tetrahydrofuran-dimethoxyethane at - 16806 produced Complexes Sir:

for interaction with six equivalent protons. The sep- aration between lines is 2.33 gauss with a line width of. 0.44 gauss.g The g-factor is 2.0027. C...
0 downloads 0 Views 303KB Size
2332

COMMUNICATIONS TO THE EDITOR

Pritchard3 and Flygare, Narath and Gwinn4 suggested the possibility of formation of the radical anion. The electron spin resonance spectrum of cyclopropane radical anion now has been ~ b s e r v e d . ~ Sodium metal6 in tetrahydrofuran’ produced no observable resonance signal from cyclopropane. Sodium in d i m e t h ~ x y e t h a n eproduced ~ a weak signal, inadequate to obtain fine structure. Potassium and rubidium metals separately in dimethoxyethane produced signals of higher intensity, and some fine structure could be observed. The use of a sodium-potassium alloy in 2 : 1 tetrahydrofuran-dimethoxyethane a t - 16806produced a completely resolved spectrum consisting of seven lines of binomial intensity (1 : 6 : 15 : 20 : 15: 6 : 1) as expected for interaction with six equivalent protons. The separation between lines is 2.33 gauss with a line width of 0.44 gauss.g The g-factor is 2.0027. Calibration was by means of peroxylamine disulfonate. The spectrum disappears on warming to - 100° but reappears on cooling again. The work on cyclopropanes is being avidly continued and is being extended to other systems. lo (3) N . Muller a n d D . E . Pritchard, J . Chem. P h y s , 31, 768 (1959). (4) W . H . Flygare. A. Narath and W . D Gwinn. ibid , 36, 200 (1961) ( 5 ) A Varian Associates V-4,500 spectrometer with 100 kc. modulation was used for this work (6) T h e metals were distilled directly into t h e quartz sample tubes on a high vacuum line. (7) Solvents were purified exhaustively until a blue solution of potassium could be supported for many days a t room temperature. (8) J K Bolton, Mol. P h y s . . 6, 219 (1963). (9) Variation in line width (presumably d u e t o electron exchange with free cyclopropane) and in intensity occurs from sample t o sample b u t the separation between lines is constant a t 2.33 gauss. (10) NOTE ADDED I N PROOF--Since submission, a number of systems have been observed t o present hazard. Some sample tubes have burst with considerable violence even though sealed under high vacuum a n d stored in liquid nitrogen. T h e lifetimes vary from minutes t o days depending on t h e compound used.

DEPARTMEKT OF CHEMISTRY KERRYW. BOWERS MASSACHUSETTS INSTITUTEOF TECHNOLOGY CAMBRIDGE 39, MASSACHUSETTS FREDERICK D. GREEXE RECEIVED JUNE3, 1963

Mossbauer Resonance Ed ects in Irono Organometallic

Complexes Sir: We wish to report the Fe57 Mossbauer resonance’ absorption results of a series of diene-iron tricarbonyl complexes and dienyl-iron tricarbonyl complex cations. In common with most other iron compounds, the Mossbauer spectra of these materials consist of two absorption lines of equal intensity. The pertinent data are listed in Table I . The center of gravity of the two lines (6) and the separation (A) are given in millimeters per second.2 Group I compounds consist of the neutral diene-iron tricarbonyls, group I1 contains the salts of the dienyliron tricarbonyl cations, while Fe(C0)6 and Fe(C0)3(PPh& are listed together as group 111. The data for Fe(CO)i3 and for cyclooctatetraene-iron tricarbony14 have been reported. Our data on splittings agree, but a comparison of the absolute chemical shift is clouded by lack of accepted standards. I t is seen from Table I that the diene-iron tricarbonyl complexes show strong quadrupole splitting and possess a chemical shift within the fairly narrow region 6 = 0.2 (1) An excellent introduction t o the subject is H Frauenfelder, “ T h e M6ssbauer Effect,” \\’ A Benjamin C o , Ne* York. N . Y ,1962. ( 2 ) S a t u r a l abundance Fe” proved adequate for these studies A copper-cohalts? source was utilized a t room temperature in a constant velocity servo-controlled loudspeaker drive system Spectra were all obtained a t T y p e 302 stainless steel absorbs, 78OK o n thin samples (40-80 mg. ‘cm ‘1) a t room temperature, a t - 0 32.5 mm ‘sec relative t o this Cu-CoST source ( 8 1 I, M Epstein. J Chem P h y s , 3 6 , 2731 (1962). ( 4 ) G K Wertheim and I< H Herber. J A m C h e m . S o c , 84, 2274 11 962)

VOl. 85

TABLE I M ~ S S B A U EPARAMETERS R FOR VARIOUSIRONORGANOMETALLIC COMPLEXES Group I 8cu, m m . / sec.

2-Methoxy-3,5-hexadiene-Fe( C0)3 -0.218 racemic-5,6-Dimethyl-1,3,7,9-tetradecaeneFez(CO)e - ,216 Alloocimine-Fe( C0)3 - ,214 meso-5,6-Dimethyl-l,3,7,9-tetradecaeneFed COh - ,208 2-Hydroxy-3,5-hexadiene-Fe(CO), - ,202 Butadiene-Fe( C0)3 - ,198 7-Acetoxy- bicycloheptadiene-Fe( c o ) ~ - ,191 1-Phenylbutadiene-Fe( CO)3 - ,178 2,4-Hexadienoic acid-Fe( C 0 ) 3 - ,178 Cyclooctatetraene-Fe( C 0 ) 3 - 140

A, mm

/

sec

1 il 1 69

1 75 1 58 1 56 1 46 2 01 1 59 1 63 1 26

Group I1 1,5-Dimethylpentadienyl-Fe(CO)3+BF41-Methyl-pentadienyl-Fe( CO)a+SbCI6Cyclohexadieny-Fe( C0)s +CIOaI-Methyl-pentadienyl-Fe( C 0 ) 3+ClOa1-Methyl-pentadienyl-Fe( C 0 ) 3+BFaCycloheptadien yl-Fe( CO)?+BF41-Methyl-pentadienyl-Fe( CO)a+PF6Bicyclooctadienyl-Fe( C0)s +BFa -

-0.126 - ,126 - ,122 - ,117 - ,117 - ,111 - ,103 - ,098

1 1 1 1 1 1 1

Group I11 Bistriphenylphosphine-Fe( C 0 ) 3 Iron pentacarbonyl

-0.324 -0.282

2 76 2 60

1 83

69 66 70 72

57 67 79

f 0.025 mm./sec. The range of ligands in complexes listed include non-conjugated dienes and substituted dienes as well as tetraene-diiron hexacarbonyl complexes; this range might therefore be considered characteristic of such complexes. The notable exception in group I is cyclooctatetraene-iron tricarbonyl ; both in chemical shift and quadrupole splitting this substance appears to be abnormal. The reasons for these deviations are not yet clear. The shift for cyclooctatetraene-[Fe(C0)3]2 is 0.05 mm./sec. less,4which places i t in a “normal” position. These shifts are small compared to those of inorganic salts which, a t B O K . , typically give 6 = +0.35 for Fe3+, 6 = 4-1.20 for Fe2+,5and 6 = -1.0 for Fe6f,6 converted to the C U - C O ~ source. ~ The cationic complexes listed in group I1 clearly form a separate group based upon their chemical shift, though they display roughly the same range of quadrupole splitting as seen in the diene-iron tricarbonyl complexes. In contrast to ferrocinium the chemical shift and quadrupole splitting of salts of the 1methylpentadienyl-iron tricarbonyl cation are relatively insensitive to the nature of the anion. The cations absorb a t higher values than do the dieneFe(C0)3 complexes; however, the effect is small and certainly would not seem large enough to support the suggestion6 t h a t these complexes be regarded now as Fez+ systems containing a carbanion as a ligand. Even considered as Fe” systems bonded to a carbonium ion ligand, the differences between group I1 complexes and group I complexes seem small. The mean difference in chemical shift between the two is 0.09 mm./ sec. Epsteinghas shown that a difference of 0.04 mm.,! sec. exists between tris-5-nitrophenanthroline and tris( 5 ) S. DeBenedetti, G Lang, and R Ingalls, P h y s . Rea Lellrrs, 6. 60 (19611 (6) G . K Wertheim and R . H Herber, J . Chem P h y s , 3 6 , 2197 (1962) (7) G K . Wertheim and R H . Herber, ibid , 38, 2106 (1963) (8) E 0 Fischer and R . D. Fischer, Angem Chem , 71, 919 (19fi0). (9) L . M . Epstein, paper delivered a t the 144th National Meeting of the American Chemical Society, Los Angeles, Callf., April I , 1963.

COMMUNICATIONS TO THE EDITOR

Aug. 5 , 1963

5-phenylphenanthroline-Fe +2 complexes at liquid nitrogen temperature ; this difference presumably involves a fairly long range effect and suggests that a greater separation between group I and group I1 complexes might have been expected. The small difference between group I and group I1 may perhaps be rationalized in terms of the duality of the bonding of ligands attached to transition metals in low oxidation states. The carbonium ion ligand might be expected to result in a decrease in electron density, including 4s, in iron in the "forward coordination" when compared to a neutral diene and therefore give rise to a positive chemical shift. However, the back donation of d electrons into the carbonium ion ligand would be stronger than for the diene; the resulting lowering of 3d electron density about iron would decrease the shielding of the 4s electron density and thereby cause a negative chemical shift. The two effects would therefore tend to nullify each other. The net result is a small decrease in effective s density and a corresponding positive shift. The dominance of forward coordination over back donation is also illustrated in the chemical shifts of Fe(C0)b and Fe(C0)3 (PPhJ2 whose stereochemistry differs from the group I and I1 complexes. The strong electron donating properties of the triphenylphosphine group result in a net negative shift compared to Fe (C0)5. In contrast to the small effects of chemical bonding on the shift 6, there exists within each group a considerable variation in splitting A . The splitting is observed to be greatest for molecules displaying axial symmetry, such as Fe(CO)b, Fe(C0)3(PPh3)2,and 7-acetoxy bicycloheptadiene Fe(C0)3. The large effects on the splitting caused by slight structural changes suggests that further analysis of the origins of electric field gradients in these molecules is desirable and may prove helpful in establishing certain stereochemical relations. Acknowledgment.-We wish to thank Mr. Richard Mazak and Mr. John Travis for some of the experimental measurements and the Alfred P. Sloan Foundation for financial support. DEPARTMENT OF PHYSICS DEPARTMEST OF CHEMISTRY THEUNIVERSITY OF TEXAS AUSTIN12, TEXAS RECEIVED JUNE 10, 1963

R. L. COLLINS R. PETTIT

Effect of Solvents on Transition States in the Reactions of t-Butoxy Radicals'

Sir :

The major products of reaction of t-butoxy radicals in the presence of two potential hydrogen donors are determined by the competition

+ CH3. k% + RiH + (CHa)3COH + R1. k, + RzH +(CH3)aCOH + RP. kd

(CH8)zC-0. (CH3)3C-O. (CH3)3C-O.

--i,

CHICOCHI

(1)

(2)

2

(3)

The rate constant ka,/'ka, may be measured directly from the relative rates of disappearance of R I H and R2H or indirectly by determining k a , / k d and k a , / k d from t-butyl alcohol/acetone ratios obtained in separate experiments in the presence or R1H and R2H alone. We wish to report a study of the effect of solvents on these competitions indicating much more significant phenomena than the small solvent effects which have been noted previously in t-butoxy radical reactions.2-4 (1) Support of this work by grants from t h e National Science Foundation is gratefully acknowledged

DETERMINATION OF

2333 TABLE I REACTION OF l-BUTOXY RADICALS CYCLOHEXANE

ka/kd I N

WITH

log Solvent Gas phase CzC13Fa CHaCS CsH6 CaHsCN CoHCl CzClr CZHZCII(trans) CzHzCIz (cis) CHCOOH

-k.Jkd 1000 (203Ib 4.29 (0.68)' 2.82 1.90 2.65 4.14 2.26 1.57 (0.65)'

(PZ),/ 00

Ed - EU .

(PZ)d

(1040)b 487 81 9 207 109 91.7 293 98.9

10 80

-4.77 - 5 04 - 5 73 - 4 63 - 4 58 -3.82 - 4 49 -4.16 -3.92 -3 6 1

40'

(625Ib 52 8 8.12 24.7 16.9 16.4 39.0 14.2 9.12 2.84

52.2

12.4

9 65 9.54 8 66 8.28 7.21 8.72 7.69 7 04 5.80

Difference in activation energies, kcal,/mole. and 30" respectively. Extrapolated values. a

' At

60, 40,

Our technique involves t-butyl hypochlorite chlorinations in which reactions 1-3 are steps in a chain sequence, followed by reaction of the radicals produced with tbutyl hypochlorite to yield CH3C1, RIC1, and R2C1, respectively, and regenerate (CH3)3C-O.. The hypochlorite was decomposed photochemically in excess (5 :1) cyclohexane of varying concentrations (0.01-0.5 M ) in each of several solvents a t 0, 25, 40, 70, and 100" and t-butyl alcohol/acetone ratios determined by gas chromatography. Ratios of k a / k d then were obtained from plots of alcohol/acetone ratios vs. cyclohexane concentration. Plots of log k a / k a us. l , / T gave excellent straight lines, and some of our ratios, activation energy differences, and ratios of PZ factors are listed in Table I , which also includes gas phase measurements obtained by a slight modification of the same technique. We see that k a / k d ratios vary by over a factor of 100 a t the lower temperatures, chiefly as a consequence of changes in the difference in activation energy for the hydrogen abstraction and @-scissionprocesses. As we have pointed out elsewhere6 solvent effects on competitive radical reactions must reflect different degrees of solvent interaction with the transition states rather than with the radicals involved (assuming any such "complexing" to be a rapid process) and solvation of the transition state for a p-scission (or other unimolecular radical reaction) is sterically more feasible than solvation of the transition state for hydrogen abstraction. Our results are consistent with this view. Aromatic solvents and the chloroethylenes6 apparently stabilize the transition state for @-scissionby 3-4 kcal. over that for attack on cyclohexane, presumably through the contribution of charge-transfer type structures. CH3 I

CH3-C-0.Ar

I

CHI

+-+CH3

CH3

I

-

C.-O.

I

CHI

CH,

I

.Ar e C H 3 . C=O Ar I CH3 +

There evidently is no major effect of simple solvent polarity, since energetics in acetonitrile are similar to those in a haloalkane, but the very large facilitation of @-scissionin acetic acid suggests that hydrogen bonding is an alternative means of transition state stabilization.' CHI

I I

CH3. . . C-0'

CH3

0

. . . .H / \ 0

( 2 ) G. A . Russell, J . Org. C h e m . , 24, 300 (1959). (3) C . Walling and B. B Jacknow, J . A m . Chem Soc , 8 2 , 6108 (1960) (4) E . L. P a t m o r e and R. J . G r i t t e r , J Org. C h e m . , 27, 4196 (1962). (5) C . Walling and A. P a d w a , J A m . Chem. Soc., 84. 1593 (1963) (6) T h e chloroethylenes were chosen since they show no significant re action with t-butoxy radicals We previously have reported a large effcct of olefinic solvents on t h e reactions of t h e 2-benzyl-2-propyloxy radical 5 (7) Evidence for hydrogen bonding of peroxy radicals has been given by F. F. R u s t and E . A. Youngman, J. Org. C h e m . , 27, 3778 (1962).