Nitrogen-15 nuclear magnetic resonance of organophosphorus

George A. Gray, and Thomas A. Albright. J. Am. Chem. Soc. , 1976, 98 (13), ... J. H. Hargis , W. B. Jennings , S. D. Worley , M. S. Tolley. Journal of...
0 downloads 0 Views 624KB Size
3857 (24)A. N. Laurent'ev, I. G. Mastennikov, V. A. Efanov, and E. G. Sochilin. Zh. Obshch. Khim., 44, 2589 (1974). (25)M. Baudler and M. Bock, 2.Anorg. Allg. Chem., 395, 37 (1973). (26)The correlation between the chemical shifl and average endocyclic bond angle exists irrespective of the question of the rigidity of the ring in solution. Rin confwrnational changes would not be manifest in ambient temperature 31Py1H]NMR spectra: four- and sixmembered rings yield singlets because of symmetry regardless of conformational changes, and five-membered rings will yield complex 3'P[1H] NMR spectra in spite of possible conformational changes not involving inversion of the phosphorus lone pair electrons.'' Clearly the average hybridization of the phosphorus atoms, as dictated by the ring size and measured by the endocyclic P-P-P bond angle, is the dominant factor in the 3'P chemical shift. (27)The chemical shifl of -2.2 ppm for (CF3P)6is a weighted average of the values reported for the five phosphorus atoms in ref 36. The weighted average from ref 22 is -4.9 ppm.

(28)E. Fluck and K. Issleib. 2. Nafurforsch. 8, 736 (1966). (29)R. P. Wells, Prog. Phys. Org. Chem., 6, 1 1 (1968). (30)J. Donahue, Acta Clystalloy., 15, 708 (1962). (31)D. F. Shriver. "The Manipulation of Air-Sensitive Compounds", McGrawHill, New York. N.Y., 1969. (32) J. L. Mills and L. C. Flukinger, J. Chem. Educ., 50, 636 (1973). (33)M. Bauder and K. Hammerstrom, 2.Naturforsch. B, 20, 810 (1965). (34)M. M. Rauhut and A. M. Semsel. J. Org. Chem., 28, 473 (1963). (35)H. J. Emeleus and J. D. Smith, J. Chem. SOC.,375 (1959). (36)H.G.Ang, M. E. Redwood, and B. 0. West, Aust. J. Chem., 25, 493 (1972). (37)A. B. Burg and P. J. Slota, J. Am. Chem. SOC., 80, 1107 (1958). (38) A. B. Burg, J. Am. Chem. Soc., 80, 1107 (1958). (39) C. S. Palmer and A. E.Scott, J. Am. Chem. SOC.,50, 536 (1928). (40)E. J. Wells, R. C. Ferguson, J. G. Hallett, and L. K. Peterson, Can.J. Chem., 48,2733 (1968),and references therein. (41)R . J. Horvat and A. Furst, J. Am. Chem. SOC.,74, 562 (1952).

5N Nuclear Magnetic Resonance of Organophosphorus Compounds. Experimental Determination and SCF-MO Finite Perturbation Calculation of 5N-3lP Nuclear Spin Coupling Constants in (Me2N)3P, (Me2N)3PS, and (Me2N)3PO George A. Gray*

la and

Thomas A. Albrightlb

Contribution f r o m the N M R Applications Laboratory. Varian Instrument Division, Palo Alto, California 94305, and the Department of Chemistry, University of Delaware, Newark, Delaware I971 1. Received October 14, 1975

Abstract: Natural abundance I5N N M R at 10.1 MHz has been used to determine the 1SN-31P nuclear spin coupling constants in (Me2N)3P, (Me*Nj3PS, and (Me2Nj3PO as +59.1, 76.0 and -26.9 Hz, respectively. Two-bond 13C-3'Pcouplings were also determined from the 25.16 MHz I3C spectra as +19.15, +3.3 and +3.4 Hz. The one-bond 3'P-170coupling was obtained from the natural abundance " 0 spectrum and has a value of 145 Hz. SCF-MO finite perturbation calculations in the CNDO/ 2 approximation were carried out on a series of model compounds giving good agreement with experiment for ' S N - 3 ' Pcouplings and the 3'P-'70 coupling using only the Fermi contact mechanism. There is a linear dependence of the calculated ISN3 ' P coupling with the bond order between the coupled atoms

Organophosphorus compounds have long been subjects of nuclear magnetic resonance investigations. In particular, structure and bonding have been primary areas of concern. In recent years I3C NMR has added yet another NMR technique toward elucidation of the above, contributing two new sensitive parameters, the I3C chemical shift and the 13C-31Pnuclear spin coupling constant. We have carried out extended programs using I3C N M R in this area and now report on studies employing t h e I5N nuclear resonance in organophosphorus compounds containing nitrogen. From the chemical point of view, phosphorus-nitrogen compounds a r e rich in synthetic and structural information.2 Spectroscopically, the group 5 and 6 elements have nuclear spin couplings which exhibit dramatic sign changes and wide ranges of magnitudes. Since it has long been the hope and expectation that molecular properties such as spin couplings could provide sensitive insights into chemical bonding and molecular electronic structure, it is imperative that their behavior, mechanisms, and mechanistic contributions be known and usable. T h e Fermi contact mechanism for spin coupling has repeatedly been shown to be dominant for homonuclear and heteronuclear coupling involving ' H and I3C,leading to, at times, unfortunate over-extension through literal "s-character" calculations based on models derived for couplings in assumed "model" systems. This is more a reflection of sought-for simplicity than actual rigor since now quite powerful theoretical methods have become available which have been successful in predicting cou-

plings, particularly one-bond couplings involving I3C. Small magnitudes and uncertain signs of I3C-l5N couplings have retarded progress in their use, although their accurate prediction should reflect well on any theoretical wave functions and spin coupling theory. Still within group 5 , phosphorus produces, in many cases, much larger couplings to nitrogen, a fact which should make theoretical methods much easier to employ in a first effort a t reproducing magnitudes and signs. T h e use of pulsed Fourier transform N M R has now made I3C N M R widespread. T h e same techniques have also been applied to natural abundance l5N NMR. In order to attain a n overview of what to expect in this class of compounds we have chosen to determine experimentally the 15N-31Pcouplings and their signs, where possible, in several bonding situations and phosphorus oxidation states. As a guide to the interpretation of the observed couplings and as a further test of theoretidal models we then attempt to reproduce them theoretically. In the early stages of this work there existed only one re.~ a few other ported value of a 31P-ISNc o ~ p l i n gSubsequently, values in highly f l ~ o r i n a t e dd, ~i a l k y l a m i n ~and , ~ organometallic6 phosphorus compounds have been documented. O u r studies have restricted substituent changes to those occurring a t the phosphorus within the highly symmetric tris(dimethylamino) substituent framework. T h e large range of substituent effects generated by the several phosphorus oxidation states a r e then examined through changes in the observed 15N-3'P couplings. Subsequently, finite perturbation SCF Gray, Albright

1

" N N M R of Organophosphorus Compounds

3858

I JPN 'Jpc I JPO

" From

+59.1 i 0.1 -I-19.15 f 0.05

76.0 f 0.2 + 3 . 3 f 0.05

-26.9 f 0.1 $3.4 f 0.1 145"

10 000 transient ''0 spectrum (G. A. Gray, to be pub-

lished).

Figure 1. Natural abundance I5N NMR spectra of [ ( C H ~ ) Z N ] ~under PO conditions of coherent low-power off-resonance proton decoupling. Upper spectrum: 15 098 transients, spectral width 512 Hz, acquisition time 4 s, 20° pulse, nodelay. Lower spectrum: 17 670 transients, 256 Hz spectral width, 20" pulse, no delay. Spectra obtained on XL-100 WG12/S-I24XL at 10.1 and 25.16 MHz (for *Jpc)locked to 10% internal C6D6 in IO-" tubes at 20 OC; no exponential weighting applied. Decoupler power 80 dB. Spectra are inverted because of significant nuclear Overhauser effect (negative because of negative Y N ) .

molecular orbital calculations a r e performed to examine the effect of conformation, substituents, and the importance of the Fermi contact coupling contribution to the directly bonded I P- s N couplings.

Experimental Section NMR Measurements. I5N N M R spectra were obtained at 10.1 MHz using a Varian XL-100 WG/S-124 XL 16K fourier transform spectrometer. Neat solutions (10% C6D6 for internal deuterium lock) were run in 10-mm tubes at -30 OC. Typical conditions are illustrated in Figure 1 where the double resonance spectra are shown for tris(dimethy1amino)phosphine oxide. This method' permitted a sign determination for the 15N-31Pcoupling in the oxide via selective irradiation of the phosphorus spin states within the 31P-'H multiplet in the proton spectrum.Since off-resonance high-frequency irradiation enhances the high-frequency component of the 15N-31Pdoublet in the ISNspectrum of the three spin ('H, ISP,31P)system the 3'P-'H and 31P-'5Nreduced couplings (KPN)are of the same sign. Since Y N is negative ' J ~ Nis of opposite sign to 3 J p H . 8 Similar consideration of 2 J ~and c 3 J p H allowed sign determination of 2Jpcfrom off-resonance studies in the I3C s p e ~ t r u mThese . ~ were performed at 25.1 MHz on the same samples as used in the ISN studies. The phosphine, sulfide, and oxide were used as received from Aldrich Chemical Co.

Calculations The geometries of the aminophosphines were taken as follows: P-H = 1.414 A, H-P-H = 93.4', and H-P-N = 97.5' a r e from methylphosphine;I0the P-N bond distance (1.70 A) and conformations of the amino groups in triaminophosphine were taken from tris(dimethy1amino)phosphine.I' For the tetravalent phosphorus compounds a standard geometry of P-H = 1.42 8, and H-P-H = 105O was chosen. C N D O / 2 (spd) optimization gave P-N bond distances of 1.69, 1.72, 1.64, and 1.66 8,for 5,6,7, and 8. A P-N bond length for 10-16 was set a t 1.66 8, from the structure of triaminophosphine oxide.'* T h e P - 0 and P-S bond lengths used were 1.51 and 1.91 A, respectively. The P-S bond length was taken from the average of several related compound^.'^ It should be noted that C N D 0 / 2 (spd) energy minimizations gave excellent geomeJ o u r n a l of the American Chemical Society

/

98:13

tries for triaminophosphine, it's oxide and s ~ 1 f i d e .Trigonal l~ bipyramidal geometry a t phosphorus was used for the phosphoranes. The P-Ne, and P-N,, bond lengths of 1.64 and 1.78 A, respectively, were taken from the x-ray structures of several related cyclic p h o ~ p h o r a n e s . T ' ~h e P-Haxial = 1.508 8, and P-He, = 1.402 8,were taken from a b initio geometry optimiz a t i o n ~The . ~ ~P-Fa, and P-F,, bond distances were assumed to be the same as those found for trifluorodimethylphosphorane (1.55 and 1.64 A, respectively).16 T h e dihedral conformations of the amino groups for the phosphoranes were taken from ref 17 and 18. A standard N-H bond distance of 1.01 8, was assumed. All calculations were done on a Burroughs B-6700 computer.

Results and Discussion Table I contains the experimental lsN-31P couplings, as well as several others of interest. Although a sign change was obse r ~ e in d going ~ ~ from tricoordinate P(II1) in (CF3)2PNH2 to the tetracoordinate P(V) in F3P=NPF2 or pentacoordinate P(V) in F3P(NH2)2, these highly fluorinated phosphorus may not be typical for most organophosphorus compounds not containing highly electronegative groups, and therefore their coupling behavior does not necessarily define a general oxidation-state dependence. The tris(dimethylamin0) substituents provide a n unchanging array at the phosphorus which experiences change only a t the unique fourth coordination site. (This is also true for the anilino dimethyl compounds recently r e p ~ r t e d : ~T) h e sign change observed here closely follows the change in reduced coupling ( I 5 N has a negative magnetogyric ratio, I3C positive) observed in the series (CH3)3P, (CH3)3PS, and (CH3)3PO which exhibits one-bond 13C-3'Pcouplings of -13.6,7a (+)56.1,17 and (+)68.3 Hz." T h e change from ~ negative in P(II1) P(V) has been explainpositive l J p to ed4a in terms of greater Fermi contact contribution resulting from substantial increase of 3s bonding character in the phosphorus hybrid orbitals, sp3 in the case of the tetrahedral P(V) in F3P==NPF2. Corroborative evidence for this viewpoint in 1 3 C - 3 1 Pcoupling has recently been obtained'* for the triwhere I J c p e q gonal bipyramidal (CH3eq)3P(CH3ax)(OCH3ax) = 116.0 H z and l J C p a x = 7.3 H z , which is consistent with phosphorus sp2 hybrid bonding orbitals within the equatorial plane and essentially p bonding to the axial ligands. T h e large two-bond 13C-31Pcoupling in the P(II1) compound is similar to the 12-1 5 H z couplings previously observed in N,N-dimethylaminophosphines.2'These are average values for the appropriate rotameric distributions. As has been shown for 2Jpcc22-24 and 2 J p N c 2 i there is a dihedral angle dependence of 2 J p ~in phosphines with respect to the orientation of the coupled carbon and the phosphorus lone pair, large and positive for a cis relationship, small and negative for trans (-7 for trans 2 J p ~+40 ~ , H z for cis).2' T h e parallels observed here, along with the consistencies observed in different oxidation states, substituent arrays, and symmetries argue that the 13C-31P coupling behavior may have significant experimental correspondence in 15N-31P couplings. This prospect should encourage the acquisition of new data to explore further their behavior in different bonding and conformational situations. On the theoretical side it is critical to

/ J u n e 23, 1976

-

+

3859 Table 11.

Calculated Nuclear Spin Couplingso Nitrogen geometry Trigonal Compound

(H2N)3P=S H4P-NH2 (equatorial NH2)

H4P-NH2 (axialNH2)

A-B

PH PN PNH PN PN H PH PN PNH PH PK PNH PF PN PNH PH PN PO PNH PN PO PNH PH PN PS PNH PN PS PNH PH PHaX PNeq PN H PH PHaX PNeq PNH P Feq PFaX PNeq PNH

lb 3d

5

7 9 10

12

Pyramidal

JAB

PSASB

-128.0 C75.6 -0.2 +95.9 -4.3 +508.8 -17.4 0.5 +397.8 +106.5 -0.3 -1566.3 +494.8 +17.3 $604.6 -6.2 $157.7 -3.9 -39.0 $330.9

0.2691 0.1315

2‘

0.1271

4‘

0.4876 0.2263

6

0.4698 0.201 3

8

0.5 154 0.2419 0.1948

11

0.265 1 0.2050

13

0.4883 0.2267 0.1706

15

0.2473 0.1781

17

0.4068 0.4843 0.2240

19

0.3762 0.4859 0.1660

21

0.1644 0.20 15 0.2963

23

$0.1

-50.8

-4.1 -28.9 -83.7 +0.6 f348.3 +482.2 -1.4 -2.9 +80.6 +588.6

16 18

20

+50.1

-9.0 -229.7 C20.5 -134.1 +12.5

22

PsAsB

16.2 f57.7 $7.1 $57.1 $3.3 +475.9 $23.3 -4.0 $359.4 $141.1 -4.8

0.2720 0.1 136

$592.6 -0.3 $147.4 -5.3 - 12.0 +26 1.4 -4.1 $539.5 -2.4 -48.3 -3.9 -9.3 -70.2 -2.1 +308. I +490.6 12.7 -5.6 100.2 $567.3 +36.5

0.51 17 0.2162 0.1945

-1

0.1230 0.4805 0.1 840

0.463 1 0.1735

0.1850 0.2048

+543.0 $7.0

14

JAB

+ +

0.2285 0.205 1 0.4886 0.21 12 0.1702 0.2158 0.1783 0.4008 0.4830 0. I944 0.3799 0.48 I5 0.1423

-8.0

-351.5 -222.8 $28.7

0.1 555 0.1949 0.1917

-10.8

Values for averages over several conformations, except where noted. In Hz. PO couplings refer to 3 1 P - ’ 7 0coupling. Without d orbitals: JPH= -197.7, Jpx = +67.9, and JPNH = $3.1. Without d orbitals: JPH = -177.5, J p x = +60.2, and J ~ N = H $13.0. Without d orbitals: J p s = +96.3 and J P ~=H -3.0. e Without d orbitals: J P N = 50.3 and J p x ~= $7.4.

assess the importance of the Fermi contact mechanism for 15N-31Pcoupling, especially since interpretation of such couplings within the contact formalism has already comm e n ~ e d .T~h,e~large magnitudes for the phosphine and oxide 1sN-31Pcouplings give ample encouragement for theoretical prediction since such predictions will not be severely hampered by very small observed couplings as a r e found, for example, for I3C-l5N couplings. With these factors in mind we here report the first systematic theoretical analysis of ‘sN-31P coupling using modern SCF-MO finite perturbation methods. Calculation of 1sN-31PCouplings. There have been several studies of the Fermi contact contribution in P-H, P-C, and P-P c o ~ p l i n g s using ~ ~ - ~molecular ~ orbital calculations within the Pople-Santry formalism.28 This formalism requires that the calculated couplings be directly proportional to the “s character” comprising the directly bonded atoms. In addition, it requires a judicious assignment of a n “average triplet exitation energy”, or a more exacting sum-over-states method. A much more promising and productive approach which avoids this problem is the finite perturbation technique29 (FPT) used

in conjunction with SCF molecular orbital wave functions in the INDO (intermediate neglect of differential overlap) or CNDO (complete neglect of differential overlap) approximations. The coupling is computed in the presence of a contact perturbation:

The coupling constant is then computed as

where p s A 2 is the diagonal spin density matrix element for the valence s orbital of atom A. This method has given reliable P-C and P-H couplings, in general, for a wide variety of organophosphorus compound^.^^-^^ Here we use the CNDO/2 approximation to examine,the effect of conformation, substituent, and the extent of contribution of the Fermi contact mechanism. The calculated nuclear spin couplings33a r e given in Table 11. A description of the geometry chosen for these compounds

Gray, Albright

/

15N N M R of Organophosphorus Compounds

3860 is given in the Experimental Section. Table I1 also lists the phosphorus 3s-nitrogen 2s bond orders. Calculations for tetravalent or pentavalent phosphorus atoms did not converge properly without d orbitals; therefore, these calculated values have not been included in Table 11. The immediate difficulty in calculating lsN-31Pcouplings for the compounds in Table I1 is that the coupling is sensitive to the planarity at nitrogen. (The P-N couplings are insensitive to rotation about the P-N bond.) It has been established that the inversion barrier a t nitrogen is markedly reduced by the substitution of a n adjacent second row atom.35Our C N D O / 2 calculations indicate that there is little difference (