G. FRAENKEL, D. G. ADAMS,
944
A =2
C
=
p
- r z - p + 3r,p - 3pr, + 2r, + 1
The parameter T , (which is known for the near-ultraviolet band of TMPD3I) changes with exciting wavelength. Consequently y is wavelength dependent. For the photoionization studies in this report the
AND
R. R. DEAN
approximate values of the polarization parameters were p = ” 5 and w = 0.95, so that, in this case, y can vary at most from 0.95 to 1.10 as T , varies from 0 to 1. Thus in this case the correction is small, and in the results reported it is assumed that y = 1. (31) A.
c. Albrecht, J . Am. Chem. SOC., 82,3813 (1960).
Nuclear Magnetic Resonance and Ultraviolet Spectroscopy of Phenylmagnesium Bromide, Phenyllithium, and Pyridine‘ by Gideon Fraenkel, David G . Adams, and Ronald R. Dean Evans Chemical Laboratory, The Ohio State University, Columbus, Ohio 43210
(Received Azkgust %.$?
1967)
Kuclear magnetic resonance spectra have been obtained for phenyllithium, phenylmagnesium bromide, and pyridine. Analysis of the spectra by a combination of deuterium substitution and computer techniques reveals that the nmr parameters for these compounds are quite similar. The coupling constants for the organometallic reagents are all of the same sign, and hydrogens ortho to the carbon-metal bond are considerably deshielded with respect to benzene. These results are explained on the basis that phenyl anion and pyridine are isoelectronic and, to the degree that the carbon-metal bonds are ionic, the same effects should account for the shifts in all three compounds. By analogy to the published discussion on shifts in pyridine, it is suggested that paramagnetic terms are mainly responsible for shifts in the organometallic reagents and n-electron density changes are not important. In fact, the shifts around C&Li and C6HsMgBrfollow the order expected from the point anisotropy approximation. Noting that mixing of the ground with the n + r* excited state in pyridine is mainly responsible for paramagnetic shifts, it is suggested that similar electronic transitions might be available to C6HjLiand C,&i\/IgBr. In fact, long tails indicative of such weak absorption is tentatively assigned uC-11 -+ r*. The similarities in the nmr and uv spectra of the above three compounds together with the way the nmr parameters change among the reagents leads to the conclusion that the magnitude of the downfield shift for the ortho hydrogens in ArM is a qualitative measure of the ionic character of the carbon-metal bond.
The study of carbanions is complicated by the fact that most of them come as the negative or partially negative part of an organometallic compound. Among the questions which concern the physical properties and behavior of these compounds are: (1) carbanioninversion phenomena,2’S(2) the ionic character of the
carbon-metal bonds,* and (3) the mean lifetime of the carbon-metal bonds between exchange^.^,^ The organometallic compounds most likely to contain carbanions are those of sodium or other alkali elements with higher atomic weights. Unfortunately, most of these are insoluble in or react with all solvents which have so far been tested and are, therefore, not susceptible to spectroscopic investigation. Organolithium and -magnesium compounds dissolve T h e Journal of Physical Chemistry
in a wide variety of solvents and behave chemically like carbanionic species;’ it would be expected that (1) Presented in part at the 147th National Meeting of the American Chemical Society, Philadelphia, Pa., April 1964, Abstracts, p N41; D . G. Adams, Ph.D. Thesis, The Ohio State University, Columbus, Ohio, 1964. (2) G. M. Whitesides, F. Kaplan, and J. D. Roberts, J . Am. Chem. Soc., 85, 2167 (1963); G. M. Whitesides, M. Witanowsky, and J. D. Roberts, ibid., 87, 2854 (1965); G. M .Whitesides and J. D. Roberts, ibid., 87, 4878 (1965); M. Witanowski and J. D . Roberts, ibid., 88, 737 (1966). (3) G. Fraenkel, D. T. Dix, and D. G. Adams, Tetrahedron Letters, 3155 (1964); G. Fraenkel and D. T. Dix, J . Am. Chem. SOC.,88, 979 (1966). (4) G. Fraenkel, D . G. Adams, and J. Williams, Tetrahedron Letters, 767 (1963). (5) D. F. Evans and M . S. Khan, Chem. Commun., 67 (1966). (6) H. 0. House, R . A. Latham, and G . M. Whitesides, J . Org. Chem., 32, 2481 (1967). (7) M. S. Kharasch and 0. Reinmuth, “Grignard Reactions of Nonmetallic Substances,” Prentice-Hall Inc., New York, N. Y., 1954; G. E. Coates, “Organometallic Compounds,” 2nd ed, John Wiley and Sons, Inc., New York, N. Y . , 1960.
NMRSPECTRAOF PHENYLMAGNESIUM BROMIDE these materials might reflect useful information about carbanions in their physical properties. This paper concerns aspects of the electronic structure of the phenyl anion which have been uncovered with nrnr spectroscopy.’ It will be shown that the results for phenyllithium and phenylmagnesium bromide bear close resemblence to those for pyridine, which is isoelectronic with the phenyl anion. While the present work was being written, certain aspects of it were published by other groups, working independently. These are Ladd’s analysis of the nmr spectra of phenyllithium, diphenylmagnesium, and diphenylzinc,8 some coupling constants for phenyllithiumJ9 and the analysis of the nmr spectrum of pyridine due to Merry and Goldstein, and Castellano, et a1.l0 These findings complement our results and strengthen the conclusions. In other physicochemical studies, much work has been done on the infrared spectroscopy of organometallic compounds. Lanpher’l found a band in the ir spectra of phenyllithium, phenylsodium, and phenylpotassium whose energy was linear with the square of the electronegativity of the corresponding metal. I n the infrared. spectroscopy of organolithium compounds Rodionov, Kocheshkov, and their colleagues found evidence for bridged bonds12and resolved dissociated species from larger aggregateP as well as one etherate from another.13 West and Glaze have reported on the Li6-Li7 isotope effects in the ir spectra of organolithium compounds.l4 Magnetic susceptibility measurements indicate phenyllithium to be ionic.I5 Various authors have reported on the aggregration of organomagnesium16 and -lithium compounds.17 Under different conditions, phenyllithium and phenylmagnesium bromide consist of mixtures of monomers and dimers. In the solid state, diphenylmagnesium dietherate and phenylmagnesium bromide dietherate are monomeric.’* Recently, Evans and Khan reported the preparation of pentafluorophenylmagnesium bromide. At low temperatures, they were able to resolve the fluorine resonance of the Grignard reagent from that of the diarylmagnesium compound. Raising the temperature increased the exchange rate among all species and averaged their shifts.6
Results and Discussion The syntheses of deuterated bromobenzenes are illustrated in Scheme I. p-Dibromobenzene was coiiverted to the mono-Grignard reagent. The latter mas hydrolyzed with DzO to give 4-deuteriobromobenzene. By two such reaction sequences, 1,3,5-tribromobenzene was transformed to 3,5-dideuteriobromobenzene. Grignard reagents were prepared from the two deuterated bromides and from bromobenzene; see the Experimental Section. These bromides were also converted to the corresponding organolithium compounds by reaction
945 Scheme I
Br
MgBr
Br
Br
I
I
,
Br
D
BrpBr M ether p,
with lithium metal19 and by halogen-lithium exchange reactions. 2o The nrnr spectra of p-deuteriophenyllithium and p-deuteriophenylmagnesium bromide are illustrated in Figures 1 and 2 with expanded reproductions of the downfield multiplets. The single absorption at T 2.74 is due to deuteriobenzene. These spectra are AA’BB’ systems,21the upfield portions of which are broadened by coupling of the deuterium to the hydrogens ortho to it. Hence, the downfield multiplets are assigned to hydrogens ortho to the carbon-metal bond. These latter multiplets were analyzed with Pople’s method21 and the fit of the nmr parameters with the observed spectra was improved with the aid of computer programs (8) J. A. Ladd, Spectrochim. Acta, 22, 1157 (1968). (9) 5. Castallano and C.Sun, J . Am. Chem. Soc., 88, 4742 (1966). (10) J. B. Merry and J. H. Goldstein, ibid., 88, 5563 (19613); C. Castellano, C. Sun, and K. Kostelnik, J . Chem. Phys., 46, 327 (1967). (11) E.J. Lampher, J . Org. Chem., 21, 830 (1956). (12) A.N. Rodionov, D. Shigorin, T. V. Talalaeva, and K. Kocheshkov, Izv. Akad. Nauk SSSR Otd. Khim. Nauk, 120 (1958). (13) A. N. Rodionov, D. N. Shigorin, E. N. Gur’yanova, and K . A. Kocheshkov, Dokl. Akad. Nauk S S S R , 125, 562 (1959); 123, 113 (1958); 128, 728 (1959); A. N. Rodionov, D. N. Shigorin, and K. A. Kocheshkov, ibid., 136, 369 (1961); T . V. Talalaeva, A. N. Rodionov, and K. A. Kocheshkov, Izv. Akad. Nauk SSSR Otd. Khim. Nauk, 1990 (1961). (14) R. West and W. H. Glaze, J . Am. Chem. SOC.,83, 3580 (1961). (15) I. B. Golovanov and A. K. Piskunov, Zh. Piz. Khim., 38, 2063 (1965). (16) E. C.Ashby and M. B. Smith, J . Am. Chem. Soc., 86, 4363 (1964). This paper contains numerous references to previous molecular weight studies. (17) (a) T . V. Talalaeva, A. N. Rodionov, and K. A. Kocheshkov, Dokl. Akad. Nauk SSSR, 154, 174 (1964); (b) R. Waake, 151st National Meeting of the American Chemical Society, New York, N. Y.,Sept 1966. (18) G. Stucky and R. E. Rundle, J . Am. Chem. Soc., 86, 4829 (1964). (19) H.Gilman and R. Jones, “Organic Reactions,” Vol. VI, John Wiley and Sons, Inc., New York, N. Y., 1951,p 353. (20) H.Gilman, W. Langham, and F. W. Moore, J . Am. Chem. SOC., 62, 2327 (1940). (21) Terminology of J. A. Pople, W. G. Schneider, and H. J. Bernstein, “High Resolution Nuclear Magnetic Resonance,” McGrawHill Book Co., Inc., New York, N. Y., 1959,Chapter 6.
Volume 7.9, Number 8 March 1968
946
G. FRAENKEL, D. G. ADAMG, AND R. R. DEAN
Table I : N m r Parameters Obtained from 1 M Solutions of Aromatic Compounds
1.450 1.976 2.007 1.975 2.362 2.307 2.361 2.360 2.743
... a 7
&0.001.
&0.02 Hz.
2.813 2.975 2.978 2,975 2.982 2.969 2.982 2.743 1.6-3.3
' S e e ref 29.
NMR
2.414 3.042
3.058 3.025 3.046 3.023 2.743
...
60Mc 33"
I
68 H z
Figure 1. Nmr spectrum, 60 Ha: top, 4-deuteriophenylmagnesium bromide; bottom; 4-deuteriophenyllithium.
and N V R I T . ~ ~The results of this calculation are listed in Table I. The shift of the para hydrogen in phenylmagnesium bromide was obtained from the spectrum of 3,5-dideuteriophenylrnagnesiumbromide. The latter gave a first-order spectrum and yielded the ortho-para coupling constant J 2 3 4under conditions where deuterium was decoupled. Spectra for phenyllithium and phenylmagnesium bromide are illustrated in Figure 3. The low-field absorptions are due to ortho hydrogens, while the other multiplet represents the meta and paya hydrogens. Using the parameters from the two deuterated compounds, together with several sets of reasonable values NMREN
The Journal of Physical Chemistry
1.77 1.62 1.53 1.65 1.62
0.96 0.84 0.51 0.84 0.98 0.70 0.96
-0.14 0.65 0.72 0.68 0.95 0.63 0.92
7.64 7.20 7.76 7.47 7.76
1.24 1.17 1.65 1.20 1.40 1.34 1.39
1.63 7.5-9.0
2.0-3.0
0.3-0.6
2.0-3.0
7.5-9.0
2.0-3.0
Reference 10, concent,rat'ionnot, given.
r D 0 M g B r 'I in Ether
k
4.82 6.52 6.86 6.50 6.98 6.80 7.00
for the missing ones, theoretical AA'BB'C spectra were computed23 and compared to the experimental spectrum of phenylmagnesium bromide. Using the theoretical spectrum most similar to the observed one, transitions were assigned to the experimental lines and energy levels calculated with XMREN. Prior to this calculation, at least two cycles were checked manually with the energy sum rule.23 Finally, the rimr parameters \yere iterated to fit the observed spectrum to give the results listed in Table I, stick diagram Figure 4A. I n the above analysis, the possibility still existed that deuterium might change the chemical shifts of hydrogens relative to the undeuterated reagents. I n fact, the proton shifts for all three Grignard reagents are identical within experimental error and n o isotope effect was observed. Diphenylmagnesium in ether gave nmr parameters almost identical with those of phenylniagnesium bromide. Analysis of the nmr spectrum of phenyllithium was accomplished as described above for the Grignard reagent, Table I, Figure 4b. These shifts for pheiiyllithium and phenylmagnesium bromide are very similar to those Ladd obtained for phenyllithium and diphenylmagnesiumJ8 respectively. However, the COUpling constants differ by as much as 0.2 H Z . ~ ~ To date, the signs of 13-H coupling constants in aromatic compounds have been reported as positive.25-28 Since the present examples are strongly coupled systems, changing the sign of one or more COUpling constants should alter the appearance of the spectra. That this is the case is evident in Figures 5 (22) J. D. Swalen, International Business Machines Corp., San Jose, Calif. (23) J. D. Swalen and C. $. Reilly, J . Chem. Phys., 37, 21 (1962). (24) Note that Ladd's analyses were accomplished by a combination of decoupling and computor techniques. (25) S. Clough, M o l . Phys., 2, 349 (1959). (26) R. Freeman, K.S. Bhacca, and C. A. Reilly, J . Chem. Phys., 38, 293 (1963). (27) J. 5 . Martin and B. P. Dailey, ibid., 37, 2594 (1962). (28) T. Schsefer, Can. J . Chem., 40, 1678 (1962).
947
NMRSPECTRA OF PHENYLMACNESIUM BROMIDE
I
I
Figure 3. Nmr spectra of (a) phenylmagnesium bromide and ( b ) phenyllithium.
-
IO Hr 7
Figure 2. Nmr spectrum, 60 Hz: top, low-field multiplet in Figure 1, top; bottom, low-field multiplet in Figure 1, bottom.
and 6, which depict stick diagrams for theoretical spectra of phenylmagnesium bromide where the signs of the smaller coupling constants have been changed one at a time. These diagrams are quite different from the final iterated solution, Figure 4a. Also, allowing any combination of two or three of the small coupling constants to be negative causes the theoretical spectra to look quite different. It is thus assumed that all the coupling constants in phenylmagnesium bromide are of the same sign. This conclusion also applies to phenyllithium. The nmr spectrum of pyridine has been discussed by several a u t h o r r ~ ~ ~ An J 0 analysis of the 40-Hz spectrum of neat pyridine based on results obtained with
deuterated pyridines was reported by Schneider, Bernstein, and P ~ p l e . Using ~ ~ their parameters, we have assigned the 60-Hz spectrum of pyridine in ether and analyzed it with the aid of computor pro~ ~ ~ I. ~ The principle grams NMREN and N M R I T , ~Table difference between this and the previous analysis31 is the very small value of J z ,noticed ~ also by the Kotvalewskis in their study of substituted pyridines.32 Very similar results have been recently published by Merry and Goldstein for neat pyridine and pyridine in water,. pH 9.l0 The nmr parameters for the organometallic reagents and pyridine are compared to typical values for covalently substituted benzenes in Table I. Each organometallic preparation yielded a single spectrum whose lines were of normal width (0.16-0.3 Hz). Mixtures of phenylmagnesium bromide and phenyllithium gave single spectra whose parameters lay between those of the two pure reagents. Hence, under our conditions, exchange amoung species is fast on the nmr time scale. However, in the pentafluorophenylmagnesium system carbon-magnesium exchange rates have been measured with the nmr line-shape t e ~ h n i q u e . ~ It is seen in Table I that the values for the lithium and magnesium reagents lie between those for pyridine and the typical substituted benzenes. Actually, the C6HSMgBrand C&&i parameters are more similar to those of pyridine than t o the last class of compounds in Table I. This is especially noticeable among the coupling constants. For instance, while in ordinary and substituted benzenes, the values of J2,4,J 2 , 6 , J3,j all lie between 2 and 3 Hz, among the organo(29) H.J. Bernstein and W. G. Schneider, J . Chem. Phys., 24, 469 (1956). (30) E. €3. Baker, ibid., 23, 1981 (1955). (31) W.G.Schneider, H. J. Bernstein, and J. A. Pople, Ann. N . Y. Acad. Sei., 70,806 (1958). (32) V. J. Kowalewski and D. G. de Kowalewski, J. Ckern. Phys., 36, 266 (1962).
Volume 7.9,Number 3 March 1968
948
G. FRAENKEL, D. G. ADAMS,AND R. R. DEAN
a
n
i b
Figure 4. Theoretical and experimental nmr spectra of (a) phenylmagnesium bromide and ( b ) phenyllithium: ortho multiplets a t left; meta and para absorption a t right.
-10
-5
5
0
IO HI
I
-10
lib
1
I 1 IO
26
1 30
, 1
35
40
45
Figure 5. Theoretical nmr spectra for phenylmagnesium bromide (meta and pura hydrogen) with the following negative coupling constants: (a) J2,6, ( b ) , J t , s , (c) Js.5, and ( d ) J ~ A . Other parameters are listed in Table I or assumed from symmetry. T h e Journal of Phvsical Chemistry
-5
0
5
10
Figure 6. Theoretical nmr spectra for phenylmagnesium bromide (ortho hydrogens) with the following negative coupling constants: ( a ) J2.6, ( b ) Jz,j, (c) Jg,j, and (d) Jg,c. Other parameters are listed in Table I or assumed from symmetry.
metallic reagents and pyridine they fall within the ranges 1.6-1.8, 0.8-1.0, 0.0-1.0, and 1.2-1.4 Hz, respectively. Furthermore, the organometallic compounds are all deshielded with respect t o benzene at the o ~ t h ohydrogens. This low-field shift is remarkable, in view of the fact that the opposite effect occurs in
949
NMRSPECTRA OF PHENYLMAGNESIUM BROMIDE aliphatic Grignard reagents and organolithium comp o u n d ~ . ~ ~ ~ ~ ~ ~ ~ The following discussion concerns the domfield shift for the ortho hydrogens in phenylmagnesium bromide and phenyllithium. The screening constant of a nucleus in a large molecule is very difficult to calculate because the diamagnetic and paramagnetic terms, which are both tend to cancel each other. Still, there are systems where contributions to the chemical shifts from different effects have been isolated and identified. For hydrogens in aromatic compounds, the main contributions to the shifts are ring currents, the effect of ~ h a r g e , ~paramagnetic ~-~~ shift~,38$~j and solvent eff e ~ t s . ~ ~ - 4These * effects will be examined to account for the anomolous shift of the ortho hydrogens in our organometallic reagents and pyridine compared to hydrogens in benzene. (1) Solvent effects in this case probably chiefly arise from specific chemical interactions in the organometallic aggregates. These will not be considered now. (2) The ring ~ u r r e n t ~ presumably ~-~l is affected only in a minor ryay by substitution of a metal for hydrogen. (3) Negative charge at C, should produce a large upfield shift42at the o ~ t h ohydrogens and smaller upfield shifts at more distant hydrogens, since the effects are most liliely transmitted by the inductive effect. Actually, the ineta and para hydrogens are slightly shielded, though this need not be due to charge; see below. (4) If the ionic carbon-metal bond may be regarded as a dipole, the latter will have only a minor influence on the chemical shifts of the ring hydrogens. Fraenkel and Kim4$have observed this effect in nmr spectra of aniline oxides and BL3 adducts of anilines. ( 5 ) The paramagnetic contribution to the chemical shift38r45 would appear to be partly responsible for the effects observed here. Employing the point-dipole a p p r o x i m a t i ~ n we , ~ ~consider ~~~ point anisotropies for different atoms around the rings. Compared to benzene, these will be most altered by the egregious change in the anisotropy of the carbon atom undergoing substitution by metal. Pyridine is isoelectronic with phenyl anion. These two species should have similar electronic structures. The nmr parameters for the aryl organometallic compounds and pyridine are also similar. To the extent that the carbon-metal bonds are partially ionic, it is instructive to mention the origins of proton shifts in pyridine. This subject has been discussed by several investigators.10~~'9-3z~5z-54 The prevailing conclusion is that the contributions due to n-electron densities and paramagnetic terms mainly account for the observed shifts.55 The contributiori of a paramagnetic term to shifts in pyridine is most evident from the fact that the a-proton shift is too low t o be accounted for by electron density effects alone. :Paramagnetic shifts in pyridine come
from magnetic mixing of the ground and nearest electronic excited states,62 the principle low-lying excited state being n + n*. According to Pople's treatment,38 the paramagnetic shift cSpsra should be inversely proportional to the lowest electronic excitation energy, AE
It seems reasonable that similar considerations to those discussed to account for shifts in pyridine should also apply to the arylorganometallic reagents. However, here variations in the n-electron densities should be relatively small, since there is no conjugative mechanism to delocalize the charge associated with C,. Hence, n-electron density shifts with respect to benzene should be fairly uniform. To summarize the above discussion, chemical shifts in phenylmagnesium bromide and phenyllithium due to different n-electron densities and the inductive effect of charges associated with the C-M bond are expected to be small. It is tentatively suggested that paramagnetic shifts are mainly responsible for the results reported here. Although this cannot be proved exactly, there are two considerations which lend weight to the above suggestion; these are now outlined below. (33) G . Fraenkel, D. G. Adams, and J. Williams, Tetrahedron Letters, 767 (1963). (34) D . F. Evans and J. P. Maher, J . Chem. Soc., 5125 (1962); H. Roos and W.Zeil, Ber. Bunsenges. P h y s i k . Chem., 67, 28 (1963). (35) N. F. Ramsay, Phys. Rev., 78, 699 (1950). (36) N . F. Ramsay, ibid., 86, 243 (1952). (37) J. A. Pople, Proc. Roy. SOC.(London), A239, 541 (1957). (38) J. A. Pople, ibid., A239, 550 (1957). (39) G . E. Johnson and F. h l . Bovey, J . Chem. Phys., 29, 1012 (1958). (40) N. Jonathan, 5. Gordon, and B. P. Dailey, ibid., 36, 2443 (1962). (41) J. S. Waugh and R. W. Fessenden, J . Am. Chem. Soc., 79, 846 (1957). D. McLachlan, and J. H. Richards, (42) G . Fraenkel, R . E. Carter, 9. ibid., 82, 5846 (1960), and references cited therein; J. I. Musher, J . Chem. P h y s . , 37, 34 (1960). (43) A. D. Buckingham, C a n . J . Chem., 38, 300 (1960). (44) B. P.Dailey and J . S. Xartin, J . Chem. Pkys., 39, 1722 (1963). (45) J. D . Das and R . Bersohn, Phys. Rev., 104, 476 (1956). (46) A. D. Buckingham, T. Shaeffer, and W. G. Schneider, J . Chem. Phys., 32, 1227 (1960). (47) R . J. Abraham, Mol. Phys., 4, 309 (1961). (48) A . A. Bothner-By, J . Mol. Spectry., 5 , 5 2 (1960). (49) G. Fraenkel and J. P.Kim, J . Am. Chem. Soc., 88, 4203 (1966). (50) H . h l . McConnell, J . Chem. P h y s . , 27, 226 (1957). (51) A. A. Bothner-By and C. ru'aar-Colin, J . Am. Chem. Soc., 80, 1728 (1958). (52) V. XI. 5 . Gil and J. iX.Murrell, T r a n s . Faraday Soc., 60, 248 (1964) (53) J. A. Elvidge and L. M.Jackman, J . Chem. Soc., 859 (1961). (54) B. P. Dailey and J. K. \Vu, J . Chem. Phys., 41, 3307 (1964). (55) These two contributions have been calculated using SCF wave functions. The results closely reproduce observed 1H and shifts in pyridine. For p y r i d i n i u m ion, ?r-electron densities are mainly responsible for the observed shifts: G. Fraenkel and T. Tokuhiro, unpublished results. I
Volume 72, Number 3 March 1968
G. FRAENKEL, D. G. ADAMS,AND R. R. DEAN
950
Anisotropy Approximation. Let us assume that the para hydrogen in phenyllithium is sufficiently distant from C, that inductive contributions to its shift are not important. The shift 6(H,) relative to benzene should be mainly paramagnetic. Using XcConnell’s point-anisotropy a p p r o x i m a t i ~ n ,eq ~ ~ 3, to assign a Ax
A~ = -3(i
3R
-3
COS^
e)
(3)
value to Ax and the parameters in Figure 7 , the meta and ortho shifts are nearly identical with those observed: calculated, +0.22 and -0.85 ppm; observed, +0.232 and -0.767 ppm, for ineta and ortho hydrogens, respectively. A similar treatment for phenylmagnesium bromide gives +0.21 and -0.79 ppm compared to the observed values of +0.24 (meta) and -0.34 (ortho) ppm, respectively. Although the close agreement of the values for phenyllithium must be fortuitious, both treatments do give the correct order for the shifts around the reagents. Ultraviolet Spectra. In pyridine, the low-lying excited state mainly responsible for the paramagnetic shifts is n + n-*.52 If the analogies between C6H&f and pyridine are correct, a transition corresponding to n + n-* in pyridine should be available t o the organometallic compounds also. Such an excitation may be written uc-31 + n-*, recognizing that the C-AI bond must be partially covalent. Accordingly, Tve have obtained uv spectra of specially purified samples of phenylmagnesiuni bromide and phenyllithium, Figure 8. In both spectra, the absorption tails off slowly toxard 300-350 Waake and Doran57 have also reported uv data for phenyllithium. By analogy to the results for pyridine, we suggest that the highwavelength tails in these uv spectra involve UC--~~Z+ T* excitation. Ordinarily, n + n-* excitation energies in pyridine and aromatic diazines increase with the strength of hydrogen bonding at nitrogen.j8 At the same time, the paramagnetic contributions to the various proton shifts should decrease. Applying the above argument to the organometallic compounds, as h l in C6HJ1- becomes more electropositive, the ionic character of the carbon-metal bond should inc,+ease,AE(uc-JI+ n-*) decrease, and 6, decrease (become more negative). Such a trend is evident among the ortho proton shifts which decrease along the series ( c 6 H ~ ) ~ Z> n c~H5MgBr ( c ~ H 6 ) ~ n I g> C6H5Li.8s6gAlthough absorption bands for the proposed (TC--I + n-* transition could not be resolved, it is qualitatively seen that the value for C6H,Li is lower than that for C6H5;\lgBr. Following the above argument, it is speculated that coordination of different solvents with 31 in ArN would be expected to alter AE((Tc--LI + n-*) and thus also the ortho shift. The principal solvent effect on these shifts should take place via specific chemical interactions.
-
The Journal of Physical Chemistry
Figure 7.
Geometry in CeHjM.
,
Figure 8. Uv spectra: -- ,phenyllithium, 0.25 M in ether (path = 0.05 m m ) ; - - phenylmagnesium bromide, 0.09 M in ether (path 0.1 mm.).
-,
Taken together, the spectral similarities-chemical shifts, coupling constants, and uv data-presented here for C6H5MgBr, C6H5Li, and pyridine strongly imply that paramagnetic terms are the most important contributions to the ortho shifts in the organometallic reagents. Proceeding down the series of organometallic compounds as 1 4 becomes more electropositive, J 2 , 3 J2,6, , and J3,5 all decrease and the most conspicuous effect is the decrease in 6,. The lowest value for this latter shift should occur for free phenyl anion, so far not reported. (56) For phenyllithium, X and E are, respectively, 261, 1000 and 330, 40; for phenylmagnesium bromide, X and e are, respectively, 255, 630 and 290, 63. These results were reproducible. On hydrolysis of these reagents, there was obtained only the weak uv absorption due to the expected concentration of benzene. (57) R. Waake and M. A. Doran, J . Am. Chem. SOC.,85, 1651 (1963). (58) NI. Kasha and G. J. Brealey, ibid., 77, 4462 (1955). (59) The ortho hydrogens in compounds such as CsHsSiCla and (Cdf6)khCh are also deshielded while the shifts of the other hydrogens are not much different from benzene: J. C. Maire and F. Hemmert, Bull. SOC.Chim. Fr., 2785 (1963). Since coupling constants have not been reported for these compounds, it is not clear whether the arguments presented in the paper apply here also.
NMRSPECTRA OF PHENYLMAGNESIUM BROMIDE The main conclusion derived from this work is that the downfield shift of the ortho hydrogens in ArRI: is a qualitative measure of the ionic character of the carbonmetal bond.
Experimental Slection Chenzicals. The following chemicals were obtained : bromobenzene, Matheson Coleman and Bell; butyllithium, Foote Mineral Co. ; lithium wire, Lithium Corp. of America; deuterium oxide, Columbia Chemical Co.; and sublimed magnesium, Dow Chemical Co. Diethyl ether was distilled from methylmagnesium bromide. Magnesium. Sublimed magnesium was milled to fine shavings, care being taken to avoid contamination by iron. The metal shavings were washed with benzene, then with three portions of diethyl ether (CP). The ether was removed in a stream of dry helium and the magnesium stored over anhydrous magnesium sulfate. 4-Deuteriobromobenxene. Magnesium turnings (2.9 g; 0.12 mol) were placed in a three-neck 500-cm3flask fitted with addition funnel, mechanical stirrer, and reflux condenser with drying tube. Diethyl ether (100 ml) was distilled into the flask from a solution of methylmagnesium bromide in ether. p-Dibromobenzene (23.6 g; 0.1 mol) was dissolved in an additional 100 ml of anhydrous ethyl ether and transferred to the addition funnel under helium. Ethylene dibromide (0.5 g; 01.003 mol) was added to the magnesium-ether mixture to initiate reaction. Stirring was begun and the dihalide-ether mixture admitted at a rate sufficient to cause continuous reflux of ether. After addition mas complete, the mixture was refluxed 30 min to ensure reaction of any bromobenzene formed by hydrolysis of p-bromophenylmagnesium bromide. Deuterium oxide (10 g; 0.5 mol) was added dropwise, followed by 100 ml of 10% ammonium chloride solution, added rapidly. The ether layer was separated, dried over anhydrous magnesium sulfate, and the ether removed by distillation. The product, 4-deuteriobromobenzene, was distilled under reduced pressure, bp 74.0-74.5' (40 imm), 75% yield. 3,6-Dideuteriobromobenxene. sym-Tribromobenzene60 (31.5 g; 0.1 mol) was treated with magnesium (3.8 g; 0.15 mol) in the same manner as the previous preparation and hydrolyzed with deuterium oxide (12 g; 0.6 mol). The product, 3,5-dideuteriobromobenzene was separated from bromobenzene by vapor chromatography in a 45% yield. The entire yield of 3,5-dibromodeuteriobenzene (10.6 g; 0.045 mol) mas treated with magnesium (1.2 g; 0.05 mol) in the same manner as above and hydrolyzed with deuterium oxide (3.8 g; 0.18 mol). The product, 3,5-dideuteriobromobenzene, was separated as above, 80% yield and 36% over-all. Syringes. The syringes used for addition of the bromides were ordinary hypodermic syringes (Xlultifit) with a locking needle; the syringe used for extractions
951 of samples of Grignard reagents was a 1-cm3 gastight syringe (Hamilton). All syringes were oven dried at 90" and stored over phosphorus pentoxide. Just before use, the syringes were flushed with dry helium. Instrumentation. Two nmr spectrometers were employed in this study: a Varian HR-60 high-resolution nmr spectrometer equipped with a Hewlett-Packard wide-range audiooscillator, Hewlett-Packard Rlodel 522D digital counter, and an NMR Specialties heteronuclear decoupler; and a Varian A-60 high-resolution nmr spectrometer. The A-60 was used for the majority of spectra not requiring decoupling. Grignard Reagents. The reaction vessel was a cylinder 25-mm 0.d. x 90 mm. An 8-mm 0.d. side arm equipped with a Teflon straight-bore stopcock served as the inlet for helides. The outer arm of the stopcock was protected with a serum cap. The top of the vessel was attached via a condenser t o the receiver side of an ether still. The glassware described above, excluding the still, was dried with flaming in a current of helium. A small Teflon-covered magnet and magnesium turnings in 100% excess were placed in the vessel and the system was connected to the ether still. Ether, 15 ml, was distilled into the vessel in an atmosphere of helium. Then, 0.1 ml of the aromatic halide was injected into the vessel through the side arm. Reaction usually set in within a few minutes, accompanied by ebullition, cloudiness, and/or color formation. The remaining halide was added at the rate of about 1 mm/min and the mixture was then heated for 30 min and stirred for 5 hr to complete the reaction. The nmr tube was dried with a flow of dry helium through a 10-in. 18-gauge hypodermic needle inserted in the tube with simultaneous warming in a Bunsen flame. After removal of the flame, the flow of helium was continued until the tube cooled to room temperature, when the needle was withdrawn and the tube closed with a serum cap. Before use, the 1-cm3 gastight syringe for sample withdrawal was flushed several times with helium. The syringe was fitted with the 10-in. needle used for drying the nmr tube and was inserted through the serum cap and stopcock on the reaction vessel. After filling the syringe, the needle was withdrawn from the reaction vessel and quickly inserted through the serum cap on the nmr tube. The tube was filled to a depth of 1 in. and cooled in Dry Ice, and the upper end was quickly sealed off. A 1.00-ml aliquot was withdrawn from the reaction vessel for acidimetric titration, using methyl red as the indicator. After titration, the mixture was extracted with 1 ml of carbon tetrachloride. The organic phase was separated, dried over anhydrous potassium carbonate, and analyzed by vapor chromatography and nmr spectroscopy. (60) Undergraduate preparation, The Ohio State University, Columbus, Ohio.
Volume 7%Number 8 March 1968
952 The only impurity present in these Grignard preparations was benzene or one of the deuterated benzenes. Yields of 96-9S% were obtained in these reactions. Phenyllithium. The reaction vessel consisted of a round-bottomed flask equipped with an S-mm 0.d. side arm, a ball joint which led, via a condenser, to an ether still and a second side arm at the bottom, 15-mm o.d., which contained a glass frit. Both side arms were protected on the outside with Teflon straight-bore stopcocks and serum caps. Lithium wire (2.3 equiv) was dipped into anhydrous ethanol and the coating scraped off under petroleum ether. The wire was cut into l-mm lengths and dropped into the reaction vessel in a stream of helium. Ether was distilled in and decanted through the frit to remove soluble impurities. Ether (10-15 ml) was again distilled in and the mixture cooled to -25" with acetone-Dry Ice. A small amount (1 mmol) of bromobenzene was added, and reaction began within 15 min, evidenced by the lithium wire turning black, then white, with formation of a white precipitate. The
The Journal of Physical Chemistry
G. FRAENKEL, D. G. ADAMS, A N D R. R. DEAN remainder of the halide (9 mmol) was added over a period of 30 min, during which time the reaction mixture was permitted to warm to 10". The mixture was then stirred 1 hr at room temperature to complete reaction. Phenyllithium was obtained in 98% yield, the principal impurity being benzene. Phenyllithium was also prepared by the halogenmetal exchange reaction by mixing equivalent quantities of bromobenzene and butyllithium in ether at 0". The nmr spectra for the two different preparations were nearly identical. Purification of Reagents. Phenyllithium and phenylmagnesium bromide were purified by crystallization from ether on the vacuum line.61
Acknou:ledgment. This research was supported by the Petroleum Research Fund, Administered by the American Chemical Society, and by the Air Force Office of Scientific Research, Grant No. AF-AFOSR-251-63. (61) S. Kobayashi, M.S. Thesis, The Ohio State University, Columbus, Ohio, 1965.