to analyse the experimental data with the aid of results obtained in the calculations. DISCUSSION OF EXPERIMENTAL AND THEORETICAL RESULTS FOR THE INFLUENCE OF N-SUBSTITUTION O N CATION-AMIDE BINDING Some critical considerations about the significance of the results seem useful at this point. We have given an outline of the accuracy of the n.m.r. experiments at the beginning of the Discussion. Comparing the shifts with methodical accuracy we can expect a correct description of the “ trends” within the investigated series;76 N .M . R . OF SOLVATION INTERACTION the numbers do not seem reliable, however, for an estimation of the influence in a quantitative sense. The necessary restriction of the quantum chemical calculations to strongly simplified models for the solvated ions, in our work on the 1 : I models, leads to some principal limitations in their applicability. In the case of the amides, however, it has been shown by a comparison of experimental ion influence on the amide’s rotational barrier and the calculated results for the 1 : 1 complexes and complexes of higher order,l* lo that the former represent a quite reliable description of the solvated ion. We could expect, therefore, an acceptable description of the substitu- tion influence, again in a qualitative sense only.Further, the average coordination numbers of the metal ions under the experimental conditions of the n.m.r. measure- ments are, according to vapour pressure measurements,2 very low (1.1-2.5), so that a good part of the metal ions actually should be present in the form of 1 : 1 complexes. The most problematic step in the comparison of experiment and calculations seems to be, therefore, the relation of calculated quantities to chemical shifts. Whereas such a relation is still quite easy to construct in the case of proton shifts, the composition of the metal shifts is highly complicated and a reliable calculation has to involve, therefore, a detailed analysis of the wavefunction, which did not seem useful for our rather unsophisticated basis set.What could be done, therefore, was to attempt a more or less “empirical” comparison with calculated quantities, which could be expected to be related to the electronic environment and thus to the shifts of the metal nuclei. Such quantities were the orbital densities at the coordination site. Considering the results of the calculations and the experimental data, we find immediately, that the orbital densities do not account at all for the observed effects. A reason for this may be that this quantity should reflect only the diamagnetic shielding component, which in the case of the metals is not as important as for hydrogen. The electron density at the cation was constant over the whole series of amides and, hence, also did not give any clues to the observed experimental effects.The electron density at the coordination site, i.e. at the oxygen atom, however, seems to be related to the shifts observed in the n.m.r. measurements. For the free iigands, methyl substitution leads to an increase in the electron density of the oxygen atom, whereas ethyl substituents have the opposite effect (cf. table 4). These effects are basically maintained in the complexes of Li-t- and Naf, but not in the Be2+- complexes, where the density is lowered. The parallel nature of this behaviour and the behaviour of the chemical shifts is obvious, taking into account, that the resonance frequencies of Li and Be should undergo a shift opposite to that of Na, upon the same change in the electronic en~ironment.~~ The correlation, that electron density decrease at the coordination site in the complex leads to a shielding effect on Li and Be nuclei and to a deshielding of the Na nucleus and vice versa, seems to be in some “ agreement ” with another empirical correlation between Na shifts and the donor numbers of various solvents,15* I 6 namely, that as the donor ability of the ligand becomes better, so the Na resonance frequencies shift downfield.Such a correlation could not be observed in the Ei, probably because of the dominance of other influences (e.g. ring currents) in the composition of the shifts.12 In the case of our very similar ligands, these influences should be almost constant, thereby enabling the observation of a correlation analogous to sodium.A satisfactory interpretation of the correlations between chemical shifts and donor site electron density on the one hand, and donor numbers on the other hand has yet to be formulated. The amount of electron density calculated for the metalB . M. RODE, T . PONTANI AND G . HECKMANN 77 binding site could influence the metal shift as a pure " neighbouring-atom-effect ", but it might have also some influence on electron transfer processes to unoccupied metal orbitals, which are supposed to contribute to the chemical shifts of these nuclei as well.7 The donor number again is a measure of the ability of the ligand to interact with Lewis acids, and it is reasonable, therefore, to expect that the electron density of the binding site is related to this macroscopic quantity.The investigation of ligands with minor structural differences, as performed here, reflect these correla- tions more clearly than the studies employing very different ligands, but they also confirm some of the tentative conclusions drawn from such investigations.'~ 16* l7 Summarizing our conclusions on the relation between metal n.m.r. shifts, donor qualities of the ligands (in this case influenced by N-substitution) and calculated quantum chemical data, we suggest that : the better the donor quality of the ligaiid, the lower is the gross population of the binding site in the complex. This lowering of electron density leads to shielding effects for Li and Be and to deshielding for Na nuclei. The donor quality of a ligand also depends on the respective acceptor, as can be seen in our examples of the values for Li and Na compared with those for Be.TABLE 5.-cOMPONENTS OF THE DIPOLE MOMENT VECTOR (IN ATOMIC UNITS) AND ANGLES MOLECULES ARE c O/O N : -2.583/0 0 : 1.299/- 1.957 FA NMF DMF NEF DEF BETWEEN DIPOLE VECTOR AND C=O BOND AXIS. THE CARTESIAN COORDINATES OF THE LIGAND x -0.863 -0.869 -0.841 -0.790 -0.573 Y 0.522 0.584 0.505 0.502 0.576 u 154.8 155.8 154.6 156.0 176.6 For the latter, both methyl and ethyl substitution iiiiproves the donor quality of the arnide, whereas for Li and Na ethyl substitution leads to better donating abilities compared with the unsubstituted amide, and methyl substitution reduces the donor quality. These conclusions only form a more or less empirical " description " of the observed relations between macroscopic and molecular quantities, and a satis- factory explanation of these relations can be expected only from much more sophisticated quantum chemical calculations, which cannot be performed at present for these large systems.A further problem was, why the n.m.r. line widths are broadened in the solvate complexes of diethylformamide of all the ions being investigated. Considering the calculated dipole moments and their components, for all of the ligand molecules, one observation could provide a possible explanation for this phenomenon (cf. table 5). The dipole vectors of all amides show approximately the same direction, except DEF, where it lies almost exactly in the C=O bond axis, which points to the metal binding site.This coincidence should allow higher polarization of the core electrons at the metal ion and hence lead to a stronger non-symmetry in its electronic environ- ment. Such non-unsymmetry, however, will necessarily lead to greater line widths. The very low coordination numbers of the metal ions in this amide also give some significance to this result for the solution, although the calculation describes an isolated 1 : 1 complex. A second factor influencing the line widths, is the exchange time for ligands between complex and bulk solvent. The viscosities of the arnides do not indicate, however, any specific difference in this sense. A satisfactory solution to this problem can be expected, therefore, only from detailed studies of the exchange kinetics, which78 N.M.R.OF SOLVATION INTERACTION could also give an explanation for the line broadening in the Na+/NEF complexes, where the calculated data do not give any information. Financial support by the " Fonds zur Forderung der Wissenschaftlichen Fors- chung" is gratefully acknowledged. Thanks are due to the computer centres of the University of Innsbruck and the Technical University of Vienna for computing time. B. M. Rode, in Metal-Ligcind Interactions in Organic and Biochemistry, The Jerusalem Symposia on Quantum Chemistry and Biochemistry (D. Reidel, Dordrecht, Holland, 1976), and references therein. Th. Pontani, Thesis (University of Innsbruck, 1976). D. Noble, in Biological Membranes (Oxford Univ. Press, 1975). M. Herlem and A. I. Popov, J. Amer. Chem. Soc., 1972, 94, 1431. R. H. Erlich, M. S . Greenberg and A. I. Popov, Spectrochim. Acta, 1973, 29A, 543. J. F. Hinton and R. E. Briggs, J. Magnetic Resonance, 1975, 19, 393. ' E. G. Bloor and R. G. Kidd, Canad. J. Chem., 1968,46,3425. * R. H. Erlich and A. I. Popov, J. Amer. Chem. SOC.. 1971,93,5620. F. W. Wherli, J. Magnetic Resonance, 1976, 23, 181 and references therein. P. W. Solwood, Magnetochemistry (Interscience, N.T /London, 2nd edn, 1956). l o B. M. Rode and R. Fussenegger, Mh. Chemie, 1977, 108, 703. l 2 A. I. Popov, Pure Appl. Chem., 1975, 41, 3. l3 I. R. Zimmermann and M. R. Forster, J. Phys. Chem., 1957, 61, 282. l4 B. M. Rode, Mh. Chemie, 1975,106,339. l6 Y . M. Cahen, C. A. Heh, P. R. Handy, E. T. Roach and A. I. Popov, J. Phys. Chern., 1975, l7 A. Saika and C . P. Slichter, J. Chem. Phys., 1954, 22, 26. R. H. Erlich, E. Roach and A. I. Popov, J. Amer. Chem. SOC , 1970,92,4989. 79, 80. (PAPER 7/525)