1331
NOTES
Sept., 1956
increased very rapidly for small [RH], but became roughly independent of it above 10 mm. pressure. Some experiments carried out a t 85” showed similar behavior. Thus, the apparent elimination of wall contamination did not lead to sensible results, and in consequence part of the reason for the failure of the experiment must lie in the inadequacy of the assumed mechanism. A significant omission would appear to be R
ks + xp-Hz --+ R +~ 1 I z
(9)
i.e. , the paramagnetic conversion by radicals without reaction, which would vary with the degree of localization of the free electron on the R radical. It was decided to test this by changing the hydrocarbon RH. Methane could not be used because it was not easily separable from hydrogen, and ethane reacted with hydrogen atoms producing methane by H f C ~ H+ B CHa
+ CHJ
(10)
a t too great a rate to apply a satisfactory correction. It was considered that propane would give a radical too similar to hexyl radical to produce an observable effect, and in consequence, fluoroform was chosen. The addition of 2 mm. CF3Hdid not alter the conversion rate significantly, but 15 mm. increased the conversion rate to y = 1.12 p (that reaction actually took place was demonstrated by exchange between D, and CF,H upon irradiation). Similar experiments with C2F6 also increased the conversion rate t o y = 1.10 p. In other words the primary effect was a slight reduction in the rate of diffusion of hydrogen atoms to the walls which was not taken into account in the reaction scheme, but the effect is relatively small and could not account for the observed independence of P/r and [RH]. The intrusion of reaction 9 can only definitely be confirmed or excluded by studying the parallel system of R H a mixture of Hz and DZwhere there is no analogous reaction. However, a number of experiments using n-hexane and a 1/1 Hz/Dz mixture showed that no results of significance were obtainable in the present apparatus because of the relative smallness of the resistance changes and the difficulty of correcting for the introduction of additional hydrogen into the system by exchange with the undeuterated hydrocarbon. We conclude that the ortho-para conversion method, which has been used in the past3to measure rate constants of reaction 7, is rendered inaccurate by the adsorption of the hydrocarbon and its decomposition products on the walls of the reaction cell and that, i n addition, diffusion effects and some complications i n the mechanisms of t,he individual reactions are also operative in pretrenting the useful exploitation of this technique. Acknowledgment.-We are indebted to Dr. A. F. ‘l’rotman-Dickenson for suggesting this problem and for his collaboration in the early stages of this work.
+
RESOLUTION OF TRIS-OXALATO METAL COMPLEXES BY F. P. DWYER AND A. M. SARGESON Contribution from the Departments of Chemastry, NoTthwestern, Evanston , Illinois and Sydney Universities, Sydney, Australia Received FebruaTy 80,1966
At various times the resolution of the labile trisoxalato complexes of aluminum, gallium and iron(II1) have been reported, but subsequent attempts. a t resolution have frequently failed. A summary of the evidence has been presented by Johnson,‘ who himself failed with the iron and aluminum complexes, and Basolo2 in a recent review. Sirice cations, hydroxyl and oxalate ions, have an accelerating influence on the rate of racemization of the tris-oxalato chromate(II1) ion8 it was suggested by Bas0104 that the conflicting evidence c,ould be due to the presence or absence of extraneous ions during the resolution. Repetition of the work, using strychnine as the resolving agent for the aluminums$6and gallium7 complexes and I-a-phenylethylamine for the iron compoundlain the presence or absence of hydrogen ions and/or oxalate ions, was fruitless. Small levorotations at various times, followed by subsequent loss of the rotation were traced to incomplete elimination and slow precipitation of the strychnine as the iodide. When the resolving agent was removed as the perchlorate no rotation was ever observed. It should be noted, however, that a dextrorotatory specimen of the tris-oxalatoaluminate ion has been claimed6 through the use of strychnine. It is also pertinent that addition of calcium chloride to a freshly prepared solution of potassium tris-oxalategallate gave an immediate precipitate of calcium oxalate-an observation also made by M ~ e l l e r . ~ When an aqueous alcoholic solution of Z-tris-1,lOphenanthroline-nickel iodide or chloride (Ni phen3)I2 was added to an aqueous solution of potassium d,Z-tris-oxalato-cobaltate(II1) the sparingly soluble green K-Z-[Ni phen3]-d-[Co(Cz0~)3].Hz0 separated immediately. The green solution remaining, on the addition of alcohol gave Z-K3Co(Cz04)3. The dextro form was obtained from the diastereoisomer after elimination of the [Ni phen3]++cation with potassium iodide and hydrogen peroxide. This procedure was shown to be a rapid general method of resolution of tris-oxalato metal complexes by the separation of the antipodes of the Cr( C Z O ~ ) ~and ” ’ Rn(CzO,),”’ ions. Though diastereoisomers of the same composition were obtained with A1(Cz04)3”’and Fe(Cz04)3”’l (hexahydrate with the iron compound instead of the usual monohydrate), no rotation was ever observed in the potassium salts precipitated with alcohol from the ( 1 ) C . H. Johnson. T r a n s . Faraday Soc., 31, 1G12 (1935). (2) F. Basolo, Chem. Reits., 62, 459 (1952). 13) D. Reese a n d C . H. Johnson, Trans. Faraday Soc., 31, 1632 (1935). ( 4 ) F. Rasolo, private communication. (5) w. W a h l , Ber., 60, 399 (1927). (fi) C;. .I. Burrows a n d K. H. Lander, J . Am. C h e r n . S o c . , 63, 3H00 (1931). (7) P. Neogi a n d N. K. D u t t , J . I n d i a n Chem. Soc.. 16, 83 (1938). ( 8 ) W. Thomas, J . Chem. Sac., 121, 196 (1922). (9) T. Moeller, yrivate communication.
NOTES
1332
remaining solution. The operation was performed a t 6-8' and not more than a minute elapsed between separation of the diastereoisomer and precipitation of the potassium salt. I n one experiment with KA1(C204)3a slight precipitate of the diastereoisomer separated from the filtrate whilst the filtration was proceeding. This could be due to a slight optical activity being lost through a secondorder asymmetric transformation. With both the iron and aluminum complexes, though an excess of resolving agent over the theoretical 1 mole of [Ni phenr]12 to 2 moles of racemic potassium salt was used, practically all was consumed. This is consistent with failure to effect a resolution being due to optical lability. Similarly no evidence was obtained for the resolution of the tris-oxalatogallate ion. The diastereoisomer was never obtained pure, and the analytical results suggest contamination with [Ni(phen) ] Cz04. Johnson' suggested that the magnetic moment of the tris-oxalatoruthenate should be indicative of whether or not resolution was possible. The moment (2.01 B.M.) indicates possibly covalent bonds. However, since Hund's rule does not necessarily apply to the second and third transitional series, the reliability of the magnetic criterion of bond type is questionable. Attempted resolution by Charonnat'O through the strychnine and quinine salts and by the method" of "active racemates" was unsuccessful. Attempts a t resolution through the N i ( ~ h e n ) ~ +salt + also were fruitless. Rapid racemization could occur through the momentary attainment of a 7- or %covalent structure containing aquo groups.. This is quite feasible with the higher transitional elements. Experimental Specimens of potassium tris-oxalatoaluminate, -ferrate(TII), -cobaltate(III) and -chromate(III), kindly supplied I)v D r . Basolo, were crystallized from warm aqueous solution by the addition of alcohol and dried over calcium chloride in the dark. Potassium tris-oxalatogallate and rhodate(II1) were prepared as d e s ~ r i b e d . ~d,~~ and I-trisI ,IO-phenanthroline-nickel (11) iodides were obtained as described previously.l3 Potassium Tris-oxalatoruthenate( 111).-The method of Charonnat,IO operating with potassium pentachloroaquoruthenate(III), was found to yield a product difficult to free from potassium chloride. The substance was best prepared by refluxing a mixture of ammonium pentabromohydroxyruthenate(1V) (2.4 g., 1 mole), oxalic acid dihydrate (1.5 g., 3 moles) and 30% formaldehyde (2 ml.) for 0.75 hr. and then adding gradually over a further 0.5 hr. potassium hicarbonnte (2.0 g., 5 moles). The refluxing was continued until the color had changed from red to olive green. On protracted heating, the color becomes brown due to the formation of the tris-oxalatoruthenate(I1)ion, but this easily reoxidixcs in the air. After filtration, the warm solution was Precipitated by the addition of alcohol. The green solid was crystallized from warm water containing drops of glacial acetic acid and a little potassium acetate, by the addition of alcohol. Cdcd. for K3Ru(Cz0d)3.3Hz0: Ru, 18.5. Found: Ru, 18.7. Potassium d- and I-Tris-oxalatocobaltate ( III).-dI-K3co(G204)3(0.7 g.) was dissolved in water (14 ml.) with potnsslum acetate (1 g.) and cooled i n ire. AII ice-cold solution of I-[Ni(phen)3]Iz(1 g. i n 12 ml. HzO 12 ml. alcohol) was
+
( I O ) R. Cliaronnst, Compl. rend., 178, 00 (1924). ( 1 1 ) XI. DelPiJine. Bull. SOC. chim., 1, 1256 (1934). (13) E . Leidiv. A n n . chim., In] 17, 307 (1913). ( 1 3 ) I". P. n n v r r and E. C. Ciyarfas, J . Proc. Ro!/. S o r . N.S.II'., 83, 2 3 2 (i!l.->fl).
..
..
-.
Vol. 60
added and the sides of the flask scratched to facilitate the deposition of the bright green diastereokomer The solution was quickly a t e r e d and cold alcohol added until the I-K3C0(CzO4)3 had precipitated. The diastereoisomer and the active salt were washed with cold absolute alcohol and dried in a vacuum desiccator. The optical isomer may be recrystallized from a little cold water and cold alcohol. The diastereoisomer was suspended in cold water (20 ml.) and potassium iodide (1 g.), acetic acid ( 2 drops, 17 N ) and hydrogen peroxide ( 2 ml., 3%) added to the suspension and well shaken. The mixture was cooled for 2-3 minutes, filtered and the d-K&o( C204)3precipitated from the filtrate with cold absolute alcohol. In 0.016% aqueous solution i t gave [ a ] %f4050°, ~ and in 0.04y0 solution [a]ms481 f1375'. These are considerably higher than those reported previously: Delt5pine,I1 [a] D 1750"; Jaeger," 250"; Johnson and Mead,I6