reaction to that employed in the preparation of (298), this can be accounted for by either structure A or structure B. However, it reacted rapidly with nitrosobenzene to give the anil oxide (242),
picric
acid
(298)H 4-NH
HN
NH
\__y (299) (242)114 -
character on the 9-carbon of the fluorene nucleus, and this fact alone indicates that there must be a contribution from the ylide structure (298B). Reaction with nitrosobenzene to give an anil oxide is a feature displayed by many nitrogen ylides (see
introduction).
The chemical behaviour of (298) in other respects was more complicated. It was thermally labile and underwent extensive decomposition on storage at room temperature for a week, even under nitrogen, although it could be kept for longer periods at -40°. Pluorenone imine (300) appeared to be one of the breakdown products, presumably arising from loss of the stable
4-NH
A
H N N H
Hf^ ~ ^ N H (300)NH _
HN
N(3oi) y KHN
NH
\
/
(302) C O ^ OHN
dihydroimidazole ring (301). The compound (298) was recovered unch^ged after 6 hr. in boiling methanolic sodium hydroxide, showing that it was resistant to hydrolysis, but on attempted recrystallisation from hot ethanol it yielded a substantial amount of an unidentified solid, which showed a shift to long wavelength in the ultra-violet spectrum (X max 310 nm.) and a much lower
115 -
mass spectrum at m/e 190 (C^^H^N), the structure (302) was
4
Ï
• I
melting point. Largely on the basis of an intense peak in the { consideredr which might arise by a Sommelet rearrangement of the
ylide form (298B) as shown to give (302). Oxidative degradation S of the rearranged product gave only fluorenone, however, whereas
(302) might have been expected to give fluorenone-l-carboxylic acid (303), which is known to be stable under the conditions of the oxidation. The problem has not yet been resolved.
Finally, the free base (298) was found to react rapidly with g-nitrobenzaldehyde at room temperature, but the product was too labile to be characterised fully. Ethylene guanidine did not appear to be lost in the reaction, however, as might have been expected from a Wittig-type mechanism, and in this respect the reaction of (298) is comparable to that of other nitrogen ylides, for example pyridinium cyanomethylide ^
- 115 -
§8. REACTIONS OF DIAZOFLUORENE WITH. GUANIDINES 1) With Arylguanidines
Because of the fact that N-fluorenyl-N',N"-diaryIguanidines, e.g. (304), and related compounds were not available by the methods described in §6, it was decided to investigate the possibility that they might be prepared by the thermal decomposition of 9-diazo- fluorene in the presence of the appropriate guanidine, since it
66,67'
has been shown that other ylides can be prepared in this
manner, for example pyridinium tetraphenylcyclopentadienylide (305) and, especially, numerous tetraphenylcyclopentadienylides containing aryl substituted heteroatoms of groups V and VI. One feature noted
N H ^ (304)
0
(305) r 3
(306) with 9-diazofluorene is the tendency to form azines. Thus with
rriphenylphosphine, 9-diazofluorene gives fluorenylidenetriphenyl- phosphinazine (306) in high yield It has also been noted
that if the decomposition reactions are carried out in the
presence of a catalyst, for instance copper bronze, then in many cases improved yields of the ylides are obtainable. This is believed to arise as a result of interaction between the filled d orbitals of the transition metal and the vacant p orbital ofz
117
the carbene which is generated from tlie diazo compound, and. gives stability to the carbene which is often described as a 'carbenoid' species. This is also believed to enhance the ability of the carbene to react with the desired substrate, namely the donor
atom of group V and VI, which may also undergo some form of complex
formation with the transition metal and further assist the
formation of the ylide, although the latter effect is believed to be less'likely with nitrogen where the d orbitals are not of a suitable energy to interact.
When 9~diazofluorene was decomposed in molten N,N'-diphenyl-. guanidine without copper bronze, a high yield of fluorenone
ketazine was obtained. The mechanism is believed to be similar to that already described for diphenyl sulphide with 9-diazo-
fluorene (see introduction, sect. 4), where the ylide, as soon as it is formed, attacks another molecule of diazo compound to give the observed product. However, when 9~diazofluorene was decomposed under similar conditions but in the presence of copper bronze, a compound with the correct composition for the ylide (307) was
obtained, along with lesser amounts of fluorenone ketazine and bifluorenylidene, the latter probably arising from slight
decomposition of the product (307). The intermediacy of (307) in the formation of the ketazine was demonstrated since the latter was formed when (307) and 9-diazofluorene were mixed and kept at room temperature for several weeks.
Owing to the low yield of the product (307) obtained and the difficulty of separation from by-products, a more practical route
118 -
for its preparation was sought. The reaction of carbodi-imides