Photochemical Oxidative Cyclisation of Stilbenes and Stilbenoids—The Mallory-Reaction

After Mallory described in 1964 the use of iodine as catalyst for the photochemical cyclisation of stilbenes, this reaction has proven its effectiveness in the synthesis of phenanthrenes, other PAHs and phenacenes with a surprisingly large selection of substituents. The “early age” of the reaction was reviewed by Mallory in 1984in a huge chapter in the Organic Reactions series, but the development has continued. Alternative conditions accommodate more sensitive substituents, and isomers can be favoured by sacrificial substituents. Herein the further developments and applications of this reaction after 1984 are discussed and summarized.


Introduction
The oxidative photocyclizations of stilbenes was discovered earlier during studies of the photochemical isomerization of stilbenes [1,2], but the reaction did not become feasible as a synthetic tool until Mallory discovered in 1964 that iodine could catalyze the reaction [3,4]. That allowed for more concentrated solutions and fewer side reactions. The reaction was thoroughly reviewed by Mallory in a large chapter in Organic Reactions in 1984 [5]. Other reviews [6][7][8][9][10] discuss various aspects and applications of the reaction. This review will focus on the reaction as a useful tool in synthesis, covering developments reported since 1984.

Oxidative Photocyclization
The reaction pathway of what should be called the Mallory-reaction is pictured in Scheme 1. Photochemical isomerization of the double bond in stilbene has been extensively studied [8,11]. From a synthetic point of view the cis/trans-isomerization occurs rapidly under the reaction conditions in such a way that different compositions of cis-and trans-stilbenes still give the same products. Thus, the stilbenes can be used as isomeric mixtures in the photocyclization, although only the cis-isomer is capable of the further cyclization. The formed dihydrophenanthrene is unstable and will, unless trapped, relax back to the stilbene. There are also examples of hydrogen-shifts at this stage under nonoxidative conditions [5]. The dihydrophenanthrene can be trapped by oxidation to form a phenanthrene, or by elimination given a suitable substituent in the ortho-position on one of the aromatics. In more concentrated solutions the stilbenes can form dimers in a [2+2] cycloaddition as well [5,8,12]. Mallory discovered that this oxidative trapping occurs much faster when traces of iodine were used together with O 2 [3], but increased concentrations of iodine did not affect the reaction rate. It has been proposed [5] that iodine is photochemically cleaved into radicals that react in a chain reaction: The hydrogen iodide is then oxidized back to iodine by oxygen. The reaction was compatible with fluoro, chloro, bromo, methoxy, methyl, trifluoromethyl, phenyl and carboxyl, but not nitro, acetyl or dimethylamino substituents. Iodo substituents were lost during the reaction. Concentrations were usually 0.01 mole/liter of stilbene [4]. Higher concentrations lead to more [2+2] cycloaddition between two stilbenes. The concentration of iodine can influence both product yields and product selectivity. A full equivalent of iodine per cyclization can prevent elimination of methanol [13] (Scheme 2).

Scheme 2.
More iodine can prevent eliminative cyclization [13]. On the other hand, the increased iodine concentration leads to formation of more hydrogen iodide that can saturate the starting stilbene and also contribute to other side reactions [14]. Other oxidants besides iodine have also been used, as reviewed by Laarhoven [10], but do not appear to have been preferred for carrying out syntheses. Representative examples of the Mallory-reaction published after 1984 are shown in Appendix I.

Katz's Conditions
Although increased amounts of iodine result in better yields in some systems, the increased concentration of hydrogen iodide causes side-reactions that limit the yields. Katz's group officially introduced new conditions in 1991 [14] to solve this problem by scavenging the formed hydrogen iodide with methyloxirane to prevent the side-reactions (they published the first reactions with methyloxirane as a scavenger in 1986 [15,16]). As a consequence the iodide could not be reoxidized by oxygen, so one equivalent of iodine was needed and the reaction could then be performed under an inert atmosphere preventing side reactions with oxygen. The combination of hydrogen iodide and light can reduce the double bond in stilbene to a saturated bond [14]. It was observed from the beginning [3,4] that catalytic amounts of iodine gave purer products and higher yields for many systems. It is not oxygen itself that is the destructive agent, but rather substances formed from oxygen during the photocyclization [14]. Table 1 compares the yields between the use of catalytic amounts of iodine and the Katz-conditions.    However, with less symmetrical starting materials this is not a viable approach. Katz-conditions will usually reduce the problem of this kind of elimination. Mallory [29] tested acidic conditions to promote elimination of methanol to control the selectivity. The reactions needed much longer reaction times. Some selectivity towards elimination was achieved with catalytic amounts of sulfuric acid, at the cost of lower yields than with oxidative conditions on the same stilbenes. The reactions were not inverted in all cases, but a good selective synthesis of 2-methylphenanthrene and 4-methylphenanthrene was obtained (Scheme 7). Oxidative cyclization of metamethylstilbene gives a 1:1 mixture of these regioisomers that are difficult to separate. However, attempts to control the cyclization into the unfavored benzo[a]anthracene failed.
It is also possible to put a good leaving group like tosyl at the bridge-double bond to promote cyclization under basic conditions [30]. Although this gave very good yields, it does not help to control the selectivity of the cyclization.  Dehydrobromination under basic conditions has been extensively studied, but it also has its limitations. Olsen [31] did a comparison between oxidative photocyclizations (2 equivalents of iodine) and elimination photocyclizations with NaOMe in methanol. Some of the results are summarized in Table 2. The yields of oxidative conditions were consistently higher, but some debrominations were observed as secondary reactions occurring after the cyclization. The basic conditions did change the selectivity, but to a lesser degree than desired. As can be seen in the last example, basic conditions can give very good selectivity, but only when the system is already inclined to react that way. Some more examples of eliminative photocyclizations are summarized in Appendix 3. However, this approach has generally been of limited use.

Reactivity Parameters
The Mallory-reaction is somewhat sensitive to steric effects of the substituents, as shown in Scheme 8, although the product distribution does not deviate much from a statistical distribution. In contrast, the aromatic ring-structure regioselectivity of the Mallory-reaction is very strong. Usually only one ring-structure is formed, even when the formation of several structures look plausible. The reaction favors ring-structures that are curled towards helicene structures as the two examples in Scheme 9 show.
Laarhoven [6,10] has evaluated reactivity parameters like free valence numbers [35] (∑F * rs ) and localization energies (∑L * rs ) for a large number of examples. He found a good correlation between these two parameters, but found free valence numbers more convenient as only one calculation is needed to evaluate all cyclization modes of a particular compound. ∑F * rs is the sum of the free valence numbers of atoms r and s involved in the cyclization in the exited state (F r = √3-∑P in which P is the bond order).
Three rules [6] for cyclization were determined: (i) Photocyclizations do not occur when ∑F * rs <1.0. (ii) When two or more cyclizations are possible in a particular compound, only one product arises if ∆(∑F * rs ) > 0.1; more products are formed if the differences are smaller.
(iii) The second rule holds when only planar or non-planar products (penta-or higher helicenes) can arise. When planar as well as non-planar products can be formed, the planar aromatic in general is the main product, provided that for its formation ∑F * rs > 1.0 Photocyclization of 1,4-distyrylbenzene [6] is a good example (Scheme 10). Another example applies to 1,3-distyrylbenzene [8]. The best discussion with several reaction examples with calculated reactivity parameters is given in a review by Laarhoven [10]. These rules should be a useful planning tool for synthesis, but no examples have been found where these rules have actually been applied in the such planning. One reason might be that the theory and calculations of these reaction parameters are not very accessible for the typical synthesis chemist with a limited background in theoretical chemistry.
Scheme 10. Calculation of ∑F * rs for the reaction indicates that methyl-substituted distyrylbenzene can undergo photocyclization, but not the unsubstituted compound. Experiments are in accordance with this [6,36].

Controlling Product Formation with Blocking Groups
Helicenes are borderline molecules in Laarhoven's cyclization rules. When they become larger than five benzene rings they become non-planar, and thus no longer favored products. Formation of planar S-shaped molecules becomes the main side reaction or even the main reaction. This led Katz's group to develop the bromo-group as a directing substituent [15,16,37]. The bromo-group also blocks its neighbor position in the cyclization (Scheme 11): Scheme 11. Br is used as a blocking group. Without the bromo-group the reaction gives 1:1 [7]helicene and the S-shaped benzo[a]naphto[1,2k]tetraphene [15]. Without blocking groups the yield of [6]helicenes also becomes low [38]. The bromo-group can even protect neighbouring methoxy-groups from elimination-cyclization. Without the bromo-groups in the example below a mixture of the desired product and products resulting from elimination reactions occurred [37] (Scheme 12):

Scheme 12.
Br also protects neighbouring methoxy-groups from elimination-cyclization. . In A) the reaction follows the natural cyclization path but the chloro-group prevents the 50:50 product mixture from meta-methyl. In B) the chloro-groups blocks the preferred cyclization path and forces the product formation. The reaction has lately also made its way into material science [46,47]. It is appropriate to end this review with a series of papers by Mallory [48-50], the latest 37 years after the publication of the use of iodine as a catalyst [3]. Here