Download PDF
Synthesis 2016; 48(15): 2361-2375
DOI: 10.1055/s-0035-1561454
short review
© Georg Thieme Verlag Stuttgart · New York

The Chemistry of Unusually Functionalized Azides

Klaus Banert*
Technische Universität Chemnitz, Organic Chemistry, Strasse der Nationen 62, 09111 Chemnitz, Germany   Email: klaus.banert@chemie.tu-chemnitz.de
› Author Affiliations
Further Information

Publication History

Received: 11 April 2016

Accepted: 18 April 2016

Publication Date:
07 June 2016 (online)

 
 PDF Download Permissions and Reprints All articles of this category


Dedicated to Professor Heinrich Lang on the occasion of his 60th birthday

Abstract

This short review describes the synthesis and reactions of organic azides bearing an adjacent additional functional group, such as azidoacetylenes, α-azido alcohols, geminal azidohalo compounds, formyl azide, 1,1-diazidoethenes, and azidocyclopentadienes. Some of these compounds were extensively investigated in previous quantum chemical studies, but experimental access was missing until quite recently; other such azides were simply overlooked for a long time. The title compounds are able to undergo a variety of intramolecular and intermolecular subsequent reactions, which can find application in synthetic chemistry.

1 Introduction

2 Ethynyl Azides

3 Geminal Azidohalo Compounds and α-Azido Alcohols

3.1 Azidohalomethanes

3.2 Discovery and Synthesis of α-Azido Alcohols

3.3 Reactions of α-Azido Alcohols

4 Formyl Azide

5 1,1-Diazidoethenes

6 Azidocyclopentadienes

7 Conclusion and Outlook


#

Key words

azides - cycloaddition - hazardous compounds - low-temperature NMR spectroscopy - photolysis - reactive intermediates - trapping reactions
1

Introduction

The azido unit belongs to the most important functional groups in organic chemistry, although it often leads to explosive properties and cannot be found in natural products. Various methods of synthesis and perspectives of application have been summarized in several books and reviews dealing with organic azides.[1] Because the chemistry of these compounds is very manifold and extensive, classification into alkyl,[2] aryl,[3] vinyl,[4] acyl, and sulfonyl azides[5] has been made. Substances that cannot be integrated into these classes, for example, azidoacetylenes, are the focus of this short review. Furthermore, vinyl and alkyl azides with an adjacent functional group at the azide-bearing carbon atom are described. In all cases, the substitution pattern is unusual and selected by the subjective view of the author. Such azides always show a high tendency for rapid secondary reactions, which are sometimes surprising, and this can explain why these azides were discovered only quite recently. Generation and succeeding reactions of the title compounds prove to be useful in organic synthesis.

Zoom Image
Klaus Banertwas born in Wattenscheid, Germany, in 1955. He studied chemistry at the Ruhr-Universität, Bochum, and received his doctoral degree in 1982, working under the direction of Prof. W. Kirmse in the field of rearrangement reactions of carbenium ions. In 1983, he went to the University of Siegen as an assistant professor (C1) and obtained the habilitation on rearrangement reactions of organic azides in 1990 and the position of an associate professor (C2) in 1991. He received calls to the Universities of Cologne and Chemnitz in 1993 and became full professor (C4) at the University of Chemnitz at this time. He is a Volume Editor for Science of Synthesis, the editor of other books on organic chemistry, and the author of several book chapters and review articles. His research interests are centered on organic syntheses combined with short-lived intermediates and mechanistic studies, especially sigmatropic rearrangement reactions, heterocumulenes (azides, iso(thio)cyanates), heterocycles, ring strain, and steric hindrance.

# 2

Ethynyl Azides

Whereas vinyl[4] and allenyl azides[4a] [6] are well known for their multifarious reactions and can be synthesized by various methods, all attempts to isolate or even to simply detect azidoacetylenes were unsuccessful until quite recently. Already in 1910, Forster and Newman analyzed the addition of bromine to vinyl azide (1) in aqueous solution in order to prepare ethynyl azide (4a) by 1,2-elimination of two molecules of hydrogen bromide (Scheme [1]).[7] However, because of the explosion-like course of the addition reaction and the high sensitivity of the product 2 to hydrolysis, 2 could not be detected and the aldehyde 5 was generated instead.

Zoom Image
Scheme 1 Unsuccessful attempts to generate azidoacetylene (4a) via dibromide 2

We showed later that the transformation 1 → 2 can be performed conveniently and quantitatively by using bromine in organic solvents at low temperature.[8] When we carefully treated 2 with wet dimethyl sulfoxide, selective hydrolysis led to the α-azido alcohol 3 in equilibrium with aldehyde 5 and hydrazoic acid. Unfortunately, all our efforts to convert 2 into azidoacetylene (4a) with the help of various bases were unsuccessful.

Since the 1950s, many groups have tried to generate 1-azidoalk-1-ynes, but these species have remained elusive like a mystery until quite recently, and several attempts to prepare such compounds failed or led to unwanted products.[4a] [9] In 1983, treatment of chloro compound 6 with sodium azide in the solvent dimethyl sulfoxide was claimed to give the sulfoximine 8 in low yield (5–8%) besides several other products (Scheme [2]).[10] The formation of 8 was explained to proceed via generation of short-lived intermediate 4b followed by liberation of dinitrogen and interception of the nitrene 7 by the solvent dimethyl sulfoxide. When the experiment with 6 was repeated 27 years later, it was shown that the real product was 11 (10% yield), resulting from the reaction route via carbene 10b, had been erroneously identified as 8.[11] The sulfoxonium ylide 11 was also formed through intermediate 10b after decay of an appropriate 3H-diazirine in the presence of dimethyl sulfoxide. The known[12] carbene trapping product 12 was isolated in low yield when the reaction of 6 with sodium azide in dimethylformamide was performed in the presence of tolane (diphenylacetylene). Furthermore, previous ab initio studies by Auer et al. demonstrated that ethynylnitrenes such as 7 do not correspond to a local minimum of energy, and additionally that loss of dinitrogen from azide 4b should generate carbene 10b.[13] [14] Not only the carbene-trapping product 11, but also the 1,2,3-triazole 9b was obtained after treatment of 6 with lithium azide in dimethyl sulfoxide in the presence of cyclooctyne.[11] The heterocyclic product 9b is more plausible evidence for obviously short-lived ethynyl azides. Because the desired substitution reaction was very slow at room temperature and accompanied by several side reactions, all our attempts to isolate or at least to directly detect such azido compounds by NMR monitoring were unsuccessful, when chloride 6 or other 1-haloalk-1-ynes were reacted with azide salts.

Zoom Image
Scheme 2 In situ generation of phenylethynyl azide (4b)

Phenyliodonium salts, such as 13, are known to add azide at the ‘wrong’ carbon atom to form azidovinylidene intermediates 14 after cleavage of iodobenzene (Scheme [3]).[15] Proof for short-lived azidovinylidenes was presented by the work of Stang and Kitamura, who studied the corresponding secondary reactions that led to insertion products.[16] When we treated the iodonium salt 13b with a highly soluble azide source like QN3 (hexadecyltributylphosphonium azide)[17] in the presence of the interception reagent 2,3-dimethyl-2-butene, not only the vinylidene-trapping product 15b (17% yield), but also the known[18] cyanocyclopropane derivative 19b (29%) were isolated.[19] The formation of 19b can be explained through a Fritsch–Buttenberg–Wiechell-like isomerization (FBW) of 14b, followed by liberation of dinitrogen from resulting 4b and trapping of carbene 10b. After repeating the experiment with the alkyne 3-hexyne instead of the olefin 2,3-dimethyl-2-butene, we found the 3-cyanocyclopropene 21b (43%) exclusively. Whereas general interception of vinylidenes by alkenes to generate methylidenecyclopropanes is well established,[20] the analogous trapping reaction with alkynes has rarely been utilized since the corresponding meth­ylidenecyclopropenes are very unstable. Therefore, it was not surprising that methylidenecyclopropenes were not detected in the reaction mixture resulting from 13b, QN3, and alkynes. On the other hand, treatment of 13b with QN3 in the presence of strain-activated cyclooctyne instead of open-chain acetylenes gave not only a small yield of 18b (7%), but also the heterocycle 9b (21%), which is the interception product of azide 4b.[19]

Zoom Image
Scheme 3 Synthesis and reactions of ethynyl azides 4

Quantum chemical calculations, in which energy barriers for the liberation of dinitrogen were analyzed for different azidoacetylenes, predicted an even greater stability of the species 4a and 4c than that of 4b, whereas ethynyl azides with donor groups, such as amino or ethylsulfanyl, should separate dinitrogen very easily.[13] Furthermore, we assumed that the nucleophilic attack of azide at the terminal carbon atom of 13a will be more rapid than the analogous reaction of 13b. Thus, we treated the iodonium salt 13a with QN3 at –40 °C in the presence of 2,3-dimethyl-2-butene and isolated the known[21] cyclopropane derivative 19a (15% yield) and the insertion product 20a (7%), which obviously resulted from carbene 10a.[19] We could not detect the corresponding azidomethylidenecyclopropane, which may form by the interception of vinylidene intermediate 14a. Therefore, we suppose that the isomerization 14a → 4a is more rapid than the FBW-like rearrangement 14b → 4b. When the reaction of 13a with QN3 was performed at –40 to –30 °C with cyclooctyne instead of 2,3-dimethyl-2-butene, we exclusively isolated the 1,2,3-triazole 9a (91%), which is a consequence of the efficient [3+2]-cycloaddition of azide 4a at the ring-strained alkyne.

Treatment of 13a with QN3 in chloroform, dichloromethane, or acetone (–40 °C/30–120 min) in the absence of any interception reagent led to solutions of azidoacetylene (4a) after careful[22] recondensation of the reaction mixtures under vacuum.[19] The yield of the desired product was approximately 40%, as determined by 1H NMR spectroscopy, and the corresponding 15N-labeled compound was similarly prepared with the help of the reagent[17] [23] Q15N3. When the analogous reaction of the iodonium salt 13c with QN3 in deuterated chloroform at –40 °C was monitored by 1H NMR spectroscopy, nearly quantitative formation of the azide 4c was observed.[24] Thus, it was not surprising that the trapping product 9c was isolated in high yield (89%) after treating the reaction mixture with cyclooctyne. After recondensation of the reaction mixture with 4c, however, 1H NMR analysis showed a reduced yield of only 10–54%, even when the recondensation was performed at –40 °C with the help of an oil diffusion pump (10–6 Torr). Obviously, 4c is less volatile than 4a; thus, the thermal stress in the recondensation process and also the intrinsically lower stability should be responsible for the significant decrease in the yield.

The ethynyl azides 4a and 4c were characterized in solution by using IR, 1H NMR, 13C NMR, and 15N NMR spectroscopy at low temperature, which provided convincing proof of the structures. Even analysis of 4a by GC-MS at room temperature is possible. In the case of 4a, experimental NMR data were compared with the corresponding δ and J values originating from quantum chemical calculations at the coupled-cluster level of theory. Generally, nice agreement between the measured and the theoretical data was found.[19] Furthermore, coupling constants 1 J(13C,1H) and 2 J(13C,1H) of 4a are significantly larger than those of acetylene. This has already been reported for other terminal alkynes with nitrogen or oxygen donor substituents.[25] The upfield shift for the signal corresponding to the C–H carbon (δ ca. 55) discloses that the azido group acts as a π donor, through which 4a becomes an electron-rich alkyne. This is also documented by the rapid conversion of bromine in deuterated dichloromethane at –80 °C. Azidoacetylene (4a) is clearly more reactive than acetylene and did not lead to a simple product of 1,2-addition. Dibromoacetonitrile (17a) was formed instead, possibly via intermediate 16a.[26]

Ethynyl azide (4a) decomposed with a half-life period of approximately 17 hours in deuterated dichloromethane at –30 °C, whereas the half-life periods of 4c in deuterated chloroform turned out to be 230 min at –30 °C and 35 min at –20 °C.[19] [24] The decay of 4a in chlorinated solvents such as chloroform or dichloromethane and in the absence of reactive partners led to C–Cl insertion products via the carbene intermediate 10a.[26] In spite of their properties as short-lived compounds, ethynyl azides 4 can be reacted intermolecularly, for instance, with cyclooctyne to afford 18a (35% yield based on precursor 13a) or 18c (95% yield based on 4c). When a cold solution of 4a in dichloromethane was warmed to ambient temperature in the presence of 2,3-dimethyl-2-butene, the expected compounds 19a (44%) and 20a (27%) were obtained. The same products were also formed with 35% and 15% yield, respectively, after photolyzing a solution of 4a and 2,3-dimethyl-2-butene in dichloromethane for 3 hours at –60 °C. Similarly, a solution of 4c in the presence of a great excess of isobutene led to 22c (69%) and 23c (5%) via thermal (–40 to +20 °C) generation of carbene 10c.

Treatment of 13a or 13c with QN3 in the high-boiling-point polar solvent propylene carbonate at –35 °C or –30 °C, respectively, enabled the transport of the volatile products 4a or 4c into the gas phase by using a vacuum line and cold traps.[24] [26] [27] Thus, gas phase IR spectra of 4a and 4c were measured to study the decay of the ethynyl azides. These investigations confirmed that 4a is more stable than 4c, not only in solution, but also in the gas phase, which is in agreement with previous quantum chemical calculations.[13] The IR spectra of both ethynyl azides showed the expected bands corresponding to C≡C, asymmetric N3, and symmetric N3 stretching modes. In the case of 4a, however, the high extinction of two equally intensive signals, found at 2196 and 2092 cm–1 in the gas phase and at 2188 and 2086 cm–1 in chloroform at –20 °C, was remarkable and can only be explained by strong couplings, Fermi resonances, and intense overtones, which is supported by high level vibrational configuration interaction calculations.[19] [27]

Zeng et al.[26] [27] were able to handle undiluted liquid azidoacetylene (4a) and to measure the melting point of –69.5 °C. But the pure compound tends to heavy explosions, especially on phase change from liquid to solid or reverse, whereas 4c turned out to be less explosive.[22] Isolation of 4a in an argon matrix and UV irradiation enabled IR analysis of the azide as well as triplet cyanocarbene (10a) and its photochemical secondary products.[26] When flash pyrolysis of gaseous argon diluted 4a at 235–315 °C was combined with argon matrix isolation techniques, the known[28] carbene 10a could be thermally generated under unusually mild conditions.

The decay of 4c or similar silylethynyl azides to give cyano(silyl)carbenes might be useful in organic synthesis, especially since another access to these carbenes is, to the best of our knowledge, unknown in the literature. In situ formation of substituted azidoacetylenes has recently been utilized to prepare a variety of products via decomposition to the corresponding cyanocarbenes and the inter- or intramolecular trapping reactions of these carbenes.[11] [19] [24] [26] [29]


# 3

Geminal Azidohalo Compounds and α-Azido Alcohols

α-Azido alcohols of type 24 were unknown until quite recently (Scheme [4]). The same is true for azidochloromethane (26a) and azidobromomethane (26b), two parent compounds of geminal azidohalo compounds. On first view, there is no obvious connection between species 24 and the geminal azidohalo compounds. However, this section illustrates how the first synthesis of azides 26 incidentally led to the generation of azidomethanol and then to the discovery of a general access to α-azido alcohols, which is extremely simple, but had been overlooked for a hundred years.

Zoom Image
Scheme 4 Formation of diazidomethane (27) from dihalomethanes 25

Single examples of geminal azidohalo compounds have been reported in the literature. But, to the best of our knowledge, a general method to prepare such compounds, for example, 26a,b has not been described. When Hassner et al. synthesized diazide 27 by treating substrates 25a or 25b with a polymeric azide reagent, they already mentioned that an intermediate such as 26a could not be detected.[30] Obviously, the substitution step 26a → 27 is more rapid than the first chlorine substitution of 25a. Thus, we started with the mixed dihalomethane 25c and hoped that the iodine displacement step 25c → 26a would not be significantly slower than the reaction 26a → 27.[31] However, we could not detect any trace of 26a after treatment of 25c with a substoichiometric amount of QN3 in chloroform and monitoring by NMR spectroscopy, which led quantitatively and under mild conditions (18 h at 20 °C) to diazide 27.

3.1

Azidohalomethanes

Zoom Image
Scheme 5 Synthesis of azidochloromethane (26a) and azidobromomethane (26b)

Previous results indicated that the synthesis of 26a and 26b can only be successful if the halide is introduced after the formation of the azido group by nucleophilic substitution. Thus, the reaction of triazide 28b, which is easily accessible from trichloride 28a,[32] with dry hydrogen chloride in chloroform or hexane led to a mixture of 26a and 27 (Scheme [5]).[31] Simple workup by treatment with anhydrous potassium carbonate and recondensation followed by separation with the help of preparative gas chromatography enabled the isolation of 26a and 27 as colorless, highly explosive[33] liquids. When the mixture of 26a and 27, resulting from 28b and hydrogen chloride, was worked up and then treated with cyclooctyne, the 1,2,3-triazoles 29a (11% yield based on 28b), 30, and 31 were obtained. Based on 27, cycloaddition product 31 could be prepared quantitatively if an excess of cyclooctyne is used. Alternative methods to synthesize 26a started with 28b and an excess of phenylsulfenyl chloride instead of hydrogen chloride (8% yield), or utilized other α-azido amines, such as (azidomethyl)dimethylamine, to introduce the chloro substituent by treatment with hydrogen chloride. But in the latter case, the yield was rather low (≤ 1%). The reaction of 28b with hydrogen bromide led to the product 26b, which was less stable than 26a and could only be characterized in solution.[31] Cycloadduct 29b was isolated in 14% yield, when 28b was reacted with hydrogen bromide followed by conversion with cyclooctyne. The synthesis of 26a,b was successful by using the special precursor 28b; however, a convenient and more general access to geminal azidohalo compounds obviously cannot be derived from these results.


# 3.2

Discovery and Synthesis of α-Azido Alcohols

Whereas cyanohydrins are easily prepared from hydrogen cyanide and carbonyl compounds, such as aldehydes 32, and have proved to be highly valuable substrates in synthetic chemistry, the addition reaction of hydrazoic acid at 32 to form α-azido alcohols 24 was completely unknown until quite recently (Scheme [6]).

Zoom Image
Scheme 6 Reactions in which α-azido alcohols 24 have been postulated as short-lived intermediates

Attempts to produce 24 by methanolysis of silyl ethers 33a exclusively led to aldehydes 32.[34] α-Azido alcohols of type 24 were postulated to be short-lived intermediates in solvolysis reactions of geminal diazides 33b, which also yielded the final products 32.[35] Recently, in situ formation of azidomethanol (24, R = H) was suggested to occur in the reaction of formaldehyde with sodium azide in the presence of acetic acid. This transformation was completed by copper(I)-catalyzed click reaction with terminal alkynes to give 1,2,3-triazoles.[36] In all of these cases, there has been no spectroscopic proof of intermediates 24, whereas compounds of types 33a [34] [37] and 33b,[38,39] and also α-azido ethers 33c [39] [40] can easily be isolated and characterized after synthesis from precursors 32 or the corresponding acetals or enol ethers. This led to the statement that hydrazoic acid does not react with aldehydes readily.[40a]

Because we did not question this presumption or, at least, we did not have any doubts about the elusiveness of α-azido alcohols 24, we incidentally obtained evidence for these compounds. When we tried to react 28b with hydrogen iodide in chloroform under apparently not completely anhydrous conditions or workup, we surprisingly obtained azidomethanol (24a) instead of the desired azidoiodomethane (26c) (Scheme [7], see also Scheme [5]).[8] The structure of 24a was confirmed not only by NMR and IR spectroscopy, but also by treatment with cyclooctyne and isolation of the stable product 34, which resulted from [2+3] cycloaddition followed by isomerization of the resulting 1H-1,2,3-triazole. NMR solutions of 24a in chloroform showed only small proportions of the cleavage products hydrazoic acid and formaldehyde (32a). Therefore, it is logical that the equilibrium of 32a/HN3 and 24a favors the α-azido alcohol. Thus, the new product 24a is very easily available by simply mixing 32a and hydrazoic acid in chloroform.

Zoom Image
Scheme 7 Synthesis of azidomethanol (24a)

The approach to α-azido alcohols 24, which is shown with a variety of examples in Table [1], can be transferred to electron-poor aldehydes like 32v,cc,dd,ee as well as to simple aliphatic, cycloaliphatic, or even aromatic aldehydes.[8] In these cases, the formation of a 1:1 adduct of type 24 is unequivocally supported by 1H NMR data. In particular, the vicinal 1H,1H coupling resulting from OH and CH groups (see Figure [1] with a 1H NMR spectrum of 24b) and the presence of diastereotopic protons or other diastereotopic groups (see, for example, 24e,g,h,i,k,u) owing to the stereogenic center of 24 exclude the alternative formation of di­azides 33b and trimers of 32 (1,3,5-trioxanes). Contrary to cycloaliphatic aldehydes 32g,h,i, cyclopropanecarbaldehyde did not react observably with hydrazoic acid at room temperature. Obviously, the three-membered ring stabilizes the partial positive charge at the formyl carbon atom, reduces the electrophilicity, and increases the thermodynamic stability of the aldehyde. Similar arguments can be used to explain the low proportion of 24ff in equilibrium with benzaldehyde (1ff) and hydrazoic acid. The latter reagent did not add to nonactivated C–C double or triple bonds of substrates 32 or products 24 (see the cases m,n,o,p,dd, and ee), and we did not observe either cleavage of esters or acetals­, or nucleophilic substitution of halides. Thus, 32q and 32r led to equilibria with 24q and 24r, respectively, but not to the formation of diazide 24s. This compound was generated from 32s and hydrazoic acid.

Zoom Image
Figure 1 1H NMR spectrum (400 MHz, CD2Cl2, –65 °C, δ scale) of 1-azidoethanol (24b)

Table 1 Equilibria of Aldehydes/Hydrazoic Acid and α-Azido Alcoholsa

R

K [L/mol]

R

K [L/mol]

R

K [L/mol]

a

H

ca. 27

m

CH2CH2C(Me)=CH2

y

(CH2)3Br

b

Me

0.93

n

CH2CH=C=CH2

0.92

z

(CH2)3OTS

c

Et

0.59

o

CH2CH2C≡CH

aa

CH2CH2CMe2N3

d

Pr

0.42

p

(CH2)3C≡CH

bb

CH2CH2CMe2Br

e

i-Pr

0.27

q

CH2Cl

12.5

cc

CO2Et

ca. 620

f

t-Bu

r

CH2Br

dd

CO2CH2CH=CH2

g

cyclobutyl

0.42

s

CH2N3

ee

CO2CH2C≡CH

h

cyclopentyl

0.30

t

CH2P+Ph3 Cl–

0.76

ff

Ph

0.0028

i

cyclohexyl

0.33

u

CH(OMe)2

gg

2-O2NC6H4

0.0205

j

Bn

0.59

v

CCl3

13.5

hh

4-O2NC6H4

0.0163

k

CHPh2

0.40

w

CH2CH2N3

ii

2,4-(O2N)2C6H3

0.1176

l

CH2CH2Ph

x

(CH2)3N3

12.8

a Measurements were performed in the range of 19.5–22.0 °C; for exact temperatures of each single example and for K values determined at –25 °C, see ref. 8.

Utilizing weighed samples of 32 and titrated solutions of hydrazoic acid in deuterated chloroform and with the help of 1H NMR spectra, the equilibrium constant K was determined at ambient temperature for some of the reactions depicted in Table [1].[8] Electron-poor aldehydes, such as 32a,[41] 32q, 32v, and especially 32cc, led to higher values of K, whereas small values of K were obtained for the aromatic substrates 32ff,gg,hh,ii. Thus, the K values of 24 behave similarly to those of the corresponding aldehyde hydrates,[42] but there is no strict correlation. Formation of an intramolecular hydrogen bond in 24gg is able to explain the greater K value of 32gg relative to that of isomeric 32hh. Aldehydes with branching at the α position such as 32e gave slightly lower K values than the linear isomers like 32d. Such a steric effect may also be responsible for the K value resulting from 32v, which is only a little higher than that of 32q, although the trichloromethyl group is a stronger acceptor than the chloromethyl unit. At lower temperatures (–25 °C versus 20 °C), significantly higher K values (increase by a factor of 3–27) were observed, and a change of the solvent, for example, to the aprotic polar dimethyl sulfoxide, can also lead to greater proportions of the α-azido alcohols 24.

Glyoxal (32jj) reacted with two equivalents of hydrazoic acid to give a 1:1 mixture of meso- and rac-24jj. A K value of 86 L2/mol–2 (CDCl3, 19.8 °C) was calculated for this transformation. Chiral carbohydrates bearing an aldehyde unit also afforded two anomeric products in the presence of hydrazoic acid. For instance, a 4:3 proportion of 24ll anomers resulted from aldehyde 32ll, whereas 24kk was formed as a 2.5:1 mixture of α-azido alcohols. These diastereomeric products should be generated under thermodynamic control since equilibria are fast at ambient temperature.

At –50 to –65 °C, the equilibrations of 32/HN3 and 24 are very slow. Thus, hydrazoic acid along with the solvent and even aldehydes, such as 32a–f,q,v, could be removed under vacuum because the corresponding adducts 24 were less volatile. The residues were dissolved in precooled solvents to give solutions of highly enriched or pure α-azido alcohols 24. Some of the products (24a–d) could be obtained as solids or viscous liquids at low temperature. The best yields of 24 were reached when the equilibrium was slowly established at reduced temperature, for example at –25 °C in a refrigerator, before the volatile substances were removed in high vacuum at even lower temperatures.[8]

α-Azido alcohols 24 have quite different properties than the silylated analogues 33a, which can simply be distilled under reduced pressure.[34] When 24 was liberated from hydrazoic acid at ambient temperature, the rapid equilibration caused complete cleavage of the substrate. In such a case, only the aldehyde 32 remained, and the formation of 24 from 32 and hydrazoic acid could easily be overlooked. For example, the reaction of acrolein (35) with excess hydrazoic acid was reported several times[43] [44] to give the azide 32w, but the generation of diazide 24w was not mentioned, although it can be detected quite simply by NMR methods (Scheme [8]).[8] Similarly, an undiscovered α-azido alcohol was involved in the reaction of vinyl azide (1) with aqueous bromine to produce bromoacetaldehyde (5) (Scheme [1]). However, this experiment was first performed in 1910,[7] and the usual methods to detect elusive intermediates were limited at that time.

Zoom Image
Scheme 8 Generation of α-azido alcohol 24w from acrolein (35)

# 3.3

Reactions of α-Azido Alcohols

α-Azido alcohols 24 include a carbon atom bearing two functional groups which are each well known for a variety of highly useful reactions. Thus, a multifarious chemistry of compounds 24 can be expected. However, the rapid equilibrium with aldehydes 32 and hydrazoic acid can lead to several unwanted side reactions if the transformation of 24 is not performed at sufficiently low temperature. In the presence of a base, for example, the corresponding salt of hydrazoic acid will be formed, and the absence of this acid will completely shift the equilibrium from 24 to 32. On the other hand, the geminal connection of OH and N3 units in 24 enables reactions that are not possible with simple alcohols and azides or azido alcohols with a greater distance between the functional groups.

1,3-Dipolar cycloaddition of 24 to yield 1,2,3-triazoles was successful not only in the presence of ring-strained cyclooctyne (Scheme [7]). In the case of 24p, a slow intramolecular cycloaddition of the azido group at the unactivated C≡C bond led to the bicyclic product 36p (Scheme [9]).[8] Irradiation of 24 with a mercury high-pressure lamp initiated loss of dinitrogen to give amides 39 and 40 most probably via intermediates 37 and 38, respectively. Both types of products were formed in similar amounts and could be easily separated by chromatography, for example, 39dd (R = CO2CH2CH=CH2, 45% yield) and 40dd (32%). Thus, the results of the photochemical conversion of 24 are different to those of the known[34] thermolysis of 33a, which exclusively occur with a proton shift to generate the corresponding N-(trimethylsilyl)carboxamides. Oxidation of α-azido alcohols 24 with the help of pyridinium chlorochromate (PCC) in chloroform led under very mild conditions to acyl azides 41b,c,j,k,q,cc,ff,gg,ii (80–89% yield based on 32) without inducing Curtius rearrangement.[8] The substrate 24jj afforded the known[45] oxalic acid diazide at –50 °C (82% isolated yield based on used HN3), and 24d was oxidized by PCC at the same temperature to provide butyryl azide (41d) in 97% yield (1H NMR). In control experiments under the latter conditions, it was shown that a mixture of butyric acid, PCC, and hydrazoic acid did not generate 41d; furthermore, butyraldehyde (32d) and PCC alone (without HN3) did not produce butyric acid at such low temperatures. Obviously, the azido group in 24 strongly influences the facility with which the hydroxy group can be oxidized to a carbonyl group. Thus, the method to prepare acyl azides 41 from aldehydes 32 via intermediates 24 under very mild conditions can be a useful alternative to known[46] procedures. In the presence of acyl chlorides, esterification of 24 is successful to isolate the products 42.[47] In some cases, enriched compounds 24, which were liberated from hydrazoic acid, gave good results, whereas with other examples, equilibrium mixtures of 24 and 32/HN3 were sufficient to obtain the products 42 with acceptable yields. Even a one-pot procedure, starting with aldehyde 32 in chloroform at –35 °C and consecutive addition of acyl chloride, concentrated sulfuric acid, and solid sodium azide led to 42 via 24. In this case, a substoichiometric amount of sulfuric acid is sufficient to generate hydrazoic acid from sodium azide because hydrogen chloride as an additional strong acid is formed in the esterification process. Acetates 42 (R1 = Me) were prepared in moderate yields from simple aldehydes like 32b,c and in excellent yields from electron-deficient substrates such as 32v,cc, whereas only poor yields were achieved with sterically hindered acyl chlorides (R1 = t-Bu).

Zoom Image
Scheme 9 Reactions of α-azido alcohols 24

The synthesis of azidochloromethane (26a) and azidobromomethane (26b) from the special precursor 28b obviously cannot be transferred from these parent species to create a general access to geminal azidohalo compounds (Scheme [5]).[48] However, attempts to prepare azidohalomethanes 26 led to the first direct detection of azidomethanol (24a) and thereafter to a very simple and general synthesis of α-azido alcohols 24 from hydrazoic acid and all types of aldehydes 32 (Table [1]). Treatment of 24 with phosphorus trichloride or phosphorus tribromide enables a return to geminal azidohalo compounds 43, and now the access is not restricted to the parent products (Scheme [9]).[49] [50] Bromo derivatives 43 are less stable than chloro analogues, and the former were prepared in acceptable yields only if 24 was derived from an electron-deficient aldehyde 32. Unstable geminal azidohalo compounds 43, especially some bromo species, tend to split off dinitrogen and hydrogen halide with formation of nitriles (R–CN). Products 43 could be utilized to introduce another functional group by nucleophilic substitution of the halogen. In the presence of bases such as potassium tert-butoxide, 43 yielded vinyl azides by 1,2-elimination of hydrogen halide.


#
# 4

Formyl Azide

Acyl azides 41 and their Curtius rearrangement to form isocyanates 44 have been known for more than a hundred years (Scheme [10]).[51] Furthermore, these azides are of great importance in modern organic synthesis as well.[1a] [5] However, spectroscopic proof or any other indication for the existence of formyl azide (41a), the parent compound of acyl azides, was missing until quite recently. Nevertheless, 41a was investigated in about a dozen theoretical studies dealing with its structure, spectroscopic data, and various reactions.[52] Quantum chemical calculations predicted activation energies for the process 41a → 44a which were significantly lower than those of the Curtius rearrangement reactions of acetyl or benzoyl azide.[52e] [f] [g] [h] [i] [j] Thus, 41a should be highly unstable at ambient or even lower temperature. Another reason why 41a has not been experimentally generated earlier is most probably the fact that formyl chloride cannot be used as a convenient precursor of 41a because this acyl chloride is a very short-lived species.[53]

Zoom Image
Scheme 10 Curtius rearrangement of acyl azides 41

Our attempts to generate 41a from formic acid (45) by using the classical Weinstock route[54] were unsuccessful (Scheme [11]). However, oxidation of azidomethanol (24a), prepared from formaldehyde (32a) and hydrazoic acid,[8] was successfully performed at –35 °C in the presence of pyridinium chlorochromate (PCC) to furnish 41a in 36% yield.[55] [56] Furthermore, cleavage of the diol 24jj, which is accessible from glyoxal (32jj) and hydrazoic acid,[8] by treatment with periodic acid at –35 °C gave 41a in 29% yield.[55] [56] Nitrosation of formylhydrazine (46) led also to the product 41a when the reaction was performed at –10 °C. The best yields of 41a were achieved by the formylation of azide salts if carried out at low temperatures (–30 to –15 °C) with a highly soluble azide source such as QN3. Reagent 47 was found to be superior to 48 [57] in the efficiency of producing 41a. Solutions of pure 41a could be conveniently obtained when such reaction mixtures were recondensed at –50 to –15 °C and 10–6 bar.[55] Although 41a proved to be highly unstable, it could be characterized in solution by UV, IR, and NMR (1H, 13C, 15N) spectroscopy. The 15N NMR data of [15N3]-41a indicated that Nγ resonated at lower field than Nβ, which is typical for electron-deficient azides.[58] In the IR spectrum of 41a, the pattern of signals pointed to strong coupling by Fermi resonance that is confirmed by high-level quantum chemical calculations.[55] Treatment of 47 with QN3 in the high boiling point solvent propylene carbonate at –15 °C for 2 hours enabled the transport of the volatile product into the gas phase at low temperatures and then isolation of solid 41a with an estimated yield of 40%.[59] Solid formyl azide (41a) sublimes at low temperature in vacuo without detectable decomposition, and melts sharply at –50.5 °C to yield a colorless liquid. Further characterization was performed with the Raman spectrum of solid 41a and the IR spectrum in argon matrix. The latter investigations indicated that a syn conformation of 41a is dominant, which is in agreement with quantum chemical calculations. Despite the risk of explosion, Zeng et al. successfully obtained single crystals of 41a that were appropriate for X-ray diffraction analysis.[59] Thus, the molecular structure of 41a in the solid state includes the lower energy syn conformation, and two different types of intermolecular C–H···O=C hydrogen bonds were also detected.

Zoom Image
Scheme 11 Syntheses of formyl azide (41a)
Zoom Image
Scheme 12Reactions of formyl azide (41a)

Even at 0 °C, solutions of 41a in deuterated chloroform liberated dinitrogen to form isocyanic acid (44a) by Curtius rearrangement (Scheme [12]). When this process was kinetically investigated with the help of 1H NMR spectroscopy, activation parameters of E a = (20.3 ± 1.1) kcal mol–1 and ln A = 27.0 ± 2.0 were calculated.[55] In less polar solvents, such as deuterated cyclohexane, and especially in the gas phase, where analysis was performed with IR spectroscopy, Curtius rearrangement of 41a turned out to be significantly slower.[55] [59] Compared with the analogous reaction of acetyl azide under the same conditions, the reaction 41a → 44a in deuterated chloroform at 25 °C was much faster (factor of 106). These results correspond well with high-level quantum chemical calculations.

Intermolecular reactions of formyl azide (41a) can only be successful if they are able to compete with the rapid Curtius­ rearrangement. Staudinger reaction of 41a with triphenylphosphine or triisopropyl phosphite resulted in the quantitative formation of phosphazenes 49a and 49b, respectively.[55] These products could be useful building blocks in aza-Wittig reactions.[60] Treatment of 41a with cyclooctyne initially gave the cycloadduct 50a, which slowly rearranged to produce the 2H-1,2,3-triazole 50b in 40% yield (based on 47 as precursor of 41a).[55] When 41a was irradiated in a CO-doped argon matrix, besides 44a, unknown formyl isocyanate was detected with the help of IR spectroscopy.[59] In this case, carbon monoxide obviously acted as a trapping reagent for the elusive formyl nitrene, which has been explored extensively by quantum chemical calculations.[61]


# 5

1,1-Diazidoethenes

Only three compounds of the class of geminal vinyl di­azides, the molecules 51a–c, have been reported in the literature (Scheme [13]). In all three cases, the 1,1-diazidoeth­enes were prepared by simple nucleophilic substitution of the corresponding dichloro compounds, which is strongly supported by the electron-withdrawing properties of the cyano and ester units. The manifold chemistry of 51a–c has been summarized several times.[62]

Zoom Image
Scheme 13 Synthesis and photolysis of 1,1-diazidoethenes 51

When we planned the synthesis of the parent compound, 1,1-diazidoethene (51d), it seemed logical to utilize the chloroethane derivative 54 as a precursor. Unfortunately, all our attempts to directly transform chloroacetaldehyde (32q) into 54, by using different well-known methods[38] [63] for the synthesis of geminal diazides from aldehydes, were unsuccessful. Thus, we treated 43q, which is available from 32q via α-azido alcohol 24q in 48% yield (see also Scheme [9]), with N,N,N′,N′-tetramethylguanidinium azide (TMGA) to obtain 54 in 77% yield.[50] The reaction of 54 with potassium tert-butoxide in diethyl ether led to the dangerously explosive 1,1-diazidoethene (51d) in 23% yield. This compound should only be handled in solution, and even then the stability is limited by a half-life period of 220 min at 40 °C in deuterated chloroform. Therefore, direct characterization of 51d was restricted to IR and NMR methods. However, the structure of 51d was additionally confirmed by treatment with cyclooctyne and formation of the cycloadduct 55 in 19% yield based on 54.

Photolysis of geminal vinyl diazides 51 should lead to 3-azido-2H-azirines, although the detection of such species has not been reported in the literature. The azido group, as a donor substituent, is able to stabilize the azirine ring by reducing the electron deficiency at the sp2 carbon atom. This effect is well documented especially for 3-amino-2H-azirines.[64] On the other hand, it is well known that the direct combination with ring strain decreases the stability of azides. When we irradiated 51d in deuterated chloroform at –55 °C with the help of a mercury high-pressure lamp, the 2H-azirine 52d was generated in up to 34% yield.[50] This heterocycle was characterized by low-temperature IR and NMR (1H, 13C) spectroscopy indicating a short half-life period of approximately 12 minutes at –40 °C. Thus, 52d is very short-lived, and attempts to trap this species in the presence of cyclooctyne were unsuccessful.

Since 3-azido-2H-azirines were postulated to act as intermediates in thermal reactions of 51a–c,[62a] we photolyzed also 51a in deuterated chloroform at –60 °C and obtained 52a, which proved to be more stable than 52d. The heterocycle 52a could be characterized by NMR data and reacted with cyclooctyne to give the cycloadduct 53 in 44–65% yield (1H NMR).[65]


# 6

Azidocyclopentadienes

Both the azido group and the cyclopentadiene structure belong to the most important functional groups in organic chemistry. However, a combination of both structures has been rarely mentioned in literature, whereas 1-azido- and 3-azido-1H-indenes[66] as well as derivatives of azidocyclopentadienide, for example, azidoferrocenes,[67] have frequently been described. The only earlier report on an azidocyclopentadiene includes the hexaazide 56b, which was synthesized by treating the chloro precursor 56a with excess sodium azide and characterized as a stable solid that was slowly degraded at 200 °C (Scheme [14]).[68] This unusual stability is obviously incompatible with the claimed structure of 56b because of the high proportion of nitrogen and especially the notoriously unstable 1,2-diazidoethene substructures.[69] When the experiment to prepare 56b was repeated, only destruction of substrate 56a, but no organic azide could be detected.[70] Milder reaction conditions and only two equivalents of sodium azide, however, led to the diazide 57 as a yellow oil, which was characterized with the help of its spectroscopic data and also by formation of the cycloaddition product 58. On treating 57 with sodium azide, only decomposition occurred. Cyclopentadienyl azides can be prepared not only by aliphatic nucleophilic substitution, but also by electrophilic transfer of an azido group as demonstrated by the synthesis of 61 from bromide 59 or hydrocarbon 60.

Zoom Image
Scheme 14 Synthesis of fully substituted azidocyclopentadienes 57 and 61

Whereas the synthesis of fully substituted azidocyclopentadienes, such as 57 and 61, proved to be straightforward, generation of azidocyclopentadienes with acidic hydrogen atoms at the five-membered ring could be accompanied by based-induced rearrangement and decay. For example, treatment of the substrate 62 with sodium azide in dimethylformamide/acetic acid directly led to the vinyl azide 65 in 91% yield via prototropic isomerization of the intermediate 64 (Scheme [15]). When the same reaction was performed without acetic acid, the more basic conditions induced decomposition of the azido group with loss of dinitrogen and exclusive formation of the imine product 66 in 77% yield. However, treatment of the diazo compound 63 with hydrazoic acid furnished in 87% yield the allyl azide 64. This compound could be isolated as pale yellow needles, but it slowly rearranged in solution to produce the enazide 65 already at ambient temperature. The parent cyclopentadienyl azides 68a and 68b are accessible from the iodo precursor 67 and N,N,N′,N′-tetramethylguanidinium azide (TMGA). These new substances can only be handled in solution, and separation of the isomers turned out to be difficult. Nevertheless, heating a solution of the 1:1 mixture at 70 °C induced the fragmentation 68a → 69 and thus led to an enrichment of 68b with 68a/68b = ca. 1:6. Owing to the slow equilibration of 68a and 68b, prolonged thermolysis resulted in quantitative transformation of 68a and 68b into 69. The 1:1 equilibration of the azidocyclopentadienes was instantaneously established, when a 1:6 mixture of 68a and 68b was treated at room temperature with a catalytic trace of 1,4-diazabicyclo[2.2.2]octane (DABCO).

Zoom Image
Scheme 15 Synthesis of azidocyclopentadienes with acidic hydrogen atoms
Zoom Image
Scheme 16 Thermal reactions of azidocyclopentadienes

The thermal stability and the structures of thermal products, resulting from 1-azido-, 2-azido-, and 5-azido-substituted cyclopentadienes, are quite different because of ring-opening, ring-closure, and ring-enlargement reactions, respectively. Thus, azide 70 proved to be highly unstable since fragmentation to furnish the nitrile 71 as a mixture of (E) and (Z) diastereomers was observed already at ambient temperature (Scheme [16]). On the other hand, the isomeric compound 72 was isolated as a relative stable lemon-yellow solid, which was converted into the indole derivative 73 by refluxing in toluene or irradiation in chloroform. Strong heating of substrate 61 in dimethyl sulfoxide gave only the pyridine 74.

The acidic hydrogen atoms in the equilibration mixture of the cyclopentadienes 68a/68b enable a condensation reaction with acetone in the presence of pyrrolidine, which led to the azidofulvene 75 (Scheme [17]). A mixture of the unstable cycloadducts 76 and 77 was obtained by the Diels–Alder reaction of 68a/68b with dimethyl acetylenedicarboxylate (DMAD). Staudinger reaction of azide 65 followed by treatment of the resulting iminophosphorane with carbon disulfide yielded the isothiocyanate 78.[70]

Zoom Image
Scheme 17 Bimolecular reactions of azidocyclopentadienes

# 7

Conclusion and Outlook

Liberation of dinitrogen from an azide is a very exothermic process. In the case of small molecules, this leads to explosive substances that have to be handled carefully. On the other hand, decay of organic azides is able to generate high-energy intermediates, for example, nitrenes, carbenes, strained heterocycles and many others, that initiate a variety of succeeding reactions with highly useful products. Even without loss of dinitrogen, however, the transformations of azides are very manifold and important. The diversity of the reactions can be further increased when azides with an adjacent functional group are utilized as substrates. Such compounds show different reactivities if compared to those of simple azides or azides with remote additional functional groups. Thus, surprising products are formed in several cases. We believe that some novel examples of unusually functionalized azides will lead to interesting chemistry in the future.


#
#

Acknowledgment

This work is part 36 in the series ‘Reactions of Unsaturated Azides’ and was generously supported by the Deutsche Forschungsgemeinschaft (BA 903/12/1–3). The author thanks Dr. A. Ihle (Chemnitz University) for his help in preparing the manuscript.



   Permissions and Reprints All articles of this category
Zoom Image
Klaus Banertwas born in Wattenscheid, Germany, in 1955. He studied chemistry at the Ruhr-Universität, Bochum, and received his doctoral degree in 1982, working under the direction of Prof. W. Kirmse in the field of rearrangement reactions of carbenium ions. In 1983, he went to the University of Siegen as an assistant professor (C1) and obtained the habilitation on rearrangement reactions of organic azides in 1990 and the position of an associate professor (C2) in 1991. He received calls to the Universities of Cologne and Chemnitz in 1993 and became full professor (C4) at the University of Chemnitz at this time. He is a Volume Editor for Science of Synthesis, the editor of other books on organic chemistry, and the author of several book chapters and review articles. His research interests are centered on organic syntheses combined with short-lived intermediates and mechanistic studies, especially sigmatropic rearrangement reactions, heterocumulenes (azides, iso(thio)cyanates), heterocycles, ring strain, and steric hindrance.
Zoom Image
Scheme 1 Unsuccessful attempts to generate azidoacetylene (4a) via dibromide 2
Zoom Image
Scheme 2 In situ generation of phenylethynyl azide (4b)
Zoom Image
Scheme 3 Synthesis and reactions of ethynyl azides 4
Zoom Image
Scheme 4 Formation of diazidomethane (27) from dihalomethanes 25
Zoom Image
Scheme 5 Synthesis of azidochloromethane (26a) and azidobromomethane (26b)
Zoom Image
Scheme 6 Reactions in which α-azido alcohols 24 have been postulated as short-lived intermediates
Zoom Image
Scheme 7 Synthesis of azidomethanol (24a)
Zoom Image
Figure 1 1H NMR spectrum (400 MHz, CD2Cl2, –65 °C, δ scale) of 1-azidoethanol (24b)
Zoom Image
Scheme 8 Generation of α-azido alcohol 24w from acrolein (35)
Zoom Image
Scheme 9 Reactions of α-azido alcohols 24
Zoom Image
Scheme 10 Curtius rearrangement of acyl azides 41
Zoom Image
Scheme 11 Syntheses of formyl azide (41a)
Zoom Image
Scheme 12Reactions of formyl azide (41a)
Zoom Image
Scheme 13 Synthesis and photolysis of 1,1-diazidoethenes 51
Zoom Image
Scheme 14 Synthesis of fully substituted azidocyclopentadienes 57 and 61
Zoom Image
Scheme 15 Synthesis of azidocyclopentadienes with acidic hydrogen atoms
Zoom Image
Scheme 16 Thermal reactions of azidocyclopentadienes
Zoom Image
Scheme 17 Bimolecular reactions of azidocyclopentadienes