2608
J. Org. Chem., Vol. 40, No. 18, 1975
Lwowski, de Mauriac, Thompson, Wilde, and Chen
Curtius and Lossen Rearrangements. 111. Photolysis of Certain Carbamoyl Azides Walter Lwowski,* Richard A. de Mauriac,l and Margaret Thompson Department of Chemistry, New Mexico State University, Las Cruces, New Mexico 88003 Richard E. Wilde* and Sis-Yu Chen Department of Chemistry, Texas Tech University, Lubbock, Texas 79409
Recieved April 24, 1975 The photo-Curtius rearrangement of several substituted carbamoyl azides gives aminoisocyanates, PhHNNCO, EtHN-NCO, Me2N-NC0, and EtzN-NCO. These can be trapped by nucleophiles. Neon-matrix isolated photolysis of dimethylcarbamoyl azide gave N,N-dimethylaminoisocyanate, whose ir spectrum was recorded and whose photodecomposition was observed. Products attributable to the carbamoylnitrenes, RR’NCON, isomers of the aminoisocyanates mentioned, were’not found. The familiar 1,2 migrations of carbon moieties from a carbon to a heteroatom often have counterparts in which a heteroatom migrates: RX-C-Y -+ RX-Y-C. The thermal Curtius rearrangement (migration of a carbon moiety from C to N) is c o n ~ e r t e d , migration ~-~ of the carbon moiety and loss of nitrogen occurring simultaneously. This is also true for the photoinduced Curtius Heteroatom migrations analogous to the Curtius rearrangement compete with alternate reactions such as the concerted displacement of nitrogen not by the heteroatom X (path a, Curtius rearrangement), but by another part of the group attached to the carbonyl (path b). Another competing reaction is the breaking of the N,-N6 bond without assistance, resulting in a nitrene (path c). 0
I
7\
R-X-----=TN-N~
s-c-0
I
1-4
(-3
-
-I
R-X-N=C=O
I
+Nz
0
I1 .. (-4 R--X-C-N-NL
path a
x-c=o RWN
R-N-N,
+ N,
path b R =phenyl)
(e.g,
0 -+
R-X-C-N
1I
+ N,
path c
(charges and valence-balancing reactions omitted)
Examples for all three paths are provided by decompositions of diaryl- and arylalkylcarbamoyl azide^.^>^ For example, the thermolysis of diphenylcarbamoyl azide gives N,N-diphenylaminoisocyanateand 1-phenylbenzimidazole by concerted paths a and b, while photolysis leads to the nitrene PhZNCON, which then cyclize^.^-^ We have communicated earlierlO the Curtius rearrangement in the photolyses of PhNHCON3, EtNHCON3, and EtzNCON3, both in proticlo and in aprotic media.ll The intermediacy of aminoisocyanates RR’N-N=C=O was inferred from the isolated end products, which are also formed when N,N-dialkylaminoisocyanates are generated thermally in the presence of the appropriate reactants.12 Some questions remain. Are carbamoylnitrenes formed together with the aminoisocyanates, as one might expect in view of the nitrene formation from diarylcarbamoyl azide^?^^^ Can the assumption of intermediate aminoisocyanates be supported spectroscopically? The present paper addresses itself to these questions.
Results Phenylcarbamoyl azide (1) was photolyzed with 254-nm light in benzene, cyclohexene, and cyclohexane solutions
H
I
I
3
4
H 2
UNHCOFHPh 5 and a search for the expected nitrene products 2-5 was made. Although the compounds 3 and 4 were prepared independently and could have been easily detected, none of the compounds 2-5 was found in the photolysis reaction mixture. What we did obtain was a complex mixture which seems to arise from dimerization of PhNHNCO, followed by tautomerizations and perhaps subsequent photoreactions. We have not studied these products further. The infrared spectra of the total reaction mixtures from the photolyses in benzene, cyclohexene, and cyclohexane were practically identical, indicating that solvent had not been incorporated in the products. In contrast to these photolyses, clean reactions were observed in methanol solution, where a 65% yield of methyl 2-phenylhydrazinecarboxylate (6) was obtained. This indicates the formation of PhNHNCO, followed by addition of methanol to give PhNHNHCOOCH3 (6). Photolysis of N-ethylcarbamoyl azide (7) in cyclohexene did not give any apparent nitrene products. Addition of EtNHCON to the double bond would give 74N-ethylcarbamoyl)-7-azabicyclo[4.1.0]heptane (8), which was prepared independently. It could have been detected easily in the photolysis reaction mixture. The amorphous product actually obtained from the photolysis in cyclohexene was very similar to that obtained by photolyzing 7 in benzeneboth photolyses did not lead to incorporation of solvent into the product mixtures, which were not studied further. Photolysis in methanol gave a 50%(not maximized) yield of methyl 2-ethylhydrazinecarboxylate (9), the apparent’ product of methanol addition to EtNHNCO. Photolysis of diethylcarbamoyl azide (10) did not give N-cyclohexyl-N’N’-diethylurea( l l ) , the product expected from C-H insertion of EtZNCON. Instead, a 18% yield of a dimer 12 of EtzN-NCO was isolated. Thermally generated EtzN-NCO gives the same dimer.12 Its structure, 1,l-diethyl-4-diethylamino-1,2,4-triazolidin-3,5-dion-l,2-aminimide, has been discussed elsewhere.ll Photolysis of 10 in
J . Org. Chem., Vol. 40, No. 18,1975 2609
Photolysis of Certain Carbamoyl Azides Table 1 Wave Numbers and Assignments of New Bands Observed after Photolysis of (CH3)2NCON3 in a Neon Matrix Cm-l
455 555 760 957 1025 1445 1456 1472 1479 1599 1608 2141 2230 2269 2899 2957 2993
5w w 5w 2 Wb 5wb 2ma i 2 ma i 2 ma i 2 ma i 2vw i 2 vw i 2 s" i 2 sb i 2 m-sa i 3 ma 5 ma i 3 ma
Assignment
*
(CH3)2"CO (CH3)z"CO
v, CH, def vgbCH, def us, CH, def v2--N=C s t r
-N=C=O str v 1 -N=C=O str v6
v1 C-H str v 5 C-H str
Cm -1
Species
i i i
* *
Table II Wave Numbers and Assignments of New Bands Observed after Photolysis of (CD3)2NCON3 in a Neon Matrix
CH,NC CH3NC CH3NC CH3NC
CH3NC (C H,),NNCO
HNCO CH3NC CH,NC CH3NC
600 1080 1115 1612 1640 1680 1830 2091 2141 2226 2241 2255 2268 2350
Species
Assionment
*
5w 5 w-m i 5wb 5 wb i 5 w' i 5w" i 5W" i 2 w-m i 2 sa f 2 m-so i 2 sb i 2 ma 2 ma 2 m-sa
f
(CD,),NNCO (CD,)@"NO
*
*
V3+S
v6+8
2v4
v2 -NSC str v 2 -N=C=O Str -N=C=O str vi C-D s t r v 5 C-D s t r vl N-D s t r
CD3NC CD3NC CDSNC
CD3NC
DNCO (CD,),NNCO C D 3NC CD,NC DNCO
* Increases in intensity upon prolonged photolysis. in intensity upon prolonged photolysis.
Decreases
late of 1,l-diethylhydrazine (by hydrolysis and decarboxylation of EtzN-NCO) in 19% yield. in intensity upon prolonged photolysis. The photolysis of N,N-dimethylcarbamoyl azide (17) in methanol solution gave a 32% yield of methyl 2,2-dimethylhydrazinecarboxylate (18) when light of 254-nm wavecyclohexene did not give the aziridine expected from the length was used. A 46% yield of 18 was obtained when we addition of EtzNCON to the double bond. The product used fluorescent uv lamps which emit most of their light mixture obtained was remarkably similar to that produced around 300 nm (RPR 3000 lamps of The Southern New by the photolysis in cyclohexane, but a 15% yield of 3,3'England Ultraviolet Co.). biscyc1ohexen:yl and some diethylurea (13) were also found. Matrix Isolation Studies. Physical evidence for the naThe last two products might well have been formed by ture of the primary photolysis product of N,N-dimethylmechanisms analogous to those found with triplet excited carbamoyl azide was obtained by matrix-isolation photolyethyl azidoformate;13 thus they furnish no proof for the intervention of a nitrene. Photolyses of 10 in solvents of difsis and infrared spectroscopy. Both (CH312NCON3 and (CD&NCON3 were photolyzed in neon matrices at 6 K ferent polarity, ether and acetonitrile, again gave no nitrene addition or insertion products, nor did we find the and the reactions monitored by ir spectroscopy. The wave knawn l-ethylimidazolidin-2-0ne,'~expected from intranumbers of the bands which appear and disappear during molecular C-€I insertion of the nitrene (H3C-CHz)zNCON. times up to 390 min are listed in Tables I and 11. After 15 The dimer 12 was obtained in 22% yield in ether solution min of photolysis, all the bands in the matrix-isolated specand in 14% yield in acetonitrile. Photolysis of 10 in methatrum of (CH&NCON3 showed a marked decrease in intennol, however, gave methyl 2,2-diethylhydrazinecarboxylate sity. However, in the 2230-cm-l region there appeared a very intense new band, as seen in Figure 1, trace b. Upon (14) in 78% crude yield (57% after purification to an undepressed mixture melting point). Low concentrations of continued photolysis, this band decreased in intensity. Based on the chemical studies discussed above, and the remethanol (or of diethylamine, see below) are sufficient to intercept the EtZN-NCO. Photolysis of a cyclohexane solusults of photolysis of methyl azidoformate in a neon mation 0.05 M in the azide 10 and 0.05 M in methanol gave a trix,I5 it is reasonable to assign the 2230-cm-l band to 41% yield of 14 (by weight after sublimation). In another (CH&N-NCO. Two weak bands at 957 and 1025 cm-1 also decreased in intensity upon prolonged photolysis and experiment, 9.73 mmol of 10 in 200 ml of cyclohexane (a 0.05 M solution) was irradiated until ca. 69% of the theorettherefore have been assigned to the (CH&N-NCO moleical yield of nitrogen had been evolved. The lamps were cule. turned off, arid 5.12 mmol of methanol was then injected As shown in Figure 1, upon photolysis a new band apinto the stirred solution. By VPC, the yield of 14 was found peared at 2141 cm-l and continued to increase in intensity to be 0.8%. Apparently, the photolysis produces an interupon prolonged photolysis, as did the band a t 2269 cm-'. It mediate that reacts readily even with low concentrations of is well known15 that the isocyanate asymmetric stretching methanol to give 14, but which reacts via a slower path in vibration of isocyanic acid occurs near 2270 cm-l. The the absence of methanol. The contention that this intermeband at 2141 cm-l is in the region of the -N=C stretching diate is EtZN-NCO is strengthened by the irradiation of vibration. In view of the data in Table I and the known16 ir cyclohexane solution, 0.25 M both in diethylamine and 10. spectrum of HsCNC I1410 (m), 1459 (m), 2166 (vs), 2966 The adduct of diethylamine to the presumed N,N-dieth(vs), and 3014cm-1 (vs)],we assign the 2141-cm-1 band to ylaminoisocyanate, N,N,2,2-tetraethylhydrazinecarboxam- H3CNC. The interaction between HNCO and H3C-CN ide, EtzN-NHCONEtz (151, was isolated in 53% yield and trapped in the matrix is responsible for the 25-cm-l loweridentified by comparison with an authentic sample. Irraing of the isocyanide stretching frequency. The C-H diation of 10 in pure tert-butyl alcohol gave a 20-40% yield stretching frequencies are also lowered somewhat by the of tert -butyl 2,2-diethylhydrazinecarboxylate(16). Irradiamatrix environment. The CH3 deformation modes listed in tion of 25.5 mmol of 10 in 500 ml of water and isolation of Table I appear in the 1400-1500-~m-~ region. These bands the oxalic acid salts gave that of diethylamine (by hydrolyincrease in intensity upon prolonged photolysis, and can be sis and decarboxylation from 10) in 35% yield, and the oxaassigned to the H3CNC molecule. The bands are surpris-
* Increases in intensity upon prolonged photolysis. b Decreases
2610
J. Org. Chem., Vol. 40,No. 18, 1975
Lwowski, de Mauriac, Thompson, Wilde, and Chen anic acid-methyl isocyanide molecule pair. The methyl isocyanide might arise from N-methylenemethanamine, which could be formed as an intermediate by @-elimination from the aminoisocyanate: HCHzNMeNCO HzC=NMe HNCO. However, no data on the photochemistry of H2C=NMe seem to be available, neither for fluid solutions nor for isolated molecules (in the gas phase or in a matrix).
+
, 2280
,I, ,
2240
2200 1 cm
,
, 2160
,
, 212C
Figure 1. Infrared spectra of (CH3)zNCONs in Ne at 6 K (a) before photolysis, (b) after 15 min photolysis, and (c) after 45 min
photolysis. ingly blue shifted with respect to the gas-phase values. However, no matrix-isolation work has been done on methyl isocyanide, and earlier gas-phase work17 reported these frequencies a t 1429 and 1467 cm-l. We believe that these assignments are reasonable. Some very weak bands have not been assigned. They may arise from the isocyanic acidmethyl isocyanide molecule pair, from dimethylaminoisocyanate, or from the products of some other decomposition occuring to a small extent. In the region below 2000 cm-l the absorption bands of the deuterio species are considerably weaker than those of their protic counterparts. Above 2000 cm-l the isocyanate asymmetric stretching vibration of DNCO is observed at 2226 cm-l in agreement with our previous study.15 The hydrogen-bonded N-D stretching vibration appears as a broad band at 2350 cm-l, and the C-D stretching frequencies of D3CNC are clearly observed at 2255 and 2268 cm-l. The -N=C stretching frequency remains practically unchanged in the deuteriomethyl isocyanide. The only anomaly seems to be the (DBC)~N-NCOisocyanate stretching frequency, which is 11 cm-l higher than that of (H3C)zNNCO. However, the azide stretching frequencies of the parent dimethylcarbamoyl azides are themselves somewhat anomalous, the matrix isolation spectrum of (HsC)2NCON3 exhibiting a single band at 2162 cm-l, while the spectrum of (D3C)2NCON3shows a doublet at 2167 and 2161 cm-l. We have no explanation for the higher frequencies of the deuterated species. In conclusion, the ir matrix-isolation studies argue strongly for an intermediate dimethylaminoisocyanate which decomposes upon prolonged photolysis to an isocy-
-
Discussion The results presented show that photolysis of alkylcarbamoyl azides (and of monophenylcarbamoyl azide) leads to intermediates identified by their chemistry and by a matrix-isolated ir spectrum as substituted aminoisocyanates RR’N-NCO. If the corresponding nitrenes RR’NCON are formed a t all, they must be unreactive toward C=C double bonds, C-H bonds, and benzene rings-a behavior not in accord with the properties of diaryl- and arylalkylcarbamoylnitrenes studies by Kametanis and Anselme.9 An earlier suggestionlo that hydrogen bonding favors the photoCurtius rearrangement of carbamoyl azides has been elegantly substantiated by Anselme.9 Photolysis of diphenylcarbamoyl azide in tetrahydrofuran solution gave a 84% yield of nitrene products, while in protic solvents both nitrene and rearrangement products were formed (e.g., in ethanol 54% nitrene and 22% rearrangement products). Kametanis photolyzed arylalkylcarbamoyl azides in tetrahydrofuran and obtained only nitrene products (e.g., in 44% yield from anisylbenzylcarbamoyl azide). Our chances for finding nitrene formation should therefore be best in aprotic media. Using a variety of these, we were still unable to detect any nitrene products. Thus arylalkylcarbamoyl azides seem to form the borderline between nitrene-forming, “rigid”, and rearranging carbamoyl azides, as far as the photolytic decomposition is concerned. The reasons for the change of reaction course with structure are not clear. Diarylcarbamoyl azides give nitrenes and do not rearrange in aprotic solvents, and give mixed reaction paths in protic solvents. Arylalkylcarbamoyl azides still give nitrenes in aprotic solvents, and it might well be that a t least some of them will rearrange when photolyzed in protic media. The azides studied here did not give detectable amounts of nitrenes, not even in various aprotic solvents. Given the detectability of l%yields of nitrene products, a difference of 2 kcal/mol in free energy of activation is all that is needed for an apparent change of reaction course. Nitrene stability and migratory aptitudes might be reflected in the respective transition states, but many other factors could be responsible as well. Unsubstituted carbamoyl azide, HtHCONs, seems to give H-0 insertion products of H2NCON upon photolysis in alcohols.18 Photolysis of the aminoisocyanates once formed from the azides occurs in the neon matrix, and most likely in fluid solutions as well. This should be less important in nucleophilic solvents, in which the aminoisocyanates are quickly intercepted. Nevertheless, it might contribute to the lesser yield of 17 obtained when 254-nm rather than 300-nm light was used. In hydrocarbon solutions, considerable photolysis of the aminoisocyanates seems likely: This might cause the formation of the intractable viscous materials formed. Photodecomposition products could react with carbamoyl azides, giving rise to their radical-chain decomposition, or do other chemical mischief. The photolysis of substituted carbamoyl azides is a good route to 2-substituted hydrazinecarboxylic esters, which are sometimes difficult to obtain, free of isomers, by acylating alkylhydrazines.lg The thermolysis of carbamoyl azides in alcohols and amines is synthetically valuable,20J1 but the temperatures required sometimes results in nucleophil-
Photolysis of Certain Carbamoyl Azides ic displacement of azide ion, or in the decomposition of the desired product.
Experimental Section Photolyses. For the matrix-isolation studies, dimethylcarbamoyl azidez2 was kept below 0' and protected from light until used. It was then introduced into a high-vacuum system and mixed with neon at a concentration of 0.2 mol % a t least 24 hr before deposition. The experimental procedure has been described previously.l5 Photolyses in fluid solution were carried out in fused silica vessels in a Rayonetz3 photochemical reactor, using low-pressure mercury lamps (emitting most of their light at 254 nm) unless otherwise specified. Phenylcarbamoyl azidez4 (1) was purified by sublimation a t 80-90° (760 mmHg), mp 107-108O (lit.20mp 103-104'). Ethylcarbamoyl azide25 (7) was distilled, bp 90' (17 mmHg), ir (CCl4) N3 a t 2150, C=O at 1700 cm-l. Diethylcarbamoyl Azide (10). Freshly distilled diethylcarbamoyl chloride (14.2 g, 117 mmol) in 150 ml of acetone and sodium azide (20 g, 300 mmol) were stirred and heated to reflux for 18.5 hr. The solution was cooled, filtered, and concentrated in vacuo. Distillation over a Vigreux column (5 in.) gave 12.3 g (74%yield) of 10, bp 47-49' (2 mm). Redistillation gave pure 10: nZ2'D 1.4625; bp 53-53.5' (3 mm);26ir spectrum ( C c 4 ) 2980 (s), 2840 (s), 2778 cm-l (w). The azide and carbonyl bands were solvent dependent: in CCL 2155 (s), 2140 (sh), and 1690 cm-'; in cyclohexane 2156 (s), 2146 (sh), and 1693 cm-l; in methanol 2155 (s), 2145 (sh), and 1665 cm-1. The uv absorption lacked maxima between 230 (cyclohexane, log e 3.63) and 260 nm (cyclohexane, log c 2.82). At 255 nm, log e in MeOH was 2.74, in cyclohexane 2.80. The NMR spectrum showed NCHz as a multiplet at 6 2.3 (hindered N-CO rotation) and CH3 at 6 1.12 (t). Heating the azide to 40' in the NMR probe led to coalescence of the NCH:, multiplet to a quartet, J = 7Hz, unchanged to 120'. The azide seems indefinitely stable when stored in a brown bottle a t room temperature. Dimethylcarbamoyl azide ( 17),22bp 53-55' (5 mm), showed two methyl signals of equal area in its NMR spectrum (22', CDCb), at 6 2.920 and 2.967, indicating a barrier to free rotation around the N-CO bond. Dimethylcarbamoyl a 2 i d e - d ~ ~was ' prepared from hexadeuteriodimethylammonium chloride, (D3C)zNHz" -C1 (EM Laboratories, 99+% dieuterated), 3.9 g (45 mmol) of which was converted to the free amine with 4 g of NaOH in 8.5 ml of water and 50 ml of dichloromethane. After 15 min of stirring, the dichloromethane phase was separated, dried over KOH, and added over 45 min a t -15' to 12 g of phosgene in 30 ml of dichloromethane. After removal of the excess of COClz and concentrating in vacuo the dichloromethane solution to one-fourth of its original volume, ether was added to precipitate 1.5 g of (D&)zNHz+ -C1. Evaporating the filtrate left 2.32 g of (D3C)zNCOCl. This was dissolved in 3 ml of acetonitrile and added to 1.4 g of NaN3 suspended in 4 ml of acetonitrile. After 2 hr of stirring, filtration, and distillation, the hexadeuterated azide was obtained in an average yield of 65%. The ir spectrum (neat,) shows the N3 band a t 2155 cm-', C-D stretching vibrations a t 2240 and 2215 cm-l, C=O a t 1685 cm-', and other strong bands ai, 1395,1256,1229,1016, and 712 cm-'. Photolysis of 2 g of dimethylcarbamoyl azide (17)in 30 ml of methanol was done in a Rayonet reactor a t Oo until about 80% of the calculated volume of nitrogqn had been evolved. Gas chromatography on a !1O-ft UCON Polar column a t 140° gave methyl 2,2dimethylhydrazinecarboxylate (18). The yield was 32% when 254-nm light was used, but 46% when fluorescent uv lamps with a peak output a t 300 nm (Rayonet RPR 3000 lamps) were employed. The compoundz6 of mp 43' had NMR signals (CDC13) a t 6 3.70 (s, 3 H) and 2.57 (s, 6 H) and ir bands a t 3270 and 1710 emi1 (in cc14).The mass spectrum showed P 118 (28%) and a base peak a t 58 (loo%, Me&&) as well as the expected other fragmentations. Methyl 2,2-dieth~lh~drazinecarboxylate ( 14)26was obtained by photolyzing (254 nm) 3.47 g (24.4 mmol) of diethylcarbamoyl azide (10)in 500 ml of degassed spectrograde methanol until 80% of the calculated volume of Nz had been evolved. The residue left after evaporating the solvent was a off-white solid, mp 85-95' (2.23 g). Sublimation a t 80-90' (760 mm) gave white crystals, mp 108--109', identical in all respects with the authentic material (see below)
