BARKER, COLLMAN, AND MARSHALL
3216
The analytical sample contained approximately 0.5% palladium as determined by spectral analysis.32 Countercurrent (32) Spectral analyses were b y Spectra-Chemical Research Laboratories, Inc., Chicago, Ill.
VOL.29
distribution in n-butylalcohol-acetic acid-isopropyl acetatewater, 10: 10: 10:27 (400 transfers), gave a slightly skewed curve ( K = 0.43) indicating a decrease in partition coefficient with increasing SOhIte Concentration. The reisolated material contained less than 0.05% palladium.
Reactions of Metal Chelates. VII. Dimethylaminomethylation and Chloromethylation of Metal Acetylacetonates' ROBERT H. BARKER,^^ JAMES P. COLLMAN, A N D ROGER L. MARSH ALL^^ Department of Chemistry, University o j North Carolina, Chapel Hill, North Carolina Received July 15, 1963 Several of the pseudo-aromatic trivalent metal acetylacetonates have been found to undergo a modified Mannich-type condensation. The resulting triamines are readily quaternixed by treatment with methyl iodide t o produce salts which are amenable to attack by some nucleophiles. This provides a new route t o reactive molecular chelates and constitutes the first reported instance of the direct introduction of an uncoordinated amino function into such a species. A method has also been devised for the synthesis of the analogous monoamine in the chromium(II1) series by use of a chelate substrate having two of its three active sites blocked by inert functions. After quaternization, this monofunctional analog seemed to be more reactive toward nucleophiles than the corresponding triamine. Another type of reactive molecular chelate was produced when the parent acetylacetonates were subjected t o chloromethylation conditions. I n this case the intermediate, presumably the expected chloromethyl chelate, could not be isolated except in the form of derivatives. All attempts to isolate the chloromethyl chelate itself resulted only in polymeric products. As in the case of the chelate amines, the best results were obtained by studying the monofunctional species.
We previously reported that metal acetylacetonate rings undergo a number of substitution reactions which are usually associated with reactive aromatic syst e m ~ . ~ - It - ~ has now been found that these chelates are also susceptible to a modified llannich-type condensation. All three of the active sites in the chelate molecule have been substituted by dimethylaminomethyl functions to produce products with drastically altered properties. This provides a new route to reactive molecular chelates and constitutes the first reported instance of the direct introduction of an uncoordinated amino function into such a species.
'9
X=CHs-C--,
R
H-C-,Br,
C1, -NO2, -SCN
substituted chromiuni(II1) acetylacetonate (A) was allowed to react with paraformaldehyde and N,N,S',N'tetramethyldianiinomethane in glacial acetic acid solution. The product B, a dark, low-melting glass, exhibited a strong infrared absorption peak at 1565 cm.-l, but no other absorption in the 1500-1600-cn~-~region. This is as expected from the empirical rule of Drydeng who observed that a series of metal acetylacetonates bearing functions other than hydrogen on the central carbon atom of the ring had a single absorption peak in this region of their infrared spectra. However, in the compounds bearing a hydrogen on this position, characteristic doublets appeared in this region of the spectra. Also, a small peak in the 12OO-cni.-' region of the spectra of the unsubstituted chelates, usually associated with a wagging mode of the ring hydrogen, characteristically disappears upon substitution of the ring. No 1200-cm.-' peak could be detected in the spectrum of the Rlannich product.
Although the conventional Mannich conditions were unsuccessful in producing the desired amines, it was found that good results could be obtained from a niodification of the procedure which Lindsay and Hauser* had previously used in the ferrocene series. The un(1) ( a ) Initially reported a t t h e 144th National Meeting of t h e American Chemical Society, Los Angeles, Calif., April, 1963; ( b ) previous paper b y J . P. Collman and hl. Y a m a d a , J . Ore. Chem., 28, 3017 ( 1 9 6 3 ) ; (e) portions of this a o r k were supported b y t h e National Science Foundation, U. S. Army Research Office, G r a n t Number nA-ORD-n-31124-G-185, a n d the Petroleum Research F u n d ; (d) abstracted in p a r t from the P h . D . Dissertations of R . H. Barker and R . L. Marshall, University of North Carolina, 1962. (2) (a) National Science Foundation Predoctoral Fellow. 1961-1962; I)ep.uttnent of Chemistry, Tulane University. New Orleans. L a . ; ( b ) Petroleum Research F u n d Fellow, 1961-1962. ( 3 ) J . P. Collrnan, R . L. Marshall, W. L. Young, and S. D . Goldby. I n o r g . Chem., 1, 704 (1962). ( 4 ) .J. P. Collman. R . A . Moss. S. D. Goldby. a n d W . S. Trahanovsky. C h e m . I n d . ( L o n d o n ) . 1213 (1960). (5) J . P. Collrnan. R. A . LIoss. H. M a l t z , a n d C. C. Heindel. J. A m . Chem. Soc.. 83,3529 (1961). (6) ,J. P. Colltiian and E. T. Kittleman, ibid.. 83,631 (1961). (7) .J. P. Collman, R . P. Blair, R. L. Marshall and L. Slade, I n o r g . Chem.. 2 , 576 (1963). ( 8 ) .J. K . Lindsay and C. R . Hauser, J . O r g . Chem.. 2 2 , 355 (1957).
A
D
C
Also indicative of the assigned structure was the nuclear magnetic resonance spectrum of the analogous product in the cobalt(II1) series. This showed the presence of three types of hydrogen atonis with peaks at T 6.92, 7.72, and 7.85 and integrated intensities of (9) R . P. Dryden and A . Winston. J . Phys. Chem , 62, 635 (1958).
NOVEMBER, 1964
REACTIONS OF METALCHELATES. VI1
1 : 3 :3, the pattern expected for the proposed structure. A siniilar pattern was observed with the crude triamine in the rhodium(II1) series even though a pure sample of this material could not be obtained. I n this latter case the resonance spectrum also exhibited a low-field peak which was in the position expected for ring hydrogens. The intensity of this peak indicated that approximately 757” substitution had occurred after 24 hr. of reaction time. Thus the product in the rhodium(111) series would be a mixture of mono-, di-, and triamines. This is apparently another example of the reduced reactivity of the rhodiuni(II1) acetylacetonate toward electrophiles. The triamine B was extremely hygroscopic and soluble in both water and n-heptane. Because of this amazing solubility range, crystallization could not be effected. Thus it was necessary to convert this material to more easily characterizable derivatives in order to completely establish its structure. This was accomplished by treatment of the suspected amine with methyl iodide to produce a pale purple precipitate which was presumably the quaternary amnionium salt C. This material was soluble in water and aqueous alcohol but not in less polar organic solvents. However, no pure sample of C could be isolated, probably owing to the presence of sonie material in which all three amine functions had not been quaternized. Such occurrences are not unprecedented in work with polyfunctional compounds.’O Further treatment of this salt with aqueous potassium cyanide produced the trinitrile D which, after purification by chromatography on an alumina column, was considerably easier to handle. Attempts to convert this trinitrile into the corresponding triacid and triamide were unsuccessful. Treatment with hydrogen over platinum oxide and palladium hydroxide produced products which were soluble in organic solvents and dilute aqueous acid but insoluble in aqueous base. These materials resisted characterization, apparently because of the difficulty in obtaining pure products from such trifunctional reactants. Similar difficulties have been encountered in attempts to displace trimethylamine from the triquaternary salt C by a variety of nucleophiles other than cyanide Phenoxide appeared to react but the product mas an intractable mixture. That acetate, chloride, hydroxide. nitrite, azide, thiocyanate, and the acetylacetonate anion all apparently failed to attack the triquaternary salt under the conditions employed is shown by the fact that no evolution of trimethylamine could be detected and no organo-soluble products could be recovered. Attempts to carry out the chloride, azide, and thiocyanate displacements under more vigorous conditions resulted in destruction of the chelate. A similar set of displacement attempts was not made on the triquaternary salt in the cobalt(II1) series since a preliniinary experiment showed that even cyanide attack failed here owing to degradation of the chelate. Only sulfide displacenient was found to be effective in both series. This, however, led to an apparently polynieric product whose structure could not be unequivocally established. It was found that nionosubstituted chelates could be obtained from the diniethylaniinoniethylation of chro(10) R
W Kluiber. J
A m Chem S o c , 83, 3030 (1961)
3217
miuni(II1) acetylacetonate using a limited amount of the diamine intermediate, but the product was again difficult to handle and could not be obtained in a reasonable state of purity. A niuch better route to the chelate monoamines appeared to be by way of the dimethylaminoniethylation of a chelate having two of its three active sites blocked by some inert function. The preparative routes leading to partially chlorinated and nitrated trivalent metal acetylacetonates were known” and it seemed that these functions would be suitable for use as blocking groups since treatment of the trichlorinated and trinitrated compounds under Mannich conditions resulted only in the recovery of unreacted starting material. Because of their ready availability, the partially chlorinated chelates were used as substrates. Treatment of the dichlorinated compounds under diiiiethylaniinomethylation conditions produced a mixture of products which was partially resolvable by column chromatography into a crude amine and another compound to which the dimeric structure E has been assigned. The structure of E was assigned on the basis of its infrared spectruni and elemental analysis. The amine, after attempted purification by Chromatography and conversion to a crystalline quaternary ammonium iodide, was shown not to be a single pure compound. Elenien-
tal analysis showed that the product expected to be F contained approximately half of the chlorine and twice the nitrogen that would be predicted for this structure. That some of this quaternary salt was formed was shown by conversion of the crude methiodide into derivatives whose only logical precursor would be the salt F. Cyanide anion attacked the salt to produce the expected iiiononitrile G which was purified by chromatography (see p. 3218, top). The p-chlorothiophenoxide and sulfide ions acted in a similar manner to produce the corresponding displacement products H and J, also purified by column chromatography. I n addition, it was found that by use of a slightly different procedure than had been used in the triquaternary salt displacements, azide would effect a similar displacement of trimethylamine to give the structure I. It would appear that the mixture of amines and quaternary salts from the dichloro chelate must have occurred through a displacement of some of the ring chlorines by one or more of the electrophilic species present in the llannich reaction. Such a reaction was certainly unexpected, but does not appear quite so surprising in view of the recently reported displacement of ring chlorines during the nitration of partially chlo(11) (a) J P Collman a n d M Ysmada, J Ore Chem 18, 3017 ( 1 9 6 3 ) . (b) J P Collman, W L Young. 111, a n d R H Barker unpublished results
3218
VOL. 29
BARKER, COLLMAN, AND MARSHALL
J
G , Z’= -CN H, Z=-S-@CI
I, Z=-Na
rinated metal acetylacetonates.11s12 I n that case also the trichlorinated compound was unaffected by the nitration conditions. I n a siniilar manner, a crude diamine was prepared in the chroniiuni(II1) series by the diniethylaniinomethylation of the monochlorinated chelate. Conversion of the crude diamine into the diquaternary salt followed by treatment with potassium cyanide produced the expected dinitrile K. I n the same way, sulfide attacked the diquaternary salt to produce a inaterial that apparently was a low molecular weight polymer.
tonate. This product closely resembled the polymer obtained from the action of paraformaldehyde on the chelate in acetic acid ~ o l u t i o n . ’ ~ Although attempts to trap the proposed reactive intermediate with several nucleophiles failed when the unsubstituted acetylacetonate was used as the substrate, this was accomplished when the substrate was the dichlorinated chelate. A large excess of iiionochloroniethyl ether and the boron trifluoride complex was used to prepare the reactive chloroniethyl coiiipound L which was not isolated but allowed to react further with such nucleophiles as absolute ethanol or methanol. Froni such treatnient the ethers A4 and K were isolated. Thus, while the formation of these products does not conclusively prove the proposed chloroniethylation and ionization mechanism, it is a t least compatible with it.
K Other low molecular weight polymers have been obtained in attempts to chloroniethylate the chelate rings by reaction with monochloromethyl ether under the influence of boron trifluoride diethyl etherate. Treatment of rhodiuni(II1) acetylacetonate under these conditions resulted not in the expected chloroniethyl chelate but rather in an amorphous, yellow solid which was apparently polymeric in nature and which was shown to contain less than 0.2% chlorine. It s e e m probable that the initially produced chloroiiiethyl chelate ionized in an s N 1 nianner, becoining an electrophile itself which then attacked other unsubstituted sites in adjacent rings to produce a branched polynier. Siniilar results were obtained from the reaction of monochloromethyl ether with chroniium(II1) acetylace-
L
M, R = CzH5N, R=CH3(12) J. P. Collman, W . L. Young, 111, R . H. Barker, a n d M. Yainada, Abstracts of Papers, 142nd National Meeting of the American Chemical Society, Atlantic C i t y , N. J.. 1962.
Experimental Dimethylaminomethylation of Tris(2,4-pentanediono)chromium(III).-To a mixture of 11.0 g. (31 mmoles) of tris(2,4-pentanediono)chromium(III)and 4.0 g. (0.12 mole) of powdered paraformaldehyde in 200 ml. of glacial acetic acid was added 12.5 ml. (0.12 mole) of N,N,N’,?;’-tetramethyldiarninomethane, prepared according to Lindsay and IIauser.8 The resulting mixture was stirred overnight a t room temperature. The acid was neutralized uith sodium carbonate and an excess of ammonium hydroxide was added to make the solution basic. Extraction of the aqueous solution with chloroform, followed by evaporation of the chloroform solution, produced the triamine B as a purple tar. This tar could be converted into a very hygroscopic glass by drying under vacuum for several hours. The crude triamine was purified by chromatography on a 6-in. column of alumina using 50:50 (v./v.) benzene-chloroform as the eluent. Owing to difficulties in handling the product, only nitrogen and chromium analyses were obtained. Anal. Calcd. for C24H42CrN306: Cr, 10.0; ?;, 8.07. Found: Cr, 10.56; S , 7.8’7. The samples for the chromium and nitrogen analyses were prepared separately. A more pure product could be prepared in lower yield by use of a n excess of the diaminomethane in the absence of paraformaldehyde. Dimethylaminomethylation of Tris(2,4-pentanediono)cobalt(III).-This was brought about in a manner very similar to that used for the chromium compound. The product was a green tar which could also be dried to a hygroscopic glass. As before, the triamine was partially purified by chromatography, but no sample of analytical purity could be obtained. Dimethylaminomethylation of Tris(2,4-pentanediono)rhodium(III).-Again the same procedure was used except that the rea(
