Preparation of Tyrian Purple (6,6′-Dibromoindigo): Past and Present

Over the past century, various synthetic approaches have been suggested to the most famous dye of antiquity, Tyrian purple (6,6′-dibromoindigo). These synthetic routes have been exhaustively surveyed and critically evaluated from the perspective of convenience, cost, safety and yield.


Introduction
6,6′-Dibromoindigo (1; see Scheme 1) is the chemical structure of the major component of Tyrian purple, the most famous dye of antiquity [1,2]. From ancient times the dye has been produced from secretions of various species of snails found off the Atlantic and Mediterranean coasts. Due to the minute amounts of dye found in the snails, the dye has always been very costly. Paul Friedländer, who in 1909 first identified the structure of the dye as 6,6′-dibromoindigo, required 12,000 Murex brandaris snails to produce 1.4 g of pure pigment [3]. Over the past century a variety of groups have undertaken to develop rational syntheses of this historic dye. Their efforts are surveyed and critically evaluated herein from the perspective of convenience, cost, safety and yield. Several possible improvements are also proposed.

The original synthesis and its elaborations
Nearly all known syntheses of 6,6′-dibromoindigo (1) are based on the oxidative coupling of a 6-bromoindole derivative, which is usually generated in situ. Thus, the first synthesis of 1, reported in 1903 by Sachs and Kempf [4] (Scheme 1), was based on the Claisen condensation of 4-bromo-2nitrobenzaldehyde (8) with acetone, in analogy to the Baeyer-Drewsen process for the manufacture of indigo [5,6]. Scheme 1. Sachs  The substituted benzaldehyde 8 was prepared, in five steps starting from 2,4-dinitrotoluene (2), in an overall yield of about 34% [4,7,8]. Under the reaction conditions, the hydroxyketone 9 spontaneously cyclizes and undergoes oxidative coupling, but it could be isolated using trisodium phosphate instead of NaOH in the Claisen condensation [8]. Interestingly, the yield of dibromoindigo in the final step was not reported.
This original Sachs and Kempf route is clearly inconvenient, due to the lengthy preparation of the bromonitrobenzaldehyde 8. Subsequent syntheses were accordingly based on shorter methods for the preparation of this aldehyde. Thus, van der Lee prepared 8 in 4 steps from p-toluidine (10) [9] (Scheme 2). In the latter procedure, 10 is nitrated in concentrated sulfuric acid to give 4-amino-2nitrotoluene (11) or its sulfate [10][11][12][13][14][15], which is then diazotized and converted to 4-bromo-2-nitrotoluene (12) in a Sandmeyer reaction or one of its variants [9,11,12,[16][17][18][19][20][21]. The substituted toluene is then condensed with amyl nitrite [22,23] to give the oxime 13. The latter is then oxidatively cleaved with ferric ammonium sulfate to give the desired aldehyde 8. This synthesis of 2 is indeed shorter, but at the expense of rather low yields. A modest improvement was achieved via a related scheme by Rottig [24], who used ethyl instead of amyl nitrite for the conversion of toluene 12 to oxime 13. Benzaldehyde 8 was subsequently condensed with acetone in the usual way to give dibromoindigo 1. However, despite these improvements, the reported overall yield of 1 based on the starting toluidine was only 5.5% [24].
An additional drawback of the above scheme is that even the modest yields of the benzaldehyde 8 obtained by reaction of 12 with alkyl nitrites could not be reproduced by later workers. As an alternative to this reaction, Barber and Stickings [25] found that oxidation of 12 with chromic acid in the presence of acetic anhydride was a more reliable method, although the yields were hardly better. The latter method had previously been used to prepare o-nitrobenzaldehyde from o-nitrotoluene in two steps through the corresponding Diacetate [26][27][28][29]. The chromic acid oxidation of 12 was subsequently used by Pinkney and Chalmers [30] and by Torimoto and coworkers [31] in their syntheses of Tyrian purple, and also by Keinan and coworkers [32]. Recently, the procedure of Pinkney and Chalmers was improved by Imming and coworkers [33], who achieved an overall 10% yield of 1 from 10 (Scheme 3). The final condensation of the benzaldehyde 8 with acetone in all these reported syntheses consistently gives yields of no more than about 50% [34]. In 1950, Harley-Mason [35], following the earlier work of Thiele [36], reported that the sodium salt of 2-nitro-1-o-nitrophenylethyl alcohol (18), obtained by the nitro-aldol condensation (Henry Reaction) of o-nitrobenzaldehyde (16) with nitromethane (17), gives a 90% yield of indigo (19)  This procedure was later used by Voss and Gerlach [37] and by Cooksey [38], who obtained 66% and 69% yields, respectively, of 1 from the benzaldehyde 8. However, Imming and coworkers reported [33,39] that, for producing dibromoindigo 1 on a large scale, this procedure was not as practical as the original Baeyer-Drewsen indigo procedure [5,6]. (12) All the foregoing syntheses of Tyrian purple rely on 4-bromo-2-nitrobenzaldehyde (8) as the key intermediate, which is prepared via 4-bromo-2-nitrotoluene (12), which is obtained in turn by diazotization of 4-amino-2-nitrotoluene (11) or its sulfate (14). While the Sandmeyer reaction gives good yields, it is rather cumbersome and comparatively expensive, as is the immediate precursor 11. It is, therefore, reasonable to consider alternative syntheses of 12; in particular, a one-step preparation from a cheap starting material. The oldest of these consists of nitration of 4-bromotoluene (20) [13,[40][41][42][43][44][45][46][47][48]. This reaction, however, usually gives a 5:4 mixture of isomeric bromonitrotoluenes 12 and 21, accompanied by variable amounts of side products, depending on the conditions of the nitration (Scheme 5). Nevertheless, Keinan and coworkers have reported a 90% yield of 12 by performing the nitration with 100% nitric acid in a mixture of acetic acid and 98% sulfuric acid (Scheme 5) [32].

Alternative preparations of 4-bromo-2-nitrobenzaldehyde (8)
As we have noted, conversion of the bromonitrotoluene 12 to the corresponding benzaldehyde 8 is plagued with inconvenient reactions or low yields. In general the oxidation of toluenes to benzaldehydes is an important industrial process for which a universally optimal procedure has not yet been found. Regarding our case, we note that in their original synthesis of Tyrian purple (Scheme 1), Sachs and Kempf [7] reported a 70% two-step conversion of 2,4-dintrotoluene (2) to 2,4-dinitrobenzaldehyde (5), by condensing p-nitrosodimethylaniline (3), followed by acidic hydrolysis of the intermediate nitrone 4 (substantially lower yields were later reported in an Organic Synthesis procedure based on this method [56]). The Sachs procedure is amenable only to the oxidation of toluenes substituted with at least two strongly electron-withdrawing groups in the ortho-and parapositions [57]. In a variation of this procedure, Barrow and coworkers [58,59] succeeded in preparing a number of aldehydes from the corresponding benzyl halides and aromatic nitroso compounds such as 3. Nevertheless, Barber and Stickings [25] were unable to prepare 8 by reaction of 12 with bromine and 3.
However, a more versatile procedure was developed by Kröhnke and coworkers [60][61][62][63][64], who found that conversion of the alkyl bromide group first to its pyridinium salt facilitates subsequent reaction with the aromatic nitroso compound to give the nitrone. An application of the Kröhnke method by Clarke [65] gave several substituted o-nitrobenzaldehydes in high overall yields from the respective toluenes. This application has served for a Organic Synthesis procedure by Kalir [66] for the preparation of o-nitrobenzaldehyde (16) in a 47-53% overall yield, and as the basis of a synthesis of 8 by Danieli and coworkers [67]. Cooksey [38] likewise reported a 55% overall yield of 8 using the Kröhnke procedure starting from 12 as part of his complete synthesis of 1 (Scheme 7). Although the Kröhnke procedure involves four steps for converting the substituted toluene to the respective benzaldehyde, it is relatively convenient. Indeed, the only step with a long reaction time is the benzylic bromination of the bromonitrotoluene 12. Scheme 7. Kröhnke conversion of 4-bromo-2-nitrotolune (8) to benzaldehyde (12). ( Two additional methods of converting o-nitro-substituted toluenes to the corresponding benzaldehydes are also worth mentioning, although they have not been specifically applied to the synthesis of 8. These methods start with the first steps of the Reissert [68,69] and Batcho-Leimgruber [70][71][72][73] indole syntheses, respectively. In the first method, the toluene is condensed with diethyl oxalate (27), and the enolate anion of the resulting phenylpyruvate ester (28) is acetylated and oxidized [74] (Scheme 8). The potassium ethoxide solution needed for the condensation can be prepared by treating an alcoholic potassium hydroxide solution with calcium oxide, thus obviating the need for using metallic potassium [75]. In the second method, the toluene is condensed with dimethylformamide dimethyl acetal (31) and the resulting β-aminostyrene is then oxidized, either catalytically with oxygen [76] or stoichiometrically with sodium periodate [77]. The latter procedure was recently reported to give a nearly quantitative yield of 4-chloro-2-nitrobenzaldehyde (33) from 4-chloro-2-nitrotoluene (30) [78] (Scheme 9). Both of the indole syntheses mentioned above have been applied to the preparation of 6bromoindole, as seen below. A completely different approach to substituted benzaldehydes, which does not proceed from the corresponding toluenes, has been developed by Beech [79]. In this synthesis, a diazotized aniline is treated with a solution of formaldoxime in the presence of copper sulfate and sodium sulfite. The intermediate benzaldoxime is then cleaved with ferric ammonium sulfate to give the free benzaldehyde. The Beech method has been applied to the synthesis of a number of substituted benzaldehydes [79,80], including those containing an ortho nitro group [81][82][83]. In particular, it was used by Dandegaonker [84] to prepare 8 in 34% yield from 4-bromo-2-nitroaniline (34) (Scheme 10).

Friedländer synthesis of Tyrian purple
In their original studies on Tyrian purple and related compounds, Friedländer and coworkers [3,11,12] reported a new synthesis of 1 from 4-bromo-2-aminobenzoic acid (43). In this procedure, aminobenzoic acid 43 is treated with chloroacetic acid (45) [119] yielding the carboxyphenylglycine derivative 46, which is cyclized in turn to the corresponding diacetylindoxyl 48. The latter is subsequently hydrolyzed and oxidized in air to give 1 (Scheme 14). The Friedländer method closely follows the Bayer process for the production of indigo (19) from anthranilic acid (44) [120], in which the disodium salt of (2-carboxyphenyl)glycine (47) is cyclized to diacetylindoxyl (49) in acetic anhydride instead of being subjected to alkali fusion, as in the earlier BASF process [121].

Alternative syntheses of 4-bromo-2-aminobenzoic acid (43) and its derivatives
As noted above, the major drawback to the Friedländer synthesis of 1 is its lengthy preparation of 4bromo-2-aminobenzoic acid (43). For this reason, Imming and coworkers have remarked that this approach is only of "historical interest" [33]. Nevertheless, a shorter route to this compound or its phenylglycine derivative 46 would indeed make it an attractive alternative to the other syntheses reviewed above.
The first alternative to the multistep syntheses of 43 described above was reported by Waldmann (Scheme 18) [133].

NaOCl
In this procedure, sodium phthalate (59) (derived from phthalic anhydride, 58) is brominated in aqueous solution to give 4-bromophthalic acid (60). The diacid closes to the corresponding anhydride 61 upon distillation, and the latter gives the corresponding imide 62 upon heating with urea. Subsequent Hofmann degradation, analogous to the production of anthranilic acid from phthalimide [134], then gives a product which was identified as 4-bromoanthranilic acid (43).
At first glance, the Waldmann procedure seems attractive since the starting material and reagents are all inexpensive and most of the reactions proceed with high yields. Yields of up to 90% of 4-bromophthalic acid (60) from bromination of phthalic acid have been reported [135], and conversion of 60 to the anhydride 61 has been reported in nearly quantitative yields [136]. Today 61 is an industrial compound which is readily available in high purity [137][138][139].

Dibromoindigo (1) from indoles
A novel synthesis of 1, in which the bromine atom is introduced directly into a previously complete indole ring system, was reported in 1930 by Majima and Kotake [238]. In their synthesis, the Grignard reagent derived from indole [239,240] (107) is treated with ethyl chloroformate and the resulting indole-3-carboxylate 108 is brominated. Saponification of the product 6-bromoindole-3-carboxylate (109), followed by oxidation with ozonized air, furnished Tyrian purple (Scheme 36). (107)  This synthesis would indeed be an attractive route to Tyrian purple (1) were it not for the fact that both the Grignard carboxylation and bromination reactions are not regioselective. Thus, both earlier and later reports indicate that only the ring nitrogen atom undergoes metalation [241,242]. In a more recent reinvestigation of the carboxylation of indole (107)  Moreover, it has also been shown that bromination of 108 followed by decarboxylation gives in fact a 45:55 mixture of 5-bromoindole (113) and 6-bromoindole (114)

77% overall
In an alternative synthesis of 114, indoline (121) is brominated in sulfuric acid in the presence of silver sulfate, giving largely 6-bromoindoline (122), accompanied by about 8% of the 4-bromo isomer [268]. The bromoindoline 122 is then dehydrogenated at -65 ºC via the azasulfonium salt with dimethyl sulfide [269], giving 114 (Scheme 42). This scheme enjoys the advantages of an inexpensive starting material and a short reaction sequence. However, the low temperatures needed for the dehydrogenation, plus the need for a chromatographic separation in order to obtain pure product, makes this method rather unsuited for the production of large quantities of 6-bromoindole (114 As noted above, the cost of the starting 6-bromoindole (114) and the reagents is the major drawback to the procedure of Tanoue and coworkers [245]. Inasmuch as 4-bromo-2-nitrobenzene (12) is the starting material for all practical syntheses of 81, it would appear more efficient to convert 12 to 4-bromo-2-nitrobenzaldehyde (8) and thence directly to 1, as detailed above, rather than to the indole 114. We have also noted that the intermediates in the two indole syntheses used for making 114 (i.e. the enolate ester of 117 in the Reissert synthesis, and 120 in the Batcho-Leimgruber synthesis) can presumably be oxidized directly to 8 (Schemes 8 and 9).
Another alternative to the procedure of Tanoue and coworkers [245], which does use 6bromoindole (114) as a starting material, might possibly be found in a recent one-pot synthesis of indigo from indole (107) which uses an organic hydroperoxide and catalysis by a molybdenum complex [270]. In this procedure, yields of up to 81% of indigo (19) have been reported when performing the oxidation with cumene hydroperoxide in t-butanol and molybdenum hexacarbonyl as a catalyst (Scheme 43). To our knowledge, this procedure has not yet been applied to the synthesis of substituted indigos such as Tyrian purple (1). (107)

Recent Convenient Low-Cost Synthesis of Tyrian Purple
Wolk and Frimer [271], have recently reported a five step synthesis of Tyrian purple (1), starting from p-dibromobenzene (40; Scheme 48). The reactions are simple, low cost, safe, high yield procedures. The first step involves the Friedel-Crafts acetylation of p-dibromobenzene (40) producing 2′,4′-dibromoacetophenone (82) as reported by Troyanov and Dibinskaya [178]. In the second step, oxidation of 2′,4′-dibromoacetophenone (82) to 2,4-dibromobenzoic acid (67), is quite straightforward. The alkaline permanganate oxidation is a standard procedure and the workup is simplified by decomposing the precipitate of manganese dioxide [272]. This is followed in the third step by an Ullmann condensation of 2,4-dibromobenzoic acid (67) with glycine (86) to give the bromocarboxyphenylglycine 46, which is the novel, key reaction in this synthesis. The condensation of 67 was done in an aqueous system using two equivalents of potassium carbonate and a mixture of copper powder and cuprous iodide as catalysts [273][274][275]. Under these conditions the reaction was vigorous at 50-60 ºC and led to crude yields of 46 of up to 89%. The fourth step involves the Claisen condensation of 46 to give the bromodiacetylindoxyl 48 in a 70% yield. In the final step hydrolysis and oxidation of diacetylindoxyl 48 gives high yields of Tyrian purple (1).
The overall yield of Tyrian purple in this five step synthesis was about 25% based on the starting pdibromobenzene (40), and has yet to be fully optimized. Although this yield is significantly lower than that achieved by Voss and Gerlach for their synthesis starting from the same compound [37], Wolk-Frimer procedure has the advantage of not requiring special techniques such as low temperatures or strictly anhydrous conditions, and is, therefore, amenable for student labs and industrial production of larger quantities.